Belt and Road Reboot

Beijing's Bid to De-Risk Its

Global Infrastructure Initiative
Full Report

November 2023

Bradley C. Parks, Ammar A. Malik,

Brooke Escobar, Sheng Zhang,
Rory Fedorochko, Kyra Solomon,
Fei Wang, Lydia Vlasto, Katherine
Walsh, and Seth Goodman
Acknowledgments
We thank Samuel Brazys, Harsh Desai, Charlotte Gardes-Landolfini, Sebastian
Horn, David Landry, Maria Repnikova, Marina Rudyak, Brad Setser, L. Teal
Emory, and Haonan Zhou for comments on an earlier version of this report; John
Custer, Sarina Patterson, Sariah Harmer, and Alex Wooley from AidData for the
editing, formatting, and graphic design of the final report; Davide Monteleone
for generously granting us permission to use his photograph for this
publication’s cover; Sasha Trubetskoy for cartographic support; and Nicholas
Costa, Taylor Liegel, Milo Yen-Goossens, Rushika Devineni, Shiying Cai, and Eli
Svoboda for data management software support. We also owe a debt of
gratitude to the large team of research assistants (RAs) who helped construct the
dataset that made this study possible—including Abena Konadu, Abigail Taylor,
Aldrin Yashko, Alec Veit, Alejandro Schnapp Mezerhane, Alex Park, Ally Swindell,
Alyson Reynolds, Andrea Phelps, Andy Shufer, Anna Garrett, Annabel Richter,
Anne Paxton, Annika Suri, Arianna Lagamma, Audrey Sun, Bailey Garber, Ben
Greenberg, Beyoncé Lightfoot, Bree Jiang, Caitlin Connor, Caitlin Noe, Camilla
Yeleussizova, Carina Rosenberg, Carolin Helmholz, Caroline Elszy, Changlin
(Bob) Zhang, Check King Tay, Christian Moore, Claire Rudinsky, Claire Schlick,
Claire Wyszynski, Clara McAllister, Collin Absher, Connor Stanton, David Zhu,
Dorothy Gao, Dory Gilmer, Elise Tsao, Elizabeth Ott, Emma Bury, Falan Kifle,
Gelila Yonas, George (Teddy) Chunias, Gujie Shen, Gwyneth Smith, Harin Ok,
Harrison Abramsohn, Helen Tieder, Henry Tieder, Holi Raparaoelina, Jack
Mackey, Jackson Humphreys, Jason Bao, Jennifer Luo, Jiaju Du, Jiayi Xu, Jiexi
Lin, Joanne Lee, Joey Lindsay, Joseph Bohley, Juhwan Cho, Kate Hentz, Katelyn
Oxer, Kathryn Spurrier, Kathryn Ziccarelli, Kayla Aviles, Kelly Shinners, Kerim
Gales, Kien-Tam Elston, Laina Lomont, Laura Lam, Lexi Olsen, Liang Geng,
Lillian Butler, Maddie Sharpe, Marianna Bonilla, Mary Trotto, Max Berckmueller,
Meveryn Chua, Mia Maccario, Micah Katahara, Michael Olyniec, Mujia He,
Nancy Lang, Nate Spiers, Nico Rullo, Noah Tran, Olivia Schlamp, Peter
Awabdeh, Pooja Muthuraj, Pooja Tanjore, Prestin Tran, Redeit Hailu, Riely
Capece, Riley Companion, Rin Tsujikawa, Rodrigo Arias, Ryan McCormich,
Sabrina Levri, Sailor Miao, Sam LeBlanc, Samantha Rofman, Sarah Kennedy,
Saranga Bansal, Sean Nguyen, Shannon McGuire, Sophie Adams-Smith, Sophie
Alden, Sophie (Grace) McGehee, Sophie Strauss, Sophie Torres, Susannah
Matthews, Taylor Perry, Tom Mooney, Victor Gedeck, Will Wu, Ye Xiao, You Wu ,
Yufei Tang, Yunlang (Bella) Li, Yuxi (Felix) Tian, Yuxin Qin, Zheyi (Serena) Liu,
Zhiying (Cindy) Zhou, Zipu Liu, Ziqi (Samuel) Yan, and Zupeng Zeng.

Over the last ten years, a large group of coauthors and collaborators—including
Christian Baehr, Ariel BenYishay, Richard Bluhm, Mengfan Cheng, Samantha
Custer, Carrie Dolan, Axel Dreher, Thai-Binh Elston, Sebastian Horn, Andreas
Fuchs, Alysha Gardner, Anna Gelpern, Siddhartha Ghose, Roland Hodler, Brook
Lautenslager, Joyce Jiahui Lin, Jacob Hall, Alexandra Joosse, Scott Morris,
Edwin Muchapondwa, Daniel Nielson, Julius N. Odhiambo, Paul Raschky,
Carmen Reinhart, Daniel Runfola, Austin Strange, Michael Tierney, Christoph
Trebesch, Rachel Trichler, Lukas Wellner, and Lincoln Zaleski—has helped refine
and expand the dataset upon which this study is based by piloting coding
procedures, recommending new sources and methods, and scrutinizing
preliminary project records.

We gratefully acknowledge financial support from the William and Flora Hewlett
Foundation, the Ford Foundation, and the Smith Richardson Foundation. The
findings, interpretations, and conclusions expressed in this study are entirely
those of the authors. They do not necessarily represent the views of the William
and Flora Hewlett Foundation, the Ford Foundation, or the Smith Richardson
Foundation. AidData’s research is guided by the principles of independence,
integrity, transparency, and rigor. A diverse group of funders supports AidData’s
work, but they do not determine its research findings or recommendations.

Citation
Parks, B. C., Malik, A. A., Escobar, B., Zhang, S., Fedorochko, R., Solomon, K.,
Wang, F., Vlasto, L., Walsh, K. & Goodman, S. 2023. Belt and Road Reboot:
Beijing’s Bid to De-Risk Its Global Infrastructure Initiative. Williamsburg, VA:
AidData at William & Mary
Executive Summary
The Belt and Road Reboot report provides myth-busting 3. In the short-run, the G7 is also stepping up its efforts
evidence about the changing nature, scale, and scope of to compete with Beijing through the Partnership for
China’s overseas development program. It also reveals new Global Infrastructure and Investment, the
insights about Beijing’s ongoing bid to de-risk its flagship India-Middle East-Europe Economic Corridor, and
global infrastructure initiative—and outflank its competitors. other initiatives. After failing to match China’s annual
The report draws upon AidData’s uniquely comprehensive and ODA and OOF commitments during the early years
granular dataset of international development finance from of the Belt and Road Initiative (BRI), the G7 outspent
China, which captures 20,985 projects across 165 low- and China by $84 billion in 2021.
middle-income countries financed with grants and loans worth
$1.34 trillion over a 22-year period.1 4. However, in the long-run, it is not clear that the U.S.
and its allies have the financial firepower to compete
dollar-for-dollar with Beijing. The G7 has a history of
Is China still the single largest official source of aid
over-promising and under-delivering net increases in
and credit to the developing world? international development spending. Beijing, by
contrast, has a real source of financial strength that
Four key takeaways
allows it to avoid making promises that it cannot
1. Contrary to the conventional wisdom, Beijing’s annual keep: foreign exchange reserves that are vastly larger
international development finance commitments than the official, foreign currency reserve holdings of
have not plummeted to nearly zero.2 It remains the its central bank.4
world’s single largest official source of international
development finance. China’s aid and credit (ODA
How has the risk profile of China’s international
and OOF) commitments to low- and middle-income
countries are now hovering around $80 billion a year.3 development finance portfolio changed?
Three key takeaways
2. Washington is beginning to close the spending gap
with Beijing. Due in large part to the U.S.
1. Repayment risk: Beijing is navigating an unfamiliar
International Development Finance Corporation
and uncomfortable role—as the world’s largest official
(DFC)’s financing of private sector projects, which has
debt collector. 55% of its loans to low- and
led to a fifteen-fold expansion in U.S. OOF,
middle-income countries have already entered their
Washington now provides approximately $60 billion
principal repayment periods and this figure will
of development finance each year to low- and
increase to 75% by 2030. Total outstanding
middle-income countries.
debt—including principal but excluding
interest—from borrowers in the developing world to
1
The latest (3.0) version of AidData’s Global Chinese Development China is at least $1.1 trillion and potentially even as
Finance (GCDF) dataset captures projects over 22 commitment years high as $1.5 trillion (in nominal USD).5 Beijing is
(2000-2021) and provides details on the timing of project
implementation over a 24-year period (2000-2023). It can be accessed finding its footing as an international debt collector at
via aiddata.org/china. a time when many of its biggest borrowers are illiquid
2
For example, the latest version of the China’s Overseas Development
or insolvent. AidData estimates that 80% of China’s
Finance (CODF) database produced by Boston University’s Global
Development Policy Center suggests that overseas development
4
finance commitments from China have plummeted by 96% since 2016, As of 2023, the official, foreign currency reserve holdings of China’s
reaching an all-time low of $3.7 billion in 2021. central bank (the PBOC) amounted to $3.1 trillion. However, this figure
3
Based upon OECD-DAC definitions and measurement criteria, excludes foreign currency reserves that the PBOC has moved off of its
AidData categorizes each project/activity in its dataset as Official balance sheet by, among other things, entrusting them to the
Development Assistance (ODA) or Other Official Flows (OOF). ODA country’s state-owned policy banks, state-owned commercial banks,
mostly consists of grants and highly concessional loans for and state-owned funds. Brad Setser of the Council on Foreign
development projects and activities that are financed by official sector Relations argues that these “hidden reserves” may be worth an
institutions. OOF mostly consists of non-concessional loans that are additional $3 trillion.
5
issued by official sector institutions. More than 90% of China’s annual Total outstanding debt from borrowers in developed and developing
international development finance commitments consist of OOF. countries to China exceeds $2.6 trillion (in nominal USD).

1
 overseas lending portfolio in the developing world is Does the G7 understand the difference between
currently supporting countries in financial distress.
BRI 1.0 and BRI 2.0—or how Beijing’s reboot of its
Overdue repayments to China are also soaring—in
absolute terms and as a proportion of total overdue “project of the century” has altered the
loan repayments to official (i.e., bilateral and competitive landscape?
multilateral) creditors.
Three key takeaways
2. Project performance risk: The cumulative number of
1. Beijing has launched a far-reaching effort to de-risk
Chinese grant- and loan-financed infrastructure
the BRI by refocusing its time, money, and attention
projects in the developing world with significant
on distressed borrowers, troubled projects, and
environmental, social, or governance (ESG) risk
sources of public backlash in the Global South. It is
exposure skyrocketed from 17 projects worth $420
learning from its mistakes and becoming an
million in 2000 to 1,693 projects worth $470 billion in
increasingly adept international crisis manager.
2021. The cumulative percentage of China’s grant-
and loan-financed infrastructure project portfolio in
2. Neither the U.S. nor its G7 allies seem to have a
the developing world with significant ESG risk
good understanding of how China is recalibrating its
exposure increased from 12% to 53% over the same
lending and grant-giving practices in response to
22-year period. Infrastructure project suspensions and
changing conditions on the ground. Consequently,
cancellations have also mounted—from nearly zero at
those who make and shape policy in Washington,
the turn of the century to 94 projects worth $56
London, Paris, Berlin, Tokyo, Rome, and Ottawa
billion in 49 countries. However, Beijing is stepping
increasingly run the risk of competing with a version
up ESG risk mitigation efforts to shield its overseas
of the BRI that no longer exists—BRI 1.0 rather than
infrastructure portfolio from the types of problems
BRI 2.0.
that have previously plagued the BRI.
3. The G7 should not underestimate the ambition of
3. Reputational risk: Beijing’s public approval rating in
China’s ongoing effort to future-proof its flagship,
the developing world plunged from 56% in 2019 to
global infrastructure initiative. Beijing is focused on
40% in 2021. Washington, on the other hand, has
giving leaders in the developing world exactly what
seen its public approval rating rise and opened up a
they want: rapid delivery of large-scale infrastructure
14 percentage point advantage over Beijing. Across
projects without unreasonably high levels of ESG risk.
the developing world, China has also struggled to
If the G7 cannot compete on this basis, its
maintain a razor-thin lead over the U.S. in media
Partnership for Global Infrastructure and Investment
coverage favorability. Yet it has proven very capable
may face a crisis of relevance.
of winning and retaining the foreign policy support of
governing elites. Across all U.N. General Assembly
votes cast between 2000 and 2021, the governments What measures has Beijing taken to reduce its
of low- and middle-income countries aligned their exposure to distressed debt in the developing
foreign policy positions with China 75% of the
world?
time—as compared to 23% with the U.S. Those who
vote with China are richly rewarded: on average, if a Seven key takeaways
foreign government chooses to increase the
alignment of its U.N. General Assembly voting with 1. In recognition of the fact that BRI 1.0 did not have
China by 10%, it can expect to see a 276% increase sufficiently robust risk management guardrails in
in aid and credit from Beijing.6 place, Beijing is fundamentally altering the
composition of its overseas lending portfolio. It is
ramping down dollar-denominated infrastructure
project lending, while ramping up RMB-denominated
6
This finding is derived from a statistical model in Dreher, A., Fuchs, emergency rescue lending to financially distressed
A., Parks, B. C., Strange, A., & Tierney, M.J. 2022. Banking on Beijing:
The Aims and Impacts of China’s Overseas Development Program.
Cambridge, UK: Cambridge University Press.

2
 borrowers.7 Beijing’s strategic objective is to ensure 4. Beijing is putting in place increasingly stringent
that its largest borrowers have enough cash on hand safeguards to shield itself from the risk of not being
to service their outstanding infrastructure project repaid. At the turn of the century, only 19% of China’s
debts. overseas lending to low- and middle-income
countries was collateralized. This figure now stands at
2. Beijing’s policy banks (China Eximbank and China 72%.9 The ability to access cash collateral without
Development Bank) have particularly high levels of borrower consent has become a particularly
exposure to non-performing loans in low- and important safeguard in China’s bilateral lending
middle-income countries. Instead of reforming these portfolio. When illiquid or insolvent borrowers fall
institutions from within, Beijing is ratcheting down its behind on their repayments, the policy banks are
use of the policy banks, while ratcheting up its use of “paying themselves” overdue principal and interest
state-owned commercial banks, such as ICBC and by unilaterally sweeping foreign currency out of the
Bank of China. In previous years, approximately escrow accounts of their borrowers. These cash
three-quarters of China’s lending to low- and seizures are mostly being executed in secret and
middle-income countries was channeled through the outside the immediate reach of domestic oversight
policy banks. However, this figure has now institutions—such as the auditor general and the
plummeted to less than one-quarter (22%). The public accounts committee within parliament—in low-
annual lending commitments of China’s state-owned and middle-income countries. After making
commercial banks to low- and middle-income withdrawals that substantially deplete the balance of
countries are now on par with those of its policy a borrower’s escrow account, an increasingly common
banks.8 practice is to require that the borrower replenish the
account as a condition for any short-term cash flow
3. Rather than relying on its own banks to vet borrowing relief. Escrow account replenishment has become a
institutions and proposed transactions, Beijing is major sticking point in debt rescheduling
increasingly outsourcing risk management to lending negotiations with the policy banks, yet it is shrouded
institutions—such as the International Finance in secrecy because of strict confidentiality
Corporation, the European Bank for Reconstruction requirements.10
and Development, Standard Chartered Bank, and
BNP Paribas—with stronger due diligence standards 5. As the number of borrowers facing liquidity and
and safeguard policies. It is dialing down its use of solvency crises has soared, Chinese state-owned
bilateral lending instruments and dialing up the creditors have introduced stronger penalties for late
provision of credit through collaborative lending repayments. The average penalty interest rate
arrangements with Western commercial banks and doubled between the early BRI period (2014-2017)
multilateral institutions. 50% of China’s and the late BRI period (2018-2021). The maximum
non-emergency lending portfolio in low- and penalty interest rate also increased from 3% to 8.7%
middle-income countries is now provided via between these two time periods. These findings
syndicated loan arrangements—and more than 80% contradict those of a previous study, which claimed
of these arrangements involve Western commercial that there is no evidence of penalty interest rates in
banks and multilateral institutions. China’s overseas lending to developing countries.

9
Beijing is taking special precautions with high-risk borrowers. At the
turn of the century, 0% of its collateralized lending commitments to
7
In the first full year of BRI implementation (2014), 65% of Beijing’s low- and middle-income countries were directed to developing
lending to low- and middle-income countries supported infrastructure countries in financial distress. By 2021, this figure increased to 74%.
10
projects. By 2021, this figure plummeted to 31%. Emergency rescue When a sovereign borrower signs an escrow account agreement or
lending represented only 13% of Beijing’s loan portfolio in low- and debt rescheduling agreement with a Chinese lender, it is not unusual
middle-income countries in 2014. However, this figure soared to 58% for the parties to agree upon an expansive set of confidentiality
by 2021. obligations that go beyond those in its original loan agreement. The
8
On average, during the pre-BRI period (2000-2013), Beijing implementation of AidData’s Tracking Underreported Financial Flows
channeled 15% of its annual lending commitments to low- and (TUFF) methodology has facilitated the retrieval and publication of a
middle-income countries through its state-owned commercial banks. significant number of unredacted escrow account and debt
This figure increased to 18% during the early BRI (2014-2017) period rescheduling agreements. The 3.0 version of AidData’s GCDF dataset
and 22% during the late BRI period (2018-2021). makes these agreements available via stable URLs.

3
 6. The repayment risk mitigation measures that Beijing Irrespective of ESG safeguard stringency, the average
is putting in place present new challenges for infrastructure project financed with Chinese aid or
borrowers in the developing world. Those who seek credit takes approximately three years to complete.
to refinance their maturing debts to China by
accepting emergency rescue loans with high interest 2. Beijing’s rivals and critics claim that it has not taken
rates and short repayment periods must be mindful meaningful steps to subject its overseas infrastructure
of the danger of swapping less expensive debt for project portfolio to more stringent ESG safeguards.
more expensive debt. Those who seek to reschedule This claim is false. By 2021, 57% of China’s grant- and
their debts to China must be prepared to ring-fence loan-financed infrastructure project portfolio in low-
foreign currency for some creditors but not others. and middle-income countries had strong de jure
Those who contract new debt from Beijing must be environmental, social, and governance safeguards in
aware of the danger of compounding arrears due to place. This represents a major departure from past
penalty interest. practice: at the turn of the century, China’s entire
grant- and loan-financed infrastructure project
7. Beijing’s go-it-alone efforts to mitigate repayment risk portfolio in low- and middle-income countries had
may undermine the international community’s efforts weak de jure environmental, social, and governance
to provide coordinated debt relief to sovereign safeguards in place.
borrowers in financial distress. In November 2020,
China agreed to participate in the G-20 Common 3. The pace of ESG safeguard reform accelerated
Framework for Debt Treatments and abide by the during the BRI 2.0 era—from 2018 to 2021.11 Over
so-called “comparable treatment” principle (i.e., the same four-year period, the annual ESG risk
reasonable burden-sharing in the way that financial prevalence rate in China’s grant- and loan-financed
losses are distributed across creditors). However, infrastructure project portfolio sharply declined from
Beijing’s latest actions suggest that it is muscling its 63% to 33%.
way to the front of the repayment line by demanding
that borrowers provide recourse to cash collateral 4. Beijing has de-risked the country’s overseas
that others lack. Paris Club, multilateral, and infrastructure project portfolio by reining in the
commercial creditors fear—with some activities of development finance institutions that lack
justification—that they are becoming junior creditors strong ESG risk management guardrails, increasing
whose loans will be repaid on a lower-priority basis. If the provision of infrastructure financing via
Beijing insists upon being treated as a senior creditor institutions that have strong ESG safeguards in place,
whose debts should be given first priority, then unwinding aid and credit relationships with countries
coordinated debt reschedulings with non-Chinese that present high levels of ESG risk, and redirecting
creditors will likely become more difficult to new infrastructure financing to countries that present
negotiate. The biggest losers in this scenario will be low levels of ESG risk.
ordinary people in the developing world who are
5. Chinese grant- and loan-financed infrastructure
denied basic public services because of a collective
projects with strong de jure ESG safeguards have
action failure among foreign creditors.
substantially lower levels of ESG risk exposure in a de
facto sense than those without such safeguards. They
What measures has Beijing taken to reduce its are also less vulnerable to suspension and
exposure to ESG risk? How are its infrastructure cancellation.

projects with strong ESG safeguards faring during 6. A particularly important finding is that Chinese grant-
implementation? and loan-financed infrastructure projects subjected to
strong de jure ESG safeguards do not face
Seven key takeaways
11
1. Beijing has earned a reputation for implementing In 2018, 26% of China’s grant- and loan-financed infrastructure
project portfolio in low- and middle-income countries had strong de
brick-and-mortar projects with lightning speed. jure environmental, social, and governance safeguards in place. By
2021, this figure had increased by 31 percentage points (to 57%).

4
 substantially longer implementation delays than development finance portfolio to “toss-up”
those subjected to weak de jure ESG safeguards.12 countries—i.e., competitive jurisdictions where
Squaring the circle between speed and safety is at neither China nor the U.S. opened up an
the center of Beijing’s BRI 2.0 strategy. insurmountable lead vis-à-vis its principal rival.

7. Beijing enjoys a stronger position in the global 3. Beijing seeks to maintain and build upon momentum.
infrastructure financing market than its bilateral and In jurisdictions where it recently made reputational
multilateral competitors realize. The fact that China gains at the expense of the U.S., it doubled down by
has put in place increasingly stringent ESG providing more aid and credit.
safeguards—without damaging its reputation for
speed—could undermine G7 efforts to outcompete it 4. China has a relatively low level of tolerance for risk in
on quality or safety grounds. Developing countries its pursuit of soft power. It devoted only 16% of its
prefer to work with lenders and donors that bankroll international development finance portfolio to
big-ticket, high-impact infrastructure projects with “moonshot” countries—those where its principal rival
reasonably robust ESG safeguards but without had momentum on its side.14 A separate, but related,
excessive implementation delays. Beijing is taking finding is that when reputational assets become
measures to meet this challenge. Whether the reputational liabilities, Beijing tends to disengage
G7—and the multilateral development banks—will from discussions about new projects and financial
do the same is an open question. commitments and refocus on managing risks within
its existing portfolio of grant- and loan-financed
projects.
What measures has Beijing taken to reduce its
exposure to reputational risk? 5. Political transitions in host countries are critical
junctures when the nature, level and pace of China’s
Six key takeaways
engagement can change significantly. If a new leader
comes to power and takes a less adversarial posture
1. In a tally of the annual number of soft power “gains”
toward China, Beijing typically springs into action and
and “losses” that China has experienced vis-à-vis the
seeks to cement bilateral relations by helping
U.S. in low- and middle-income countries since the
incumbents take credit for high-profile infrastructure
first full year of BRI implementation (2014), Beijing’s
projects.
losses outnumbered its gains—by a substantial
margin.13 It experienced more losses than gains
6. Given that Beijing tends to disengage rather than
vis-à-vis Washington on three different measures of
double down in countries where there are strong
soft power: public opinion, media sentiment, and
indications of BRI backlash, Beijing’s competitors may
elite support.
be able to lure such countries back into the West’s
orbit. However, doing so would require that the G7
2. Across all three measures of soft power, Beijing
act quickly when these windows of opportunities
devoted nearly two-thirds of its entire international
arise and adapt their programming to address the
12
As a point of comparison, World Bank projects subjected to the unmet needs of partner countries.
organization’s most stringent environmental and social safeguards take
more than 7 years, on average, to move from the proposal stage to
the commencement stage. On average, it takes World Bank projects
another 6 years to move from the commencement stage to the
completion stage.
13
We measure the relative gains and losses experienced by China on a
country-by-country basis between 2014 and 2021 on three different
measures of soft power (public approval, media sentiment, and elite
support). For example, to measure the relative gains or losses in public
approval, we (1) calculate the difference between the public approval
rating for China in a given year and the prior year; (2) calculate the
difference between the public approval rating for the U.S. in a given
year and the prior year; and (3) calculate the “double difference”
14
between (1) and (2) to determine if China experienced a greater gain Similarly, Beijing has assigned a lower level of priority to “toss-up”
or loss in public support than the U.S. in the same country-year. countries where momentum recently shifted in favor of the U.S.

5
Acronyms
AEs Advanced Economies CIDCA China International Development
AfDB African Development Bank Cooperation Agency
AGTF Africa Growing Together Fund CLA Chinese Loans to Africa
AI Artificial Intelligence CMEC China Machinery Engineering
AIIB Asian Infrastructure Investment Bank Corporation
AL Awami League CNPC China National Petroleum Corporation
AsDB Asian Development Bank CODF China’s Overseas Development Finance
AVIC Aviation Industry Corporation of China CPC Communist Party of China
B3W Build Back Better World CPEC China-Pakistan Economic Corridor
BCRA Central Bank of Argentina (Banco CR China State Railway Group Co. Ltd.
Central de la República Argentina) CRS Creditor Reporting System
BCS Bilateral Currency Swap CSIS Center for Strategic and International
BIS Bank of International Settlements Studies
BNP Bangladesh Nationalist Party DAC Development Assistance Committee
BOP Balance of payments DBZ Development Bank of Zambia
BPC Botswana Power Corporation DFC U.S. International Development
BRI Belt and Road Initiative Finance Corporation
BRT Bus Rapid Transit DLP Debt Limits Policy
BU Boston University DRS Debtor Reporting System
BUILD Better Utilization of Investment Leading DSF Debt Sustainability Framework
to Development Act DSRA Debt Service Reserve Account
CBRC China Banking Regulatory Commission DSSI Debt Service Suspension Initiative
CCDI Central Commission for Discipline EBRD European Bank for Reconstruction and
Inspection Development
CCTV Closed-Circuit Television EIA Environmental Impact Assessment
CDB China Development Bank EMDEs Emerging Market and Developing
CFK Cristina Fernández de Kirchner Economies
CGD Center for Global Development EMP Environmental Management Plan
CGSP China-Global South Project EPC Engineering, Procurement, and
CGT Common Ground Taxonomy Construction
CHICO China Henan International Cooperation EPCF Engineering, Procurement,
Group Company Construction and Financing
China Eximbank Export-Import Bank of China ESG Environmental, Social, and Governance
CIC China Investment Corporation
ESIA Environmental and Social Impact IMEC India-Middle East-Europe Economic
Assessment Corridor
EU European Union IMF International Monetary Fund
EUR Euros IPSF International Platform on Sustainable
EURIBOR Euro Interbank Offered Rate Finance
FASB Financial Accounting Standards Board JPY Japanese Yen
FOCAC Forum on China–Africa Cooperation JV Joint Venture
FPIC Free, Prior, and Informed Consent KCHP Néstor Kirchner and Jorge Cepernic
G-24 The Group of 24 Hydroelectric Power Plant Project
G7 The Group of 7 KHPC Kariba Hydro Power Company (Private)
GAAP Generally Accepted Accounting Limited
Principles KRC Kenya Railways Corporation
GBP Pound Sterling LAC Latin America and the Caribbean
GCDF Global Chinese Development Finance LBS Locational Banking Statistics
Dataset LHC Lahore High Court
GCL Government Concessional Loan LIBOR London Inter-Bank Offered Rate
GDELT Global Database of Events, Language, LICs Low-Income Countries
and Tone LMICs Lower-Middle Income countries
GDP Gross Domestic Product LPR China Loan Prime Rate
GWP Gallup World Poll LSF Liquidity Support Facility
HIA Heritage Impact Assessment M&A Mergers And Acquisitions
HICs High-Income Countries MCDF Multilateral Cooperation Center for
HSBC Hongkong and Shanghai Banking Development Finance
Corporation Limited MCPP Managed Co-Lending Portfolio
IADB Inter-American Development Bank Program
ICBC Industrial and Commercial Bank of MDB Multilateral Development Bank
China MICs Middle-Income Countries
ICT Information and Communications MMD Movement for Multi-Party Democracy
Technology MOF Ministry of Finance
IDS International Debt Statistics MOFCOM Ministry of Commerce
IFAD International Fund for Agricultural MOU Memorandum of Understanding
Development MUFG Mitsubishi UFJ Financial Group
IFC International Finance Corporation NATO North Atlantic Treaty Organization
IFI International Financial Institution NDRC National Development and Reform
IFRS International Financial Reporting Commission
Standards NGO Non-Governmental Organizations
NOC No Objection Certificate
NORINCO China North Industries Group SAFE State Administration of Foreign
Corporation Limited Exchange
NPV Net Present Value SAIS-CARI China-Africa Research Initiative at the
ODA Official Development Assistance Johns Hopkins School of Advanced
ODI Overseas Development Institute International Studies
OECD Organisation for Economic SCP Supreme Court of Pakistan
Co-operation and Development SDR Special Drawing Rights
OFC Offshore Financial Center SHIBOR Shanghai Interbank Offered Rate
OOF Other Official Flows SIA Structural Impact Assessment
P4I Partnerships for Infrastructure Sinosure China Export & Credit Insurance
PAP Project-Affected Persons Corporation
PBC Preferential Buyer’s Credit SPV Special Purpose Vehicle
PBOC People’s Bank of China TAZARA Tanzania-Zambia Railway Authority
PF Patriotic Front TPDC Tanzania Petroleum Development
PGII Partnership for Global Infrastructure Corporation
and Investment TUFF Tracking Underreported Financial Flows
PIIE Peterson Institute for International U.S. United States
Economics UMICs Upper Middle-Income Countries
PIMCO Pacific Investment Management UN United Nations
Company UNDRIP UN Declaration on the Rights of
PLAD Political Leaders’ Affiliation Database Indigenous Peoples
PPG Public and Publicly Guaranteed Debt UNGA United Nations General Assembly
PxF Pre-Export Financing USD United States Dollars
Quad Quadrilateral Security Dialogue WDPA World Database on Protected Areas
RAP Resettlement Action Plan YoY Year-on-Year
RMB Renminbi
S&P Standard & Poor
Table of Contents

Chapter 1: Belt and Road Reconstruction—From Fire-Fighting to

Future-Proofing......................................................................................................1
Section 1: Has Beijing’s global infrastructure initiative become an asset or a
liability?.............................................................................................................1
Section 2: Myth-busting evidence about the scale and composition of China’s
overseas development program—and how its rivals have responded............ 5
Section 3: Repayment risk from Beijing’s perspective.................................... 16
Section 4: Project performance risk from Beijing’s perspective......................20
Section 5: Reputational risk from Beijing’s perspective.................................. 26
Section 6: The balancing act of portfolio risk management...........................37
Chapter 2: The Road to Repayment for the World’s Largest Official Debt
Collector.............................................................................................................. 45
Section 1: Debunking the myth that Beijing’s overseas lending program has
collapsed........................................................................................................ 45
Section 2: Major changes in Beijing’s overseas lending portfolio during the
BRI 2.0 era...................................................................................................... 64
Section 3: Beijing’s crisis-time approach to repayment risk mitigation—Fool
me once, shame on you. Fool me twice, shame on me................................. 81
Section 4: What have we learned about Beijing’s bid to reboot its overseas
lending program?......................................................................................... 116
Chapter 3: Redesigning the Belt and Road for Safety and Speed....................119
Section 1: Beijing’s journey from skeptic to advocate of ESG risk
management................................................................................................ 119
Section 2: Measuring the scope and severity of ESG risk exposure in China's
infrastructure project portfolio......................................................................126
Section 3: Measuring the stringency of ESG safeguards in China’s
infrastructure project portfolio with new sources of contractual evidence...139
Section 4: Do ESG safeguards produce a project performance payoff or
penalty?........................................................................................................ 157
Section 5: Decoding Beijing’s ESG risk mitigation strategy......................... 164
Section 6: Is Beijing course-correcting or virtue-signaling?..........................173
Chapter 4: Reputational Rehabilitation on the Belt and Road..........................176
Section 1: China’s quest for soft power during the BRI era.......................... 176
Section 2: Vying for soft power: China’s play for public opinion, media
sentiment, and elite support........................................................................ 182
Section 3: From observations to reactions................................................... 192
 Section 4: Country case studies................................................................... 198
Section 5: Beijing’s approach to reputational risk management.................. 236
References......................................................................................................... 239
Appendix: Supplementary Material.................................................................. 270
Section A-1: Figures and tables referenced in the report............................ 270
Section A-2: How AidData measures concessionality and intent................. 323
Section A-3: How does AidData measure the cumulative stock of official
financial flows from China to LICs and MICs?.............................................. 325
Section A-4: Additional details on the BIS-based estimates of China’s
international lending portfolio......................................................................326
Section A-5: How does AidData categorize Chinese lending to different
types of borrowers?...................................................................................... 330
Section A-6: How did we identify projects that rely on contractors sanctioned
for fraudulent and corrupt behavior?........................................................... 331
Section A-7: How did we identify ESG risks and risk mitigation measures in
AidData project descriptions?...................................................................... 333
Section A-8: How did AidData identify ESG risk mitigation measures in
infrastructure financing agreements?........................................................... 335
Section A-9: Assigning ESG Safeguard Stringency Ratings to Infrastructure
Financing Agreements and the 8 Infrastructure Financing Instrument
Categories.................................................................................................... 338
Section A-10: How does the composition of the coding sample of
infrastructure financing agreements compare to the composition of China’s
entire grant- and loan-financed infrastructure project portfolio in LICs and
MICs?............................................................................................................354
Section A-11: ESG risk exposure and de jure ESG safeguard protection in
China’s overseas infrastructure project portfolio tables................................355
Section A-12: Soft power gains and losses during early and late BRI periods...
......................................................................................................................365
Section A-13: China’s official sector lending portfolio in LICs and MICs: a
comparison of AidData, IDS, and CODF......................................................379
Section A-14: Comparison of AidData and IDS estimates of PPG and
non-PPG debt exposure to China................................................................ 386
Boxes

Box 1a: How AidData identifies when China’s borrowers are experiencing
financial distress................................................................................................... 19
Box 1b: How AidData measures grassroots, media, and elite support for China
and the U.S. in the developing world.................................................................. 27
Box 2a: What is the true scale of China’s overseas lending program?................ 51
Box 2b: The investment agency behind the curtain—China’s State Administration
of Foreign Exchange (SAFE)................................................................................ 60
Box 2c: Should emergency liquidity support from PBOC swap lines be treated as
short-term or long-term debt?.............................................................................97
Box 3a: China’s experience with the application of World Bank social safeguards
to the Western Poverty Reduction Project in Qinghai....................................... 150
Box 3b: De jure versus de facto application of ESG safeguards to the Lahore
Orange Line Metro Train Project........................................................................153
Box 4a: How AidData measures relative gains and losses in Chinese soft power...
........................................................................................................................... 185
Figures and Tables

Figure 1.1: The global distribution of Chinese ODA- and OOF-financed projects
in LICs and MICs.................................................................................................... 6
Figure 1.2: Official financial flows from China to the developing world,
2000-2021..............................................................................................................8
Figure 1.3: Cumulative official financial flows from China to the developing
world, 2000-2021.................................................................................................10
Figure 1.4: Official financial flows from China and the G7 to the developing
world during the BRI era, 2014-2021...................................................................11
Figure 1.5: Official financial flows from China, the U.S., and G7 countries to the
developing world during the BRI era, 2014-2021................................................12
Figure 1.6: Official financial flows from the U.S. to the developing world,
2000-2021............................................................................................................13
Figure 1.7: Change in sectoral composition of official financial flows from China
to the developing world, 2014-2021................................................................... 15
Figure 1.8: Percentage of China’s portfolio of loan commitments supporting
countries in financial distress, 2000-2021............................................................ 17
Figure 1.9: China’s share of overdue repayments owed to official creditors,
2000-2021............................................................................................................18
Figure 1.10: Canceled or suspended Chinese infrastructure projects, 2000-2021..
21
Figure 1.11: Early versus late BRI: Chinese government-financed infrastructure
projects that are behind schedule....................................................................... 22
Figure 1.12: Early versus late BRI: Commencement and completion delays.......23
Figure 1.13: Infrastructure projects facing environmental, social, or governance
risks...................................................................................................................... 26
Figure 1.14: China versus the U.S.: Public approval rates................................... 30
Figure 1.15: China versus U.S.: Media sentiment................................................ 31
Figure 1.16: China versus the U.S.: UN voting alignment................................... 32
Figure 1.17: China’s soft power losses vis-à-vis the U.S.......................................34
Figure 1.18: Public support for China: BRI countries vs. non-BRI countries........ 35
Figure 1.19: Media sentiment toward China: BRI countries vs. non-BRI countries..
............................................................................................................................. 36
Figure 1.20: UN voting alignment with China: BRI countries vs. non-BRI countries
............................................................................................................................. 37
Figure 2.1: Official sector lending commitments from China to LICs and MICs by
source.................................................................................................................. 48
Table 2.1: Comparison of AidData- and BIS-based estimates of China’s
international lending portfolio............................................................................. 53
Figure 2.2: Composition of China’s loan portfolio by currency of denomination 59
Figure 2.3: Cumulative loan commitments to LICs and MICs by financial
institution and capital injections from SAFE, 2000-2021..................................... 61
Figure 2.4: Loans from China within their repayment periods, 2000-2050..........66
Figure 2.5: Loans from China reaching their (originally scheduled) final
repayment dates, 2000-2050...............................................................................67
Figure 2.6: Composition of loan portfolio by financial instrument...................... 70
Figure 2.7: Composition of loan portfolio by creditor category.......................... 71
Figure 2.8: Chinese emergency rescue lending by countries’ level of debt
exposure to China................................................................................................72
Table 2.2: Average sovereign risk ratings and gross reserves for recipients of
different Chinese loan types................................................................................ 73
Figure 2.9: Rescue lending and debt rescheduling events for the top 50
borrowers in financial distress, 2000-2021...........................................................75
Figure 2.10: RMB-denominated rescue lending to countries in distress and
countries not in distress....................................................................................... 78
​Table 2.3: Illustrative escrow account balances linked to loans from China
Eximbank, CDB, and ICBC...................................................................................82
Figure 2.11: Percentage of loans that have reached maturity and number of
loans rescheduled................................................................................................ 89
Figure 2.12: Loan rescheduling for countries in and not in financial distress...... 90
Figure 2.13: Percentage of lending portfolio using fixed or variable interest rates
95
Figure 2.14: Composition of variable interest rate lending portfolio.................. 96
Figure 2.15: Loan portfolio backed by collateral................................................. 98
Figure 2.16: Early versus late BRI: penalty interest rates..................................... 99
Figure 2.17: Average penalty interest rates in select countries......................... 101
Figure 2.18: Composition of non-emergency lending portfolio........................104
Figure 2.19: Percentage of variable interest rate lending to countries in and not
in financial distress............................................................................................. 105
Figure 2.20: Composition of lending to countries in and not in financial distress...
........................................................................................................................... 107
Figure 2.21: Repayment risk in emergency lending portfolio versus project
lending portfolio................................................................................................ 108
Figure 2.22: Repayment risk in bilateral, multilateral, and syndicated lending
portfolio............................................................................................................. 109
Figure 2.23: Composition of non-emergency lending portfolio by channel of
delivery.............................................................................................................. 110
Table 2.4: Illustrative set of syndicated loan agreements with Chinese and
non-Chinese bank participants.......................................................................... 112
Figure 3.1: Infrastructure projects with significant environmental, social, and
governance risk exposure.................................................................................. 131
Figure 3.2: ESG risk prevalence in overseas infrastructure portfolio from China to
LICs and MICs.................................................................................................... 132
Figure 3.3: Environmental risk prevalence in overseas infrastructure portfolio
from China to LICs and MICs............................................................................. 133
Figure 3.4: Social risk prevalence in overseas infrastructure portfolio from China
to LICs and MICs................................................................................................134
Figure 3.5: Governance risk prevalence in overseas infrastructure portfolio from
China to LICs and MICs..................................................................................... 135
Figure 3.6 A global map of China’s infrastructure project portfolio in LICs and
MICs with significant environmental, social, governance (ESG) risk exposure.. 138
Table 3.1: De jure ESG safeguard stringency in China’s overseas infrastructure
portfolio by type of financing instrument.......................................................... 144
Figure 3.7: De jure ESG safeguard stringency in China’s overseas infrastructure
portfolio............................................................................................................. 147
Figure 3.8: De jure safeguard stringency in China’s overseas infrastructure
portfolio by safeguard type............................................................................... 148
Figure 3.9: Infrastructure project portfolio with de jure vs. de facto ESG risk
mitigation...........................................................................................................153
Figure 3.10: Monetary value of project suspensions and cancellations by de jure
ESG safeguard strength..................................................................................... 162
Figure 3.11: Proportion of infrastructure project portfolio facing significant ESG
risk exposure by level of safeguard stringency.................................................. 164
Figure 3.12: Composition of infrastructure project portfolio: reliance upon the
policy banks and weak de jure ESG safeguards................................................ 167
Table 3.2: China’s grant- and loan-financed infrastructure portfolio by type of
financing instrument over time.......................................................................... 167
Figure 3.13: Composition of infrastructure project portfolio: Use of syndicated
loans and strong de jure ESG safeguards..........................................................171
Figure 3.14: Proportion of infrastructure portfolio allocated to LICs/MICs with
high ESG risk prevalence rates.......................................................................... 172
Figure 3.15: Proportion of infrastructure portfolio allocated to LICs/MICs with
low ESG risk prevalence rates............................................................................173
Figure 4.1: Bangladesh example: Is China losing or gaining ground vis-à-vis the
U.S. in 2017?...................................................................................................... 185
Table 4.1: Bangladesh example: Calculating an average double-delta score for
soft power measures..........................................................................................186
Figure 4.2: China’s public approval gains and losses vis-à-vis the U.S.............. 188
Figure 4.3: China’s media sentiment gains and losses vis-à-vis the U.S............ 190
Figure 4.4: China’s elite alignment gains and losses vis-à-vis the U.S............... 191
Table 4.2: Expected versus observed allocation of Chinese development finance
by soft power cohort..........................................................................................193
Figure 4.5: Beijing’s aid and credit prioritization strategy..................................196
Figure 4.6: Official financial flows from China to Zambia, 2014-2021............... 203
Figure 4.7: China versus the U.S. in Zambia: Public approval rates...................204
Figure 4.8: Official financial flows from China to Bangladesh, 2014-2021........ 215
Figure 4.9: China versus the U.S. in Bangladesh: Public approval rates............216
Figure 4.10: China versus the U.S. in Bangladesh: Media sentiment scores..... 217
Table 4.3: Development finance commitments from China to Bangladesh by
executive administration.................................................................................... 218
Table 4.4: Chinese ODA and OOF infrastructure project milestones during the
second half of 2018 in Bangladesh....................................................................221
Figure 4.11: Official financial flows from China to Argentina, 2014-2021......... 225
Figure 4.12: Argentina’s U.N. voting alignment with China and the U.S........... 227
Table 4.5: Estimated percentage of Argentina’s foreign currency reserves derived
from PBOC swap facility.................................................................................... 231
Appendix Figures and Tables

Table A1: AidData’s Global Chinese Development Finance Dataset, Version 3.0
at a glance......................................................................................................... 270
Figure A1: Composition of official financial flows from China to the developing
world, 2000-2021...............................................................................................271
Figure A2: Stock of official financial flows from China and the G7 to the
developing world during the early and late BRI periods................................... 272
Figure A3: Sectoral composition of official financial flows from China to
developing world, 2014 vs. 2021.......................................................................273
Figure A4: Number of financially-distressed low- and middle-income countries
with outstanding debt to China......................................................................... 274
Figure A5: Sovereign overdue repayments owed to official sector creditors in
China..................................................................................................................275
Figure A6: Project life cycle............................................................................... 276
Figure A7: Average length of completion delays across all Chinese ODA- and
OOF-financed infrastructure projects................................................................ 277
Figure A8: China’s soft power gains and losses vis-à-vis the U.S.......................278
Figure A9: China versus the U.S.: Public disapproval rates............................... 279
Figure A10: Composition of China’s overseas lending portfolio by level of public
liability................................................................................................................280
Figure A11: Proportion of China’s overseas lending portfolio provided via buyer’s
credits................................................................................................................ 281
Figure A12: Proportion of China’s overseas lending portfolio provided via
Government Concessional Loans (GCLs) and Preferential Buyer’s Credits (PBCs)...
........................................................................................................................... 281
Figure A13: Proportion of China’s overseas lending portfolio provided via
infrastructure project loans................................................................................ 282
Figure A14: Composition of China’s overseas lending portfolio by emergency
and non-emergency lending instrument............................................................283
Figure A15: Proportion of RMB-denominated rescue lending to countries in and
not in financial distress.......................................................................................284
Figure A16: Composition of China’s overseas lending portfolio by currency of
denomination for countries in financial distress.................................................285
Figure A17: Composition of China’s overseas loan portfolio by currency of
denomination for countries not in financial distress.......................................... 286
Figure A18: Early versus late BRI: weighted average interest rates...................286
Figure A19: Early versus late BRI: weighted maturity lengths........................... 287
Figure A20: Early versus late BRI: weighted average grant element.................287
Figure A21: Weighted average grant element of overseas lending from China.....
........................................................................................................................... 288
Figure A22: Weighted average maturity lengths for countries in and not in
financial distress................................................................................................. 289
Figure A23: Weighted average interest rates for countries in and not in financial
distress............................................................................................................... 290
Figure A24: Weighted average grant elements for countries in and not in
financial distress, 2014-2021..............................................................................291
Figure A25: Grant element across China’s portfolio to countries in and not in
financial distress................................................................................................. 291
Figure A26: Composition of China’s overseas lending portfolio by type of
lending instrument.............................................................................................292
Figure A27: Composition of China’s overseas lending portfolio by type by
creditor category............................................................................................... 293
Figure A28: Proportion of China’s overseas lending portfolio involving a
multilateral institution........................................................................................ 293
Figure A29: Composition of China’s overseas lending with and without credit
enhancements to countries in and not in financial distress............................... 294
Figure A30: Composition of China’s non-emergency overseas loan portfolio by
repayment risk................................................................................................... 294
Figure A31: Annual loan commitments to LICs and MICs by financial institution
and capital injections from SAFE, 2000-2021.................................................... 296
Figure A32: Proportion of China’s syndicated overseas lending that involves
non-Chinese creditors........................................................................................297
Figure A33: Proportion of Chinese overseas syndicated lending involving a
multilateral institution........................................................................................ 298
Figure A34: Average financial commitments from Chinese state-owned creditors
to different types of syndicated loans................................................................298
Figure A35: Chinese state-owned bank contributions to syndicated loans in LICs
and MICs by participant cohort......................................................................... 299
Figure A36: Composition of China’s overseas project lending portfolio by
channel of delivery.............................................................................................299
Figure A37: Official lending commitments by lending institution type, 2000-2021
........................................................................................................................... 300
Figure A38: Cumulative percentage of infrastructure portfolio with significant
environmental, social, or governance risk exposure..........................................301
Figure A39: Infrastructure projects located within environmentally sensitive areas
302
Figure A40: Infrastructure projects located within socially sensitive areas........ 303
Figure A41: Infrastructure projects located in geographical areas vulnerable to
political capture and manipulation.................................................................... 304
Figure A42: Infrastructure projects involving contractors sanctioned for
fraudulent and corrupt behavior........................................................................ 305
Figure A43: Infrastructure projects that encountered significant environmental,
social, or governance challenges over time.......................................................306
Figure A44: ESG risk prevalence in overseas infrastructure portfolio from China
to LICs and MICs................................................................................................307
Figure A45: Environmental risk prevalence in overseas infrastructure...............308
Figure A46: Social risk prevalence in overseas infrastructure portfolio............. 309
Figure A47: Governance risk prevalence in overseas infrastructure portfolio... 310
Figure A48: Infrastructure projects supported by one or more de facto
environmental, social, or governance risk mitigation efforts............................. 311
Figure A49: De jure ESG safeguard stringency in China’s overseas infrastructure
portfolio with ESG risk exposure........................................................................312
Figure A50: Infrastructure project portfolio with de jure vs. de facto ESG risk
mitigation...........................................................................................................313
Figure A51: Infrastructure project portfolio with de jure vs. de facto ESG risk
mitigation, by year of project completion......................................................... 314
Figure A52: Infrastructure project portfolio with de jure vs. de facto ESG risk
mitigation, by year of project commencement..................................................315
Figure A53: Proportion of infrastructure project financing facing significant
environmental risks by whether the project financing had strong de jure
environmental safeguards versus those with weak de jure ESG safeguards......316
Figure A54: Proportion of infrastructure project financing facing significant social
risks by whether the project financing had strong de jure environmental
safeguards versus those with weak de jure ESG safeguards............................. 317
Figure A55: Proportion of infrastructure project financing facing significant
governance risks by whether the project financing had strong de jure
environmental safeguards versus those with weak de jure ESG safeguards......318
Figure A56: Composition of infrastructure project portfolio: Reliance on China
Eximbank and weak de jure ESG safeguards.....................................................319
Figure A57: Composition of infrastructure project portfolio: Reliance on China
Development Bank and weak de jure ESG safeguards......................................320
Figure A58: Proportion of China’s infrastructure project portfolio behind
schedule by de jure ESG safeguard strength.....................................................320
Figure A59: Average length of commencement delays in infrastructure projects
by de jure ESG safeguard strength....................................................................321
Figure A60: Average length of completion delays in infrastructure projects by de
jure ESG safeguard strength.............................................................................. 321
Figure A61: Average time to complete an infrastructure project by de jure ESG
safeguard strength............................................................................................. 322
Figure A62: ESG map - project count................................................................323
Figure A63: Official sector PRC lending to LICs and MICs versus Y1 HICs....... 329
Table A2: Contractors sanctioned for fraudulent and corrupt behavior that were
involved in Chinese ODA- and OOF-financed infrastructure projects during their
debarment periods............................................................................................ 332
Table A3: Criteria to evaluate environmental, social, and governance safeguards
in Chinese grant and loan contracts.................................................................. 336
Table A4: Safeguard stringency ratings for infrastructure financing agreements at
the ESG domain-contract dyad level................................................................. 339
Table A5: Sample of infrastructure financing agreements used to code the de
jure stringency of ESG safeguards..................................................................... 341
Table A6: Summary of environmental, social, and governance safeguard clauses
in the contract sample for the pre-BRI and early BRI period (2000-2017)......... 344
Table A7: Summary of environmental, social, and governance safeguard clauses
in the contract sample for the late BRI period (2018-2021)...............................346
Table A8: Environmental Social Governance Safeguard Questions.................. 348
Table A9: Composition of coding sample of infrastructure financing agreements
versus China’s entire grant- and loan-financed infrastructure project portfolio. 354
Table A10: Regional distribution of environmental, social, and governance risk
exposure in China’s overseas infrastructure project portfolio............................ 355
Table A11: Distribution of environmental, social, and governance risk exposure
in China’s overseas infrastructure project portfolio by host country income level...
356
Table A12: Country-by-country distribution of ESG risk exposure and de jure
ESG safeguard protection in China’s overseas infrastructure project portfolio,
2000-2021..........................................................................................................357
Table A13: China’s soft power gains and losses vis-à-vis the U. S. during early
and late BRI periods...........................................................................................365
Table A14: Country-by-country breakdown of expected versus observed
development finance allocations during late BRI period...................................372
Table A15: Official sector lending commitments from China to LICs and MICs:
Country-by-country comparison of AidData, IDS, and CODF in constant 2021
USD millions.......................................................................................................380
Table A16: Country-by-country comparison of AidData and IDS estimates of PPG
and non-PPG debt exposure to China in constant 2021 USD millions.............. 388
Chapter 1: Belt and Road Reconstruction—From
Fire-Fighting to Future-Proofing

Section 1: Has Beijing’s global infrastructure initiative

become an asset or a liability?
Two competing narratives about China’s overseas development program are
gaining traction. The first is that Beijing sits atop a mountain of foreign exchange
reserves and has gained the upper hand in a zero-sum, great power competition
by using its flagship infrastructure program—the Belt and Road Initiative
(BRI)—to win the loyalty of foreign leaders and generate favorable international
public sentiment. According to this narrative, China is an international lender of
first resort that developing countries cannot afford to alienate or antagonize. It
bankrolls big-ticket infrastructure projects—such as roads, railways, bridges,
tunnels, power plants, and telecommunication systems—that neither the U.S.
nor its allies will support. It plies foreign leaders with lavish spending on vanity
projects like statues, theaters, museums, convention centers, stadiums,
presidential palaces, and parliamentary buildings. It is also a deep-pocketed
lender of last resort that is willing and able to bail out borrowers when they are
underwater or sailing into strong macroeconomic headwinds. Strategists and
decision-makers in Western capitals often invoke this argument—about Chinese
strength—to make the case that Beijing is “outmaneuvering us in overseas
theaters,” “eating our lunch,” or “winning the global battle for hearts and
minds.”1

But a second narrative—a counter-narrative about Chinese weakness—is gaining

currency in Washington, London, and Brussels. The BRI, in this alternative
reading, is more of a liability than an asset, and Beijing is on a perilous path, in
danger of spending its way into international disrepute. Developing countries

1
The 2018 National Defense Strategy of the U.S. asserted that “[t]he central challenge to U.S. prosperity
and security is the reemergence of long-term, strategic competition…[with] revisionist powers” (U.S.
Department of Defense 2018). It called upon the U.S. government to "out-think, out-maneuver, out-partner,
and out-innovate revisionist powers” (U.S. Department of Defense 2018). In October 2018, the U.S.
Congress passed the Better Utilization of Investment Leading to Development (BUILD) Act, establishing a
“full service” development finance institution to help the U.S. government compete with China around the
globe. Then, in September 2019, it authorized the creation of a “Countering Chinese Influence” fund.
1
that borrowed extensively from China are now saddled with debts for oversized
infrastructure projects that generate insufficient revenue. The grace periods on
these loans are expiring, forcing Beijing into an unfamiliar and uncomfortable
role as the developing world’s largest official debt collector. On one hand, it
wants to position itself at the front of the repayment line by requiring that
borrowers grant it sources of leverage—such as cash collateral in escrow
accounts—that other official creditors do not possess. On the other hand, it
wants to characterize itself as a global champion of “South-South cooperation”
that privileges solidarity with low-income countries (LICs) and middle-income
countries (MICs). Another source of vulnerability is that politicians in the Global
South increasingly feel compelled to cancel or mothball high-profile BRI projects
because rising levels of public antipathy toward China are making it difficult to
maintain close relations with Beijing. International media outlets are also training
their sights on an array of problems in the BRI project portfolio, such as
overpricing, corruption, habitat destruction, and involuntary displacement of
vulnerable and marginalized populations.

The fundamental difference between these narratives is that one assumes China
is playing offense and the other assumes China is playing defense. Which one is
a better characterization of reality? Or is it possible that both—or neither—are
true? Beijing is clearly aware that it faces a BRI “buyer’s remorse” problem.2 But
have Chinese lending and grant-giving institutions learned from their past
mistakes and recalibrated their policies and practices? If so, how?

One of the first signs that Beijing was considering a major course correction
came in October 2016 when an official with China’s National Development and
Reform Commission (NDRC) told a London-based newspaper that “these days
we need viable projects and a good return. We don’t want to back losers”
(Financial Times 2016). Then, in May 2017, Xi Jinping announced that “[w]e will
[...] strengthen international cooperation on anti-corruption in order to build the
Belt and Road Initiative with integrity.”3 He delivered a similar message in
September 2018: financing from China was “not to be spent on any vanity
projects but in places where they count the most.”4
2
Euractiv 2023.
3
Belt and Road Forum for International Cooperation 2017.
4
Quoted in Shepherd and Blanchard 2018.
2
Calls for a major overhaul of the country’s flagship global infrastructure
initiative—dubbed “BRI 2.0”—grew louder over the course of the next two
years. In April 2019, at the Second Belt and Road Forum for International
Cooperation, Xi announced that the next phase of the BRI (“BRI 2.0”) would be
“open, green and clean” (Ministry of Foreign Affairs of the People’s Republic of
China 2019). He also said that China would “adopt widely accepted rules and
standards and encourage participating companies to follow general
international rules and standards in project development, operation,
procurement and tendering and bidding” (Ministry of Foreign Affairs of the
People's Republic of China 2019). Then, in 2020, Beijing signaled interest in
“multilateralizing” the BRI and harmonizing some of its policies and practices
with prevailing international development finance rules and standards.5

However, it is unclear if these rhetorical commitments have translated into

significant changes in the way that Beijing administers its overseas development
program. Some “China watchers” have expressed skepticism that Beijing is
sincerely interested in financing, designing, and implementing infrastructure
projects in coordination and collaboration with multilateral institutions and
Western powers. One reason to question the credibility of Beijing’s commitment
to BRI 2.0 is that the country’s comparative advantages (scale, speed, and
near-term economic impact) vis-à-vis OECD-DAC donors and multilateral
development banks may not be fully compatible with BRI multilateralization.6
Another reason is that China has a long track record of engaging in strictly
bilateral, “go-it-alone” efforts to manage overseas crises.7 Intensifying
competition in the global infrastructure finance market presents an additional
challenge: the U.S., the U.K., and the other members of the G7 are now
promoting a Partnership for Global Infrastructure and Investment

5
Morris et al. 2020; Dreher et al. 2022. In 2020, Beijing teamed up with eight multilateral institutions to
establish the Multilateral Cooperation Center for Development Finance (AIIB 2021).
6
We address this issue at greater length in Chapter 3. Also see Parks 2019; Malik et al. 2021; and Dreher et
al. 2022.
7
For example, when China’s biggest borrowers have experienced financial distress, it has not called for
coordinated debt reschedulings with all major creditors (Gardner et al. 2021; Bon and Cheng 2021).
Instead, it has discreetly provided emergency rescue loans to ensure that such borrowers are sufficiently
liquid to continue servicing their existing BRI project debts (Horn et al. 2023a, 2023b). Beijing has also
spurned multiple invitations to join the Paris Club (the main venue for sovereign debt restructurings) and
insisted upon loan contracts with clauses that expressly exclude Chinese debts from the Paris Club and
other collective restructuring initiatives, thereby granting its state-owned creditors sole discretion to decide
if, when, and how they will grant debt relief (Dreher et al. 2022; Gelpern et al. 2022).
3
(PGII)—previously known as the Build Back Better World (B3W) initiative—that
they characterize as an alternative, higher quality option for countries that want
to undertake infrastructure projects based on the principles of sustainable and
transparent financing, public sector mobilization of private capital, consultation
and partnership with local communities, and strict adherence to environmental,
social, and governance (ESG) safeguards.8 Therefore, if Beijing wanted to
protect the BRI brand by spreading reputational risk across a larger group of
donors and lenders, it would have to secure the buy-in of a set of actors who do
not seem to be particularly interested in collaboration (Parks 2019; Dreher et al.
2022).

Regardless of whether China can convince traditional donors and lenders to

jump on the BRI bandwagon, there are indications that a reboot of the
infrastructure initiative is underway. In November 2017, the country's top
banking regulator—China Banking Regulatory Commission (CBRC)—introduced
new rules that require China Development Bank (CDB) and China Eximbank to
put in place more robust environmental and social risk management procedures
(CBRC 2017a, 2017b). Then, in 2021, China’s State Administration of Foreign
Exchange (SAFE)—the ultimate source of funding for most of China’s
state-owned policy banks, commercial banks, and investment funds (see Box 2b
in Chapter 2)—announced that it would prioritize “adopting MDB’s ESG criteria”
and “incorporating ESG principles into the whole project investment process
from decision-making to post-investment management” (SAFE 2021: 54). At the
same time, SAFE’s parent organization (the PBOC) called for a reduction in
collateralized lending to overseas borrowers that present high levels of
repayment risk (Chen 2023: 1173).9

In this chapter, we document the evolving nature, scale, and composition of

China’s overseas development program. We also explain why China has pivoted
into an international crisis manager role and how it is seeking to manage three
8
The European Union, Japan, Australia, and India have also launched a set of complementary
initiatives—called Global Gateway, Partnerships for Infrastructure (P4I), the Blue Dot Network, and the
India-Middle East-Europe Economic Corridor (IMEC)—that are designed to increase choice and
competition in the global infrastructure finance market.
9
At the third Belt and Road symposium in November 2021, Xi Jinping also called for better risk
management in overseas infrastructure projects and discouraged Chinese companies from going to
“chaotic and dangerous” places (State Council of the People's Republic of China 2021).

4
different types of risk in its overseas project portfolio: (1) repayment risk, (2)
ESG-related project performance risk, and (3) reputational risk.

In Chapters 2 and 3, we identify the specific measures that Beijing has taken to
de-risk its overseas development finance portfolio, and evaluate whether these
changes are consistent with its rhetorical commitments to multilateralization and
harmonization with the prevailing international development finance rules and
standards. Then, in Chapter 4, we explain how China has used aid and credit
instruments to respond to the soft power gains and losses it has experienced in
LICs and MICs. Our findings suggest that the ambition of Beijing’s effort to
“future-proof” its overseas development finance portfolio—and its flagship,
global infrastructure initiative—should not be underestimated.

Section 2: Myth-busting evidence about the scale and

composition of China’s overseas development
program—and how its rivals have responded
AidData maintains the world’s most comprehensive dataset of international
development finance commitments from China to the developing world. The
latest (3.0) version of our Global Chinese Development Finance (GCDF) dataset
captures 20,985 projects and activities in 146 countries supported by financial
and in-kind transfers worth $1.34 trillion from official sector institutions in
China.10 It covers every low-income, lower-middle income, and upper-middle
income country and territory across every major world region, including Africa,
Asia, Oceania, the Middle East, Latin America and the Caribbean, and Central
and Eastern Europe (including 19 additional countries or territories where
systematic searches were undertaken but no Chinese government-financed
projects and activities were identified). The dataset tracks projects and activities
over 22 commitment years (2000-2021), and it includes details on the timing of
project/activity implementation over a 24-year period (2000-2023). 17,957
records in the dataset represent formally approved, active, and completed
projects and activities. The remaining 3,028 records in the 3.0 dataset represent
(1) projects and activities that secured official financial or in-kind commitments

10
The $1.34 trillion figure excludes the short-term “rollover” facilities described in Box 2c and Section A-3).
It increases to $1.5 trillion when such facilities are included in the tally.
5
from China but were subsequently suspended or canceled; (2) projects and
activities that secured pledges of financial or in-kind support from official sector
institutions in China but never reached the formal approval (official commitment)
stage; and (3) so-called “umbrella” records that are designed to support
multiple subsidiary projects and activities. Figure 1.1 presents the global
distribution of approved, active, and completed projects and activities that were
financed with Official Development Assistance (ODA) and Other Official Flows
(OOF) from China between 2000 and 2021.11

Figure 1.1: The global distribution of Chinese ODA- and OOF-financed projects in LICs and MICs

Figure 1.1
The global distribution of Chinese ODA- and OOF-financed
projects in LICs and MICs

Notes: This map shows the geographic locations of projects supported by Chinese ODA and OOF
commitments across all LICs and MICs between 2000 and 2021. Only projects from the 3.0 version of

11
All projects and activities in AidData’s GCDF dataset must qualify as official financial flows (ODA or OOF).
For ease of exposition, in the remainder of this report, we refer to all such projects and activities as
“projects.” The definitions of and measurement standards for ODA and OOF are described in Section A-2
of the Appendix.
6
AidData’s GCDF dataset that have physical footprints or involve specific locations are represented.
Goodman et al. (2023) describes the process by which these point, polygon, and line vector data were
generated.

When AidData released the 2.0 version of the GCDF dataset in September
2021, it provided evidence that China was outspending the U.S. in the
developing world on at least a 2-to-1 basis.12 The 3.0 version of the GCDF
dataset demonstrates that China remains the single largest source of
international development finance in the world. In 2021, official financial flows
(ODA and OOF commitments) from China to LICs and MICs amounted to $79
billion (see Figure 1.2).13 None of China’s peers or rivals have overseas lending
and grant-giving programs that are comparable in scale. In 2021, no single
member of the G7 provided official financial flows to LICs and MICs in excess of
$61 billion.14 China also outspent all multilateral sources of international
development finance. The single largest multilateral source of international
development finance in 2021 was the World Bank, with international
development finance commitments worth approximately $53 billion.

According to the conventional wisdom among think tanks and media outlets,
Beijing made a concerted effort to rein in the BRI after 2017 (Lu 2023; Olander
2023; Do Rosario and Savage 2023) and its annual development finance
commitments plummeted to nearly zero by 2021 (Gallagher and Ray 2020;
Hwang et al. 2022; Ray 2023; Myers and Ray 2023; Moses et al. 2023).15
However, the 3.0 version of AidData’s GCDF dataset debunks the myth that
Beijing’s overseas development program has experienced a total collapse. With

12
At the time, we estimated that China’s average annual development finance commitments amounted to
$85.4 billion between 2013 and 2017 and average annual development finance commitments from the U.S.
amounted to $37 billion over the same five-year period (Malik et al. 2021). However, the latest (3.0) version
of the GCDF dataset demonstrates that China was outspending its rivals by a wider margin than we
previously understood: average annual development finance commitments from China amounted to $117
billion between 2013 and 2017 and average annual development finance commitments from the U.S.
amounted to $40.6 billion over the same five-year period (in constant USD 2021 prices). These historical
revisions imply that Beijing was outspending Washington on a nearly 3-to-1 basis during the early years of
the BRI.
13
To see the annual percentage of China’s official financial flows to LICs and MICs provided via ODA and
OOF, see Figure A1 in the Appendix.
14
Since the first full year of BRI implementation in 2014, no member of the G7 has outspent China. The
individual members of the G7 spent between $2 billion and $61 billion a year on overseas development
activities between 2014 and 2021. China spent between $74 billion and $142 billion a year between 2014
and 2021.
15
In a recent stock-taking exercise, Nikkei Asia concluded that “initial optimism [for the BRI] has been
replaced by disappointment over mismanagement, debt crises and corruption that have left many projects
unfinished or incapable of fulfilling their promised potential (Aamir et al. 2022).
7
more complete data on the full range of China’s lending and grant-giving
activities in LICs and MICs, it shows a far less dramatic decline in overseas
spending during the late BRI period: official financial flows (ODA and OOF
commitments) from China to LICs and MICs declined from $115 billion in 2018
to $104 billion in 2019 and $74 billion in 2020, before increasing to $79 billion in
2021 (see Figure 1.2).16

Figure 1.2: Official financial flows from China to the developing world, 2000-2021

Figure 1.2

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (see
Section A-2 of the Appendix for details). The Vague (Official Finance) is a residual category for official
financial commitments from China that could not be reliably categorized as ODA or OOF because of
insufficiently detailed information.

In Chapter 2, we set the record straight by demonstrating that a sharp decline in

dollar-denominated, bilateral lending for public investment projects did indeed
take place between 2018 and 2021, but it was offset to a significant extent by

16
In Chapter 2, we explain why AidData’s estimates of official sector lending commitments from China to
LICs and MICs are substantially different from those recorded in other publicly available databases, such as
the China's Overseas Development Finance Database, the Chinese Loans to Latin America and the
Caribbean Database, the Chinese Loans to Africa Database, the China's Global Energy Finance Database,
the China Overseas Finance Inventory Database, and the World Bank's International Debt Statistics (IDS).
8
contemporaneous increases in RMB-denominated, bilateral emergency rescue
lending and increases in non-emergency lending via syndication and
multilateralization. We also provide evidence that, although Beijing is reducing
its reliance on the policy banks, it is ratcheting up its use of the country’s
state-owned commercial banks and central bank.

The scale of China’s ambition as a global development financier becomes even

clearer in an analysis of “stocks” rather than “flows.”17 Figure 1.3 provides
evidence that the inflation-adjusted value of Beijing’s 21st-century overseas
development finance portfolio reached the $745 billion mark by the first full year
of BRI implementation (2014) and surpassed the $1 trillion mark by the fourth full
year of BRI implementation (2017). The portfolio continued to expand in
subsequent years—so much so that it exceeded $1.3 trillion in 2021.18 Figure 1.4
presents the cumulative amount of Chinese ODA and OOF from 2014 to 2021
and compares it to cumulative amounts of ODA and OOF from each member of
the G7 over the same time period. China’s overseas development program is
vastly larger than that of any G7 member, including the U.S. Beijing has outspent
Washington in LICs and MICs on a more than two-to-one basis since the BRI was
launched.19 Between 2014 and 2021, official financial flows from China to the
developing world amounted to $680 billion, while those from the U.S. amounted
to $319 billion.20 Over the same time period, China also outspent the single
largest source of multilateral development finance—the World Bank—on a more
than two-to-one basis.21

17
The rationale for evaluating “stocks” (cumulative financial commitments) rather than “flows” (annual
financial commitments) is that grants and loans are issued at specific points in time, but they disburse over
many years and support projects implemented over many years.
18
Section A-3 in the Appendix provides guidance for those who wish to use the 3.0 version of AidData's
GCDF dataset to estimate cumulative stocks of official financial flows from China to LICs and MICs.
19
China’s average annual international development finance (ODA/OOF) commitments amounted to $85
billion between 2014 and 2021 (excluding the short-term "rollover" facilities described in Box 2c and
Section A-3). This figure rises to $105 billion if one includes short-term "rollover" facilities. Average annual
international development finance (ODA/OOF) commitments from the U.S. amounted to $40 billion
between 2014 and 2021.
20
The $680 billion figure excludes the short-term "rollover" facilities described in Box 2c and Section A-3. It
increases to $841 billion when such facilities are included in the tally.
21
Total international development finance (ODA and OOF) commitments from the World Bank amounted
to $307 billion from 2014-2021.
9
Figure 1.3: Cumulative official financial flows from China to the developing world, 2000-2021

Figure 1.3

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (as
described in Section A-2 of the Appendix). The Vague (Official Finance) is a residual category for official
financial commitments from China that could not be reliably categorized as ODA or OOF because of
insufficiently detailed information. This figure excludes short-term "rollover" facilities from the tally of
official financial commitments (see Box 2c and Section A-3 in the Appendix).

10
Figure 1.4: Official financial flows from China and the G7 to the developing world during the BRI era, 2014-2021

Figure 1.4

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (as
described in Section A-2 of the Appendix). The Vague (Official Finance) is a residual category for official
financial commitments from China that could not be reliably categorized as ODA or OOF because of
insufficiently detailed information. G7 ODA and OOF data represent gross disbursements from the
OECD-DAC. This figure excludes short-term “rollover” facilities from the tally of official financial
commitments (see Box 2c and Section A-3 in the Appendix).

At the same time, there is evidence that the U.S. is beginning to close the
overseas development spending gap with China (see Figure 1.5). During the
early BRI period (2014-2017), China outspent the U.S. on a nearly three-to-one
basis. However, during the late BRI period (2018-2021), Beijing spent $2.33 for
every overseas development dollar spent by Washington. The gap narrowed
even more during the last year for which reliable data are available: Beijing only
spent $1.30 for every overseas development dollar spent by Washington in
2021.22

22
While official financial flows (ODA and OOF) from the U.S. to LICs and MICs amounted to $61 billion in
2021, official financial flows from China to LICs and MICs amounted to $79 billion during the same year.
11
Figure 1.5: Official financial flows from China, the U.S., and G7 countries to the developing world during the BRI era, 2014-2021

Figure 1.5

Notes: This figure measures official financial flows (ODA and OOF commitments) from China, the U.S. and
G7 countries to LICs and MICs from 2014 to 2021. AidData relies on OECD-DAC measurement criteria to
make ODA and OOF determinations (as described in Section A-2 of the Appendix). The U.S. and G7 ODA
and OOF data represent gross disbursements from the OECD-DAC.

Figure 1.6 demonstrates that the U.S. gained ground on China in 2021 due to a
fifteen-fold (1,423%) increase in OOF expenditure.23 In 2020, OOF represented
just 4% of the U.S. international development finance portfolio; the remaining
96% consisted of ODA. However, one year later, the percentage of U.S.
international development finance provided via OOF soared to 36%. This major
compositional change in U.S. development expenditure suggests that
Washington is seeking to compete with Beijing via emulation rather than
differentiation.24

23
U.S. OOF amounted to $1.4 billion in 2020 and $21.8 billion in 2021.
24
Relatedly, Zeitz (2021) provides evidence that the World Bank is seeking to compete with China via
emulation rather than differentiation.
12
Figure 1.6: Official financial flows from the U.S. to the developing world, 2000-2021

Figure 1.6

Notes: U.S. ODA and OOF flows reflect gross disbursements (as OOF commitment data are not published
by the OECD-DAC for individual DAC members).

The $20.3 billion increase in OOF that took place in 2021 was the result of an
expansion in the overseas activities of the U.S. International Development
Finance Corporation (DFC).25 In October 2018, the U.S. Congress passed the
Better Utilization of Investment Leading to Development (BUILD) Act, which
established the DFC as a “full service” development finance institution to help
the U.S. “outcompete” China around the globe.26 However, the DFC did not
become fully operational until 2021. One of its earliest (attempted) transactions
was a $2.8 billion loan in January 2021 to help the Government of Ecuador
repay some of its outstanding debts to China ahead of schedule, in exchange
for a commitment to exclude Chinese companies from its telecommunications
networks.27 The proposed borrowing terms of the DFC loan were

25
In July 2023 correspondence with AidData, the USAID office responsible for ODA and OOF reporting to
the OECD-DAC confirmed that the full operation of the DFC in 2021 led to the major increase in U.S. OOF
in 2021. See also DFC 2021a and DFC 2022. The DFC's transaction-level data, which is organized by fiscal
year rather than calendar year, can be accessed via https://www.dfc.gov/our-impact/transaction-data.
26
Akhtar and Brown (2021); Dreher et al. (2022).
27
Sevastopulo and Long (2021); Landers et al. (2021).
13
non-concessional in nature: an 8-year maturity, a 1-year grace period, and an
interest rate of LIBOR plus a 2.25% margin.28 At the time, DFC CEO Adam
Boehler said that the loan would “refinance predatory Chinese debt and help
Ecuador improve the value of its strategic assets.”29

But U.S. spending patterns do not provide the full picture because Washington
is seeking to outcompete China by partnering with its allies in London, Paris,
Berlin, Tokyo, Rome, and Ottawa. G7 efforts to compete with China are
gathering steam. During the early BRI period (2014-2017), China and the G7
were effectively matching each other dollar-for-dollar: for every overseas
development dollar that China spent, the G7 spent $1.09.30 However, during the
late BRI period (2018-2021), the G7 stepped up its efforts, spending $1.47 for
every overseas development dollar spent by China (see Figure A2 in the
Appendix).31 By 2021, the G7 was outspending China on a nearly two-to-one
basis (see Figure 1.5).32

As the U.S. and its allies seek to compete with China by rolling out flagship
infrastructure programs (like the Partnership for Global Infrastructure and
Investment and the India-Middle East-Europe Economic Corridor Initiative) and
ramping up non-concessional lending (OOF), a strategic pivot is underway in
Beijing. Figure 1.7 provides evidence of major changes in the sectoral
composition of China’s overseas development finance portfolio between 2014
and 2021.33 Beijing was for the most part focused on providing credit for

28
Bruni (2021). The January 2021 framework agreement between the DFC and Government of Ecuador can
be accessed in its entirety via
https://www.dropbox.com/s/oy377uc6wz8u9oe/Ecuador%20DFC%20January%202021%20Framework%20
Agreement.pdf?dl=0.
29
DFC (2021b).
30
The G7 failed to match China’s ODA and OOF commitments in 2014 and 2016 (see Figure 1.5).
31
While China spent $93 billion a year on average between 2018 and 2021, the G7 spent $137 billion a
year on average.
32
In 2021, official financial flows (ODA and OOF) from all members of the G7 (combined) to LICs and MICs
reached $163 billion, while official financial flows from China to LICs and MICs amounted to $79 billion.
33
According to Figure A3, the percentage of China’s overseas development finance portfolio that
supported three “hardware” sectors (Energy; Industry, Mining, and Construction; and Transport and
Storage) declined from 68.3% in 2014 (the first full year of the early BRI period) to 30.6% in 2021 (the last
year of the late BRI period for which we have complete data). The percentage of China’s overseas
development finance portfolio that supported the “Banking and Financial Services” sector and “General
Budget Support” sector increased from 13.0% to 58.9% between 2014 and 2021. These two sectors
capture emergency lending from China’s central bank (PBOC) to LIC and MIC central banks via currency
swap arrangements, as well as emergency lending to finance LIC and MIC finance ministries via China’s
State Administration of Foreign Exchange (SAFE), state-owned policy banks, state-owned commercial
banks, and state-owned energy companies
14
large-scale infrastructure projects during the early BRI (or BRI 1.0) era. Yet, as we
explain at greater length in Chapter 2, it ramped down infrastructure project
lending and ramped up emergency rescue lending during the late BRI (or BRI
2.0) period. Beijing made this course correction in order to adapt to a new
reality: the fact that many of its largest borrowers were having serious difficulty
repaying their infrastructure project debts.34

Figure 1.7: Change in sectoral composition of official financial flows from China to the developing world, 2014-2021

Figure 1.7

Notes: This figure, which relies on 3-digit OECD sector codes from the 3.0 version of AidData’s GCDF
dataset, presents proportional changes in the sectoral composition of Chinese ODA and OOF
commitments (measured in constant 2021 USD) to LICs and MICs between 2014 and 2021. Figure A3 in the
Appendix provides supplementary evidence on sectoral changes over the same time period.

There are glass-half-full and glass-half-empty ways of interpreting G7 efforts to

compete with China. The glass-half-full view is that Beijing is ceding its
leadership in the global infrastructure space and the G7 is seeking to fill the
vacuum, which may help address the large, unmet infrastructure financing needs
of developing countries. The glass-half-empty view is that the G7 is misreading
the demand signal from the Global South and seeking to compete with a
version of the BRI (BRI 1.0) that no longer exists. After recently going on a
borrowing spree for big-ticket infrastructure projects, low-income and
middle-income countries may have less appetite for expensive infrastructure and
more appetite for balance of payments support that will keep them afloat during

34
Beijing effectively created a backstop for highly exposed Chinese creditors by channeling emergency
rescue loans to the subset of BRI participant countries that present major balance sheet risks (see Horn et
al. 2023a, 2023b as well as the replication exercise that we conduct in Chapter 2).
15
a time when global economic conditions are highly unfavorable (as interest rates
rise, the dollar strengthens, local currencies weaken, and economic growth
slows).

Section 3: Repayment risk from Beijing’s perspective

Beijing’s pivot away from infrastructure project lending and toward emergency
rescue lending is as much about supply-side pressures as it is about
demand-side pressures. After the Belt and Road Initiative was launched, Chinese
state-owned creditors went on a lending spree, issuing thousands of loans for
big-ticket infrastructure projects spread across 129 countries in the developing
world. However, they did so without strong risk management guardrails in place.
They lent to borrowers with bad credit ratings or no credit ratings (like Laos,
Tajikistan, Zambia, South Sudan, Suriname, Zimbabwe, Pakistan, and Argentina);
banked on borrowers being able to repay loans with the cash proceeds from
natural resource exports (like Angola, Equatorial Guinea, Ecuador, Venezuela,
Congo-Brazzaville, and Turkmenistan); and issued dollar- and euro-denominated
loans to countries (like Russia, Belarus, Myanmar, Sudan, Iran, and Cuba) that
would later be unable to transact in those currencies due to international
sanctions.

Now, Chinese state-owned creditors are saddled with many underperforming

loans and want to ensure that their overseas borrowers are sufficiently liquid to
continue servicing their existing infrastructure project debts.35 They are
responding to this challenge in two ways, which we document in greater detail
in Chapter 2: debt reschedulings that provide short-term cash flow relief to
borrowers and emergency rescue loans that help borrowers shore up their
foreign exchange reserves and repay existing (infrastructure project) debts.

35
Chinese state-owned creditors have for the most part responded to this challenge via cash flow relief
(emergency rescue loans, grace period extensions, and/or maturity extensions), which suggests that they
think their borrowers are illiquid but not insolvent (Horn et al. 2023a, 2023b). In a recent interview with
Muyang Chen of Peking University, one policy bank official characterized the rising tide of debt distress in
the developing world as “essentially a problem of liquidity” (Chen 2023: 1772).
16
Figure 1.8: Percentage of China’s portfolio of loan commitments supporting countries in financial distress, 2000-2021

Figure 1.8

Notes: MOFCOM interest-free loan commitments (which are typically issued without a credible expectation
of repayment) and emergency rescue loan commitments (responses to financial distress) are excluded from
the calculation. To determine if a country was experiencing financial distress in a given year, we use the
binary measure described in Box 1a.

17
Figure 1.9: China’s share of overdue repayments owed to official creditors, 2000-2021

Figure 1.9

Notes: Sovereign arrears capture principal and interest arrears (i.e., overdue repayments) on PPG debt to
China Eximbank, China Development Bank, and China’s Ministry of Commerce contracted by a subset of
LICs and MICs that participate in the World Bank’s Debtor Reporting System (DRS). Each country-year
observation is given equal weight in a given year to generate global averages. Years in which a country
maintained diplomatic relations with Taiwan are excluded. The data are drawn from the World Bank’s
International Debt Statistics.

Figures 1.8, 1.9, and A4 in the Appendix highlight the scope and severity of the
challenge. Until 2008, Beijing never had to deal with more than 10
financially-distressed countries with unpaid debts to Chinese state-owned
creditors; however, by 2021, at least 57 countries with outstanding debt to China
were in financial distress (see Figure A4).36 In 2000, 34% of China's overseas

36
According to the World Bank’s International Debt Statistics (IDS), sovereign arrears from LICs and MICs to
official sector creditors in China have also sharply increased in absolute terms (see Figure A5). However,
due to the credit coverage and underreporting issues that affect IDS data (see Figure 2.1, Table A15, Horn
et al. 2021, and Malik et al. 2021), the absolute amounts in Figure A5 should be interpreted with caution.
18
lending portfolio supported borrowers in financial distress.37 By 2021, that figure
skyrocketed to 79% (see Figure 1.8).38

Box 1a: How AidData identifies when China’s borrowers are experiencing
financial distress
Figure A4, Figure 1.8 and other graphs and tables in this report require a measure of when
countries have borrowed from China under normal circumstances and when countries have
borrowed from China during periods of financial distress. In order to determine whether and
when a borrower country experienced a financial distress episode, we identify if it met any one
of the following conditions in a given year:

● It registered a score of 5 or less on the sovrate index produced by the World Bank;

● Its overdue repayments on public debt to China were equal to or greater than 10% of its
of total outstanding public debt to China (as measured by the World Bank); or

● One or more of its official sector loans from China—that are within their originally
scheduled repayment periods—showed signs of financial distress (as measured by the
3.0 version of AidData’s GCDF dataset).

The World Bank’s sovrate index is a measure of repayment risk—based on average sovereign
credit ratings from Moody’s, Standard & Poor’s, and Fitch—that varies from 0 to 21, with higher
scores indicating lower levels of sovereign credit risk (Kose et al. 2022). Countries with scores of
0-5 are in “C and D territory” on the measures produced by the Big Three credit rating
agencies—i.e., they are in default or at a high risk of default (Teixeira et al. 2018; Séri et al.
2021).39 The data on overdue loan repayments to China are drawn from the World Bank’s
International Debt Statistics (IDS), which is based on voluntary reporting by 119 low-income and
middle-income countries. We use these data to measure whether a country’s principal and
interest arrears on public and publicly guaranteed (PPG) debt to official sector creditors in China
are equal to or greater than 10% of its total outstanding PPG debt to official sector creditors in
China.40 We also use a variable from the 3.0 version of AidData’s GCDF dataset that measures on

37
Figure 1.8 draws inspiration from Figure 1 in Horn et al. (2023b), which relies on an earlier (2.0) version of
AidData’s GCDF dataset. Both figures seek to measure the percentage of official sector Chinese lending to
LIC and MIC borrower countries in distress, but in somewhat different ways. The shape of the line in Figure
1.8 is different from the one in Horn et al. (2023b), in that the onset of exceptionally high levels of distress
(in excess of 50%) begins in 2011 rather than 2022. Figure 1.8 also suggests that nearly 80% of China’s
overseas lending to LICs and MICs is now supporting countries in distress. Horn et al. (2023b) estimate that
the figure is closer to 60%, although they use a version of the GCDF dataset which captures 3,030 official
sector loans from China rather than the 4,890 official sector loans from China captured in the 3.0 version
(including 4,776 for approved, active, and completed projects/activities and 114 for suspended and
canceled projects/activities).
38
During the pre-BRI period (2000-2013), 32% of China's overseas lending portfolio supported borrowers in
financial distress on average each year. This figure increased to 79.7% during the early BRI period
(2014-2017) and remained unchanged (79.7%) during the late BRI period (2018-2021).
39
The sovereign credit ratings produced by Moody’s, Standard & Poor’s, and Fitch vary between AAA and
D. Ratings of BB or lower are considered to be “junk territory.”
40
We exclude all observations in years when a country maintained diplomatic relations with Taiwan (since
the reported arrears may be to Taiwan).
19
a loan-by-loan basis whether the borrower had difficulty making repayments or showed signs of
financial distress during the repayment period.41

Our measure of whether a given borrower country experienced a financial distress episode is
reset to 0 (“turns off”) in a given year if the country's sovrate score exceeds 5.0, its overdue
repayments on public debt to China are no longer equal to or greater than 10% of its of total
outstanding public debt to China, and none of its loans from official sector creditors in China
(that are within their originally scheduled repayment periods) show signs of financial distress.

Another useful gauge of the health of China’s overseas lending portfolio is the
extent to which its borrowers are falling behind on their repayments to Chinese
creditors (in comparison to other external creditors). Figure 1.9 tracks the
percentage of all overdue payments (principal and interest arrears) from
low-income and middle-income governments to all official creditors that are
owed to creditors in China. It shows that a rapidly growing proportion of
overdue loan repayments are owed to Chinese state-owned creditors. This
figure more than doubled—from 7% in 2000 to 19% in 2021.42

Section 4: Project performance risk from Beijing’s

perspective
Beijing has channeled an extraordinary amount of aid and credit to the
developing world for large-scale infrastructure projects. According to the 3.0
version of AidData’s GCDF dataset, which includes a new measure of whether
projects involve the construction, rehabilitation, expansion, or maintenance of
physical infrastructure, official sector donors and lenders in China issued 4,800
grants and loans (worth $825 billion) for infrastructure projects in 140 developing
countries between 2000 and 2021.43 Infrastructure projects are notoriously
difficult to implement (for reasons that we discuss at greater length in Chapter

41
Once this dummy variable is set to 1 (“turns on") in a particular country-year, it retains a value of 1 (“stays
on”) for that country until the end of the loan’s originally scheduled repayment period. The repayment
period is calculated for each loan commitment (regardless of whether it was subsequently suspended or
canceled) based on the originally scheduled first repayment date (estimated by adding the grace period to
the commitment date) and the originally scheduled final repayment (maturity) date (estimated by adding
the maturity to the commitment date). In cases where the grace period is unknown, the average grace
period across all official sector loans from China to the same borrower country is applied. The variable
never “turns on” for MOFCOM interest-free loans (which are typically issued without a credible expectation
of repayment) or emergency rescue loans (responses to financial distress).
42
Sovereign arrears from LICs and MICs to official creditors in China have also sharply increased in absolute
terms (see Figure A5).
43
For more on the definition and measurement of the “infrastructure” variable, see Custer et al.
(2023).
20
3). Therefore, an important part of Beijing’s portfolio management challenge is
dealing with project performance risk.

The 3.0 version of AidData’s GCDF dataset is different from other publicly
available datasets of Chinese development finance in that it captures project
suspensions and cancellations (see Figure A6). We have previously encountered
criticism for publishing data on infrastructure—and non-infrastructure—projects
backed by official commitments that are subsequently suspended or canceled.44
However, we maintain that it is important to systematically track these projects
and subject them to analysis. Shielding suspended and canceled projects from
public scrutiny leaves analysts and decision-makers with an incomplete picture
of Beijing’s overseas development program. It also limits opportunities to learn
from failure.

Figure 1.10: Canceled or suspended Chinese infrastructure projects, 2000-2021

Figure 1.10

44
Users of the 3.0 version of AidData’s GCDF dataset who wish to exclude suspended and
canceled projects from their analysis can easily do so by using the “Recommended for
Aggregates” filter or the “Status” variable.
21
Notes: Based on Chinese ODA- and OOF-financed projects (including those canceled or suspended since
2000) marked as “infrastructure” in the 3.0 version of AidData’s GCDF dataset. No projects that were
committed in the year 2000 were subsequently canceled or suspended.

Figure 1.10 demonstrates that suspensions and cancellations of infrastructure

projects have rapidly accumulated over time. By 2021, 94 infrastructure projects
in 49 countries secured Chinese grant and loan commitments worth $56 billion
that were subsequently suspended or canceled. These projects are likely viewed
by Beijing as evidence of BRI buyer’s remorse, as 80% of the suspensions and
cancellations that took place between 2014 and 2021 were in BRI participant
countries.45 However, this method of measurement is a conservative,
lower-bound estimate of the total number of infrastructure projects that China
unsuccessfully sought to bankroll and build, as it excludes those that were
shelved before securing financial commitments (Lu et al. 2023b).46

Figure 1.11: Early versus late BRI: Chinese government-financed infrastructure projects that are behind schedule

Figure 1.11

Notes: “Behind schedule” is defined as Chinese ODA- and OOF-financed projects where the actual
implementation start date took place 3 months or more after its originally scheduled implementation start
date, as well as projects where the actual completion date took place 3 months (or more) after its originally
scheduled completion date.

Another source of implementation risk and potential cause for concern is the
increase in the proportion of Beijing’s infrastructure project portfolio running

45
To calculate this figure, we first determine which countries had officially joined the BRI by the
end of 2021, and then calculate the share of suspended and canceled projects in BRI participant
countries between 2014 and 2021.
46
Take for example the project to build a 97-km “oil road” connecting Masindi-Biso, Kabaale-Kiziranfumbi
and Hohwa-Nyairongo-Kyarushesha-Butoole in Uganda (captured via ID#96073 in the 3.0 version of
AidData’s GCDF dataset). In December 2021, Uganda’s Ministry of Finance withdrew its request for
parliamentary authorization to contract a loan from China Construction Bank for the project. Therefore,
AidData status-codes the project as “Pipeline: Pledge” rather than a suspension or cancellation of a
financial commitment.
22
behind schedule between the early BRI period and the late BRI period (see
Figure 1.11).47 However, this measure of whether a project is running behind
schedule captures two different types of delays: commencement delays and
completion delays.48 More specifically, it measures whether a project’s (a) actual
implementation start date took place 3 months after its scheduled
implementation start date or longer, and/or (b) its actual completion date took
place 3 months after its scheduled completion date or longer. When this
summary metric is unbundled into its constituent parts, a more complex and
nuanced picture of China’s overseas infrastructure project portfolio emerges.
Figure 1.12 demonstrates that, while commencement delays have increased,
completion delays have not. The average commencement delay increased by 32
days between the early BRI period and the late BRI period, while the average
completion delay shrank by 59 days over the same two time periods.49

Figure 1.12: Early versus late BRI: Commencement and completion delays

Figure 1.12

Notes: This figure is based on active and completed infrastructure projects financed with
Chinese ODA and OOF. Delays are calculated by taking the difference (in calendar days)
between the originally scheduled project implementation start date/completion date and the
actual project implementation start date/completion date (respectively).

The fact that a shrinking proportion of infrastructure projects launch on their

originally scheduled commencement dates could be related to several different

47
Figure 1.11 demonstrates that not only the proportion but also the overall size of Beijing’s infrastructure
project portfolio (in constant 2021 USD) running behind schedule increased between the early BRI period
and the late BRI period.
48
This measure is derived from the “Deviation from Planned Implementation Start Date” and “Deviation
from Planned Completion Date” variables in the 3.0 version of AidData’s GCDF dataset.
49
Figure A7 provides evidence of a steady decline in average completion delays—from 571 days
in 2000 to 220 days in 2021.
23
factors. One possibility is that Chinese contractors and/or their host country
counterparts are increasingly expected to comply with environmental, social, or
governance (ESG) standards prior to the start of project implementation—or
they are underestimating the difficulty of meeting these standards before
infrastructure projects can get underway.50 A separate but related possibility is
that Chinese lenders are asking their borrowers to meet more ESG conditions
(so-called “conditions precedent”) prior to the date of the first loan
disbursement, which typically precedes the start of project implementation. We
address these issues at greater length in Chapter 3. Another possibility, which
we address in Chapter 4, is that host country politicians are increasingly reluctant
to “claim credit” for infrastructure projects financed by China via high-profile
groundbreaking ceremonies.

At the same time, not all of the lights on Beijing’s project performance
dashboard are “flashing red.” Chinese lenders and contractors have evidently
learned how to reduce the likelihood that overseas infrastructure projects will
not be finished on time—and the length of any delays that do take place during
project implementation. These improvements could be the result of better
planning (more realistic forecasting of how long it takes to complete projects) or
fewer/smaller scope of work deviations by the contractors responsible for
project implementation. They also call attention to an important measure that
Beijing can take to slow or reverse the rising tide of BRI “buyer’s remorse”:
timely completion of projects that improve the provision of public services in
host countries. In new work with collaborators from Heidelberg University, the
University of Göttingen, and the University of Hong Kong, we provide causal
evidence that the completion of Chinese development projects increases
popular support for the Chinese government in host countries (Wellner et al.
forthcoming).51 We also show in the same study that host country residents are
more likely to report satisfaction with the delivery of public services upon the
completion of Chinese development projects.
50
Typically, the proceeds from a grant or loan from a Chinese state-owned entity are used by the recipient
to finance a commercial contract between a “project owner” in the host country and a contractor from the
financier’s country of origin. The commercial contract typically specifies an expected implementation start
date and an expected completion (implementation end) date, but contractors and/or project owners can
negotiate scope of work and timeline modifications.
51
One potential reason why project completion dates are consequential for reputations is that they "erase
any uncertainty about whether a project will actually reach completion” (Wellner et al. forthcoming). The
same study does not find that project commencement consistently delivers a public opinion dividend.
24
However, Beijing cannot afford to rest on its laurels. Another key finding from
the same study is that these effects erode over time: on average, we find that
the completion of one additional Chinese development project increases public
support for the Chinese government by approximately 3 percentage points in
the short run but only 0.2 percentage points in the longer run (Wellner et al.
forthcoming).

Beijing also faces a daunting set of ESG risks in its overseas development
program. In Chapter 3, we develop and analyze a new composite measure of
ESG risk that captures whether a given Chinese grant- or loan-financed
infrastructure project (1) took place in an area that is environmentally sensitive,
socially sensitive, or vulnerable to political capture and manipulation; (2) relied
on contractors sanctioned by other international financiers for fraudulent and
corrupt behavior; or (3) encountered a significant environmental, social, or
governance problem before, during, or after implementation. Figure 1.13, which
draws upon this measure, indicates that the cumulative number of Chinese
grant- and loan-financed infrastructure projects in LICs and MICs with significant
ESG risk exposure increased from 17 projects in 16 countries in 2000 to 1,693
projects in 125 countries in 2021. As of 2000, Beijing had issued grants and
loans worth $420 million for infrastructure projects in LICs and MICs that faced
one or more significant ESG risks. This figure increased on a cumulative basis to
$470 billion in 2021. The ESG risk prevalence rate, which we define as the
annual percentage of China’s grant- and loan-financed infrastructure project
portfolio (measured in constant 2021 USD) with significant environmental, social,
or governance risk exposure, also increased from 12% in 2000 to 33% in 2021
(see Figure 3.2).52

52
The average ESG risk prevalence rate reached 54% during the early BRI (2014-2017) period and 47%
during the late BRI (2018-2021) period (see Chapter 3 and Figure 44).

25
Figure 1.13: Infrastructure projects facing environmental, social, or governance risks

Figure 1.13

Notes: The presence of significant ESG risk exposure is based on a project-level composite measure that is
described in Section 2 of Chapter 3.

Section 5: Reputational risk from Beijing’s perspective

Managing the risk profile of China's overseas development program also
requires a focus on grassroots, media, and elite support in host countries.53 In
2014, Xi Jinping made the case for reputational risk management in China’s
international activities, arguing that “[w]e should increase China’s soft power,
give a good Chinese narrative, and better communicate China’s message to the
world” (People’s Daily 2014). Reinforcing this point, a senior official from China’s
Ministry of Commerce noted that “the work of foreign aid relates to China’s
image. We cannot tolerate any negligence or projects of poor quality”
(MOFCOM 2014). More recently, at a September 2023 conference on the 10th
anniversary of the BRI, China's Ambassador to the European Union
acknowledged that "[w]hen it comes to the BRI, whether it's good or bad, we

53
Foreign powers undertake reputational risk management efforts that focus on the general public and the
media because they believe that more favorable sentiment can “filter up and influence elite policy to be
more amenable to [their own] interests” (Brazys and Dukalskis 2019: 567).
26
need to listen more to the receiving countries. We know there is a lot of criticism
in the Western media and also from Western governments sometimes but we
care more about the reactions from the receiving states, especially in the Global
South” (Euractiv 2023).

Figure 1.14 presents data from the Gallup World Poll (GWP) on average levels of
public support for the Chinese government and the U.S. government across the
developing world during the early BRI (2014-2017) period and the late BRI
(2018-2021) period. The GWP data show similar levels of public support for
Beijing and Washington during the early BRI period. Both countries saw their
approval ratings in the Global South fluctuate between 50% and 60% between
2014 and 2017. However, as the initial momentum behind China’s flagship
global infrastructure initiative waned and countries re-evaluated the risks and
rewards of their continued participation during the late BRI period, global public
opinion vis-à-vis China soured. Beijing suffered a 16 percentage point loss
between 2019 and 2021; its public approval rating in low-income and
middle-income countries plunged from 56% in 2019 to 40% in 2021.54
Washington, by contrast, saw its approval rating in the Global South rise by 7
percentage points in 2021, thereby opening up a 14 percentage point
advantage over its rival.

Box 1b: How AidData measures grassroots, media, and elite support for China
and the U.S. in the developing world

This report relies on three different measures of Chinese and American soft power in low-income
and middle-income countries: (1) public opinion, (2) media sentiment, and (3) elite support.

We rely on the Gallup World Poll (GWP) for data on public approval of China and the U.S. The
GWP is the most systematic effort to consistently collect public opinion data in every major
world region over time. It provides annual (repeated cross-section) data from 2006 to 2021 for
more than 140 countries. The data are probability-based and nationally representative of the
resident population of 15 years and older. To facilitate our analysis, we first transform the
respondent-level data from WP156 and WP151 (“Do you approve or disapprove of the job
performance of the leadership of China?” and “Do you approve or disapprove of the job

54
Figure A9 in the Appendix provides evidence that this decline is not due to increased indifference toward
China (i.e., the absence of active approval). It is because of an increase in active disapproval—from 44% in
2019 to 60% in 2021. High levels of active disapproval likely reflect multiple factors, including concerns
about the local effects of Chinese development projects and how Beijing handled the COVID-19 pandemic
(Silver et al. 2020; Blair et al. 2022a).
27
performance of the leadership of the United States?”) into two binary indicators that assume
values of 1 if a respondent approves of the leadership of China or the leadership of the U.S.,
respectively.55 We then calculate the percentage of respondents who approved of the leadership
of China or the leadership of the U.S., respectively, at the country-year level.

We measure media sentiment toward China and the U.S. by calculating two sets of average
scores—one for China and one for the U.S.—at the country-year from the Global Database of
Events, Language, and Tone (GDELT) 1.0 Event Database. We rely on the AvgTone variable,
which is algorithmically calculated based on the tone of hundreds of millions of news articles
across nearly 200 countries. This measure varies from -100 to +100, with positive scores
indicating favorable media coverage related to government actors in mainland China (or the
U.S.) and negative scores indicating unfavorable coverage related to government actors in
mainland China (or the U.S.).56 For most countries at most times, AvgTone scores vary between
-10 and +10, with values of 0 indicating neutral media coverage.

We measure elite support for China and the U.S. by assessing the extent to which other
governments align their votes in the United Nations General Assembly (UNGA) with each of
these foreign powers. The UNGA is a venue in which governments have an opportunity to stake
out foreign policy positions that are similar or dissimilar to those adopted by China (or the U.S.).
Our measure of UNGA voting alignment with China (or the U.S.) is based on
“idealpointdistance” estimates between each country in the 3.0 version of AidData’s GCDF
dataset and China (or the U.S.). These estimates are drawn from Version 29.0 of the United
Nations General Assembly Voting Data (Bailey et al. 2017) and they are inverted, such that
higher values indicate higher levels of UNGA voting alignment with China (or the U.S.).57

The global competition for favorable media coverage did not play out in quite
the same way. Figure 1.15 demonstrates that, in absolute terms, China
outperformed the U.S. on this dimension of soft power in the developing world
during both the early BRI period and late BRI period.58 This pattern is consistent
with evidence that Beijing’s grassroots image management strategy involves
proactive use of public diplomacy tools to generate more favorable media

55
We also generated two binary indicators that assume values of one if a respondent disapproves of the
leadership of China or the leadership of the U.S., respectively. We dropped all "Don't Know" and "Refused
to Answer" observations.
56
GDELT event records are recorded in a dyadic format, with two actors and an action performed by Actor
1 on Actor 2 (e.g., the provision of aid from one country to another country, a leader from one country
visiting another country). For the purposes of our analysis, we restrict Actor 1 to the LICs and MICs that are
covered by the 3.0 version of the GCDF dataset and Actor 2 to China and the U.S. We also restrict our
analysis to event records where the actor2type1code variable is set to GOV (in order to ensure that we are
measuring media sentiment about the Chinese Government and the U.S. Government, respectively).
57
Although the “one country, one vote” rule applies in UNGA, we report population-weighted estimates of
UNGA voting alignment in chapters 1 and 4 because we use this measure as a proxy for elite support of
China and the U.S. (rather than as a direct measure of an empirical phenomenon of interest). We also report
population-weighted estimates of grassroots support (via Gallup World Poll) and media support (via GDELT)
in chapters 1 and 4. This approach is based on the assumption that large countries are more important to
China (and the U.S.) than small countries, regardless of the soft power outcome that is being sought.
58
However, in relative terms, the U.S. gained ground on China between 2014 and 2021 (see Figure 1.17.
28
reporting about China (Brazys and Dukalskis 2019; Custer et al. 2018, 2019). But
Beijing’s advantage over Washington was hardly insurmountable; by 2020 and
2021, China had lost ground to the U.S. and was struggling to maintain a
razor-thin lead in media coverage favorability.

China’s outsized influence in the Global South is most clearly evident in the
United Nations General Assembly (UNGA), where voting patterns are often used
as a proxy for the extent to which governing elites in developing countries align
their foreign policy positions with those of the U.S. or China. Figure 1.16
demonstrates that countries in the Global South consistently vote with China
rather than the U.S. in UNGA. Although there are some natural foreign policy
affinities between China and countries in the Global South, Beijing has a
well-established track record of using its largesse to buy votes in international
organizations.59 In joint work with our longtime collaborators from Heidelberg
University, the University of Göttingen, the University of Hong Kong, and William
& Mary, we show in a new book called Banking on Beijing that when countries
vote with China in the UN General Assembly, they are richly rewarded. Our
statistical model results imply that if a low-income or middle-income
government chooses to increase the alignment of its UNGA voting with China
by just 10%, it can expect to see a 276% increase in aid and credit (ODA and
OOF commitments) from Beijing, on average (Dreher et al. 2022).60

59
To be sure, China is not the only major power that has used foreign aid and credit to influence the foreign
policy positions of developing countries (Alesina and Dollar 2000; Kuziemko and Werker 2006; Vreeland
and Dreher 2014; Rose 2018).
60
We thank Axel Dreher, Andreas Fuchs, Austin Strange, and Mike Tierney for generating and sharing
supplementary evidence derived from a statistical model in the fifth chapter of Banking on Beijing.
29
Figure 1.14: China versus the U.S.: Public approval rates

Figure 1.14

Notes: Average public approval ratings for China and the U.S. are weighted by country population. The
construction of this variable is described in greater detail in Box 1b.

30
Figure 1.15: China versus U.S.: Media sentiment

Figure 1.15

Notes: Average media sentiment ratings for China and the U.S. are weighted by country population. The
construction of this variable is described in greater detail in Box 1b.

31
Figure 1.16: China versus the U.S.: UN voting alignment

Figure 1.6

Notes: Average UNGA voting alignment scores for China and the U.S. are weighted by country population.
The construction of this variable is described in greater detail in Box 1b.

Figure 1.17 tallies the annual number of soft power “gains” and “losses” that
China experienced vis-à-vis the U.S. on a country-by-country basis between
2014 and 2021.61 On all three measures of soft power (public opinion, media
sentiment, and elite support), China has experienced more losses than gains
vis-à-vis the U.S. since 2014 (the first full year of BRI implementation). Public
opinion in the developing world has moved in a particularly unfavorable
direction for Beijing. During the early BRI period, 39% of the country-level public

61
Figure 1.17 present the percentages of LICs and MICs in which China experienced relative gains or losses
in popular support, media sentiment, and UNGA voting alignment vis-à-vis the U.S. The percentages are
reported over two time periods: early BRI (2014-2017) and late BRI (2018-2021). To measure the relative
gains or losses in popular support, we follow a three-step calculation for each country: (1) calculate the
difference between the public approval rating for China in a given year and the prior year; (2) calculate the
difference between public approval rating for the U.S. in a given year and the prior year; and (3) calculate
the “double difference” between (1) and (2) to determine if China experienced a greater gain or loss in
public support than the U.S. in the same country-year. For relative gains and losses in media sentiment and
UNGA voting alignment, the same three-step calculation was followed using the average media sentiment
score for each country-year from the GDELT 1.0 Event Database (related to government actors from
mainland China or the U.S.) and the average “idealpointdistance” estimate between each country and
China (or the U.S.) in a given year.
32
opinion changes that China experienced were relative losses rather than relative
gains (i.e., public opinion toward the U.S. improved at a faster rate than public
opinion toward China, or public opinion toward the U.S. declined at a slower
rate than public opinion toward China). However, during the late BRI period, this
figure shot up to 66%. By 2021, nearly 85% of the country-level public opinion
changes that China experienced were relative losses rather than relative gains
(see Figure A8 in the Appendix). Over time, rising levels of public antipathy
toward China and expanding popular support for the U.S. have widened
Washington’s soft power advantage over Beijing. The battle for hearts and minds
in the developing world was effectively a toss-up during the early years of the
BRI: Beijing and Washington achieved a similar number of public opinion gains
and losses on a country-by-country basis.62 However, during the late BRI period
(2018-2021), Beijing’s losses outnumbered its wins—by a significant margin.63

Figures 1.18 and 1.19 suggest that Beijing has suffered less acute public opinion
and media sentiment losses in BRI participant countries, which is consistent with
new research on the international image-enhancing effects of Chinese aid and
credit (Wellner et al. forthcoming, 2023; Brazys and Dukalskis 2019).64 However,
even in BRI participant countries, the trend lines have moved in a direction that
should provide cold comfort to Beijing. Figure 1.20 provides additional grounds
for concern, since it demonstrates that governing elites in BRI participant
countries are taking foreign policy positions that are increasingly out of
alignment with those of China. Souring media sentiment and declining levels of
public support may be making it more difficult for governing elites to maintain
close relations with Beijing.

62
During the first three years of the early BRI period (2014-2016), 55% of the public opinion changes that
China experienced at the country-year level vis-à-vis the U.S. were relative gains and 45% were relative
losses. Over the full early BRI period (2014-2017), 61% of the public opinion changes that China
experienced at the country-year level vis-à-vis the U.S. were relative gains and 39% were relative losses (see
Figure A8).
63
During the late BRI period (2018-2021), 39% of the public opinion changes that China experienced at the
country-year level vis-à-vis the U.S. were relative gains and 61% were relative losses (see Figure A8).
64
BRI participant countries include those countries that have signed MOUs with China to join its Belt and
Road Initiative. A country is assigned to the BRI participant cohort in the year it signed the MOU and every
year thereafter.
33
Figure 1.17: China’s soft power losses vis-à-vis the U.S

Figure 1.17

Notes: China’s relative gains and losses in popular support, media sentiment, and UNGA voting alignment
vis-à-vis the U.S are calculated on a country-by-country and year-by-year basis. The construction of these
measures are described in greater detail in Box 1b and footnote 61.

34
Figure 1.18: Public support for China: BRI countries vs. non-BRI countries

Figure 1.18

Notes: Average public approval ratings for China are weighted by country population. The
construction of this variable is described in greater detail in Box 1b.

35
Figure 1.19: Media sentiment toward China: BRI countries vs. non-BRI countries

Figure 1.19

Notes: Average media sentiment ratings for China are weighted by country population. The construction of
this variable is described in greater detail in Box 1b.

36
Figure 1.20: UN voting alignment with China: BRI countries vs. non-BRI countries

Figure 1.20

Notes: Average UNGA voting alignment scores for China are weighted by country population. The
construction of this variable is described in greater detail in Box 1b.

Section 6: The balancing act of portfolio risk management

Managing the risk profile of China’s overseas development program is a
multifaceted challenge. First, China has a high and rapidly rising level of
exposure to non-performing loans in LICs and MICs. Its exposure to distressed
debt is heavily concentrated among the most important Belt and Road
participants, which raises the question of whether its efforts to get repaid will
conflict with its foreign policy (and soft power) objectives (Hancock and Hill
2022).65 The crux of the matter, as described by a senior official from China’s
central bank, is that “debt reduction can shape China’s image as a responsible,
major global player, [but] it often leads to big [financial] losses and even causes a
moral hazard—debtors may refuse to make repayments if they know [that we
may write off their debts]” (Chengjun 2021). Second, a large share of China’s

65
Beijing’s exposure to distressed debt in LICs and MICs also raises the question of whether certain
borrowers are “too big to fail” and whether Chinese creditors will strictly adhere to the principle that every
loan must be repaid in full (an issue we address at greater length in Chapter 2).
37
infrastructure project portfolio has significant ESG risk exposure, and while there
are many good reasons to de-risk the portfolio by adopting stronger safeguards,
doing so could result in China losing a competitive edge in the global
infrastructure market. China is no longer “the only game in town” for countries
seeking external sources of infrastructure financing, and the other major players
in the market already offer high-quality infrastructure projects that benefit from
strong ESG safeguards. China has historically outcompeted G7 countries and
the multilateral development banks on two key dimensions: speed and
convenience. If these differentiators are no longer applicable, China may face a
different type of problem: insufficient demand for Chinese infrastructure
financing. Third, rising levels of public antipathy toward Beijing and a souring
media environment have left China increasingly dependent on the good graces
of political leaders in the Global South. One way to address this challenge would
be to dust off an old playbook and cater to the parochial interests of host
country politicians by plying them with lavish spending on pet projects—like
presidential palaces—and amenities in major urban centers (such as museums,
theaters, convention centers, and stadiums). An alternative approach would be
to double down on public diplomacy efforts—such as scholarships, sister city
initiatives, and content-sharing partnerships with local radio stations, television
channels, and newspapers—to generate more favorable media coverage and
influence public sentiment in BRI participant countries.

Beijing is clearly aware of the need to pivot and assume a “fire-fighting” role. It
is rapidly refocusing its time, money, and attention on distressed borrowers,
troubled projects, and sources of public backlash in the Global South. However,
a longer-term reinvention of the BRI is also underway. As Beijing learns from
past mistakes, it is recalibrating its lending and grant-giving practices and
making efforts to future-proof its flagship, global infrastructure initiative. The
ambition of BRI 2.0 remains poorly understood—and underestimated—by those
who make and shape policy in G7 countries. Washington, London, Paris, Berlin,
Tokyo, Rome, and Ottawa are, for the most part, still formulating policy on the
basis of evidence from the BRI 1.0 era.

The primary goal of this report is to explain how China is handling its new role as
an international crisis manager in the short-run while engaging in a longer-run

38
effort to future-proof the BRI. In the next three chapters, we attempt to answer
the following questions:

● How is China coping with the rising tide of debt distress? What measures
is it taking to reduce its exposure to non-performing loans?

● Is China stepping up its ESG risk mitigation efforts? If so, where, when,
and how? Are its infrastructure projects with and without strong ESG
safeguards faring differently during implementation?

● How does China manage reputational risk? What measures does it take to
preserve grassroots, media, and elite support in host countries? Are
Chinese development finance institutions learning from their past
mistakes and recalibrating their policies and practices in BRI “buyer’s
remorse” countries?

● What are China’s tolerance levels for repayment risk, project performance
risk, and reputational risk?

However, we also want to introduce researchers, policymakers, and journalists to

the “art of the possible” with the 3.0 version of AidData’s GCDF dataset. There
are many additional questions that can now be answered because of the
dataset’s uniquely comprehensive scope and unprecedented granularity.66
Several new and improved features of the dataset merit special attention:

1. Donor and lender coverage: The 3.0 dataset captures projects and
activities in LICs and MICs supported by 791 official sector donors and
lenders in China. It also identifies the participation of 1,225 co-financing
institutions—including Western commercial banks, multilateral
development banks, and OECD-DAC development finance institutions
that have chosen to collaborate or coordinate with Beijing—in Chinese
grant- and loan-financed projects and activities. A new feature of the 3.0
dataset is the inclusion of two “flag” variables that allow for easy

66
Table A1 presents a broad view of how the 3.0 version compares to the 2.0 version of the GCDF dataset.

39
 identification of projects/activities that involve (a) non-Chinese financiers
or (b) multilateral institutions.

2. Borrower and recipient coverage: The 3.0 dataset identifies 5,037

receiving (borrowing) institutions and categorizes each one by type
(government agency, state-owned bank, state-owned company, special
purpose vehicle/joint venture, intergovernmental organization, private
sector, etc.), country of origin (recipient country, China, or a third country),
and, when applicable, role (direct borrower or indirect borrower through
an on-lending arrangement). In recognition of the fact that special
purpose vehicles and joint ventures play an important role in China’s
overseas lending program and often blur the lines between public and
private debt (Malik et al. 2021; Malik and Parks 2021), we have also
added two new variables to the latest version of the dataset: a measure of
the extent of host government ownership and a measure of the extent of
Chinese government ownership whenever the borrowing institution
(receiving agency) is a special purpose vehicle or joint venture.67 The 3.0
dataset also identifies 422 institutions (“accountable agencies”) that have
supported Chinese loan-financed projects and activities by providing
repayment guarantees, insurance policies, and collateral which can be
seized in the event of default.68

3. Financial instrument coverage: The 3.0 dataset allows users to easily

differentiate between the 10,291 grant-financed projects/activities and
4,776 loan-financed projects/activities. However, given that Beijing relies
on an increasingly diverse set of debt instruments to finance its overseas
development program in LICs and MICs, AidData has introduced a new
loan categorization scheme in the latest version of the dataset that allows
users to isolate specific types of loan instruments, including but not
limited to bilateral loans, syndicated/club loans, interest-free loans,

67
These new variables are called “JV/SPV Host Government Ownership” and “JV/SPV Chinese
Government Ownership.” The 3.0 dataset captures 851 loan commitments worth $315 billion (in constant
2021 USD) to borrowing institutions that are categorized as special purpose vehicles or joint ventures.
68
Consistent with the 2.0 version of the GCDF dataset, each accountable agency in the 3.0 dataset is still
categorized by type and country of origin. However, unlike the 2.0 version of the GCDF dataset, each
accountable agency in the 3.0 dataset is also categorized by role (guarantor, insurance provider, or
collateral issuer).
40
 government concessional loans, preferential buyer’s credits, public
investment loans, balance of payments (BOP) loans, M&A loans, working
capital loans, inter-bank loans, refinancing loans, deferred payment
agreements, and pre-export financing (PxF) agreements.

4. Borrowing terms and conditions: There is no other publicly available

dataset of China’s overseas loan commitments with global coverage from
2000-2021 that identifies borrowing terms and conditions at the
transaction level. The 3.0 dataset identifies 2,699 interest rates, 3,315
maturity lengths, 1,854 grace periods, 498 commitment fees, 480
management fees, and 2,537 grant elements across 4,776 loans in Africa,
Asia, Oceania, Eastern and Central Europe, the Middle East, and Latin
America and the Caribbean. It also identifies 668 loans backed by
third-party repayment guarantees, 529 loans supported by credit
insurance policies, and 1,015 loans underpinned by one or more sources
of collateral. Three additional dataset features are worth noting. First, in
light of recent changes to the OECD’s grant element method of
measurement (see Section A-2) and the growing importance of the World
Bank/IMF-based method of measurement to determine if and when
governments have complied with the non-concessional borrowing limits
specified in their World Bank and IMF programs, we now provide three
different grant element measures: one based on the OECD’s cash-flow
methodology, one based on the the OECD’s grant-equivalent
methodology, and another based on the latest (post-2013) World
Bank-IMF methodology.69 Second, in order to facilitate more accurate
calculation of the “all-in” price of Chinese debt, we have introduced
variables that measure (a) the default (penalty) interest rate that applies to
a loan in the event of default (i.e., non-payment of principal, interest, or
fees on their scheduled payment dates), and (b) the cost of credit
insurance.70 Third, to support future research on debt service to China, we
69
The OECD’s cash-flow methodology assumes a fixed, 10 percent discount rate. Its grant-equivalent
methodology uses fixed discount rates that depend on the income level of the borrowing country (9% for
LDCs and other LICs and 6% for UMICs). The World Bank-IMF methodology assumes a fixed, 5 percent
discount rate. For more on the IMF’s Debt Limits Policy (DLP), see
https://www.imf.org/en/Topics/sovereign-debt/debt-limits-policy.
70
For example, Argentina’s Ministry of Economy and Public Finance signed a $4,714,350,000 syndicated
term facility (loan) agreement on August 1, 2014 with Bank of China, China Development Bank, and ICBC
for the 1,740 MW Néstor Kirchner and Jorge Cepernic Hydroelectric Power Plant Construction Project. The
41
 have included two new measures in the 3.0 dataset: the calendar day on
which the borrower was originally scheduled to make its first loan
repayment and the calendar day on which the borrower was originally
scheduled to make its last loan repayment. Whenever possible, we have
also documented disbursements, repayments, and amounts outstanding
in the dataset’s “description” field.

5. Spatial and temporal granularity: The 3.0 dataset provides an

unprecedented level of detail on project commencement
(implementation start) dates and project completion (implementation
end) dates. It identifies precise, calendar day-level commencement dates
for 11,286 projects (backed by financial commitments worth $767 billion)
and calendar day-level completion dates for 11,542 projects (backed by
financial commitments worth $606 billion). By way of comparison, the 2.0
version of the GCDF dataset identified calendar day-level
commencement dates for 5,539 projects (backed by financial
commitments worth $504 billion in constant 2021 USD) and calendar
day-level completion dates for 6,061 projects (backed by financial
commitments worth $383 million in constant 2021 USD). The 3.0 dataset
also provides data on the originally scheduled project commencement
dates and completion dates, which has paved the way for the
introduction of two new measures (“Deviation from Planned
Implementation Start Date” and “Deviation from Planned Completion
Date”) of to what degree projects ran (or are running) ahead of schedule
or behind schedule.71 Another important value addition to the 3.0 dataset
is the level of geographical detail regarding where projects take place. As
we describe in greater detail in Goodman et al. (2023), for 9,497 projects

loan agreement, which was later amended on January 27, 2015 and again in mid-2022, specifies a default
(penalty) interest rate of 1.5%. As a credit enhancement, the borrower purchased a buyer’s credit insurance
policy from Sinosure worth approximately 7.1% of the face value of the loan ($502,976,000) (see Project
ID#59723, 59724, 37002 in the 3.0 version of the GCDF dataset).
71
The “Deviation from Planned Implementation Start Date” variable captures the difference between the
“Planned Implementation Start Date” and the “Actual Implementation Start Date” when values are
recorded for both variables. It captures the difference as the number of calendar days, whereby positive
values represent cases where the project started implementation ahead of schedule and negative values
represent cases where the project started implementation behind schedule. The “Deviation from Planned
Completion Date” variable captures the difference between the “Planned Completion Date” and the
“Actual Completion Date” when values are recorded for both variables. It captures the difference as the
number of calendar days, whereby positive values represent cases where the project was completed ahead
of schedule and negative values represent cases where the project was completed behind schedule.
42
 that have physical footprints or involve specific locations, the 3.0 dataset
extracts point, polygon, and line vector data via OpenStreetMap URLs
and provides a corresponding set of GeoJSON files and geographic
precision codes.72 72% (6,919) of these projects include “precise” or
“approximate” geocodes; the remaining 28% (2,578 projects) are
measured at an administrative unit level.73 Measuring the spatio-temporal
rollout of project implementation with a high level of precision is
important because it creates new opportunities to identify
cause-and-effect relationships in rigorous ways.74

6. Sectoral coverage: The 3.0 dataset systematically tracks provision of

official financial flows from China to LICs and MICs across all sectors.
Every project/activity is assigned a 3-digit sector code based on OECD
definitions and measurement criteria. This “methodological crosswalk” is
important because it allows for cross-financier comparisons—at global,
regional, national, and subnational scales—since most official sources of
international development finance (including OECD-DAC members and
multilateral institutions) use the same criteria. It also facilitates analysis of
sectoral patterns and trends over space and time.

7. Qualitative detail: The 3.0 dataset provides detailed project narratives

that “tell the story” of each project in the “description” field. The average
length of each project narrative increased from 144 words in 2.0 dataset
to 169 words in the 3.0 dataset. Whereas the project narratives in the 2.0
dataset consisted of 1.93 million words (roughly the same number of
words one would find in 19 full-length books), the project narratives in the
3.0 dataset consist of 3.48 million words (roughly the same number of
words one would find in 34 full-length books).75 As we will demonstrate in
72
Users who wish to conduct analysis at higher levels of spatial aggregation can find the ADM1s (provinces)
and ADM2s (districts) that correspond to these project locations in the 3.0 version of the GCDF dataset.
73
A project with “precise” geocodes is one for which we have highly precise boundaries of the project’s
geofeature(s). A project with “approximate” geocodes is one identified within a 5-km radius of the precise
boundaries of the project’s geofeature(s). For more details, see Goodman et al. (2023).
74
To better understand how highly precise data on the spatio-temporal rollout of Chinese grant- and
loan-financed projects make it possible to estimate the causal effects of such projects on intended and
unintended outcomes, see Dreher et al. (2019, 2022), Marty et al. (2019), Blair et al. (2022), Baehr et al.
(forthcoming), Isaksson and Kotsadam (2018a, 2018b), Isaksson (2020), Martorano et al. (2020), Iacoella et
al. (2021), Malik et al. (2021), Bluhm et al. (2020), Anaxagorou et al. (2020), Wellner et al. (forthcoming), and
Asmus et al. (forthcoming).
75
A typical, full-length book includes 100,000 words.
43
 Chapter 3, these narratives are useful in that they document the risks and
challenges that arose during project design and implementation (e.g.,
bankruptcies, scandals, protests, labor strikes, and criminal investigation)
and how funding, receiving, implementing, and accountable institutions
responded to these risks and challenges.

8. Scale, diversity, quality, and transparency of sourcing: The 3.0 dataset

was assembled with 147,703 sources (including 99,393 unique sources in
more than a dozen languages, of which 51,597 are official sources). By
way of comparison, the 2.0 dataset was assembled with 91,356 sources
(including 63,464 unique sources in more than a dozen languages, of
which 34,075 were official sources). Whereas the average record in the 1.0
dataset was based upon 3.6 sources and the average record in the 2.0
dataset was based on 6.8 sources, the average record in the 3.0 dataset is
based upon 7.0 sources. 87% of the records in the 3.0 dataset are
underpinned by at least one official source. To expose our coding and
categorization determinations to public scrutiny and promote replicable
research findings, we disclose all of the sources that were used to
construct the dataset at the individual record level, including hundreds of
unredacted grant, loan, debt forgiveness, and debt rescheduling
agreements (that AidData has never previously published).

44
Chapter 2: The Road to Repayment for the
World’s Largest Official Debt Collector

Section 1: Debunking the myth that Beijing’s overseas

lending program has collapsed
The conventional wisdom is that Beijing has responded to the deteriorating
performance of its overseas lending program via retreat. Eric Olander,
co-founder of the China-Global South Project (CGSP), recently summarized the
state of expert opinion, noting that “[t]here was a time when Chinese lending to
developing countries rivaled the World Bank” but “[t]hose days are now long
gone as Chinese overseas development lending has been on a steady
downward trajectory” (Olander 2023). Scott Kennedy of the Center for Strategic
and International Studies (CSIS) told Foreign Policy magazine earlier this year
that the BRI was a “shadow of its former self” (Lu 2023).76 Elliot Wilson of
Euromoney magazine claims that “Chinese overseas lending to the developing
world has collapsed” (Wilson 2022).

However, the conventional wisdom is mostly wrong. The empirical evidence that
we present in this chapter paints a different picture—one in which China is
behaving more like an international crisis manager than a country admiring its
problems and sticking its head in the sand. Beijing is rebalancing its overseas
lending portfolio by adopting a wide-ranging set of de-risking measures. It is
ramping down the provision of long-term, dollar-denominated bilateral loans to
sovereign borrowers for public investment projects, while at the same time
ramping up the provision of loans that are RMB-denominated, short- or
medium-term in nature, unrelated to public investment projects, and/or
involving multiple Chinese and/or non-Chinese banks. It is ratcheting down its
use of the policy banks (China Development Bank and China Eximbank), while at
the same time ratcheting up its use of the central bank (People’s Bank of China),
state-owned commercial banks (such as Industrial and Commercial Bank of
China, Bank of China, and China Construction Bank), and syndicated loan
76
The same Foreign Policy magazine article notes that “experts say China's lending for BRI projects has
plummeted” (Lu 2023).
45
arrangements with non-Chinese banks (such as Standard Chartered, BNP
Paribas, the International Finance Corporation, and the European Bank of
Reconstruction and Development). It is also putting in place stronger safeguards
to protect itself from borrowers that present high levels of repayment risk. So,
another way of reading the evidentiary record is that Beijing is behaving like a
yield-maximizing investment portfolio manager (see Box 2b).

The de-risking measures that Beijing is implementing are poorly understood

among those who make and shape policy in Western capitals, which means that
Washington, London, and Brussels increasingly run the risk of competing with a
version of the BRI that no longer exists—BRI 1.0 rather than BRI 2.0. Our aim in
this chapter is to set the record straight. We do so by first debunking the
popular myth that China’s overseas lending program has experienced a total
collapse. The rest of the chapter is focused on demystifying the purpose, nature,
and scope of Beijing’s ongoing BRI reboot.

The “Beijing in retreat” storyline has gained traction because it is consistent with
the topline Chinese lending commitment trends that are recorded in several
publicly available databases, such as the China’s Overseas Development Finance
Database, the Chinese Loans to Latin America and the Caribbean Database, the
Chinese Loans to Africa Database, the China’s Global Energy Finance Database,
the China Overseas Finance Inventory Database, and the World Bank’s
International Debt Statistics (IDS). Consider the following claims that have been
made based upon these data sources:

● In December 2020, Boston University’s Global Development Policy Center

released the China’s Overseas Development Finance (CODF) Database,
which provides data on China Eximbank and CDB lending commitments
to governments, inter-governmental bodies, majority state-owned entities
and minority state-owned entities with sovereign guarantees. The
research team responsible for the database claimed at the time that it
provided evidence of “China’s overseas development finance
commitments” declining from $75 billion in 2016 to $3.9 billion in 2020
(Gallagher and Ray 2020). Then, in January 2023, the same research team
used an updated version of the CODF database to argue that China’s

46
 overseas development finance commitments remained at an
exceptionally low level ($3.6 billion) in 2021 (Ray 2023).

● The World Bank’s International Debt Statistics (IDS) show a sharp decline
in official sector loan commitments from China to public sector
institutions in low-income and middle-income countries—from $31.5
billion in 2016 to $7 billion in 2021.

● According to the Chinese Loans to Latin America and the Caribbean

Database jointly produced by Inter-American Dialogue and Boston
University’s Global Development Policy Center, “the LAC region saw a
precipitous decline in loans from CDB and Eximbank between 2015 and
2020, when lending ceased altogether” (Myers and Ray 2023: 1). They
also concluded that Chinese development finance commitments to the
LAC region remained stubbornly low in 2021, identifying only one ($204
million) CDB loan commitment in that year.77

● According to the Chinese Loans to Africa (CLA) database, which was

initially developed by the China-Africa Research Initiative at the Johns
Hopkins School of Advanced International Studies (SAIS-CARI) and is now
maintained by Boston University’s Global Development Policy Center,
Chinese lending commitments to African governments and state-owned
entities plunged from a high of $28 billion in 2016 to a low of $1.9 billion
in 2020. In a summary of their findings, the researchers responsible for the
CLA database conclude that “[c]ompared to previous years, the number
of loans and the total value of loan commitments decreased dramatically
in 2020” (Hwang et al. (2022: 2).78

The latest (3.0) version of AidData’s Global Chinese Development Finance

(GCDF) dataset paints a different picture of China’s overseas development

77
AidData has categorized this $204 million (EUR 175 million) CDB loan as an informal pledge of
financial support rather than a formal loan commitment, based on evidence that it gathered via
direct correspondence with the Debt Management Division of Trinidad and Tobago's Ministry of
Finance. See ID#95549 in the 3.0 version of AidData’s GCDF dataset for more details.
78
An update of the CLA database was published at the time that this report was going to press.
It shows a continued decline in Chinese lending commitment—to $1.2 billion in 2021 and
$994.4 million in 2022 (Moses et al. 2023).
47
finance portfolio (see Figure 1.2 in Chapter 1). It demonstrates that Beijing is still
the single largest source of international development finance in the world, with
annual ODA and OOF commitments to LIC and MICs now hovering around $80
billion. Although it provides evidence of China’s annual international
development finance commitments falling between 2016 and 2020, it also
shows an increase in 2021 (returning to a level that is roughly comparable to the
first full year of BRI implementation).

Figure 2.1: Official sector lending commitments from China to LICs and MICs by source

Figure 2.1

Notes: Figure 2.1 compares the total size of official sector lending commitments from China to LICs and
MICs across three datasets: the 3.0 version of AidData’s GCDF dataset, Boston University’s China’s
Overseas Development Finance Database (CODF), and the World Bank’s International Debt Statistics
(representing commitments to official creditors in China).

Figure 2.1 provides a comparison of official lending commitments from China to

borrowers in low-income and middle-income countries, as measured by three
different sources: the 3.0 version of AidData’s GCDF dataset, the World Bank’s
IDS, and the CODF database produced by Boston University's Global
Development Policy Center.79 AidData captures lending commitments worth
79
To ensure comparability, we convert the IDS data series and the CODF data series into constant 2021 USD.

48
$1.28 trillion between 2000 and 2021, while CODF and IDS capture lending
commitments worth $605 billion and $378 billion, respectively.80 Whereas
AidData records a 45% decline in lending commitments between 2016 and
2021, CODF and IDS record substantially larger declines—96% and 78%,
respectively. CODF and IDS record $3.7 billion and $7.1 billion, respectively, in
new official lending commitments from China in 2021. AidData captures $75.5
billion in new official lending commitments from China in the same year.

There are several reasons why the estimates from AidData’s GCDF dataset, BU’s
CODF dataset, and the World Bank’s IDS are widely divergent. First, although all
three sources provide data on public and publicly guaranteed debt (PPG) debt
from Chinese state-owned creditors,81 they provide different levels of
geographical coverage.82 AidData’s GCDF dataset covers 126 countries, while
BU’s CODF dataset covers 96 countries and the World Bank’s IDS covers 89
countries.83 Second, there are temporal coverage differences: whereas AidData’s
GCDF dataset and the World Bank’s IDS provide data for 22 commitment years
(2000-2021), BU’s CODF dataset provides data for 14 commitment years
(2008-2021). Third, in the subset of LICs and MICs for which CODF, IDS, and
GCDF data are available, there are differences in how much PPG debt from
Chinese state-owned creditors is captured. Table A15 demonstrates that, in the
subset of LICs and MICs for which CODF or IDS data are available, AidData’s
GCDF dataset captures $947 billion of lending commitments84 from official

80
The $1.28 trillion figure excludes short-term “rollover” facilities to refinance maturing debts. When
short-term “rollover” facilities are included in the tally, AidData captures lending commitments from China
worth $1.44 trillion between 2000 and 2021.
81
PPG debt consists of (a) long-term external obligations of public debtors, including the national
government, a political subdivision (or an agency of either), and autonomous public bodies; and (b)
long-term external obligations of private debtors that are guaranteed for repayment by a public entity
(World Bank 2000).
82
The 3.0 version of AidData’s GCDF dataset provides comprehensive coverage of Chinese ODA and OOF
commitments across 165 LICs and MICs between 2000 and 2021. However, only 134 of these countries
secured loan commitments from official sector creditors in China over the same time period. An even
smaller subset (126 countries) secured loan commitments between 2000 and 2021 from official sector
creditors in China that qualify as PPG debt.
83
The IDS data capture official sector lending commitments from China to 89 low-income and
middle-income countries (excluding the People’s Republic of China) from 2000-2021. The IDS data do not
allow users to differentiate between the People’s Republic of China (“China”) and the Republic of China
(“Taiwan”). As such, we exclude all loan commitments that 9 additional countries reported during the years
when they maintained diplomatic relations with Taiwan.
84
The $947 billion figure excludes short-term “rollover” facilities. When short-term “rollover” facilities are
included in the tally, AidData captures lending commitments to LICs and MICs that qualify as PPG debt
worth $1.09 trillion between 2000 and 2021.
49
sector institutions in China to LICs and MICs that qualify as PPG debt.85 In total,
BU’s CODF dataset captures $605 billion and the World Bank’s IDS captures
$378 billion of lending from official sector institutions in China that qualifies as
PPG debt.86 These topline differences reflect widely varying levels of lending
institution, borrowing institution, and debt instrument coverage (which we
discuss at greater length below). Table A15 calls attention to the fourth and final
difference: neither BU’s CODF dataset nor the World Bank’s IDS provide any
coverage of lending commitments from official sector institutions in China to
LICs and MICs that qualify as non-PPG debt. AidData’s GCDF dataset, by
contrast, captures $333 billion that does not (clearly) qualify as PPG debt: $67
billion of potential public sector debt (i.e., debt contracted by a minority
state-owned institution in the borrowing country without a public sector
repayment guarantee), $216 billion of private sector debt, and $50 billion that is
not allocable due to a lack of sufficient information for categorization
purposes.87

One additional source of information is worth considering: newly published data

from the Bank of International Settlements (BIS) on total outstanding credit from
Chinese banks to LICs and MICs from 2015 to 2021 (see Box 2a). These data are
particularly valuable because they are based on direct reporting from Chinese
banks about their cross-border claims. Although the BIS does not currently allow
for its Chinese lending data to be disaggregated by borrower countries, one can
derive lower-bound and upper-bound estimates of total outstanding credit from
Chinese banks to LICs and MICs based on new research by Cerutti et al. (2023).
As we describe at greater length in Box 2a, lower-bound estimates based on the
BIS data indicate that total outstanding credit from Chinese banks to LICs and
85
In the 96 countries that are covered by the CODF dataset, BU captures $605 billion and AidData captures
$824 million of lending commitments (excluding short-term “rollover” facilities) between 2008 and 2021
from official sector institutions in China that qualifies as PPG debt. When short-term “rollover” facilities are
included in its tally, AidData captures $987 million in the same set of countries.
86
For a country-by-country comparison of AidData and IDS measures of PPG debt exposure to China (that
are normalized by host country GDP), see Table A16.
87
The “Level of Public Liability” variable in the 3.0 version of AidData’s GCDF dataset captures the extent
to which the host government may eventually be liable for debt repayment. Each loan record is assigned to
one of six categories: (1) Central government debt, (2) Central government-guaranteed debt, (3) Other
public sector debt, (4) Potential public sector debt, (5) Private debt, or (6) Unallocable. The sum of the first
three categories is equivalent to PPG debt. The fourth category captures loans to special purpose vehicles
(SPV) or joint ventures (JV) that are minority-owned by one or more public sector institutions in the host
country and that do not benefit from a central government repayment guarantee or a repayment guarantee
from a state-owned entity other than the central government in the host country. For more information
about the “Level of Public Liability” variable, see Section A-5 in the Appendix.
50
MICs increased in nominal terms from $644 billion in 2015 to $1.16 trillion in
2021. Upper-bound estimates based on the BIS data indicate that total
outstanding credit from Chinese banks to LICs and MICs increased in nominal
terms from $878 billion in 2015 to $1.58 trillion in 2021. AidData’s estimates of
the cumulative size of China’s overseas lending program in LICs and MICs are
consistent with the more conservative (lower-bound) BIS-based estimates.88
According to the 3.0 version of the GCDF dataset, China’s cumulative overseas
lending commitments increased from $620 billion in 2015 to $1.03 trillion in
2021 (in nominal USD).89 Neither AidData’s GCDF dataset nor the BIS-based
estimates support the argument that China’s overseas lending program has
experienced a total collapse.

Box 2a: What is the true scale of China’s overseas lending program?
The overall size of China’s overseas lending program is a subject of ongoing debate and
controversy. Horn et al. (2019) conducted pioneering work, producing a $393 billion estimate of
total outstanding cross-border credit from Chinese banks to LICs and MIC borrowers in 2017. At
the time, their estimate was criticized by the IMF as being exaggerated (IMF 2020). Bräutigam
and Acker (2020) also published a critique, concluding that they “agree[d] with the IMF.” Horn et
al. (2020a) then issued a response to their critics, arguing that “our numbers are substantially
below comparison figures [from the PBOC and other official sources] and likely a lower bound
estimate of the true extent of Chinese overseas lending.”90 They explained that “[d]espite our
best efforts to merge data from multiple sources, we still miss substantial amounts of Chinese
overseas lending.”

There was also an important development around the time that the Horn et al. (2019) study was
published: China began reporting to the Bank of International Settlements (BIS) on the
cross-border claims of its banks.91 At the time, the Locational Banking Statistics (LBS) published
by the BIS indicated that total outstanding credit from Chinese banks to overseas borrowers was

88
A key caveat, as we explain in Section A-4, is that the BIS data are represented as amounts outstanding,
which is effectively equivalent to cumulative disbursements minus cumulative repayments (i.e., credit stocks
rather than credit flows). Consequently, cumulative lending commitments usually exceed amounts
outstanding.
89
These nominal USD figures exclude short-term “rollover” facilities to refinance maturing debts. If
short-term “rollover” facilities are included in the tally, China’s cumulative overseas lending commitments
increased from $630 billion in 2015 to $1.17 trillion in 2021 (in nominal USD). In general, we prefer to report
China’s overseas lending commitments in constant 2021 USD. However, since the BIS data are recorded in
nominal USD, an apples-to-apples comparison with the 3.0 version of AidData’s GCDF dataset requires use
of nominal USD. In Table 2.1, we report China’s cumulative overseas lending commitments to LICs and
MICs from 2015-2021 in both nominal and constant 2021 USD.
90
At the time, China’s State Administration of Foreign Exchange (SAFE) had published (2017) data on
“China’s International Investment Position” that identified $637 billion of total outstanding cross-border
credit. By 2021, this figure rose to $988 billion. See https://www.safe.gov.cn/en/2018/0928/1459.html.
91
Although China has joined the list of countries reporting to the BIS, its data (unlike the data of many of
other BIS reporting countries) are not made publicly available on a bilateral basis.
51
more than double ($919 million) the size of the Horn et al. (2019) estimate.92 Since then,
additional BIS reporting by China and new research by Cerutti et al. (2023) has made it possible
to generate updated estimates of total outstanding credit from Chinese banks to borrowers in
LICs, MICs, high-income countries (HICs), and other overseas jurisdictions.93 In Section A-4, we
provide a step-by-step description of how these BIS-based measures of total outstanding credit
from Chinese banks to overseas borrowers are derived. We also discuss a number of important
caveats and considerations regarding the BIS cross-border lending data.

The estimates, which we report in Table 2.1, demonstrate that China’s total outstanding
credit—measured in nominal terms—to borrowers in LICs, MICs, HICs, and other overseas
jurisdictions soared from $1.45 trillion in 2015 to $2.63 trillion in 2021. Conservative,
lower-bound estimates based on BIS reporting also indicate that total outstanding credit from
Chinese banks to LIC and MIC borrowers effectively doubled in nominal terms between 2015
and 2021—from $644 billion to $1.16 trillion. This measure of the stock of LIC and MIC debt to
Chinese banks is remarkably similar to our own measure of the stock of LIC and MIC debt to
Chinese banks.94 According to the 3.0 version of AidData’s GCDF dataset, China’s cumulative
overseas lending commitments to LIC and MICs increased in nominal terms from $620 billion in
2015 to $1.03 trillion in 2021.95

Table 2.1 provides several additional insights. One is that China’s total outstanding credit to
LICs, MICs, and HICs soared from $1.45 trillion in 2015 to $2.63 trillion in 2021. Another is that
total outstanding credit from Chinese banks to HIC borrowers effectively doubled over the same
six-year period. Between 2015 and 2021, this figure rose from approximately $330 billion to
$600 billion. A third is that total outstanding credit from Chinese banks to “other” borrowers
increased from $314.3 billion in 2015 to $568.3 billion in 2021.

While the LBS data from the BIS are extremely valuable for cross-validation purposes (since they
provide credible estimates of total outstanding credit from Chinese banks to LIC, MICs, and
HICs), they do not make it possible to track disbursements, repayments, and amounts
outstanding on a loan-by-loan basis. AidData recently launched a new data collection initiative
and research program to address this major evidentiary gap. For the time being, in the 3.0
version of the GCDF dataset, we have documented disbursements, repayments, and amounts
outstanding in the “description” field for a subset of countries. However, in the future, we intend
to publish loan-level data on disbursements, repayments, and amounts outstanding for a more
complete set of countries. We also intend to make these data available in a more user-friendly
format.

92
On this point, see Zhou and Cerutti 2018.
93
BIS classifies most HICs as “Advanced Economies” (AEs) and most LICs and MICs as Emerging Market
and Developing Economies (EMDEs). For the sake of clarity and consistency, we use the LIC, MIC, and HIC
acronyms. For more on the BIS country classification system, see Cerutti et al. (2023).
94
This BIS-based measure of China’s lending portfolio in LICs and MICs is based on amounts outstanding.
AidData’s measure of China’s lending portfolio in LICs and MICs is based on cumulative lending
commitments, which usually exceed amounts outstanding. See Section A-4.
95
According to BU’s CODF dataset, China’s cumulative overseas lending commitments to LIC and MICs
increased in nominal terms from $308 billion in 2015 to $498 billion in 2021. According to the World Bank’s
IDS, China’s cumulative overseas lending commitments to LIC and MICs increased in nominal terms from
$188 billion in 2015 to $293 billion in 2021.
52
 Table 2.1: Comparison of AidData- and BIS-based estimates of China’s international lending portfolio

Table 2.1
Comparison of AidData- and BIS-based estimates of China’s
international lending portfolio
AidData Total BIS Total to BIS Total to BIS Total
BIS Total to BIS Total to
to LICs and BIS Total to BIS Total to "Other" OFC to LICs,
LICs and LICs and
MICs HICs (Based HICs (Based Borrowers Borrowers MICs,
Year MICs (Based MICs (Based
Nominal USD on 22.6% on 22.8% (Based on (Based on HICs, and
on 60.4% on 44.31%
(Constant Assumption) Assumption) 21.6% 30.33% “Other”
Assumption) Assumption)
2021 USD) Assumption) Assumption) Borrowers

620.7
2015 878.7 644.7 328.8 331.7 314.3 441.3 1,454.9
(823.4)

721.4
2016 1,046.1 767.4 391.4 394.9 374.1 525.3 1,731.9
(942.2)

805.7
2017 1,197.7 878.7 448.2 452.1 428.3 601.4 1,983.0
(1,039.3)

889.4
2018 1,348.3 989.1 504.5 509.0 482.2 677.1 2,232.3
(1,130.6)

951.0
2019 1,367.0 1,002.8 511.5 516.0 488.8 686.4 2,263.2
(1,199.8)

986.8
2020 1,492.7 1,095.1 558.5 563.5 533.8 749.6 2,471.4
(1,239.8)

1,027.1
2021 1,589.2 1,165.9 594.6 599.9 568.3 798.0 2,631.1
(1,280.1)
Notes: The BIS data are reported in current (nominal) USD. For the sake of comparability, the amounts
recorded in the “AidData Total to LICs and MICs” column are also reported in current (nominal) USD,
though the constant 2021 USD amounts are reported in parenthesis. The totals from AidData exclude
short-term “rollover” facilities (see Box 2c and Section A-3 in the Appendix).

There are three main reasons why AidData’s GCDF dataset challenges the
conventional wisdom about the total collapse of China’s overseas lending
program.

1. Lending institution coverage: Unlike other publicly available datasets that

measure Chinese development finance, the 3.0 version of AidData’s
GCDF dataset does not exclusively track the overseas lending activities of
China’s government agencies (MOFCOM and CIDCA) and its state-owned

53
 policy banks (China Eximbank and CDB).96 Based on OECD-DAC
reporting directives, it tracks the overseas lending activities of all
government and state-owned creditors in China, including state-owned
commercial banks (such as Bank of China, ICBC, China Construction Bank,
Bank of Communications, China CITIC Bank, Bank of Shanghai, Postal
Savings Bank of China, China Merchants Bank, and Harbin Bank),
state-owned companies (such as CNPC, CMEC, Poly Technologies,
NORINCO, and AVIC), state-owned funds (such as the Silk Road Fund
and China Investment Corporation), state-owned policy banks (CDB and
China Eximbank), and government agencies (such as MOFCOM, PBOC,
and SAFE). In total, 180 Chinese lending institutions are included in the
3.0 version of AidData’s GCDF dataset. By way of comparison, two
Chinese lending institutions are included in Boston University’s CODF
database (China Eximbank and CDB) and only a handful of Chinese
lending institutions are included in the IDS.97 The breadth of AidData’s
lending institution coverage is particularly consequential because, in
recent years, the LIC and MIC lending operations of China’s central bank
(PBOC) and state-owned commercial banks have expanded while those of
CDB and China Eximbank have contracted (see Figure 2.7).98

96
The China’s Overseas Development Finance Database, the Chinese Loans to Latin America and the
Caribbean Database, and China’s Global Energy Finance Database track the overseas lending activities of
two state-owned policy banks (CDB and China Eximbank). They do not track the overseas lending activities
of China’s state-owned commercial banks. Nor do the World Bank IDS data capture loans from Chinese
state-owned commercial banks. As explained by Horn et al. (2021: 15), “the World Bank's definition [of
official sector lending] does not cover lending by commercial banks such as the Industrial and Commercial
Bank of China (ICBC) or the Bank of China (BoC), despite the fact that they are state-owned. These banks
are official creditors according to our (OECD) definition (they are owned and controlled by the Chinese
government), but they are not bilateral creditors according to the World Bank’s definition, because they are
not a ‘public enterprise’ in a narrow sense, in contrast to the policy banks such as China Ex-Im Bank or
CDB.”
97
IDS includes loan commitments from government agencies (such as MOFCOM and CIDCA) and
state-owned policy banks (such as China Eximbank and CDB), but excludes loan commitments from
state-owned companies, state-owned funds, and state-owned commercial banks (Horn et al. 2021: 15). For
the most part, IDS also appears to exclude loan commitments from the PBOC and SAFE, which is a PBOC
subsidiary (see Box 2c). This is true despite the fact that IDS seeks to capture all loans from “the general
government, central government; state and local government; [and] central bank and public enterprise”
(World Bank 2020b: 4).
98
On average, during the pre-BRI period (2000-2013), Beijing channeled 15% of its annual lending
commitments to low- and middle-income countries through its state-owned commercial banks. This figure
increased to 18% during the early BRI (2014-2017) period and 22% during the late BRI period (2018-2021).
On average, Beijing channeled 70% of its annual lending commitments to low- and middle-income
countries through its policy banks during the pre-BRI period. This figure dropped to 60% during the early
BRI period and 30% during the late BRI period. On average, during the pre-BRI period, Beijing channeled
only 3% of its annual lending commitments to low- and middle-income countries through PBOC/SAFE. This
figure increased to 14% during the early BRI period and 43% during the late BRI period. See Figure 2.7 for
more details.
54
 2. Borrowing institution coverage: OECD-DAC reporting guidelines specify
that ODA and OOF are designed to capture official sector financial flows,
where the “official sector” refers to the sources rather than the
destinations of such flows (Horn et al. 2021: 23). AidData adheres to this
reporting standard to ensure that its measures of Chinese development
finance are comparable to OECD-DAC sources of international
development finance. As such, the 3.0 version of AidData’s GCDF dataset
captures lending from all Chinese government and state-owned creditors
to all public sector and private sector borrowers in low-income and
middle-income countries, regardless of whether they secured sovereign
repayment guarantees. By contrast, other publicly available
datasets—such as Boston University’s CODF dataset and the World Bank’s
IDS—exclusively track Chinese loans to government institutions, majority
state-owned entities, and borrowing institutions that benefit from
sovereign repayment guarantees (i.e., public and publicly guaranteed
debt owed to China). This coverage difference is consequential because a
significant percentage of China’s overseas lending portfolio is channeled
to special purpose vehicles, joint ventures, private sector institutions, and
minority state-owned entities (see Figures 2.18 and A10). While these
loans typically do not appear on government balance sheets in
low-income and middle-income countries, many of them benefit from
implicit host government liability protection, which has blurred the
distinction between public debt and private debt.99 In total, the 3.0
version of the GCDF dataset captures $525 billion of lending
commitments to 661 borrowing institutions that qualify as central
government or central government-guaranteed debt, $421 billion of
lending commitments to 455 borrowing institutions that qualify as
another type of public sector debt, $67 billion of lending commitments to
85 borrowing institutions that qualify as potential public sector debt, $216
billion of lending commitments to 724 borrowing institutions that qualify
as private sector debt, and $50 billion that is not allocable due to a lack

99
As Horn et al. (2021: 4) explain, "[w]hile the distinction between private and public sector recipients is
clear in principle, it tends to be blurry in practice, in particular in developing countries and during financial
crises. Private debt often turns into public debt once a crisis hits and many of the loans [to private sector
borrowers] might have explicit or implicit government guarantees.” On this point, also see Malik et al.
(2021) and Malik and Parks (2021).
55
 of sufficient information for categorization purposes.100 By contrast, BU’s
CODF dataset captures $605 billion of lending commitments from official
sector institutions in China to 150 borrowing institutions that qualify either
as central government debt, central government-guaranteed debt or
another type of public sector debt.101

3. Debt instrument coverage: For the most part, existing Chinese

development finance datasets—such as BU’s CODF dataset and the
World Bank’s IDS—track long-term, dollar-denominated bilateral loans for
public investment projects. However, China’s overseas lending program is
supported by a more diverse set of financing instruments, some of which
are RMB-denominated, short- or medium-term in nature, unrelated to
public investment projects, and/or involving multiple Chinese and/or
non-Chinese banks. They include syndicated loans with multiple Chinese
and/or non-Chinese bank participants, loans entrusted to multilateral
administrators, short- or medium-term liquidity support facilities (LSFs) to
provide balance of payments (BOP) support, currency swap facilities,
foreign currency deposit loans, pre-export financing (PxF) facilities,
deferred payment facilities, EPCF arrangements, interbank loan
agreements, and revolving credit facilities, among other things.
Consistent with OECD-DAC measurement standards, the 3.0 version of
AidData’s GCDF dataset tracks the full set of debt instruments used by
Chinese state-owned entities in low-income and middle-income
countries.102 Here again, the breadth of instrument coverage is becoming
increasingly important, as Chinese state-owned creditors pivot away from
the traditional approach of issuing long-term, dollar-denominated
100
These figures exclude the short-term “rollover” facilities that are described in Box 2c and Section A-3.
When such facilities are included, the 3.0 version of the GCDF dataset captures the following lending
commitment amounts: central government or central government guaranteed debt ($687 billion), other
public sector debt ($422 billion), potential public sector debt ($67 billion) and private sector debt ($216
billion). There is an additional $50 billion in debt that is not allocable due to a lack of sufficient information
for categorization purposes.
101
For its part, IDS captures $378 billion of lending from official sector institutions in China to borrowing
institutions that qualifies either as central government debt, central government-guaranteed debt or
another type of public sector debt (i.e., PPG debt). However, the publicly available IDS data cannot be
disaggregated by borrowing institution. For a country-by-country comparison of AidData and IDS measures
of PPG and non-PPG debt exposure to China, see Table A16
102
The OECD-DAC uses a broad definition of what qualifies as a debt instrument. According to the latest
version of its Converged Statistical Reporting Directives for the Creditor Reporting System (CRS) and the
Annual DAC Questionnaire, "[d]ebt instruments require the payment of principal and/or interest at some
point(s) in the future” (OECD 2023b: 12).
56
 bilateral buyer’s credits and concessional loans for public investment
projects (see Figures A11, A12, and A13) and toward short- and
medium-term bilateral emergency rescue loans as well as long-term
syndicated loans and loans entrusted to multilateral administrators.103

The “scope parameter” differences between the GCDF dataset and other
publicly available datasets matter for a simple but important reason: exclusively
tracking the lending activities of China Eximbank and CDB and the subset of
loans from these two policy banks that qualify as PPG debt is not a useful way to
monitor the changing scale and composition of Beijing’s overseas lending
portfolio if the portfolio has shifted toward new creditors, new borrowers, and
new lending instruments. In this chapter, we demonstrate that such changes
have already taken place, which means that a continued focus on PPG debt from
China Eximbank and China Development Bank would be analogous to the
proverbial drunkard who insists upon searching for his keys beneath the
lamppost “because that’s where the light is.”104

There are several supply-side factors that have likely prevented China’s overseas
lending program from collapsing. First, Beijing’s state-owned banks have high
levels of international exposure to default risk, which means they have an
interest in ensuring that their biggest borrowers are sufficiently liquid to continue
servicing their existing debts. One way of providing liquidity relief to borrowers
is via short- or medium-term bridge loans.105 Second, even though Chinese
banks are increasingly reluctant to issue long-term infrastructure project loans to
government borrowers due to the rising tide of sovereign debt distress, Chinese
companies with significant international operations have an interest in securing

103
During the early BRI period (2014-2017) and late BRI period (2018-2021), an increasing proportion of
China's official sector lending to LICs and MICs consisted of emergency rescue loans, including those of the
“rollover” variety (see Figures 2.6 and A14 and Box 2c). The percentage of China’s non-emergency lending
program in LICs and MICs that was channeled via multilateral institutions or syndicated loan arrangements
increased from 33.7% in 2014 to 51.5% in 2021 (see Figure 2.23).
104
According to the 3.0 version of AidData’s GCDF dataset, the percentage of official sector lending
commitments from China to LICs and MICs that were channeled through the policy banks (CDB and China
Eximbank) plummeted from 87% in 2009 to 22% in 2021 (see Figure A27). As Mingey and Kratz (2021) put
it “[t]he issue ultimately is one of scope. The [...] focus on policy bank loans obscures changes in China’s
lending patterns—whether a shift in the source of loans to emerging market governments from policy
commercial banks, or shifts in the destination of loans from governments to private infrastructure vehicles
and corporates.”
105
Other options include grace period extensions, maturity extensions, and interest rate reductions. For
more on this topic, see Horn et al. (2023a, 2023b).
57
new business and preserving market share in the countries where they operate.
They therefore have incentives to ensure that new sources and types of
financing—such as syndicated loans that pool credit risk across Chinese and
non-Chinese lenders, loans to special purpose vehicles and joint ventures rather
than sovereigns, and deferred payment or EPCF agreements that involve more
risk-sharing between Chinese companies and their overseas clients—are brought
to bear in support of new infrastructure projects in overseas markets. Third, to
sustain high levels of domestic economic growth, China has a long-run need to
secure natural resources that it lacks in sufficient quantities at home, which is an
important reason why Beijing allows its overseas borrowers to collateralize and
repay loans with the money they earn from natural resource exports to China.
Fourth, China is seeking to position itself as a major world power and project
influence around the globe, which serves as a counterweight to financial
pressures for retrenchment.

However, there is probably no factor more important than the overall size of
China’s foreign exchange reserves. China would not have become the world’s
largest official creditor to the developing world—larger than the World Bank, the
IMF, and all Paris Club creditors combined—if not for its massive stockpile of
foreign exchange reserves (Dreher et al. 2021, 2022). When Beijing adopted the
“Going Out” strategy at the turn of the century, it prioritized
dollar-denominated lending to overseas borrowers to deal with a foreign
currency oversupply challenge: annual trade surpluses led to a rapid
accumulation of dollar reserves, which created a risk of currency (RMB)
appreciation and prompted China’s State Administration of Foreign Exchange
(SAFE) to search for international assets where it could invest its surplus dollar
reserves and get a good return.106 SAFE used these funds in the early 2000s to
recapitalize several state-owned banks, which in turn ramped up
dollar-denominated lending to overseas borrowers (see Figure 2.2 and Box 2b).
However, until the 2008 Global Financial Crisis, SAFE parked the lion’s share of
its surplus dollar reserves in U.S. government securities. Then, in 2008 and 2009,
international asset prices plummeted and the U.S. Federal Reserve weakened
the dollar via quantitative easing. Beijing’s traditional investment strategy of

106
For decades, China has sought to avoid a rapid appreciation of its currency in order to sustain high
levels of export growth, which it regards as essential to achieve high-income country status.
58
parking surplus dollar reserves in U.S. government securities became less
attractive and it launched a search for higher-yield (undervalued) overseas
assets. SAFE entrusted a larger proportion of its surplus dollar reserves to the
country’s state-owned policy banks, state-owned commercial banks, and
state-owned investment funds and tasked them with the pursuit of higher
investment returns via dollar-denominated international lending (see Box 2b).107

Figure 2.2: Composition of China’s loan portfolio by currency of denomination

Figure 2.2

Notes: The “Other” category includes all other currencies of denomination, including EUR, GBP,
and local currencies in low-income and middle-income countries.

The central role that foreign exchange reserves have played in the dramatic
expansion of China’s 21st century overseas lending program raises several
important questions. Are China’s foreign exchange reserves rising or falling? If
they are rising, how are they being invested? Are Chinese state-owned lenders
being recapitalized with these reserves and tasked with using these reserves to
extend foreign currency-denominated loans to overseas borrowers?
107
However, SAFE is not the only source of foreign exchange that China's state-owned commercial banks
have drawn upon to support foreign exchange-denominated overseas lending activities. Their balance
sheet data demonstrates that they also have access to domestic foreign exchange deposits (Setser 2023a).
59
China’s foreign currency reserves remain vast—approximately $3.1 trillion as of
2023. However, this figure only refers to the official, foreign currency reserve
holdings of China’s central bank (the PBOC). It does not account for the
country’s so-called “hidden reserves,” which include foreign currency that the
PBOC has moved out of its official reserve holdings (and off of its balance sheet)
by entrusting them to Chinese state-owned policy banks (like CDB and China
Eximbank), state-owned commercial banks (like BOC, ICBC, and China
Construction Bank), and state-owned funds (like the Silk Road Fund and CIC).
Brad Setser of the Council on Foreign Relations argues that the country’s
“hidden reserves” may be worth an additional $3 trillion.108 He also provides
evidence that these additional reserves have rapidly increased over the last
decade, which may explain why the 3.0 version of the GCDF dataset does not
show a major contraction in the overall size of China’s overseas lending program
in LICs and MICs.109 Additionally, it may explain why China’s overseas lending
program in high-income countries (HICs) and offshore financial centers (OFCs)
nearly doubled between 2015 and 2021 (see Box 2a and Table 2.1).

Box 2b: The investment agency behind the curtain—China’s State

Administration of Foreign Exchange (SAFE)

China’s State Administration of Foreign Exchange (SAFE) is a vice ministry-level institution and a
subsidiary of the PBOC; its original mandate in 1955 was to act as the country’s foreign
exchange regulatory authority. However, prior to the opening up of China’s economy in the late
1970s, the country had limited foreign exchange reserves and the PBOC had limited central
bank responsibilities.110

Despite its inauspicious beginnings, SAFE was eventually made responsible for the management
of the world’s largest stockpile of foreign exchange reserves. It also became one of the most
important investors in the world.111 Due to recurring current account surpluses and capital
account surpluses, China’s stock of foreign exchange reserves skyrocketed from from $200 billion
in 2000 to $1 trillion in 2006, $2 trillion in 2009, and $3 trillion in 2011. Keeping all of these
reserves (mostly USD) onshore posed a currency appreciation risk and threatened to undermine

108
Setser (2023a) argues that China’s “hidden reserves” consist of (a) non-reserve foreign exchange assets
that SAFE has provided to the policy banks, state-owned commercial banks, and state-owned investment
funds to lend and invest abroad; (b) foreign assets that state-owned commercial banks have purchased to
match their domestic foreign currency deposit base; and (c) foreign exchange that CIC purchased off the
balance sheet of SAFE.
109
See Setser 2023a, 2023c.
110
At that time, the PBOC was a state-owned commercial bank operating under the supervision of the
Ministry of Finance. The PBOC did not officially become China’s central bank until 1983.
111
Wei 2013; Liu 2023.
60
the country’s export-led economic growth strategy, so SAFE was tasked with investing surplus
dollars in overseas assets that would yield attractive returns within acceptable risk levels.112

Figure 2.3 below presents cumulative lending commitments of China’s state-owned policy banks,
state-owned commercial banks, and the Silk Road Fund to LICs and MICs, in relation to the
timing of SAFE’s investments in these organizations. It does the same for Sinosure-backed
lending commitments to LICs and MICs. One can see that large cash injections from SAFE have
generally preceded increases in the overseas lending activities of China’s state-owned policy
banks, state-owned commercial banks, and the Silk Road Fund. An increase in Sinosure-backed
lending to overseas borrowers also followed SAFE’s recapitalization of the state-owned credit
insurance agency in 2010.

Figure 2.3: Cumulative loan commitments to LICs and MICs by financial institution and capital injections from SAFE, 2000-2021

Figure 2.3
Cumulative loan commitments to LICs and MICs by financial institution and capital injections
from SAFE, 2000-2021

112
China’s foreign exchange management law requires that the country’s official reserves be invested in
highly liquid and low-risk assets that can be used to address balance of payment needs. However, foreign
exchange reserves that are entrusted to a state-owned entity for investment purposes fall outside the
official (IMF) definition of foreign exchange reserves. They can therefore be invested in higher-risk, illiquid
assets (Liu 2023).
61
Notes: This table presents cumulative lending commitments to LICs and MICs (in constant 2021 USD) from
selected Chinese state-owned policy banks, state-owned commercial banks, and state-owned funds. It also
presents cumulative lending commitments to LICs and MICs (in constant 2021 USD) that are backed by
credit insurance from Sinosure. The vertical dashed lines represent years in which a SAFE capital injection is
known to have taken place. These figures exclude short-term "rollover" facilities to refinance maturing
debts (see Box 2c and Section A-3 in the Appendix).

SAFE’s first major investment came in December 2003, when it capitalized the Central Huijin
Investment Corporation (Central Huijin) with $45 billion, which in turn bought equity stakes in
two state-owned commercial banks: Bank of China ($22.5 billion) and China Construction Bank
($22.5 billion). In April 2005, Central Huijin also bought an equity stake in ICBC for $15 billion.
SAFE injected an additional $150 billion into China’s state-owned commercial banks—by
swapping USD for RMB held by the banks—in late 2005 and 2006.113 The first known
recapitalization of a state-owned policy bank came in July 2005, when SAFE injected $5 billion
into China Eximbank. Then, in December 2007, Central Huijin—a wholly-owned subsidiary of
SAFE—injected $20 billion into CDB. Six months later, SAFE agreed to provide additional
funding (worth an estimated $166.5 billion) to CDB via entrusted loan agreements to support the
overseas activities of Chinese companies.114 Under these agreements, CDB acted as a custodian
of funds for SAFE, disbursing loans to borrowers, supervising the use of the funds, and
managing repayments.115 SAFE eventually expanded its use of entrusted loan agreements to
other state-owned banks.116 By 2010, it had “taken initial steps toward giving policy and
commercial banks authority to handle loans for intergovernmental cooperation projects” and
moved “beyond its traditional role of managing foreign exchange reserves, effectively
[becoming] a foreign-currency lender” (Yuzhe 2010).117

Then, in April 2010, Central Huijin injected $11.7 billion (RMB 80 billion) into China Eximbank
and Sinosure to help them clean up bad loans on their balance sheets. Fourteen months later, in
June 2011, China Investment Corporation—another state-owned entity responsible for investing
the country's foreign exchange reserves—injected an additional $3.15 billion (RMB 20 billion)
into Sinosure. Then, in December 2014, SAFE injected $40 billion into the Silk Road Fund
through a subsidiary known as Buttonwood Investment Holding Company Ltd.118 SAFE injected
$48 billion into CDB and $45 billion into China Eximbank in July 2015 through wholly-owned
subsidiaries (known as Wutongshu Investment Platform Co. Ltd, Sycamore Tree Investment
Platform, and Buttonwood Investment Holding Company Ltd.). SAFE also purchased equity

113
Central Huijin injected RMB 130 billion ($19 billion) into the Agricultural Bank of China in November
2008. However, it was no longer a wholly-owned subsidiary of SAFE at the time.
114
Nine foreign reserve entrusted loan agreements were signed by SAFE and CDB in May 2010.
115
However, as a fiduciary acting on behalf of SAFE (in exchange for a commission), it does not assume any
of the risks or rewards of the entrusted loans. As such, CDB records entrusted loans as off-balance sheet
items.
116
According to Liu (2023: 174), “SAFE does have a minimum return target for the foreign exchange
entrusted loans of about 2.5 percent, calculated as a spread of several basis points above the international
benchmark bank-lending rate, LIBOR. In 2012 and 2013, an interest rate of 2.5 percent was not particularly
low in an environment of quantitative easing in the EU and the United States; at the end of 2012 the US
ten-year Treasury note, a proxy for the risk-free rate, was only 1.5 percent.”
117
At that time, SAFE also agreed to "act as the organizer and primary arranger of syndicated loans under
entrust agreements” (Yuzhe 2010).
118
In May 2017, the Chinese Government announced that it would inject another RMB 100 billion into the
Silk Road Fund. The entity responsible for the injection is unknown.
62
stakes in ICBC and Bank of China—via Wutongshu Investment Platform Co. Ltd—during the last
quarter of 2015.119

Since 2015, SAFE has not publicly announced any additional cash infusions into China’s
state-owned policy banks, state-owned commercial banks, or its state-owned credit insurance
agency. An ongoing source of speculation and debate is whether SAFE is worried about
“throwing good money after bad.”120 There are some indications that the country’s largest
state-owned banks and Sinosure, which are stewarding foreign exchange reserves from SAFE,
may have high levels of cross-border exposure to non-performing loans on the balance sheets of
Chinese banks.121 In November 2017, China Banking Regulatory Commission (CBRC)—the
country's top banking regulator—publicly called upon CDB and China Eximbank to put in place
more robust risk management procedures (Xueqing 2017).122 Then, in 2018, Sinosure’s Chief
Economist took the extraordinary step of publicly admonishing China’s policy banks for their
“downright inadequate” due diligence procedures (Pilling and Feng 2018). Another sign of
potential peril is the fact that the annual overseas lending commitments of CDB, China
Eximbank, Bank of China, and ICBC sharply declined after they received large cash infusions
from SAFE in 2015 (see Figure A31). One potential explanation for this unusual pattern is that
SAFE’s money was used to clean up bad debts on the balance sheets of these banks instead of
supporting new overseas lending commitments.

SAFE is discreet about the returns that it has earned via overseas investments, due to domestic
political concerns about whether China’s “xuè hàn qián” (the hard-earned money of Chinese
workers) is being effectively stewarded. It “tries to limit its investments outside [U.S. government
securities] to amounts small enough to hide from the public in case the bets go bad” (Wei 2013).
However, in 2021, SAFE disclosed that it had earned an average annual return on its foreign
exchange reserve portfolio of 3.35% between 2008 and 2017 and 4.11% between 1998 and

119
The bank (re)capitalization information in Box 2b was drawn from SAFE 2004, 2017a, 2017b; Ma 2006;
PBOC 2007, 2012; Ying 2008; China Daily 2009; Reuters 2009, 2010a, 2010b, 2016; Parson 2010; Shan
2011; Yuan 2014; Tangjun et al. 2014; Jia 2015; Xinhua 2015; Xiao 2016; Xie and Lamar 2016; Chen 2014;
Kong and Gallagher 2017; Office of the Leading Group for Promoting the Belt and Road Initiative 2019;
Embassy of the People's Republic of China in the Republic of the Philippines 2019; Liu 2023; and Setser
2023a.
120
SAFE has a particularly high level of exposure to non-performing loans on China Eximbank’s balance
sheet. According to a bond prospectus that China Eximbank issued in March 2017, “the ownership of the
Ministry of Finance in China Eximbank is approximately 10.7% while [...] SAFE owns approximately 89.3% of
China Eximbank through its investment platform” (Export-Import Bank of China 2017: 20). SAFE is also
exposed to non-performing CDB loans, but its exposure is related to its use of CDB as a fiduciary for
entrusted loans and CDB’s own balance sheet. SAFE holds a 27.1% ownership stake in CDB through
Buttonwood Investment Holding Company Ltd. It purchased this stake on July 15, 2015, when Buttonwood
Investment Holding Company Ltd. injected $48 billion of share capital into CDB (CDB 2020).
121
During the primary data collection effort for the 3.0 version of the GCDF dataset, AidData uncovered
evidence of Sinosure in recent years providing credit insurance for loans issued by non-Chinese banks for
projects being implemented by Chinese contractors. These loan commitments are not included in the 3.0
version of AidData’s GCDF dataset.
122
Zhou Minyuan, the head of CBRC's policy banks supervision department, announced at the time that it
“required both banks to fully identify overseas business risks, step up compliance management, completely
understand the operational and financial status of their clients as well as the laws and regulations of host
countries, strictly observe the local environmental and industrial regulations, and strengthen communication
with local regulators” (Xueqing 2017). CBRC also “demanded the banks enhance capital supervision via
on-site inspections and investigations, effectively prevent and control overseas business risks by taking
risk-sharing measures, prudentially evaluate the feasibility and compliance of relevant guarantee measures,
and improve their emergency response mechanism” (Xueqing 2017).
63
2017 (SAFE 2021: 53). At the time, it characterized the rate of return as being at a “relatively
good level” (SAFE 2021: 53). SAFE has not released any data on average annual returns for 2018
or any subsequent years.

Another important question that remains unanswered is the extent to which SAFE owns the
bonded debt of the LIC and MIC governments of Emerging Market and Developing Economies
(EMDEs)—either through direct purchases or secondary market purchases (i.e., via investment
funds like PIMCO, Blackrock, AllianceBernstein, Fidelity Investments, and Amundi Asset
Management). There are several reasons to believe that SAFE’s holdings of bonded EMDE
sovereign debt may be significant. First, when SAFE agreed to buy $300 million of bonded debt
from the Government of Costa Rica in 2008, it attempted to hide the purchase (Anderlini 2008).
The Deputy Administrator of SAFE, Fang Shangpu, sought and secured a written assurance from
the Government of Costa Rica that it would “take necessary measures to prevent the disclosure
of the financial terms of this operation and of SAFE as a purchaser of the bonds” (SAFE 2008).123
The purchase was only made public because of a court order. When SAFE was asked to
comment on the matter, it said that there was nothing unusual about the purchase because it
“owns bonds issued by many other governments” (Batson 2008). Second, SAFE’s Chief
Investment Officer from 2010 to 2014 was a bond trader at PIMCO (and right-hand man of
PIMCO co-founder Bill Gross) from 2006 to 2009 (Wei 2013). Third, SAFE opened an office on
Fifth Avenue in New York City in 2013, and shortly thereafter it became an open secret among
the world’s largest asset managers—like PIMCO and BlackRock—that SAFE was one of their
most important confidential clients.

Section 2: Major changes in Beijing’s overseas lending

portfolio during the BRI 2.0 era
Our goal in the remainder of this chapter is to explain how China is responding
to the rising tide of debt distress in the developing world and identify the
measures it is taking to de-risk its overseas lending portfolio. However, before
we do so, it is important to understand some of the major changes in Beijing’s
overseas lending portfolio that have recently transpired. Three changes, which
have become hallmarks of the BRI 2.0 era, merit special attention:

1. China’s new and unfamiliar role as the world’s largest official debt
collector

123
Also see Government of Costa Rica (2007, 2008). We have made the official correspondence between
SAFE and Costa Rica’s Ministry of Finance in January 2008 accessible via
https://www.dropbox.com/scl/fi/a20fdxb2lecowt8tp5lym/2-January-2008-SAFE-Letter-to-Minister-of-Financ
e-of-Costa-Rica.pdf?rlkey=y9ywdrrhap4rdkt0cyqj4mj4y&dl=0 and
https://www.dropbox.com/scl/fi/48qdeslzwim9t6dqhyk0m/7-January-2008-Letter-from-Minister-of-Finance-
of-Costa-Rica-to-SAFE.pdf?rlkey=pr3p4advi1wfzzwjr5t8gva5m&dl=0.
64
 2. The rise of emergency rescue lending and the fall of infrastructure project
lending

3. A strategic pivot from USD-denominated bilateral lending to

RMB-denominated bilateral lending

China’s new and unfamiliar role as the world’s largest official debt
collector
When the BRI was initially launched, Beijing differentiated itself from its peers
and competitors by financing big-ticket infrastructure projects—like high-speed
railways and next-generation telecommunication networks—that virtually no one
else was willing to bankroll. Many of the loans that were issued for these projects
included lengthy grace periods, so most borrowers didn’t need to worry about
making large debt service payments for five, six, or seven years. With a lot of
“easy money” sloshing around, many borrower countries saw their economic
growth rates soar (Dreher et al. 2021, 2022). China’s popularity around the globe
also soared: the high-water mark of public support for China in the developing
world—after the introduction of the BRI—was in 2019.124

However, as Figure 2.4 demonstrates, the age of easy money is in the rear-view
mirror for an expanding set of borrowers. By 2020, 40% of official sector loans
from China to low-income and middle-income countries had entered their
principal repayment periods (following the expiration of their grace periods).
This figure increased to 55% in 2023, and we expect it will reach approximately
75% by 2030 and 100% by 2049.

Another way of thinking about the looming repayment challenge is to track the
percentage of loans in China’s LIC and MIC portfolio that have already reached
their (originally scheduled) final repayment dates. In 2014 (the first full year of
BRI implementation), this figure stood at only 17% (see Figure 2.5). By 2023 (the
tenth full year of BRI implementation), it had increased to 44%. We expect this

124
At that time, the population-weighted average level of public support for China in the
developing world was 55.66%. On the causal relationship between the receipt of Chinese aid
and credit and public support for China, see Wellner et al. (forthcoming, 2023).
65
figure to rise to 52% by 2025 and approach nearly 100% by 2040 (see Figure
2.5).

Figure 2.4: Loans from China within their repayment periods, 2000-2050

Figure 2.4

Notes: This graph shows the cumulative percentage of official sector loans from China to LICs and MICs (as
measured by the number of loans) that are within their repayment periods). To determine when each loan
will enter repayment, each loan’s grace period is added to its commitment date. This figure represents
when loans will reach their repayment period according to their original borrowing terms, although many
loans have been rescheduled (often involving an extension of the loan’s grace period and/or maturity).
MOFCOM interest-free loan commitments (which are typically issued without a credible expectation of
repayment) are excluded from the calculation.

66
Figure 2.5: Loans from China reaching their (originally scheduled) final repayment dates, 2000-2050

Figure 2.5

Notes: This graph shows the cumulative percentage of official sector loans from China to LICs and MICs (as
measured by the number of loans) that have reached maturity. To determine when each loan will reach
maturity, each loan’s maturity period is added to its commitment date. This figure represents when loans
will reach their maturity according to the original borrowing terms, although many loans have been
rescheduled (often involving an extension of the loan’s grace period and/or maturity). MOFCOM
interest-free loan commitments (which are typically issued without a credible expectation of repayment) are
excluded from the calculation.

What does all of this mean in practical terms? It means that a rapidly growing
percentage of borrowers in the Global South are making large debt service
payments (that are for the most part denominated in U.S. dollars) to Beijing at a
time when interest rates are rising, the U.S. dollar is strengthening, local
currencies are weakening, and global growth is slowing. Many borrowers do not
have enough hard currency on hand to meet their repayment
obligations—especially on loans with LIBOR-based interest rates that increased
by five-and-a-half percentage points between January 2022 and September
2023.125

125
Average 6-month LIBOR skyrocketed from 0.426% in January 2022 to 5.892% in September 2023.
67
It also means that Beijing finds itself in an unfamiliar and uncomfortable role—as
the world’s largest official debt collector. Some of China’s state-sponsored
tabloids and research institutions are attempting to deflect criticism by blaming
the U.S. Federal Reserve for the sharp increase in LIBOR-based interest rates
(e.g., Xueqing 2022; Qing 2023).126 However, this position will be difficult to
defend. When Beijing decided to make LIBOR central to its dollar-denominated
overseas lending program (see Figure 2.14), it did not do so at the behest of the
U.S. or any other foreign power.127 It did so on its own—and for its own
profit-making purposes (see Box 2b). Now, the grace periods on most of China’s
LIBOR-based loans are expiring, and Beijing is learning a difficult lesson: it’s
easier for great powers to behave like commercial creditors in times of plenty
than in times of want. If China was a commercial bank, it would be easier to
demand that financially-distressed LIC and MIC borrowers draw upon their
limited U.S. dollar reserves to make increasingly large debt service payments.
However, China is a global superpower that has to balance a wide range of
competing equities, including reputational risk and repayment risk (see Chapter
4 for more on this key tension and tradeoff).

The rise of emergency rescue lending and the fall of infrastructure

project lending
China’s overseas lending program has become synonymous with the BRI—an
overland “Belt” of road, rail, port, and pipeline projects that seeks to create an
infrastructure corridor from China to Central Asia and Europe, and a “Maritime
Silk Road” that seeks to link China to South and Southeast Asia, the Middle East,
and Africa through deep-water seaport construction projects along the littoral
areas of the Indian Ocean. However the 3.0 version of AidData’s GCDF dataset

126
For example, in April 2023, China’s state-owned tabloid, Global Times, published an op-ed identifying
“the US’ irresponsible monetary policy [as] the root of African debt problems.” It argued that “[r]elying on
dollar hegemony, the US has implemented three rounds of quantitative easing, cut interest rates to near
zero, and flooded Africa and emerging markets with low-interest dollars. It then arbitrarily and aggressively
raised interest rates, boosted the U.S. dollar exchange rate, attracted the return of dollars, as a result,
African countries have to face liquidity shortages, broken funding chains, currency depreciation,
skyrocketing debt repayment costs denominated in dollars, a surge in sovereign debt, and exacerbated
debt problems. The unfair financial system led by the US is the root of Africa’s debt problems.” (Qing
2023).
127
According to the 3.0 version of AidData’s GCDF dataset, 45% of China’s dollar-denominated official
sector lending to LICs and MICs is based on 6-month LIBOR or another LIBOR-based interest rate and 67%
of China's variable interest rate official sector lending to LICs and MICs is based on 6-month LIBOR or
another LIBOR-based interest rate (see Figure 2.14).
68
highlights the importance of not conflating China’s flagship, global infrastructure
initiative with its overseas lending program.128

During the pre-BRI period (2000-2013), Beijing provided an extraordinary

amount of credit for infrastructure projects in LICs and MICs.129 In total, the 3.0
version of AidData’s GCDF dataset identifies 2,217 loan-financed projects worth
$628 billion during this fourteen-year period.130 66% of these projects (worth
$412 billion) sought to construct, rehabilitate, expand, or maintain physical
infrastructure.131 However, Figures 2.6 and A13 provide evidence that, since the
BRI was launched, a rapidly shrinking proportion of China’s overseas lending to
LICs and MICs has supported infrastructure projects.132 This was true during the
early BRI period (2014-2017) and during the late BRI (2018-2021) period.
Infrastructure project lending commitments as a share of total lending
commitments to LICs and MICs fell from 65% in 2014, to 50% in 2017, 49% in
2018, and 31% in 2021.

128
Table 2.1 also calls attention to this point by spotlighting the vast scale of China’s overseas lending to
high-income countries (HICs) and offshore financial centers (OFCs).
129
84% of China’s infrastructure project lending from 2000-2014 was provided by CDB and China
Eximbank, and 32% was provided via buyer’s credits. However, Figures A11 and 2.7 demonstrate that a
rapidly shrinking percentage of China’s overseas lending program in the developing world is provided via
the country’s policy banks and buyer’s credits.
130
To generate this estimate, we identify all loans in the 3.0 version of the GCDF dataset that are
categorized as “investment project loans.” For more on the definition and measurement of the investment
project loan variable, see Custer et al. (2023).
131
To generate this estimate, we identify all loans in the 3.0 version of the GCDF dataset that are
categorized as “investment project loans” and that involve the construction, rehabilitation, expansion, or
maintenance of physical infrastructure. For more on the definition and measurement of the infrastructure
variable and the investment project loan variable, see Custer et al. (2023).
132
Figure A26 in the Appendix provides another version of this graph that presents project lending rather
than infrastructure project lending over time. It shows a very similar pattern: project lending commitments
as a share of total lending commitments fell from 78% in 2014 to 56% in 2017, and then from 50% in 2018
to 32% in 2021.
69
Figure 2.6: Composition of loan portfolio by financial instrument

Figure 2.6

Notes: For details on how infrastructure project lending is measured, see footnote 131.

Given that Beijing did not dramatically scale back the overall size of its overseas
lending program in LICs and MICs, how and why did it continue to lend record
amounts to developing countries at a time when it was ratcheting down Belt and
Road project lending? Figure 2.6 provides a clear answer: Beijing was ramping
up its emergency rescue lending activities while it was ramping down its
infrastructure project lending activities.133 In 2013, one year before the first full
year of BRI implementation, emergency rescue lending represented only 5% of
China’s overseas lending to LICs and MICs. By 2021, 58% of China’s overseas
lending to LICs and MICs consisted of emergency rescue lending. The People’s
Bank of China (PBOC)—China’s central bank—is by far the most important
financier of international emergency rescue lending operations, which explains

133
Consistent with Horn et al. (2023a, 2023b), we define emergency rescue loans as all loans from Chinese
state-owned entities to government borrowing institutions in low-income and middle-income countries that
can be used for at least one of three purposes: repaying existing debts, financing general public
expenditures, or shoring up foreign exchange reserves. Such loans include borrowings via currency swap
agreements, liquidity support facilities, foreign currency term financing facility agreements, deposit loans,
commodity prepayment facilities, and so-called “sovereign loans” (主权贷).
70
why it had assumed a dominant role in Beijing’s LIC and MIC lending portfolio
by 2020 (see Figure 2.7).134

Figure 2.7: Composition of loan portfolio by creditor category

Figure 2.7

In March 2023, a team of researchers from AidData, the World Bank, the Harvard
Kennedy School, and the Kiel Institute for the World Economy published a study
that explains why Beijing has undertaken rescue lending operations worth nearly
$250 billion in 22 countries (Horn et al. 2023a).135 They find that most of these
operations have taken place in BRI participant countries with high levels of
outstanding (infrastructure project) debt to Chinese banks and companies. They
also find that bailouts from Beijing are directed to distressed government
borrowers at times when their foreign exchange reserve levels are low and their
credit ratings are weak.

134
In 2013, the PBOC and SAFE (its subsidiary) were responsible for only 6% of China’s official sector
lending commitments to LICs and MICs. By 2021, that figure reached 54% (see Figure 2.7 and Figure A37).
135
The authors of the study include Sebastian Horn of the World Bank; Brad Parks, Executive Director of
AidData and Research Professor at William & Mary’s Global Research Institute; Carmen Reinhart, former
World Bank Group Chief Economist and current Professor at the Harvard Kennedy School; and Christoph
Trebesch, Director at the Kiel Institute for the World Economy.
71
Figure 2.8: Chinese emergency rescue lending by countries’ level of debt exposure to China

Figure 2.8

Notes: The three categories (high, medium, or low debt exposure to China) are constructed by ranking
countries according to total non-emergency lending commitments from official sector institutions in China
between 2000 and 2021. Countries that did not receive any non-emergency lending commitments are
excluded from the calculation. The calculation of average emergency rescue loan commitments excludes
short-term "rollover" facilities to refinance maturing debts (see Box 2c and Section A-3 in the Appendix).

The 3.0 version of AidData’s GCDF dataset provides an opportunity to conduct a

basic replication exercise with four years of additional data (2018-2021),
substantially revised data for 2000-2017, and more precise measurements of
China’s project, infrastructure project, and non-emergency lending activities. In
Figure 2.8, we reproduce Figure 5 from the Horn et al. (2023a) study with
updated data on emergency and non-emergency lending commitments and
confirm that China’s international emergency lending operations are
concentrated in countries that accumulated large amounts of debt to China for
non-emergency purposes between 2000 and 2021.136 We can also confirm that
every country that received an emergency rescue loan from China is a
participant in the BRI.137 In Table 2.2, we replicate Table A4 from the Horn et al.
(2023a) study and confirm that Chinese emergency rescue loans are issued to
sovereigns at times when reserves are low and at times when borrowers have

136
Figure 2.8 uses a new dummy variable (“rescue”) in the 3.0 version of the GCDF dataset for emergency
rescue loans, which captures any loan that allows a sovereign debtor to (1) service existing debts, (2)
finance general budgetary expenditures and/or (3) shore up foreign reserves.
137
As of 2021, this was true of all emergency rescue loan recipients other than Argentina and Malawi. Then,
in 2022, Argentina and Malawi joined the BRI.
72
very weak credit ratings.138 By contrast, we find that Chinese project loans and
Chinese infrastructure project loans are issued to borrowers at times when they
have relatively strong credit ratings and reserve adequacy ratios.139 All of these
findings are consistent with those of Horn et al. (2023a) and support their
characterization of Beijing’s crisis management response as one of “Bailouts on
the Belt and Road.”140

Table 2.2: Average sovereign risk ratings and gross reserves for recipients of different Chinese loan types

Table 2.2
Average sovereign risk ratings and gross reserves for
recipients of different Chinese loan types

Average
Gross Reserves (in
Type of Loan Sovereign Moody’s Rating Fitch Rating S&P Rating
Months of Imports)
Risk Rating
Rescue Loans 5.7 Caa1 CCC CCC+ 4.7

Project Loans 9.4 Ba3 BB- BB- 6.2

Infrastructure 9.2 Ba3 BB- BB- 5.9

Project Loans
Notes: This table presents the average sovereign risk rating and level of gross reserves (in months of import
cover) for countries that received emergency rescue loans, project loans and infrastructure project loans
from China between 2000 and 2021. The averages are weighted by the commitment amounts of the
emergency rescue, project, and infrastructure project loans from the 3.0 version of AidData’s GCDF dataset.
The data on gross reserves in months of imports are drawn from the World Bank’s World Development
Indicators. The sovereign risk ratings data are derived from the World Bank’s sovrate index using the
conversion scale in Séri (2021). Sovrate is a measure of repayment risk that varies from 0 to 21, with higher
scores indicating lower levels of sovereign credit risk (Kose et al. 2022).

138
The sovereign risk ratings produced by Moody’s take one of 21 categorical values, where Aaa represents
the lowest level of risk and C represents the highest level of risk. The sovereign risk ratings produced by
Fitch take one of 21 categorical values, where AAA represents the lowest level of risk and D represents the
highest level of risk. The sovereign risk ratings produced by S&P take one of 21 categorical values, where
AAA represents the lowest level of risk and C/D represents the highest level of risk. For more details, see
Séri (2021).
139
Horn et al. (2023a) uses loan commitments for non-emergency purposes as a proxy for “project loans”
and relies on the 2.0 version of AidData’s GCDF dataset. Table 2.2 uses the 3.0 version of AidData’s GCDF
dataset, which allows for more precise measurement of project loans and infrastructure project loans. For
more on these measurements, see footnote 130 and footnote 131.
140
Emergency rescue loans and debt reschedulings are similar in that they both provide cash flow relief to
insufficiently liquid borrowers. In this way, they can both be used to “bail out” a borrower (Horn et al.
2023b).
73
In Figure 2.9, we take the analysis one step further. We first use the country-year
level measure of financial distress (that we introduced in Chapter 1) to identify
Beijing’s 50 largest LIC and MIC borrowers that experienced financial distress at
some point between 2000 and 2021.141 We then identify the timing of bailouts
(emergency rescue loans) and debt reschedulings (cash flow relief) from Beijing
in relation to the onset and duration of financial distress episodes.

141
In order to differentiate between repayment risks and repayment risk mitigation efforts, we modify the
criteria for the financial distress measure. Instead of using all loan records where the description field in the
3.0 GCDF dataset indicates that the borrower had difficulty making repayments or experienced financial
distress, we exclude all observations for which the only source of evidence of the borrower having difficulty
making repayments or experiencing financial distress is an attempted or actual debt rescheduling.
74
Figure 2.9: Rescue lending and debt rescheduling events for the top 50 borrowers in financial distress, 2000-2021

Figure 2.9
Rescue lending and debt rescheduling events for the top 50
borrowers in financial distress, 2000-2021

Notes: This figure provides an overview of the timeline of when countries experienced financial distress
(blue shading), when China provided rescue lending (circles), and when China rescheduled existing loan
repayments (X’s). A circle indicates that at least one rescue loan was provided by China to the respective
country that year, and an X indicates that at least one loan was rescheduled by China for the respective
country that year. Countries included in this list represent the top 50 borrowers in distress, ordered by the

75
size of their cumulative lending portfolio as of 2021. See footnote 141 for details on how the financial
distress index was modified to differentiate between repayment risks and repayment risk mitigation efforts.

Figure 2.9 demonstrates that 83% of China’s emergency rescue loans (including
short-term “rollover” facilities) were issued in years when the recipients of these
loans were in financial distress.142 Similarly, 80% of China’s debt reschedulings
took place in years when borrowers in the participating countries experienced
financial distress.143 Figure 2.9 also provides evidence of Beijing repeatedly
targeting debt reschedulings and emergency rescue loans to the same BRI
participant countries with high levels of debt exposure to China.144 The serial
nature of these cash flow relief efforts suggests that Beijing’s biggest borrowers
may not only have short-term liquidity problems, but also long-term solvency
problems.

Another important implication of these results is that China is increasingly

behaving like an international crisis manager. It has effectively created a safety
net for financially distressed sovereigns that are participating in the BRI—and, by
extension, their highly exposed Chinese creditors.145 It has also taken a
differentiated approach across countries that present varying levels of risk to the
Chinese banking sector, whereby countries that present a high level of balance
sheet exposure get new money (via balance of payments support) and countries
that present a low level of balance sheet exposure get cash flow relief (via debt
reschedulings) but no new money (Horn et al. 2023a, 2023b).146 These actions

142
By comparison, 49% of China’s non-emergency loans were issued in years when the recipients of these
loans were in financial distress. This difference is also observable during the BRI era (2014-2021). Over this
eight-year period, 83% of China’s emergency rescue loans (representing 86% of China’s emergency rescue
lending portfolio in monetary terms) were issued in years when the recipients of these loans were in
financial distress. During the BRI era (2014-2021), 63% of China’s non-emergency loans (representing 67%
of China’s non-emergency rescue lending portfolio in monetary terms) were issued in years when the
recipients of these loans were in financial distress. These percentages reflect the distress marker that was
modified to differentiate between repayment risks and repayment risk mitigation efforts (see footnote 141).
143
This finding is also consistent with Horn et al. (2023a) and Horn et al. (2023b).
144
According to the underlying data that was used to construct Figure 2.9, 100% of the countries that
received serial debt reschedulings and 100% of the countries that received serial emergency rescue loans
are formal participants in the BRI. All of these countries rank among Beijing’s 50 largest LIC and MIC
borrowers and either benefited from debt reschedulings in two or more consecutive years or emergency
rescue loans in two or more consecutive years.
145
It is important to keep in mind that, from a historical perspective, countercyclical official lending is the
norm rather than the exception. As Horn et al. (2020b) explain, “[d]uring the course of the 1930s, the
United States joined European states in extending official loans to states with balance-of-payments
problems, in particular through the US Export-Import Bank and the US Treasury’s Exchange Stabilization
Fund, which was established in 1934” (Horn et al. 2020b: 7).
146
For more on this point, see Horn et al. (2023a) and Horn et al. (2023b).
76
are difficult to reconcile with the increasingly popular “Beijing in retreat”
narrative that we previously described.

A strategic pivot from USD-denominated bilateral lending to

RMB-denominated bilateral lending
As we noted in Section 1 of Chapter 2, China dramatically increased its foreign,
dollar-denominated lending activities in the immediate aftermath of the 2008
Global Financial Crisis. Figure 1.2 in Chapter 1 demonstrates that the single
largest year-on-year increase in official sector lending commitments from China
to LICs and MICs was between 2008 and 2009, and Figure 2.2 provides
evidence that nearly all of these commitments were denominated in dollars.

But the 3.0 version of AidData’s GCDF dataset also calls attention to three
additional changes that took place after the 2008 Global Financial Crisis. First,
Chinese state-owned creditors were already holding a substantial amount of
distressed, dollar-denominated debt by the time the BRI was announced in 2013
(see Figures 1.8 and A4). Second, from 2013 onward, the dollar’s importance in
China’s overseas lending portfolio steadily declined: the share of new lending
commitments denominated in dollars fell sharply, from 93% in 2013 to 44% in
2021 (see Figure 2.2). Dollar-denominated loans were replaced by
RMB-denominated loans: the share of new lending commitments denominated
in RMB soared from 6% in 2013 to 50% in 2021 (see Figure 2.2). Third, the
RMB-denominated loans that Beijing issued were predominantly emergency
rescue loans to countries in financial distress (see Figures 2.9, 2.10, and A15).147

147
In the Appendix, we replicate Figure 2.2 for two different cohorts: one for countries in financial distress
and another for countries not in financial distress (see Figures A16 and A17). One can see an increase in
RMB-denominated lending across both cohorts, but the increase is more substantial for countries in
financial distress. Figure 2.10, which includes short-term, roll-over loan amounts, shows a similar pattern in
China’s RMB-denominated rescue lending portfolio.
77
Figure 2.10: RMB-denominated rescue lending to countries in distress and countries not in distress

Figure 2.10

Notes: To determine if a country experienced financial distress in a given year, we use the binary measure
that is described in Box 1a in Chapter 1.

On one hand, this strategy makes sense. During the 1930s and after World War
II, the U.S. became a major international lender of last resort, providing
dollar-denominated emergency rescue loans to borrowers with large
outstanding dollar-denominated debts to U.S. companies and banks through the
U.S. Federal Reserve, the U.S. Treasury’s Exchange Stabilization Fund, and the
U.S. Ex-Im Bank (Horn et al. 2020b). These activities helped the dollar eventually
become a dominant currency for reserve holdings and international financial
transactions. Now, Chinese state-owned policy banks, state-owned commercial
banks, and state-owned enterprises have high levels of exposure to overseas
borrowers that are in default or teetering on the edge of default, and the
institution with a mandate to protect the health of China’s financial sector and
internationalize the RMB (the PBOC) is ramping up the provision of emergency
rescue loans to ensure that its overseas borrowers are sufficiently liquid to
continue servicing their outstanding debts to Chinese creditors.

78
On the other hand, the PBOC’s decision to provide RMB-denominated rescue
loans is puzzling because the borrowers being bailed out need USD more than
RMB to repay their outstanding debts to Chinese creditors. One potential
explanation is that the PBOC is heeding Winston’s Churchill's advice to “never
let a good crisis go to waste.” For many years, it sought to internationalize the
RMB—without achieving much progress due to tight capital controls and an
insufficiently deep and liquid RMB bond market outside of mainland China.148
However, the rising tide of debt distress in the Global South has made two
groups of countries more willing to increase their RMB reserve holdings: (1)
countries facing severe liquidity and/or solvency problems that would like to
avoid borrowing from the traditional lender of the last resort (the IMF) because
of concerns about onerous policy conditionalities (like Venezuela, Belarus, and
Laos); and (2) countries facing severe liquidity and/or solvency problems that
have decided to seek IMF support but need additional support in order to stay
afloat (like Argentina, Mongolia, and Sri Lanka).149

Why has the PBOC channeled RMB-denominated emergency rescue loans to

both groups of countries since 2013? The logic of doing so in the first country
cohort seems relatively straightforward: the proceeds from an
RMB-denominated emergency loan can be used to service previously contracted
loans from Chinese creditors that were denominated in RMB. For the second
cohort of countries, Beijing’s intentions are opaque and poorly-understood. But
recent events in Argentina may provide a clue. In mid-2023, the dollar reserve
holdings of the country’s central bank (BCRA) were perilously low, and it was
urgently seeking bridge funding to avoid defaulting on its repayment
obligations under a $44 billion loan agreement with the IMF. The PBOC stepped
into the breach, helping BCRA make three large debt service payments to the
IMF in June 2023, July 2023, and October 2023 (Nugent 2023a, 2023b; Do
Rosario 2023b; Do Rosario and Strohecker 2023). It did so by allowing BCRA to
use RMB drawings under a swap line between the two central banks worth
approximately $9.3 billion. BCRA was able to use short-term RMB swap debt

148
As of 2022, the RMB accounted for less than 3% of global currency reserves and less than 2.5% of global
payments. Key impediments to RMB internationalization include tight capital controls that inhibit the free
movement of the currency and the absence of a large offshore market for investors to purchase safe,
RMB-denominated assets that are comparable to U.S. Treasury bonds (Bertaut et al. 2023).
149
See Horn et al. (2023a, 2023b).
79
from the PBOC to repay the IMF “without touching [its] dollar reserves” (do
Rosario and Otaola 2023) for two reasons: (1) money is fungible, and (2) IMF
loans can be repaid with multiple currencies (including USD, EUR, RMB, JPY,
GBP, and SDR). In other words, BCRA repaid its debt to the IMF in RMB, which
allowed it to preserve its dollar reserve holdings.150 The case of Argentina also
calls attention to a separate, but closely related, point: Beijing has an
encompassing interest in ensuring that its largest borrowers with
dollar-denominated debts to Chinese creditors do not exhaust their dollar
reserve holdings.151

In this way, the PBOC’s willingness to serve as an international lender of last

resort suggests that it may be seeking to kill several birds with one stone:
preserving the dollar reserve holdings of its biggest borrowers, encouraging
greater use of the RMB in cross-border transactions, and laying the groundwork
for the RMB to eventually achieve global reserve currency status.152 The fact that
the PBOC played an instrumental role in helping the IMF’s single largest
borrower meet its repayment obligations with RMB also suggests that former
U.S. Treasury Secretary Larry Summers may have been overly optimistic when he
was asked about the risk of de-dollarization and responded that: “you cannot
replace something with nothing. […] Europe’s a museum, Japan’s a nursing
home, and China’s a jail. We don’t need to worry about those currencies being
some kind of major threat to us.”153

150
In August 2023, Argentina’s central bank (BCRA) decided to repay some of its outstanding swap debt to
the PBOC with the proceeds from an IMF loan disbursement (Do Rosario 2023a).
151
There is also a potential demand-side explanation for why sovereigns in financial distress—like Argentina
and Sri Lanka—are willing to contract RMB-denominated swap debt from the PBOC. Despite significant
restrictions on the free and flexible use of PBOC swap drawings, central banks can use these RMB drawings
as a “window dressing” device to temporarily inflate their gross international reserves (Horn et al. 2023a).
This approach might help avert credit rating downgrades and borrowing cost increases. However, it can
also free up otherwise encumbered foreign exchange reserves to facilitate dollar-denominated debt service
to Chinese creditors (other than PBOC) and non-Chinese creditors. For example, Brad Setser has argued
(see https://twitter.com/Brad_Setser/status/1602151579150438401) that the receipt of PBOC swap debt in
Sri Lanka freed up foreign exchange reserves to facilitate a $400 million payment on a maturing Eurobond
and dollar-denominated debt service to China Eximbank.
152
It is also noteworthy that Beijing has changed its public messaging about the advisability of
dollar-denominated overseas lending and borrowing (e.g., Qing 2023).
153
Indeed, there is evidence that signing a currency swap agreement with the PBOC results in a 14%
increase in the probability of a country using the RMB for international payments (Bahaj and Reis 2022).
Large-scale borrowing via PBOC swap lines can also have the direct effect of changing the currency
composition of a country’s reserve holdings. After nearly a decade of RMB drawdowns through its swap line
with PBOC, approximately 50% of Argentina’s reserve holdings consisted of RMB (Douglas 2022).
80
Section 3: Beijing’s crisis-time approach to repayment risk
mitigation—Fool me once, shame on you. Fool me twice,
shame on me.
In the next section of this chapter, we analyze the 3.0 version of AidData’s GCDF
dataset to better understand how Beijing is seeking to de-risk its overseas loan
portfolio. We see evidence of Chinese state-owned lenders taking the following
risk mitigation efforts:

1. Sweeping cash out of escrow accounts

2. Providing short-term cash flow relief in exchange for escrow account

replenishment

3. Lending with higher interest rates, shorter repayment periods, more

safeguards, and more severe penalties for default

4. Taking a differentiated approach with borrowers that present high and low
levels of repayment risk

5. Scaling down bilateral lending operations and scaling up lending

operations via syndication and multilateralization

Risk mitigation strategy #1: Sweeping cash out of escrow accounts

In March 2021, a team of researchers from AidData, the Center for Global
Development (CGD), the Kiel Institute for the World Economy, and the Peterson
Institute for International Economics (PIIE) published a study entitled “How
China Lends,” which demonstrated that Chinese state-owned lenders have a
preference for sources of collateral that do not require liquidation through a
costly, time-consuming, and uncertain judicial process (Gelpern et al. 2021,
2022). More specifically, the study found that Chinese state-owned lenders
prefer to collateralize on foreign currency deposits in escrow accounts that they
control and can unilaterally debit (without having to initiate judicial proceedings
to try to recover an overdue debt by seizing or liquidating a physical asset). It
also found that borrowers are typically required to maintain a minimum cash

81
 balance in a special type of escrow account—known as a “Repayment Reserve
Account” or “Debt Service Reserve Account” (DSRA)—equivalent to one year’s
worth of principal and interest repayments.

At the time that the study was published, there was no hard evidence of
borrowers complying with these escrow account conditions. Nor was there any
hard evidence of Chinese lenders sweeping cash out of these escrow accounts
in order to deal with nonpayment or late payment by overseas borrowers. The
3.0 version of the GCDF dataset provides such evidence.154

​Table 2.3: Illustrative escrow account balances linked to loans from China Eximbank, CDB, and ICBC

Table 2.3
Illustrative escrow account balances linked to loans from
China Eximbank, CDB, and ICBC
Aggregate Cash Balance
Country Lender Borrower of Escrow Accounts Corresponding Loan(s)
(Maximum)
Tanzania China Ministry of $60.3 million $920 million (2012), $275 million
Eximbank Finance, TPDC (2012)
Guinea ICBC Central €76.35 million €559.4 million (2018)
Government
Republic of China Central $338 million ~20 loans under $1.6 billion (2006)
Congo Eximbank Government and $1 billion (2012) framework
agreements
Suriname China Central $9.3 million $98.4 million (2016)
Eximbank Government,
Telesur
Ghana CDB Central $71.2 million $850 million (2012), $150 million
Government (2013), $210.6 million (2019), $185.5
million (2019)
Malawi ICBC Reserve Bank of $32 million $66 million (2021)
Malawi
Myanmar CDB Myanmar Oil $77.1 million (in € €452.7 million (2010)
and Gas equivalent)
Enterprise
Zimbabwe China Ministry of $17.2 million $319 million (2013)
Eximbank Finance, KHPC
Angola CDB Ministry of $1.5 billion $15 billion (2015)
Finance
Kenya China National $250 million $1.9 billion (2014), $1.6 billion (2014)
Eximbank Treasury, KRC

154
Escrow account cash balances are recorded in the 3.0 version of the GCDF dataset’s “collateral” and
“description” fields. Cash sweeps out of escrow accounts recorded in the GCDF dataset’s “description”
field.
82
 Aggregate Cash Balance
Country Lender Borrower of Escrow Accounts Corresponding Loan(s)
(Maximum)
Ecuador CDB Ministry of $113 million $1 billion (2010)
Finance
Ghana China Central $27.2 million $293.5 million (2007)
Eximbank Government
Zimbabwe CDB Econet Wireless $12.4 million $93 million (2014)
Zimbabwe
Botswana ICBC BPC $33 million $825 million (2009)

Zambia CDB DBZ $6 million $30 million (2015)

Notes: This table provides examples of escrow account cash balances linked to loans from Eximbank, CDB,
and ICBC (project ID#59752, 59733, 65116, 65115, 60219, 59273, 55437, 73140, 30578, 58586, 60039,
98520, 34468, 62674, 66847, 31777, 37103, 35865, 183, 62601, 40, 52190 in the 3.0 version of AidData’s
GCDF dataset). The escrow account balance information is drawn from the collateral field and description
field. Escrow account balances vary over time. This table records the maximum observed account balances.

​Table 2.3 provides an illustrative set of escrow account cash balances linked to
China Eximbank, CDB, and ICBC loans in 15 countries. There are several
important points to keep in mind about these balances. First, although there is
some evidence of borrower noncompliance with the escrow account conditions
in their Chinese loan agreements, compliance seems to be the norm rather than
the exception. Borrowers subject to such conditions typically maintain escrow
account cash balances that are sufficient to cover the cost of 1 to 3 semi-annual
principal and interest payments. These amounts are usually equivalent to 5-10%
of the face value of the loan supported by the escrow account. Second, it is not
uncommon for minimum cash balances—and minimum cash balance
requirements—to change based on a loan’s actual or expected amount
outstanding at different points in time over the lifetime of the loan. The cash
balance (requirement) is usually at its lowest point during the grace period when
the loan has not fully disbursed and at its highest point when the loan has fully
disbursed but no repayments have been made. Some, but not all, Chinese
lenders allow borrowers to incrementally reduce the amount of cash in their
escrow accounts as repayments are made and the total amounts outstanding
shrink.155

155
In the 3.0 version of AidData’s GCDF dataset, there is more evidence that CDB follows this practice than
China Eximbank. See, for example, project ID#37103, 55437, 58839, and 58842.
83
Third, notwithstanding the “1 to 3 semi-annual debt service payments” rule of
thumb, minimum cash balance requirements can be adjusted based on the
borrower’s risk profile and/or the lender’s level of exposure.156 Chinese
state-owned creditors may, for example, use a “portfolio-wide approach” to
compensate for a high level of exposure to a risky borrower. China Eximbank’s
collateralized lending arrangement with the Government of Congo-Brazzaville is
a case in point. The borrower is required to keep a minimum cash balance
equivalent to 20% of its total outstanding debt under multiple China Eximbank
loan agreements in an offshore, lender-controlled escrow account (République
du Congo 2018). The 3.0 version of AidData’s GCDF dataset demonstrates that
the Congolese authorities have for the most part complied with this
requirement: the cash balances in their China Eximbank-controlled escrow
account were $338 million in 2017 and $266.6 million in 2020.157 Fourth, while
the amounts of foreign currency that Chinese state-owned creditors ask
borrowers to ring-fence in escrow accounts are not necessarily large enough to
be consequential during normal times, the significance of these funds can
increase during periods of financial distress, when borrowers are strapped for
hard currency and seeking a coordinated debt restructuring with multiple
creditors. Non-Chinese creditors often lack access to foreign currency that is
ring-fenced for their exclusive use and they fear—with some justification—that
Chinese creditors have positioned themselves at the front of the repayment line
by demanding that borrowers grant them access to cash collateral that other
creditors lack (and that can be unilaterally debited in a moment’s notice).
Consequently, they may not be willing to participate in a coordinated debt
rescheduling unless all creditors agree to abide by the so-called “comparable
treatment” principle—i.e., ensure that there is reasonable burden sharing in the
156
Gelpern et al. (2021, 2022) analyze the terms and conditions in 100 loan contracts issued by Chinese
state-owned creditors and 142 foreign loan contracts issued by 28 non-Chinese (commercial, bilateral, and
multilateral) creditors to government borrowers in LICs and MICs. They find that 30% of the Chinese loan
contracts include escrow or revenue account provisions, but only 2% of the non-Chinese loan contracts (one
from AfDB, one from Commerzbank, and one from Agence Française de Développement) include such
provisions. In the rare cases when non-Chinese creditors require government borrowers to maintain
minimum cash balances in escrow accounts, the amounts that they require are similar to the amounts
required by Chinese creditors. There is only one contract in the sample of 142 non-Chinese loan contracts
analyzed by Gelpern et al. (2021, 2022) that clearly specifies a minimum cash balance requirement: a $56.6
million loan agreement between Commerzbank AG Paris Branch and the Government of Cameroon that
requires the borrower to (initially) maintain a escrow account cash balance equivalent to one year’s worth of
principal and interest payments (i.e., two semi-annual debt service payments). The unredacted loan
agreement can be accessed in its entirety via
https://docs.aiddata.org/ad4/pdfs/how_china_lends/CMR_2015_121.pdf.
157
See ID#60219 and #59273 in the 3.0 version of the GCDF dataset.
84
way that financial losses are distributed across creditors (Buchheit and Gulati
2023).158

During our review of the primary sources that underpin the dataset (including
escrow account agreements, repayment mechanism agreements, and the
audited financial statements of borrowing institutions), we also discovered that
Chinese lenders have put in place several different safeguards (or “lines of
defense”) to minimize the risk that these escrow accounts will not fulfill their
intended risk mitigation purposes. The first safeguard is a requirement that the
borrowers initially meet their minimum cash balance requirements by depositing
funds into the repayment reserve (or DSRA) accounts with revenues that are
already at their disposal (rather than project revenues, which are typically
minimal or nonexistent when project loans are first issued). The second
safeguard is automaticity in the way that the repayment reserve (or DSRA)
accounts are replenished after withdrawals have taken place. In a typical escrow
account agreement between a Chinese lender, borrower, and escrow account
bank, if the borrower misses a principal and/or interest payment and the lender
sweeps cash out of the repayment reserve account (in order to satisfy its desire
to be repaid in a timely manner), the escrow account bank is responsible for
immediately replenishing the repayment reserve account with cash from another
escrow account that is often referred to as the “revenue account” or “sales
collection account.” Chinese lenders typically require that their borrowers
deposit all project revenues—or all of the revenues generated by the underlying
infrastructure asset (e.g., a toll road, an airport, a telecommunications network)
supported by the project—into this additional escrow account.159 In some cases,
Chinese lenders will go one step further and require that a fixed percentage of
all revenues of the borrowing institution be deposited in the revenue account.160
The third safeguard is that Chinese lenders usually possess exclusive authority to
freeze the revenue account (without the consent of the borrower) and prevent

158
If there is a perception that China wants to be treated as a senior creditor whose debts need to be given
first priority and other creditors are being pushed to the back of the repayment line, a collective action
failure among creditors (i.e., no coordinated debt rescheduling) becomes more likely (Wigglesworth and Yu
2023; Ferry and Zeitz 2023).
159
See, for example, ID#59753 in the 3.0 version of AidData’s GCDF dataset. Alternatively, the borrower
may be required to deposit into this account cash proceeds from sales under a long-term commodity
purchase agreement. See, for example, ID#35865 in the 3.0 version of AidData’s GCDF dataset.
160
See, for example, ID#55437 in the 3.0 version of AidData’s GCDF dataset.
85
the borrower from making withdrawals from the account.161 The fourth safeguard
is that, in the event of a missed interest or principal payment, Chinese lenders
are entitled under the terms of most escrow account agreements to “pay
themselves” by withdrawing an equivalent amount of cash out of the repayment
reserve account and/or the revenue account (without borrower consent).162

The latest version of the GCDF dataset also provides evidence that, when
borrowers default on their repayment obligations, Chinese lenders do in fact
“pay themselves” by unilaterally making cash withdrawals from the escrow
accounts that they established for risk mitigation purposes. Consider, for
example, the $98.4 million loan that China Eximbank issued to the Government
of Suriname and Telesur—the state-owned telecommunications company of
Suriname—in 2016 for a National Broadband Network Project.163 As a source of
cash collateral, the lender asked its borrower to maintain a minimum balance in
a USD repayment reserve account of $2.9 million (equivalent to total payable
interest for one year or two semi-annual interest payments) during the loan’s
grace period and $9.3 million (equivalent to total payable principal for one year
or two semi-annual principal payments) during the loan’s repayment period. It
also required that Telesur deposit at least 50% of its organizational funding
(including broadband user revenues from the National Broadband Network
Project) in a local currency revenue account. Then, during the COVID-19
pandemic, the Government of Suriname defaulted on its sovereign debt
obligations multiple times in 2020 and 2021. By the end of 2021, the
Government of Suriname had accumulated principal and interest arrears to
China Eximbank worth approximately $61 million (IMF 2021). Then, in February
2022, the Government of Suriname and Telesur missed a scheduled interest
payment on the China Eximbank loan for the National Broadband Network

161
More precisely, Chinese lenders typically possess the exclusive authority to instruct the escrow account
bank to freeze the revenue account and prevent borrower withdrawals. See, for example, ID#59753 in the
3.0 version of AidData’s GCDF dataset
162
By way of example, see ID#37103, #31777, #59753, and #35865 in the 3.0 version of AidData’s GCDF
dataset.
163
On December 30, 2016, China Eximbank and the Government of the Republic of Suriname signed a
$98.4 million preferential buyer’s credit (PBC) agreement for the Suriname National Broadband Network
Project. The loan officially went into effect in November 2017 after several preconditions (so-called
“conditions precedent”) were met by the borrower, including but not limited to the signing of a repayment
mechanism (escrow account) agreement. The loan carries the following borrowing terms: a 15-year
maturity, a 5-year grace period, a 3% interest rate, a 0.4% management fee, and a 0.4% commitment fee.
The Government of Suriname on-lent the proceeds of the loan to Telesur on May 12, 2017. For more
details, see ID#55437 in the 3.0 version of AidData’s GCDF dataset.
86
Project. The lender responded by immediately withdrawing $1.47 million (the
monetary value of one semi-annual interest payment) from the repayment
reserve account in order to cover the cost of the missed interest payment. It also
instructed a local Surinamese escrow account bank (FinaBank N.V) to
immediately (a) block Telesur’s access to the local currency revenue account, and
(b) replenish the repayment reserve account with funds in the revenue
account.164 These actions proved consequential for domestic and international
reasons. The decision to deny the state-owned telecommunications company
access to at least 50% of its organizational funding instigated questions among
local stakeholders about why a foreign lender possessed such extraordinary
authority. The revelation that China Eximbank paid itself by executing a cash
sweep out of an offshore, lender-controlled escrow account also proved
controversial because the Government of Suriname was pursuing a coordinated
debt rescheduling with all of its major external creditors at the time, many of
whom were unaware that China Eximbank had recourse to ring-fenced foreign
currency reserves (and under the impression that the Government of Suriname
was cash-strapped and unable to make loan repayments to any external
creditors).165

Most of these cash sweeps are done in secret. These are exceptionally difficult
to monitor because the lender is debiting cash from an escrow account that is
typically domiciled outside of the borrower country or inside the borrower
country but beyond the immediate reach of domestic oversight
institutions—such as the auditor general and the public accounts committee
within parliament.166 Also, the legal agreements that grant Chinese lenders the
authority to conduct these cash sweeps effectively represent “side agreements.”
164
Under the original escrow account arrangement that was finalized in 2017, Telesur was required to
deposit at least 50% of its funding (including broadband user revenues from the National Broadband
Network Project) in the local currency-denominated revenue account. See ID#55437 in the 3.0 version of
AidData’s GCDF dataset.
165
In March 2022, the IMF reported that the February 2022 “[p]ayment from the repayment reserve account
for the Telesur loan will be reflected in the eventual debt restructuring with [China Eximbank] to ensure
there is comparability of treatment with other official creditors” (IMF 2022).
166
In most sovereign debt transactions, finance ministries are the borrower country counterparts to escrow
account agreements with Chinese creditors. These agreements typically impose expansive confidentiality
obligations on borrowers (finance ministries). By way of illustration, the escrow account agreement for the
China Eximbank loan captured in Project ID#59753 of AidData’s GCDF dataset (Version 3.0) says that “[t]he
obligation of confidentiality shall endure in perpetuity. […] The Parties [to the escrow account agreement]
shall not at any time during the terms of this Deed release any statement to the press or make any other
public statement of any nature which could reasonably be expected to be published in any media
regarding the relationship or the subject matter of this Deed […].”
87
The authority to conduct a cash sweep is almost never identified in loan
agreements between Chinese lenders and their overseas borrowers, which are
more readily accessible to domestic oversight institutions. Instead, loan
agreements with Chinese state-owned creditors typically cross-reference another
agreement—often known as an “account management agreement,” an “escrow
account agreement,” or a “repayment mechanism agreement”—that grants the
lender such authority. Finance ministries rarely disclose these side agreements
to auditors in the executive branch, overseers in the legislative branch, or
international organizations (like the IMF) with surveillance responsibilities unless
they are pressed to do so. These agreements are shrouded in secrecy for a
purpose: collateralizing on cash deposits in lender-controlled escrow accounts is
rare among official creditors, so (sovereign) borrowers are reluctant to disclose
that they have granted these sources of leverage (debt seniority) to Chinese
state-owned creditors but not other official creditors. In the vanishingly rare
instances in which cash sweeps are discovered, it can become more difficult for a
distressed sovereign to get all major creditors to participate in a coordinated
debt rescheduling governed by the “comparable treatment” principle. If some
creditors have recourse to ring-fenced sources of foreign currency and others do
not, any promise by the sovereign borrower to abide by the “comparable
treatment” principle is rendered less credible.167

Risk mitigation strategy #2: Providing short-term cash flow relief in

exchange for escrow account replenishment
Figure 2.11 demonstrates that we have entered a new era of “reschedulings
galore.” The 3.0 version of the GCDF dataset records 98 Chinese debt
reschedulings in 2020 and 2021, which is more than double the number of
Chinese debt restructurings recorded in the Rhodium Group database over the
same time period (Mingey and Wright 2023). Chinese debt reschedulings for
LICs and MICs have increased in tandem with the percentage of loans in China’s
LIC and MIC portfolio that have already reached their (originally scheduled) final
repayment dates (see Figure 2.11). They have also for the most part taken place
in financially distressed countries (see Figures 2.9 and 2.12).

167
Suriname is a case in point.
88
Figure 2.11: Percentage of loans that have reached maturity and number of loans rescheduled

Figure 2.11

Notes: To determine when each loan will reach maturity, each loan’s maturity period is added to its
commitment date. This figure represents when loans reached their final maturity dates according to the
original borrowing terms, although many loans have been rescheduled (often involving an extension of the
loan’s grace period and/or maturity). MOFCOM interest-free loan commitments (which are typically issued
without a credible expectation of repayment) are excluded from the calculation.

89
Figure 2.12: Loan rescheduling for countries in and not in financial distress

Figure 2.12

Notes:To determine if a country experienced financial distress in a given year, we use the binary measure
that is described in Box 1a in Chapter 1. See footnote 141 for details on how the financial distress index
was modified to differentiate between repayment risks and repayment risk mitigation efforts.

These empirical patterns underline the fact that many of China’s overseas
borrowers are insufficiently liquid to meet their repayment obligations and in
need of debt relief. How is Beijing responding to this challenge? The 3.0 version
of AidData’s GCDF dataset demonstrates that Chinese lenders are generally
willing to defer principal and/or interest payments for several years, thereby
providing short-term cash flow relief (i.e., “breathing room”) to their
borrowers.168 However, as yield-maximizing surrogates of the state, Chinese
lenders are ultimately focused on protecting the bottom line (Chen 2020a,

168
There is an important distinction between cash flow relief and debt relief. Chinese state-owned creditors
are substantially more willing to provide cash flow relief than debt relief, which is traditionally defined in the
academic literature as “the reduction in the net present value of the debtor’s outstanding obligations due
to the restructuring agreement” (Horn et al. 2022a: 14). The importance of this distinction came into
sharper resolution after China agreed to participate in the G20’s Debt Service Suspension Initiative (DSSI).
In April 2021, the President of China Eximbank publicly clarified that “debt suspension [...] is neither debt
reduction nor debt forgiveness. One should not take the opportunity [of the Covid-19 pandemic] to harm
China’s interests and take advantage of China” (The Export-Import Bank of China 2021).
90
2020b, 2023; Dreher et al. 2021, 2022).169 As such, they are generally unwilling
to accept significant financial losses in net present value (NPV) terms (Ministry of
Finance of the People's Republic of China 2020; Bon and Cheng 2021; Gardner
et al. 2021; Horn et al. 2022b).170

The 3.0 version of AidData’s GCDF dataset also sheds light on the key role that
escrow account replenishment plays in debt rescheduling negotiations with
Chinese creditors. Once a borrower has defaulted on its repayment obligations
and its Chinese lender has exhausted the funds in a repayment reserve account
(DSRA), the Chinese lender will typically instruct the escrow bank to (a)
immediately replenish the account with funds from the revenue (sales collection)
account, and (b) block the borrower from making any withdrawals from the
revenue (sales collection) account. However, this approach is not foolproof, since
a financially distressed borrower can stop making deposits into the revenue
(sales collection) account, thereby eliminating the Chinese lender’s second line
of defense. Consequently, as more borrowers seek debt relief, Chinese lenders
are demanding a credible protection against (another) default in exchange for
short-term cash flow relief.

The Government of Angola’s 2020 debt rescheduling with China Development

Bank is a case in point.171 In late 2015, CDB issued a $15 billion loan to the
Government of Angola and required that the borrower maintain a minimum
balance of $1.5 billion in an escrow account as a source of cash collateral. Then,
the Angolan Government faced a cash crunch—due to a sharp decline in oil
prices—and had difficulty servicing its dollar-denominated debts. CDB agreed to
defer principal payments as part of a debt reprofiling agreement in 2020 and

169
During debt rescheduling negotiations, Chinese lenders have traditionally provided cash flow relief to
borrowers so long as there was no significant net present value (NPV) reduction in total repayments to the
lender over the lifetime of the loan (i.e., the NPV of debt service payments after rescheduling was not lower
than the NPV of debt service payments before rescheduling). However, change may be afoot. In January
2023, IMF Managing Director Kristalina Georgieva announced that "[w]e have reached an understanding in
principle that China will de facto accept NPV reduction on the basis of significant stretching of the
maturities and reduction of interest. [...] In China there is not yet a consensus to take upfront haircuts”
(Mfula 2023). Reductions in a loan’s NPV can be achieved via substantial interest rate reductions, lengthy
maturity/grace period extensions, and/or face value reductions to loan principal (so-called “face value
haircuts” or “principal haircuts”).
170
Reinhart and Trebesch (2016) study debt relief operations during two periods—1920-1939 and
1978-2010—and find that economic growth generally increased following debt stock reductions (face value
reductions to loan principal) but not debt flow reductions (cash flow relief via maturity/grace period
extensions and interest rate reductions).
171
See ID#95415 and ID#53063 in the 3.0 version of AidData’s GCDF dataset.
91
use the cash in the escrow account to satisfy the borrower’s interest payment
obligations from 2020 to 2022. However, in anticipation of the escrow account
balance being depleted to nearly zero by mid-2022, CDB demanded that the
borrower replenish the account to $1.5 billion by 2023.

Risk mitigation strategy #3: Lending with higher interest rates,

shorter repayment periods, more safeguards, and more severe
penalties for default
One way to go about de-risking an overseas lending portfolio—on a going
forward basis (“future-proofing”)—is to reduce the provision of concessional
credit. Lending on below-market terms is risky from a creditor perspective for
multiple reasons.172 First, any subsidy from a creditor to a borrower implies that
the creditor will achieve a smaller investment return than it would in the absence
of the subsidy. Second, most concessional loans carry low interest rates that are
fixed rather than floating, so a creditor that provides such loans must bear all of
the risk associated with its “cost of funds” increasing over time.173 As the “cost
of funds” increases (for example, due to an increase in interbank lending rates or
an increase in the cost of borrowing via bonds issued on capital markets), the
creditor’s incentive to offer concessional loans with low, fixed interest rates
weakens. Third, concessional loans from official creditors are subject to
substantially larger haircuts (investor losses) than loans from commercial
creditors.174 Therefore, if SAFE—a central government institution with
discretionary authority to entrust surplus foreign exchange to “official creditors”
like China Eximbank or “commercial creditors” like Bank of China (see Box 2b
for more on SAFE’s role)—wanted to de-risk the country’s overseas loan
portfolio, it might seek to minimize future losses (i.e., future-proof the portfolio)

172
If a creditor issues a loan to a borrower at an interest rate that is lower than its own borrowing terms (i.e.,
“cost of funds”) at the time the loan is issued, the creditor is lending to the borrower at a “below-market”
rate. There is significant “cost of funds” variation across Chinese banks and loans denominated in different
currencies (Chen 2020a). With respect to RMB-denominated loans, the state-owned commercial banks
mobilize funds by accepting deposits (with the deposit rate at 1-2%), while the state-owned policy banks
mobilize funds by issuing bonds with yields of 3-5%. With respect to foreign currency-denominated loans,
the state-owned policy banks can borrow from PBOC at a 2-3% interest rate.
173
Relatedly, the bank bears all of the downside risk associated with an increase in the opportunity cost of
funds.
174
Depending on the discounting approach that is used, there is a 19.4 percentage point minimum
difference and 38.6 percentage point maximum difference in average haircuts (investor losses) for official
creditors versus commercial creditors (Schlegl et al. 2019).
92
by reining in the amount of concessional credit provided via official creditors and
prioritizing the provision of non-concessional credit via commercial creditors.175

The 3.0 version of AidData’s GCDF dataset provides evidence that Beijing is
indeed moving in this direction. Figure A18 demonstrates that the weighted
average interest rate of official sector lending from China to LICs and MICs
increased from 4.5% during the early BRI period to 5% during the late BRI
period. Consistent with this shift toward harder lending terms, the weighted
average repayment period (maturity) declined from 11 years during the early BRI
period to 6 years during the late BRI period (see Figure A19). The weighted
average grant element—a summary measure of financial concessionality
(discussed at greater length in Section A-2 in the Appendix)—fell by 9
percentage points between the early BRI period and the late BRI period: from
25% to 16% (see Figure A20).176

In order to gauge whether Beijing is making a concerted effort to rein in the

provision of concessional cross-border credit, it is also worth considering if it has
increased or reduced use of the two primary instruments that it has established
for concessional lending to overseas borrowers: government concessional loans
(GCLs) and preferential buyer’s credits (PBCs) from China Eximbank.177 Figure
A12 provides evidence that GCLs and PBCs have indeed fallen out of favor, as
they represent a rapidly dwindling percentage of new overseas lending
commitments from China to LICs and MICs (18% in 2014 and 3% in 2021).178

175
Some Chinese state-owned lenders participate in debt restructurings that are (loosely) coordinated with
official creditors from other countries, while others do not and wish to be treated as commercial creditors
(Gardner et al. 2021; Horn et al. 2022b). Recent events in Zambia suggest that Beijing considers the claims
of its official creditors to include those of China Eximbank and those of other Chinese state-owned creditors
that are insured (“guaranteed”) by Sinosure. Beijing evidently considers CDB debts that are not backed by
a Sinosure credit insurance policy to be the claims of a commercial creditor (Setser 2023b).
176
The weighted average grant element of Chinese lending to overseas borrowers declined by 19
percentage points—from 30% in 2014 to 11% in 2021 (see Figure A21).
177
GCLs are RMB-denominated loans that are issued to government institutions and provided on
below-market terms (usually 20-year maturities, 5-year grace periods, and 2% interest rates). PBCs are
USD-denominated loans that are issued to government institutions on terms that are more generous than
prevailing market terms, but slightly more expensive (higher interest rates, shorter maturities, and shorter
grace periods) than GCLs (Morris et al. 2020; Horn et al. 2021; Dreher et al. 2022).
178
The fact that these two concessional lending windows are increasingly inaccessible has significant
implications for sovereign borrowers. The terms and conditions in PBC and GCL contracts are fairly
standardized (Gelpern et al. 2022) and reasonably well-understood among debt management officers and
transaction lawyers in LIC and MIC finance ministries (e.g., Banco Central de Bolivia 2016; NEDA 2017;
Economic Relations Division of the Government of the People's Republic of Bangladesh 2023). They
include low (fixed) interest rates, long maturities, generous grace periods, no penalty interest provisions,
and strict requirements to use Chinese contractors and keep all terms and conditions confidential, among
93
During the pre-BRI and early BRI era, GCLs and PBCs were Beijing’s “workhorse”
lending instruments. China Eximbank issued 669 GCLs and PBCs worth $121
billion from 2000 to 2017.179 All of these loans carry low, fixed interest rates and
nearly 100% of them qualify as concessional loans under OECD-DAC
measurement standards.180 However, Figure 2.13 provides evidence of a shift
away from fixed interest rate lending and toward variable interest rate lending:
whereas 60% of new lending commitments from China to LIC and MIC
borrowers carried variable interest rates in 2014, this figure increased to 90% by
2021. These portfolio-level summary statistics suggest that an effort is underway
to transfer more risk from lenders to borrowers. Risk-based loan pricing models
usually rely on a variable interest rate, which in turn is based on a floating market
reference rate—such as the London Interbank Offered Rate (LIBOR) or Euro
Interbank Offered Rate (EURIBOR)—and a premium that accounts for the risk
profile of the borrower. It is also important to keep in mind that SAFE has tasked
Chinese state-owned lenders with maximizing investment returns within
acceptable risk levels, which is more difficult to do via fixed interest lending
when variable (floating market) interest rates are increasing and pushing up the
opportunity cost of funds for lenders (see Box 2b).

other things. However, the terms and conditions that are included in more “exotic” Chinese lending
instruments (like PBOC swap contracts, CDB liquidity support facilities, and accounts receivable financing
arrangements with Chinese state-owned commercial banks) are not well-understood.
179
Between 2018 and 2021, China Eximbank issued 49 GCLs worth $9 billion and 40 PBCs worth $22
billion.
180
99.6% (260 of 261) of the GCLs that were issued between 2000 and 2017 with grant element
observations met the OECD’s 25% grant element threshold of concessionality. 99.6% (247 of 248) of the
PBCs that were issued between 2000 and 2017 with grant element observations met the OECD's 25%
grant element threshold of concessionality.
94
Figure 2.13: Percentage of lending portfolio using fixed or variable interest rates

Figure 2.13

Notes: Variable interest rates that Chinese state-owned creditors use as benchmarks include LIBOR,
EURIBOR, SHIBOR, BADLAR, CIRR, JIBOR, LPR and BADCOR. We exclude all loans for which we cannot
determine if a fixed or variable interest rate was applied.

Together, the shift from dollar-denominated lending to RMB-denominated

lending and the shift from fixed interest rate lending to variable interest rate
lending have resulted in a dramatic turn away from LIBOR-based lending and
toward SHIBOR-based lending. Figure 2.14 demonstrates that the percentage of
China’s variable interest rate lending to LICs and MICs based on LIBOR sharply
declined from 100% in 2008 to 29% in 2021.181 The fall of LIBOR was
accompanied by the rise of SHIBOR. The percentage of China’s variable interest
rate lending to LICs and MICs based on SHIBOR soared from 0% in 2008 to 70%
in 2021.182

181
China's LIC and MIC lending program was almost exclusively dollar-denominated in 2009 (see Figure
2.2).
182
SHIBOR and LPR are both variable interest rates set by the PBOC. Figure 2.14 combines all loans with
SHIBOR- and LPR- based interest rates into a single category. However, LPR is not yet widely used in
China’s overseas lending portfolio. There are only 2 LPR-based loan records in the 3.0 version of the GCDF
dataset. The shift that has taken place is largely a shift from LIBOR to SHIBOR.
95
Figure 2.14: Composition of variable interest rate lending portfolio

Figure 2.14

Notes: LIBOR refers to the London Interbank Offered Rate. SHIBOR refers to the Shanghai Interbank
Offered Rate. LPR refers to the China Loan Prime Rate. The “other” category includes loans with variable
interest rates, such as EURIBOR, BADLAR, CIRR, JIBOR, and BADCOR.

Another important reason why concessionality at the portfolio level has fallen
over time is that some of Beijing’s biggest sovereign borrowers have swapped
less expensive debt for more expensive debt. Horn et al. (2023a) provide
evidence that the interest rates on China’s emergency rescue loans exceed the
interest rates on the existing debt stocks of the borrowers. They also provide
evidence that China’s emergency loans have very short maturities (in many
cases, 1 year or less), although it is not unusual for such loans to be “rolled
over” when they reach their official maturity dates (see Box 2c for more on this
issue).183 As such, refinancing with bailouts from Beijing typically does not
reduce the net present value of a borrower’s debt stock, which highlights an
emerging tension between those providing and those receiving new loans:

183
Rollover debt comes in two varieties: (1) loans that reach their original contractual maturity
dates and secure final maturity date extensions; and (2) loans that are repaid on their original
contractual maturity dates and reissued (with similar or different face values and borrowing
terms) and assigned new maturity dates.
96
financially distressed borrowers want cheap credit that will help them “ride out
the storm,” but Chinese state-owned creditors are unsure if their biggest Belt
and Road borrowers are illiquid or insolvent, so they are pricing in a higher risk
of default when they provide additional credit.

Box 2c: Should emergency liquidity support from PBOC swap lines be treated
as short-term or long-term debt?

Between 2016 and 2021, the PBOC used its swap line network to provide nearly $150 billion in
emergency liquidity support to central banks in LICs and MICs (Horn et al. 2023a). These
borrowings have created a new measurement challenge for organizations that (a) track the
international lending activities of external creditors, and (b) monitor levels of public debt
exposure in the developing world.

The reporting directives of the OECD’s Creditor Reporting System (CRS) specify that “[l]oans with
a maturity of one year or less are not reportable in DAC statistics” (OECD 2021: 51).184 Similarly,
governments that participate in the World Bank’s Debtor Reporting System (DRS) are asked to
report their long-term debt repayment obligations to external creditors on an annual basis.185
Long-term debt is defined in the DRS reporting manual as debt “with an original contractual or
extended maturity of more than one year […]” (World Bank 2000: 4).

Based on a narrow interpretation of the prevailing international reporting rules, PBOC swap line
borrowings should not be reported to the DRS or the CRS. Nearly all of these borrowings carry
de jure maturities of one year or less (i.e., they are initially scheduled for repayment in 12 months
or less). However, central banks that borrow from the PBOC frequently see their final maturity
dates extended—or they repeatedly receive short-term loans to refinance maturing debts. Horn
et al. (2023a) provide evidence that the de facto maturity of the average PBOC swap line
borrowing is 3.5 years.

The custodians of the DRS are aware of the gap between the de jure and de facto maturities of
PBOC swap line borrowings and the underreporting of PBOC swap debt. In October 2020, they
sounded the alarm, questioning whether “currency swap arrangements that represent loans from
other central banks are reflected in external debt stocks of low- and middle-income countries”
(World Bank 2020a: 13). At that time, they clarified that “[t]he DRS […] considers one-year
[central bank] deposits that are consistently rolled over (de facto) to be long-term debt” (World
Bank 2020a: 13). They also emphasized that “[t]he transparency of data on debt must evolve to
keep pace with an ever-changing creditor landscape and with new and increasingly complex
debt-like instruments and data requirements,” and that one of their top priorities is

184
Since 1973, the OECD’s Creditor Reporting System (CRS) has collected and published data on
official sector financial flows (ODA and OOF) from DAC and non-DAC countries.
185
The World Bank’s Debtor Reporting System (DRS) has served as the primary international
reporting system for public debt since 1951. It supports the publication of a widely used data
source: the World Bank’s International Debt Statistics (IDS).
97
“incorporating Central Bank deposits and currency swaps lines into the DRS dataset” (World
Bank 2021: 29).186

The 3.0 version of AidData’s GCDF dataset captures the full range of China’s international rescue
lending operations. Figure A14 demonstrates that an increasing proportion of China’s official
sector lending to LICs and MICs consisted of “rollover” emergency rescue loans during the early
BRI period (8%) and the late BRI period (34%).

Another way of mitigating elevated levels of default risk is to attach stronger

repayment safeguards to new loans. Beijing appears to be following this
strategy. Figure 2.15 demonstrates that the percentage of China’s overseas
lending portfolio supported by collateral increased from 42% during the pre-BRI
and early BRI period (2000-2017) to 56% in the late BRI period (2018-2021). By
2021, nearly three-quarters (72%) of China’s overseas lending to LICs and MICs
was collateralized.

Figure 2.15: Loan portfolio backed by collateral

Figure 2.15

Notes: Collateralized lending commitments are identified with the “collateralized” field in the
3.0 version of the AidData’s GCDF dataset.

186
The custodians of the CRS, by contrast, have not addressed this issue—most likely because China does
not participate in its reporting system.
98
There is also some evidence that, in order to reduce the likelihood of default
and/or minimize post-default losses, Chinese state-owned creditors are
including stiffer penalties for default in their contracts with borrowers. The 3.0
version of AidData’s GCDF dataset includes a new variable that measures the
default (penalty) interest rates that apply to individual loans from Chinese
state-owned entities. Figure 2.16, which draws upon the data, demonstrates that
the average default (penalty) interest rate more than doubled between the early
BRI period and the late BRI period (0.584% from 2014-2017 and 1.175% from
2018-2021).187 The maximum (observed) default (penalty) interest rate also
increased—from 3% during the early BRI period to 8.7% during the late BRI
period.188 These findings are difficult to reconcile with those of SAIS-CARI, which
has concluded that they “see no evidence of penalty interest rates” in China’s
overseas lending to developing countries (Acker et al. 2020: 31).189

Figure 2.16: Early versus late BRI: penalty interest rates

Figure 2.16

Notes: Default (penalty) interest rates are identified with the “Default Interest Rate” field in the 3.0 version
of the AidData’s GCDF dataset.

187
In most Chinese loan agreements that include such provisions, the default (penalty) interest rate applies
to overdue principal and/or overdue interest amounts. See, for example, Section 6.9 of the China Eximbank
buyer’s credit loan agreement for Phase 1 of the Standard Gauge Railway Project (accessible via
https://www.dropbox.com/s/5j3alwun2tv8wk2/SGR%20BCL%202014.pdf).
188
It is also important to keep in mind that the creditor composition of China’s overseas lending portfolio is
shifting away from financiers that use no or low penalty interest rates. Interest-free loans from MOFCOM,
government concessional loans from China Eximbank, and preferential buyer’s credits from China Eximbank
do not carry penalty interest rates—and they are all on the decline (see Figures 2.7 and A12). However,
lending from Chinese state-owned commercial banks carries penalty interest rates in the 0.5% to 3%
range—and it is on the rise (see Figure 2.7).
189
Lest there be any confusion about whether sovereign borrowers are responsible for making penalty
interest payments to their Chinese creditors, the Government of Sudan disclosed that it owed $127 million
of penalty interest to Chinese creditors as of March 31, 2022 (CBOS 2023). The Government of Zimbabwe
and the Government of Serbia have also acknowledged incurring penalty interest under their loan
agreements with China Eximbank (NBS 2007; MOFED 2022).
99
Figure 2.17 zooms in on four sovereign borrowers—Ghana, Uganda, Guyana,
and Serbia—for whom we have reasonably complete data over time on the
default (penalty) interest rates attached to loans from Chinese state-owned
creditors. One can see that, in all of these countries, default (penalty) interest
rates varied between 0-0.33% until 2015. These rates remained mostly stable
from 2016 to 2020, although there was a slight uptick in two countries. However,
by 2021 or 2022, default (penalty) interest rates reached 2% in Serbia, 2% in
Uganda, 3.5% in Guyana, and 4.8% in Ghana. These new risk mitigation
measures by Beijing pose a challenge to borrowers in the developing world that
is rarely taken into consideration: those who continue to contract new debt from
Beijing must be aware of the danger of compounding arrears due to penalty
interest.

100
Figure 2.17: Average penalty interest rates in select countries

Figure 2.17

Notes: This chart shows unweighted average default (penalty) interest rates on loans from official sector
institutions in China to government and state-owned borrowing institutions in Ghana, Serbia, Guyana and
Uganda. The absence of a value in a given country-year indicates missing penalty interest rate data.
Observations are drawn from 2000-2021 GCDF (Version 3.0) data and preliminary 2022 GCDF data. In
addition to loan commitments, pledged loans and suspended/canceled loan commitments are included
since borrowers may be more likely to reject loan offers with high penalty interest rates and suspend or
cancel loan commitments with terms that they perceive to be onerous.

101
The 3.0 version of AidData’s GCDF dataset also provides evidence that when
borrowers fail to honor their repayment obligations, Chinese lenders will seek to
recover the penalty interest that they are owed by sweeping cash out of escrow
accounts (when they have recourse to such collateral). By way of illustration,
consider how China Eximbank responded to an overdue debt repayment from
Tanzania Petroleum Development Corporation (TPDC)—a parastatal that is
wholly-owned by the Government of Tanzania—for the Songo Songo to Dar Es
Salaam Gas Pipeline and Natural Gas Processing Plants at Mnazi Bay and Songo
Songo Project. As a source of collateral for a $275 million buyer's credit loan
(BCL) that China Eximbank issued in 2012, TPDC deposited approximately $60
million in escrow accounts accessible to the lender.190 However, by the first half
of 2017, TPDC had accumulated arrears to China Eximbank. The lender
responded by sweeping cash out of one of the escrow accounts between July
2017 and June 2018. According to TPDC’s audited financial statements, it
withdrew the funds “as a penalty for [the borrower’s] late repayment of due
installment” (The Controller and Auditor General of the National Audit Office of
Tanzania 2018: 76).191

Risk mitigation strategy #4: Taking a differentiated approach with

borrowers that present high and low levels of repayment risk
Previous research indicates that when creditors are confronted with a sudden
wave of defaults or a sharp increase in non-performing loans, they may seek to
rebalance the risk profiles of their asset portfolios by issuing fewer loans to
potentially risky borrowers—or by pulling back from an entire asset class with a
new risk profile (Kaminsky and Reinhart 2000; Longstaff et al. 2011; Brooks et al.
2015; Gilchrist et al. 2022). Consistent with this expectation, the 3.0 version of
AidData’s GCDF dataset provides evidence of a compositional shift in

190
According to TPDC’s audited financial statements, the escrow accounts “were opened to secure
repayment of principal and payment of interest and fees under the loan agreements” and the minimum
cash balances in these accounts functioned as sources of “collateral.”
191
In June 20, 2012, China Eximbank and Tanzania’s Ministry of Finance signed a $1,225,327,000 financing
agreement for the Songo Songo to Dar Es Salaam Gas Pipeline and Natural Gas Processing Plants at Mnazi
Bay and Songo Songo Project. Two loans were issued to Tanzania’s Ministry of Finance: (1) a $275 million
buyer’s credit loan (BCL) with an interest rate of 6-month LIBOR plus 430 basis points, a 1.83-year
(22-month) grace period, and a 12.83-year (154-month) maturity, and (2) a $920 million preferential buyer’s
credit (PBC) with a 20-year maturity, a 7-year grace period, and a 2% interest rate. The BCL and PBC
proceeds were then on-lent from Tanzania’s Ministry of Finance to the Tanzania Petroleum Development
Corporation (TPDC). See ID#59733 and 59752 in the 3.0 version of AidData’s GCDF dataset.
102
non-emergency lending from public sector to private sector borrowers. This
move away from the sovereign debt “asset class” is particularly noticeable
between 2020 and 2021—when the effects of the COVID-19 pandemic were
most acute. The share of China’s non-emergency lending commitments to
private sector borrowers in LICs and MICs soared from 21% in 2020 to 54% in
2021, while the share devoted to public sector borrowers shrank from 67% in
2020 to 43% in 2021 (see Figure 2.18).192

192
Here we define public sector loans as the sum of central government debt, central
government-guaranteed debt, and other public sector debt (as described in Section A-5 in the
Appendix), which is consistent with the IDS definition of PPG debt. In the Appendix, we replicate
this graph for China’s total (emergency and non-emergency) lending commitments to borrowers
in LIC and MICs (see Figure A10). The incorporation of China’s emergency lending commitments
disguises the pivot away from public sector lending and toward private sector lending (as
emergency lending commitments are large and exclusively channeled to sovereign borrowers).
103
Figure 2.18: Composition of non-emergency lending portfolio

Figure 2.18

Notes: This graph shows the composition of China’s non-emergency lending portfolio (as measured in 2021
constant USD) in LICs and MICs according to the extent to which host governments may eventually be
liable for debt repayment. Central government debt and other public sector debt represent loans where
the borrower is a government agency or a wholly- or majority-owned state entity. Central government debt
represents loans that have a sovereign guarantee from the host government. Potential public debt
represents loans to entities (including special purpose vehicles or joint ventures) where the host
government has a minority stake. Private debt captures loans to private entities.

Another potential de-risking strategy is to take a differentiated approach across

borrower countries that present varying levels of repayment risk. Figures A22
and A23 provide evidence that suggests Beijing is adopting this approach:
Chinese state-owned creditors have lent to financially distressed countries at
higher interest rates and with shorter repayment periods than countries that are

104
not in financial distress. Consequently, the concessionality level (weighted
average grant element) of official sector lending from China is consistently lower
in countries experiencing financial distress than in countries not experiencing
financial distress (see Figure A24).193

Figure 2.19: Percentage of variable interest rate lending to countries in and not in financial distress

Figure 2.19

Notes: Variable interest rates include LIBOR, EURIBOR, SHIBOR, BADLAR, CIRR, JIBOR, LPR and BADCOR.
To determine if a country experienced financial distress in a given year, we use the binary measure that is
described in Box 1a.

Chinese state-owned creditors have also changed the way that they lend to
financially distressed countries over time by shifting toward variable interest rate
lending (see Figure 2.19). This approach to lending follows a risk-based pricing
model by adding a borrower-specific margin—that accounts for the credit profile
characteristics of the borrower—to a market-based reference interest rate.194 In

193
Between 2000 and 2021, the weighted average grant element of official sector lending from China to
countries experiencing financial distress was ten percentage points lower than in countries not experiencing
financial distress (see Figure A25 in the Appendix).
194
Risk-based loan pricing models (a) charge risky borrowers higher prices (i.e., attach larger risk premia to
borrowers that present a high probability of default), and (b) account for the expected magnitude of
post-default losses (i.e., how much money the creditor expects to lose if the borrower defaults).
105
the early 2000s, there was no evidence whatsoever of any variable interest rate
lending by Chinese state-owned creditors to LICs or MICs in financial distress.
However, by 2021, more than 80% of China’s variable interest rate lending was
directed to countries in financial distress.

The 3.0 version of AidData’s GCDF dataset also reveals that China is increasingly
collateralizing loans to countries in dire financial straits: the share of China’s
collateralized lending portfolio directed to countries in financial distress
increased from zero at the turn of the century to 74% by 2021 (see Figure
2.20).195 This change is part of a broader pattern in China’s use of credit
enhancements, which AidData defines as lending backed by a credit insurance
policy, a third-party repayment guarantee, and/or collateral. Figure A29 divides
China’s LIC and MIC lending portfolio into three categories: (1) credit-enhanced
lending to countries in financial distress; (2) credit-enhanced lending to countries
not in financial distress; and (3) lending that is not credit-enhanced. It shows
almost no proportional increase in credit-enhanced lending to countries that are
not in financial distress, but a large proportional increase in credit-enhanced
lending to countries that are in financial distress.

195
Figure 2.20 shows the proportion of official sector lending commitments from China (in constant 2021
USD) to LICs and MICs that falls into three categories: (1) loans that are collateralized and are going to a
borrowing country that is in distress at the time of the loan commitment, (2) loans that are collateralized but
are going to a borrowing country that is not experiencing financial distress at the time of the loan
commitment, and (3) loans that are not collateralized at the time of commitment. The underlying data from
Figure 2.20 demonstrate that, while 51% of China's collateralized lending commitments to LICs and MICs
were directed to developing countries in financial distress during the pre-BRI period (2000-2013), this figure
increased to 82% in the early and late BRI periods (2014-2021).
106
Figure 2.20: Composition of lending to countries in and not in financial distress

Figure 2.20

Notes: To determine if a country experienced financial distress in a given year, we use the binary measure
that is described in Box 1a.

Given that the large-scale provision of emergency rescue loans to financially

distressed borrowers has increased Chinese bank exposure to repayment risk,
another potential portfolio rebalancing strategy is to direct project (or
nonemergency) loans to less risky borrowers. As a preliminary test of whether
Chinese state-owned creditors may be responding in this way, we first measure
the percentage of China’s emergency rescue loan portfolio supporting borrower
countries with credit ratings in “C and D territory”—i.e., countries that Moody’s,
Standard & Poor’s, and Fitch identify as in default or presenting a high risk of
default. We then measure the percentage of China’s project loan portfolio
supporting borrower countries with credit ratings in C and D territory. Figure
2.21 provides these summary statistics, which demonstrate that the overall credit
quality of China’s project loan portfolio is substantially better than that of its
emergency rescue loan portfolio.196

196
Project loans are defined as those in the 3.0 version of AidData’s GCDF dataset that are categorized as
investment project loans (IPLs).
107
Figure 2.21: Repayment risk in emergency lending portfolio versus project lending portfolio

Figure 2.21

Notes: Countries in default or at a high risk of default represent LICs and MICs with scores of 5 or less on
the sovrate index (see Box 1a). Countries not at a high risk of default represent LICs and MICs with scores
above 5 on the sovrate index. The World Bank’s sovrate index is a measure of repayment risk that varies
from 0 to 21, with higher scores indicating lower levels of sovereign credit risk (Kose et al. 2022).
Country-year observations without official sector Chinese lending commitments or sovrate scores are
excluded from the figure. MOFCOM interest-free loan commitments (which are typically issued without a
credible expectation of repayment) are excluded from the calculation.

We also find evidence of portfolio rebalancing over time. According to Figure

A30, China reduced the percentage of its non-emergency loan portfolio
supporting borrower countries with credit ratings in C and D territory—from
7.9% during the early BRI period to 0.8% during the late BRI period. One
potential way of shifting the non-emergency (and project) lending portfolio away
from countries that present high levels of repayment risk is via syndication and
multilateralization. Figure 2.22 demonstrates that when Beijing outsources risk
management (including the loan origination process), less credit is channeled to
countries that are in default or present a high probability of default. However, is
there any evidence that Beijing is actually pivoting away from its bilateral
lending institutions and toward multilateral lending institutions and syndicated
loan arrangements? This is the next question that we attempt to answer.

108
Figure 2.22: Repayment risk in bilateral, multilateral, and syndicated lending portfolio

Figure 2.22

Notes: Countries in default or at a high risk of default represent LICs and MICs with scores of 5 or less on
the sovrate index (see Box 1a). Countries not at a high risk of default represent LICs and MICs with scores
above 5 on the sovrate index. The World Bank’s sovrate index is a measure of repayment risk that varies
from 0 to 21, with higher scores indicating lower levels of sovereign credit risk (Kose et al. 2022).
Country-year observations without official sector Chinese lending commitments or sovrate scores are
excluded from the figure. MOFCOM interest-free loan commitments (which are typically issued without a
credible expectation of repayment) are excluded from the calculation.

Risk mitigation strategy #5: Scaling down bilateral lending

operations and scaling up lending operations via syndication and
multilateralization
If China’s bilateral development finance institutions do not have sufficiently
strong risk management guardrails in place, one option is to reform these
institutions from within. However, these institutions are politically powerful and
slow to change (Chen 2020a, 2020b, 2023). Another option to improve the risk
profile of the country’s overseas lending program is to dial down the provision of
credit through bilateral channels and dial up the provision of credit through
syndicated and multilateral lending channels.

Syndicated loans allow a group of lenders (a “syndicate”) to pool their funds and
share credit risk. When a transaction is financed through a syndicated
arrangement, all members of the syndicate must agree to a common set of
contractual terms and conditions, including the standards and safeguards that
will govern the transaction. Given that each lender has its own standards and
safeguards, the members of a syndicate can seek to reconcile (hybridize) their
109
respective standards and safeguards. However, it is more common for the
members of the syndicate to defer to the standards and safeguards of one
member of the syndicate. In most syndicated loan arrangements, a “lead
arranger”—sometimes referred to as the “arranging lender” “lead lender,”
“lead manager,” or “underwriter”—establishes the transaction’s key terms and
conditions, which cannot be amended without the consent of all members of the
syndicate.197 It is also customary for all members of the syndicate to use a
common set of due diligence standards to vet the borrowing institution and the
proposed transaction.198 Here again, the lead arranger is usually responsible for
identifying and applying the due diligence standards (Dennis and Mullineaux
2000; Ivashina 2009).

Figure 2.23: Composition of non-emergency lending portfolio by channel of delivery

Figure 2.23

Notes: All emergency rescue loans are excluded. The “loans that involve multilateral institutions” category
include loans where a multilateral agency has some formal role, such as an entrusted loan agreement or a
co-financing arrangement.

197
However, other terms and conditions can usually be amended with the consent of the “majority
lenders.”
198
Sufi (2007) demonstrates that lead arrangers reduce the costs of due diligence for all other syndicate
participants.
110
Figure 2.23 provides evidence that, at the turn of the century, China’s
non-emergency lending program in LICs and MICs exclusively consisted of
bilateral loans—i.e., loans issued by a single lender to a single borrower.
However, over time, Beijing has moved away from this approach, ramping up its
use of syndicated loan arrangements. It began experimenting with this more
collaborative way of issuing credit during the pre-BRI period and early BRI
period, but made syndication central to the country’s overseas lending strategy
during the late BRI period. By 2021, 50% of China’s non-emergency lending
program in LICs and MICs consisted of syndicated loan commitments.199

This empirical pattern flies in the face of the conventional wisdom that
“[e]merging economies’ debt to China is [...] non-marketable” (Kondo et al.
2022).200 That was certainly true 25 years ago, but it is no longer the case: half of
China’s non-emergency lending portfolio in the developing world now consists
of syndicated loans (Figure 2.23).

China's state-owned commercial banks are particularly focused on this approach

to cross-border lending. The 3.0 version of AidData’s GCDF dataset
demonstrates that they are more heavily engaged than the country’s policy
banks in syndicated lending to LICs and MICs. In 2021, 84% of China’s
state-owned commercial bank lending to LICs and MICs relied on syndicated
loan instruments and the remaining 16% relied on bilateral loan instruments. By
comparison, only 36% of China’s policy bank lending to LICs and MICs relied on
syndicated loan instruments and the remaining 64% relied on bilateral loan
instruments.

Another important aspect of the country’s pivot from bilateral lending to

syndicated loans is the growing involvement of non-Chinese banks and
multilateral institutions in the syndicates that are being established. Figure A32
demonstrates that roughly 80% of the syndicated loans from official sector

199
49% of China’s non-emergency lending program in LICs and MICs consisted of bilateral loan
commitments in 2021. Beijing’s pivot toward lending via syndicated loans is especially noticeable in the
project lending portfolio (see Figure A36 for supplementary evidence).
200
A syndicated loan constitutes marketable debt in that the arranging lender responsible for originating
and structuring the transaction can distribute/sell part of the loan to other banks and nonbank institutions
through a marketing and syndication process and the loan can be traded on secondary markets.
111
creditors in China to LICs and MICs involve non-Chinese bank participants.201 A
non-trivial percentage of these loans also involve multilateral institutions (see
Figure A33).202 Multilateral and non-Chinese bank participation could have
far-reaching consequences if their standards and safeguards prevail as the ones
that all other syndicate members must follow.203 In Table 2.4, we provide
metadata for an illustrative set of syndicated loan agreements involving Chinese
state-owned creditors and non-Chinese creditors. One can see that syndicated
loans with Chinese and non-Chinese participants frequently rely on Western
commercial banks and multilateral institutions to serve as lead arrangers, which
is consistent with the notion that Beijing is de-risking its overseas loan portfolio
by outsourcing risk management.204 In Chapter 4, where we address this issue at
greater length, we find that Chinese participation in syndicated loan agreements
with non-Chinese banks and multilateral institutions consistently results in
stronger rather than weaker risk management standards and safeguards.

Table 2.4: Illustrative set of syndicated loan agreements with Chinese and non-Chinese bank participants

Table 2.4
Illustrative set of syndicated loan agreements with Chinese
and non-Chinese bank participants

Total
Value of Chinese
Country Project Year Lead Arranger Number of
Loan Participants
Participants
Sierra Leone Port Elizabeth II 2017 $659 ICBC ICBC, China 2
Upgrading and million Eximbank
Expansion Project

201
Although there are more syndicated loans with Chinese and non-Chinese participants (833) than there
are syndicated loans with exclusively Chinese participants (312), syndicated loans with Chinese and
non-Chinese participants tend to be smaller ($138.8 million on average) than syndicated loans with
exclusively Chinese participants ($545.5 million on average). These summary statistics are drawn from
Figures A34 and A35.
202
When one accounts for multilateral participation in syndicated loans and entrusted loan agreements with
multilateral institutions, the multilateralization of China’s LIC and MIC lending portfolio comes into sharper
resolution (see Figure A28).
203
We provide evidence that suggests this is indeed the case in Chapter 3.
204
With an earlier vintage of the GCDF dataset, Joosse et al. (2023) use social network analysis to map the
international network of non-Chinese financiers that facilitate the participation of Chinese state-owned
creditors in syndicated loan agreements. They conclude that the ten most important network “brokers”
include two multilateral institutions (Inter-American Development Bank and African Development Bank) and
eight Western commercial banks (BNP Paribas, Standard Chartered, Sumitomo Mitsui, MUFG, Citibank,
ING, Deutsche Bank, and HSBC).
112
 Total
Value of Chinese
Country Project Year Lead Arranger Number of
Loan Participants
Participants
Iraq Basrah Natural Gas 2021 $260 International ICBC, Bank of 9
Liquids Facility million Finance China
Construction Project Corporation (IFC)
Kazakhstan Almaty Ring Road 2020 $585 European Bank for Bank of China 5
Project million Reconstruction and
Development
(EBRD)
Argentina La Castellana Wind 2017 $64.05 International SAFE through 2
Power Project million Finance the Managed
Corporation (IFC) Co-lending
Portfolio
Program
Bangladesh Unit 3 of 220 MW 2017 $196.7 Standard Chartered Bank of China 4
Sirajganj Combined million Bank
Cycle Power Plant
Project
Notes: This table provides examples of syndicated loans with Chinese state-owned participants (ID#62223,
62224, 95921, 92613, 98022, and 69033 in the 3.0 version of AidData’s GCDF dataset). The “Year” column
captures the financial commitment year. The “Value of Loan” column captures the aggregate monetary
value of all syndicated loan tranches/contributions. The “Chinese Participants” column captures all official
sector Chinese participants in the syndicate. The “Total Number of Participants” column captures the total
number of Chinese and non-Chinese creditors that participated in the syndicate.

Figure 2.23 above highlights another interesting trend: a modest increase in

Chinese lending that is channeled via multilateral institutions during the BRI era
(2014-2021). During the five year period that preceded the BRI (2009-2013), 2%
of China’s non-emergency lending to LICs and MICs was channeled via
multilateral institutions. This figure doubled (to 4%) during the first eight years of
BRI implementation (2014-2021).205

One of China’s initial forays into entrusted lending via multilateral institutions
began in 2013 and 2014, with the creation of the Africa Growing Together Fund
(AGTF) at the African Development Bank, the China Co-financing Fund for Latin
America and the Caribbean at the Inter-American Development Bank, and the
Managed Co-Lending Portfolio Program (MCPP) at the International Finance
Corporation (IFC). The MCPP, which was launched by the IFC in partnership with
China’s State Administration for Foreign Exchange (SAFE), helps illustrate the
logic of an entrusted loan agreement with a multilateral institution. Rather than
directly lending to borrowers in developing countries, SAFE entrusted $3 billion
to the IFC and “leverage[d] IFC’s project pipeline and due diligence skills to [...]

205
Figure 2.23 excludes emergency loans since they are exclusively provided via bilateral channels.
113
co-lend to projects or groups of projects alongside IFC on commercial terms”
(World Bank Group 2020: 35). Beijing’s decision to outsource loan management
to a multilateral institution was evidently motivated by a desire for stronger
safeguards and attractive investment returns (see Box 2b for more on SAFE’s
mandate to maximize investment returns on the country’s surplus foreign
exchange reserves). An ex-post evaluation of the MCPP concluded that “[m]ost
of these borrowers [had] the capacity to meet their financial commitments and
[were] less vulnerable to nonpayment than other speculative projects" (World
Bank Group 2020: 36).

Think tanks, research institutions, and media outlets in G7 countries have in

recent years trained their sights on China’s bilateral lending activities—in
particular, infrastructure projects financed by China Eximbank and China
Development Bank (e.g., Dyer et al. 2011; Sanderson and Forsythe 2013;
Gallagher and Ray 2020; Myers and Ray 2023; Ray 2023; Do Rosario and Savage
2023). Yet there has been virtually no discussion about Beijing’s strategic pivot
away from bilateral lending and toward collaborative lending arrangements
involving Chinese and non-Chinese banks. The reason why this change in
China’s overseas lending portfolio has not received serious attention is simple:
Chinese bank participation in syndicated loans (with non-Chinese participants)
and loans entrusted to multilateral institutions have gone undetected because
they are mostly excluded from other publicly available Chinese development
finance datasets.206 Beijing has not been especially secretive about this change
in the composition of its overseas lending strategy. It was hiding in plain sight;
those seeking to track China’s overseas lending activities either did not detect it
or did not think it was important enough to track.

206
Here again, one is reminded of the proverbial drunkard who insists upon searching for his keys beneath
the lamppost “because that’s where the light is.” Chinese creditor contributions to syndicated loans are not
systematically tracked in the China’s Overseas Development Finance Database, the Chinese Loans to Latin
America and the Caribbean Database, the Chinese Loans to Africa Database, the China’s Global Energy
Finance Database, the China Overseas Finance Inventory Database, or the World Bank’s International Debt
Statistics (IDS). However, Figure A32 demonstrates that official sector lending from China is increasingly
channeled via bank syndicates that include Chinese and non-Chinese banks.
114
There are several reasons why Beijing’s pivot away from bilateral lending
instruments is crucial to understand its de-risking strategy.207 First, since most of
the syndicated loans in question involve non-Chinese bank participants (see
Figure A32), Beijing is increasingly able to outsource risk management to
lending institutions with stronger rules and standards.208 Second, multilateral
institutions have particularly strong risk management guardrails in place (see
Chapter 3), so the use of entrusted loan agreements with multilateral institutions
necessarily involves the application of a more stringent set of safeguards. These
are effectively de-risking “shortcuts.” Bilateral development finance institutions
have rules and standards that have evolved over decades via accretion. It is
unlikely that these institutions are going to dispense with these safeguards or
dramatically change them in the short-run. So, delegating borrower selection
and loan preparation to a credible third party is an attractive shortcut to de-risk a
loan portfolio (on a going forward basis).209 Second, participation in a syndicated
loan agreement or an entrusted loan agreement with a multilateral development
bank (MDB) is a rapid and reliable de-risking strategy because MDBs enjoy de
facto preferred creditor status (Schlegl et al. 2019) and confer this benefit to all
other lenders that participate in their syndicated loan agreements and entrusted
loan agreements (Gurara et al. 2020).210 An added benefit of this approach is
that loans involving MDBs are generally exempt from rescheduling and shielded
from large haircuts (investor losses).211 From a risk mitigation perspective,
syndicated loan agreements also have two perversely attractive features to
official creditors: (1) unanimous consent requirements can make them more
difficult and time-consuming to reschedule (Buchheit 1985, 1991; Gelpern

207
As we explain at greater length in Chapter 3, Beijing has positioned its bilateral lending institutions as
alternative sources of financing for LICs and MICs that would prefer not to deal with OECD-DAC donors or
multilateral development banks (Parks 2019; Malik et al. 2021; Dreher et al. 2022). Therefore, the pivot
toward syndication and multilateralization represents a major departure.
208
In November 2017, the country's top banking regulator—the China Banking Regulatory Commission
(CBRC)—called upon Chinese state-owned banks to “prevent and control overseas business risks by taking
risk-sharing measures” (Xueqing 2017, emphasis added). The shift toward syndication and
multilateralization during the late BRI period (highlighted above in Figure 2.23) may constitute evidence of
bank responsiveness to CBRC’s policy guidance.
209
On this point, see Dennis and Mullineaux (2000).
210
In a typical A/B syndicated loan arrangement involving an MDB, the lender-of-record is the MDB; it
keeps a part of the loan for its own account (the “A-loan”) and it sells participation in the remainder of the
loan (the “B-loan”). As the lender-of-record, the MDB confers its status as a de facto preferred creditor to
all B-loan participants.
211
On this point, see Cordella and Powell (2021).
115
2016); and (2) they are generally subject to smaller haircuts (financial losses) than
bilateral loans (Cruces and Trebesch 2013; Schlegl et al. 2019).212

Section 4: What have we learned about Beijing’s bid to

reboot its overseas lending program?
In this chapter, we have presented evidence that challenges the conventional
wisdom about the total collapse of China’s overseas lending program. With
more complete data on the full range of China’s lending activities in LICs and
MICs, we have shown that Beijing remains the world’s largest official creditor: in
nominal terms, the world owes China about $2.6 trillion and the developing
world owes China at least $1.1 trillion and perhaps as much as $1.5 trillion. We
have also shed light on the ways in which Beijing is rebalancing its cross-border
credit portfolio—much like a yield-maximizing investment portfolio manager that
is navigating an environment in which a growing number of borrowers are
illiquid or insolvent. Beijing has ramped down the provision of long-term,
dollar-denominated bilateral loans to sovereign borrowers for public investment
projects, while at the same time ramping up the provision of RMB-denominated
emergency rescue loans that are short- or medium-term in nature. It has reduced
its reliance upon the policy banks, while making greater use of state-owned
commercial banks, the central bank, syndicated loan arrangements with
non-Chinese banks, and multilateral administrators.

What can we say, by way of conclusion, about Beijing’s efforts to de-risk its
overseas lending portfolio and its determination to ensure that LIC and MIC
borrowers repay their debts? The evidence at hand does not suggest that it is
ready to take financial losses in order to minimize diplomatic blowback and
reputational damage.213 Quite the opposite: Beijing appears to be stiffening its
resolve and preparing for a long and difficult slog. It is sweeping cash out of the
escrow accounts of its overseas borrowers, requiring that borrowers replenish

212
Bilateral loans from official creditors are typically rescheduled through the Paris Club, while syndicated
loans are typically rescheduled through London Club reschedulings (regardless of whether the syndicate
members include official creditors or commercial creditors).
213
Gong Chen, the founder of Anbound (a Beijing-based think tank) and a BRI adviser to the central
government, recently told Nikkei Asia that “widespread debt evasion and avoidance [by BRI participants]
would have a significant impact on China's financial stability” and “we are concerned that some countries
may try to avoid paying back their debt by utilizing geopolitics and the ideological competition between
East and West” (Aamir et al. 2022).
116
escrow accounts in exchange for short-term cash flow relief, introducing stronger
penalties for late repayments, and channeling emergency rescue loans with high
interest rates and short repayment periods to financially distressed borrowers (to
make sure they have enough cash on hand to service their existing infrastructure
project debts). Only time will tell if Beijing has enough “steel in its spine” to stay
the course, but its actions to date suggest that it intends to do whatever it takes
to protect the bottom line.

A final point bears emphasis. We freely concede that this chapter has only
scratched the surface of what can be done with the 3.0 version of AidData’s
GCDF dataset to uncover new insights about Beijing’s contemporary lending
activities and practices in the Global South. The dataset can—and should—be
used to answer an array of additional questions, such as:

● How have the terms and conditions in Chinese loan contracts changed (or
not) since Beijing endorsed the Common Framework in November 2020?

● What are the consequences of borrower non-compliance with escrow

account conditions?

● How do Chinese creditors select their preferred partners in syndicated

loan agreements?

● Why does Beijing prefer to channel loans through some multilateral

institutions rather than others?

● What are the supply-side and demand-side drivers of China’s emergency

lending program? Has this changed over time, and if so, how and why?

● What are the supply-side and demand-side drivers of China’s

non-emergency lending program? Has this changed over time, and if so,
how and why?

Although there are commercial incentives to put the GCDF dataset behind a
paywall and professional incentives to withhold release until it is introduced in
leading, peer-reviewed academic journals, we remain fully committed to the

117
principle that all past, present, and future versions of the GCDF dataset should
be treated as public goods rather than private goods.214 Our hope is that the
dataset will catalyze a knowledge multiplier effect and facilitate evidence-based
decision-making.215

214
In this respect, we are carrying forward the “open research” tradition of the original, interdisciplinary
group of researchers who developed the Tracking Underreported Financial Flows (TUFF) methodology that
underpins AidData’s GCDF dataset. See Chapter 3 of Dreher et al. (2022).
215
If past is prologue, we expect that this approach will deliver a significant payoff. To date, the GCDF
dataset has been used in more than 500 research publications (Wooley 2023).
118
Chapter 3: Redesigning the Belt and Road for
Safety and Speed

Section 1: Beijing’s journey from skeptic to advocate of ESG

risk management
For the better part of the last two decades, Beijing sought to position itself as an
alternative source of infrastructure financing for governments that would prefer
not to deal with OECD-DAC donors or multilateral development banks. It
characterized itself as a demand-driven financier of South-South cooperation
and tacitly encouraged its foreign counterparts to bypass the rules and
standards of OECD-DAC donors and multilateral development banks by
fast-tracking large-scale infrastructure projects with as little “red tape” and
“hassle factor” as possible. It offered fast and flexible project preparation
procedures; a streamlined set of loan and grant approval processes; the ability
to issue contracts without competitive bidding requirements; and less stringent
environmental, social, and governance (ESG) safeguards.216

China’s value proposition was compelling. Between 2000 and 2021, 140 LICs
and MICs accepted $825 billion of aid and credit from Beijing for 4,800
infrastructure projects.217 China became the developing world’s go-to banker for
big-ticket infrastructure projects because it demonstrated three comparative
advantages vis-à-vis OECD-DAC donors and multilateral development banks
(MDBs): scale, speed, and impact. Beijing bankrolled large-scale infrastructure
projects that its peers and competitors were unwilling or unable to support.218 It
financed 1,385 infrastructure projects with grants and loans worth $100 million

216
At the same time, Beijing spurned nearly all invitations to follow the prevailing set of international
development finance rules and norms, and it admonished Western donors and multilateral lenders for their
“one-size-fits-all” policies (Malik et al. 2021; Dreher et al. 2022).
217
These figures only refer to active and completed projects. Between 2000 and 2021, Beijing also issued
grants and loans worth $56 billion for 94 infrastructure projects in 49 LICs and MICs that were subsequently
suspended or canceled.
218
China was faster than its competitors at finalizing loan agreements for large-scale infrastructure projects.
In 2008, Senegal’s then-President, Abdoulaye Wade, wrote in the Financial Times that “with direct aid,
credit lines and reasonable contracts, China has helped African nations build infrastructure projects in
record time. … I have found that a contract that would take five years to discuss, negotiate and sign with
the World Bank takes three months when we have dealt with Chinese authorities” (Wade 2008).
119
or more between 2000 and 2021.219 China also earned a reputation for
implementing brick-and-mortar projects with lightning speed: the average
Chinese government-financed infrastructure project between 2000-2021 took
only 2.7 years to complete.220 Similar projects financed by the World Bank and
regional development banks usually took 5-10 years to complete (Bulman et al.
2017; Lagarda et al. 2018; Duggan et al. 2020; World Bank 2023c). China’s
overseas development projects—in particular, those involving the construction
of infrastructure such as roads, bridges, tunnels, railways, and ports—also
generated significant economic benefits in a politically relevant timeframe.221 In
an evaluation of 4,304 projects in 138 LICs and MICs over a 15-year period,
Dreher et al. (2021) find that the average project increased economic growth by
0.95 percentage points two years after securing funding approval from
Beijing.222 Bluhm et al. (2020) provide evidence that China’s connective
infrastructure project portfolio was especially effective at promoting
spatially-inclusive economic development—by decentralizing economic activity
within the provinces and districts where they were implemented.223

219
Between 2000 and 2021, Beijing financed 735 infrastructure projects with grants and loans worth at least
$250 million in 89 LICs and MICs, 377 infrastructure projects with grants and loans worth at least $500
million in 68 LICs and MICs, and 163 infrastructure projects with grants and loans worth at least $1 billion in
47 LICs and MICs. All of these figures include infrastructure projects that secured grant or loan
commitments from China, which were subsequently suspended or canceled.
220
With the 3.0 version of AidData’s GCDF dataset, we calculate the average amount of time needed to
complete a Chinese grant- or loan-financed infrastructure project by measuring the average number of
calendar days between actual project implementation start dates and actual project completion dates. We
only include active projects and completed projects that secured official commitments from China in our
analysis.
221
This finding implies that if a host country chose to accept three additional Chinese ODA or
OOF-financed development projects, it could reasonably expect to boost its economic (GDP) growth by
2.85 percentage points within two years of Beijing agreeing to bankroll the projects. For more on the
socioeconomic impacts of Chinese grant- and loan-financed development projects, see Bluhm et al. (2020),
Martorano et al. (2020), Dreher et al. (2022), Mandon and Woldemichael (2023), and Wellner et al.
(forthcoming, 2023).
222
More specifically, Dreher et al. (2021a) find that one additional Chinese ODA- or OOF-financed project
increases economic growth by between 0.41 and 1.49 percentage points (pp) two years after the funding
for the project is approved, on average. 0.95 pp represents the midpoint of this range.
223
Bluhm et al. (2020) find that, on average, Chinese ODA and OOF-financed connective infrastructure
projects reduce economic concentration (as measured by the Gini coefficient of nighttime light output in all
9.3 km square grid cells within a particular subnational locality) by about 2.2 percentage points. They also
provide evidence that these projects have effectively relocated economic output from dense areas like city
centers to their immediate peripheries (i.e., peri-urban and suburban areas). The installation of connective
infrastructure can accelerate spatially-inclusive economic development by making it easier for firms to reach
more distant markets and individuals to commute or relocate to places of work. It also can also lower the
cost of consumer goods and inputs, promote the development of new businesses, increase land values,
boost agricultural production, and facilitate knowledge and technology spillovers.
120
Beijing’s track record of bankrolling and building big-ticket infrastructure projects
with record speed and near-term economic impact changed the nature of
policymaker demand in the Global South. Through its Listening to Leaders
program, AidData has repeatedly surveyed thousands of senior and mid-level
governmental officials across 140 LICs and MICs, and these leaders now report a
strong preference for working with Beijing rather than its competitors on
infrastructure projects (Custer et al. 2021; Horigoshi et al. 2022; Blair et al.
2022b).

Still, the fact that China—seemingly overnight—became the Global South’s

infrastructure financier of first resort is remarkable. Large-scale infrastructure
projects are notoriously difficult to implement. They often encounter major
delays and cost overruns and raise concerns about pollution, biodiversity loss,
the displacement of nearby residents, and the preservation of cultural
heritage.224 If they are not carefully designed and implemented, they can also
lead to public protests, lawsuits, labor strikes, corruption scandals, and
allegations of political favoritism.225

Beijing initially sought to overcome these challenges by asking political leaders,

rather than technocrats, to propose infrastructure projects and oversee their
implementation (Dreher et al. 2019, 2022).226 It relied on presidents and prime
ministers and their deputies to resolve complex and contentious matters that
could delay or derail projects—for example, the grievances and compensation
demands of those displaced or harmed by construction activities (Dreher et al.
2022). It also dispensed with many of the costly and cumbersome requirements
of Western donors and multilateral development banks (MDBs). Rather than
insisting upon strict adherence to international ESG safeguards (such as those

224
On the difficulty of implementing these types of projects on time and on budget, see Flyvbjerg et al.
(2002).
225
On this point, see Isaksson and Kotsadam (2018a, 2018b), Isaksson (2020), Dreher et al. (2019, 2022),
Anaxagorou et al. (2020), Iacoella et al. (2021), and Baehr et al. (forthcoming).
226
China’s demand-driven project selection system also encourages collusion between host country
politicians and Chinese contractors. As Zhang and Smith (2017: 2335) explain, “Chinese firms and host
governments enter into an informal alliance in which China’s companies persuade host governments to
raise new aid projects with China while the contractors promise to help behind the scenes to secure
financing. The projects are reverse-engineered to suit the political needs of local politicians and the
commercial strategies of Chinese contractors.” Similarly, Bräutigam (2019) argues that “[t]he Achilles Heel
of China’s bank financing model is that it relies heavily on Chinese companies to develop projects together
with host country officials. This creates strong incentives for kickbacks and inflated project costs. Particularly
in election years, companies and public works ministers may collude to get projects approved.”
121
described in the IFC’s Performance Standards on Environmental and Social
Sustainability), Beijing’s state-owned banks requested compliance with national
rules and standards (Export-Import Bank of China 2017; Chen and Landry 2018;
Baehr et al. forthcoming). To facilitate rapid mobilization as soon as loan or grant
applications were approved, they issued no-bid contracts to Chinese companies
with an established, on-the-ground presence (Bräutigam 2019).227 The absence
of any requirement or expectation to coordinate with other donors and lenders
eliminated additional obstacles to implementation (Bourguignon and Platteau
2015; Furukawa 2018).

But speed and convenience came at a cost: at least 54% of China’s overseas
infrastructure project portfolio from 2000 to 2017 had significant environmental,
social, or governance risk exposure (see Figure A38 in the Appendix).228 In some
cases, multi-billion dollar public investment projects were suspended or
rescoped because of insufficient inattention to environmental and social impact
assessments (see Lu et al. 2023b and Box 3b). In other cases, journalists and civil
society organizations uncovered evidence of Chinese companies and host
government officials colluding by artificially inflating sole-source contract prices
for construction projects and sharing the illicit proceeds (Malik et al. 2021: 67;
Dreher et al. 2022: 11-12). In still other cases, local grievances were not
addressed in a timely or thorough manner and they metastasized into
reputational liabilities (that we analyze at greater length in Chapter 4).229

By the end of the early BRI era, the authorities in Beijing seemed to coalesce
around the idea that sustaining elite and public support for its flagship global
227
Sole-source procurement is the rule rather than the exception in most of CDB and China Eximbank’s
overseas loan agreements. In fact, these agreements almost always reference a specific commercial
contract with a specific Chinese firm and strictly instruct the borrower to exclusively use the proceeds of the
loan to finance the pre-selected commercial contract that is referenced in the loan agreement (Gelpern et
al. 2021, 2022).
228
This figure represents the cumulative percentage of China’s grant- and loan-financed infrastructure
project portfolio (measured in constant 2021 USD) between 2000 and 2017 with significant environmental,
social, or governance risk exposure. The average annual ESG risk prevalence rate, as defined in Figure 3.2),
was 47% between 2000 and 2017. Between 2000 and 2017, 1,403 infrastructure projects in LICs and MICs
supported by grants and loans from China worth $383 billion (in constant 2021 USD) presented a significant
environmental, social, or governance risk (see Figure 1.13).
229
Many of these problems probably could have been avoided via a variety of mechanisms: environmental
and social impact assessments to ensure that indigenous peoples are granted free, prior, and informed
consent (FPIC) and avoid siting projects near endangered habitats; competitive bidding rules to ensure
good value-for-money; blacklisting procedures to avoid hiring contractors with a track record of
participating in corrupt and collusive behavior; and grievance mechanisms to make it easier to identify and
respond to the concerns of local stakeholders (Parks 2019; Dreher et al. 2022).
122
infrastructure initiative would require more effective ESG risk management and
mitigation. One of the first signs that change was afoot came in November 2017
when the China Banking Regulatory Commission (CBRC)—the country's top
banking regulator—issued a new set of rules, requiring CDB and China
Eximbank to put in place more robust environmental and social risk
management procedures (CBRC 2017a, 2017b).230 By 2018, the authorities were
planning a transition “from a hazily defined BRI 1.0 to a more fine-tuned BRI
2.0” (Ang 2019). On August 27, 2018, in the run-up to the fifth anniversary of
the BRI, Xi Jinping used a Chinese painting metaphor to call for “a switch from
xieyi, freehand painting for outlining broad strokes, to gongbi, the careful
inscription of details” (Ang 2019). Then, in April 2019, he gave a speech at the
Second Belt and Road Forum for International Cooperation where he
announced that China would “adopt widely accepted rules and standards and
encourage participating companies to follow general international rules and
standards in project development, operation, procurement and tendering and
bidding” (Ministry of Foreign Affairs of the People's Republic of China 2019). He
conveyed during the same speech that “in pursuing Belt and Road cooperation,
everything should be done in a transparent way, and we should have zero
tolerance for corruption” (Ministry of Foreign Affairs of the People's Republic of
China 2019).231

One year later, Beijing signaled preliminary interest in “multilateralizing” the BRI
by co-financing, co-designing, and co-implementing infrastructure projects with
Western and multilateral development finance institutions and subjecting these
projects to stronger safeguards.232 It teamed up with eight multilateral
institutions—the World Bank, the Inter-American Development Bank, the Asian
Development Bank, the European Bank for Reconstruction and Development,

230
In 2018, Beijing also financed the creation of a China-IMF Capacity Development Center to train
government officials on debt sustainability frameworks (DSFs) in low-income countries and other
BRI-related policy issues (Morris et al. 2020).
231
In January 2021, Hu Huaibang, the former chairman of CDB, was sentenced to life in prison for taking
$13 million in bribes. Then, in September 2021, He Xingxiang, a CDB vice president, was placed under
investigation by China’s Central Commission for Discipline Inspection (CCDI) for “severe discipline and law
violations” (Wilson 2022). One year later, Li Li, the former President of the Beijing Branch of the China
Eximbank was expelled from the Chinese Communist Party due to corruption charges. Then, in March
2023, Liu Liange resigned from his position as chairman of Bank of China and CCDI investigated him on
suspicions of corruption and graft (Wong and Zhai 2023).
232
There is some evidence of international financial institution (IFI) leaders trying to steer China in this
direction (e.g., Kim 2017; Lagarde 2019).
123
the European Investment Bank, the International Fund for Agricultural
Development (IFAD), Corporación Andina de Fomento, and the Asian
Infrastructure Investment Bank—to establish a Multilateral Cooperation Center
for Development Finance (MCDF).233 The Center’s mandate is to (a) invest in
more upstream project preparation work; (b) build the capacity of lenders and
borrowers to more effectively manage and mitigate risks related to debt
sustainability, procurement, corruption, and environmental and social issues; and
(c) facilitate greater information-sharing and coordination between Chinese and
non-Chinese development finance institutions (AIIB 2021).234

Then, in 2021, SAFE—the ultimate source of funding for most of China’s

state-owned policy banks, commercial banks, and investment funds (see Box 2b
in Chapter 2)—announced that it would prioritize “adopting MDB’s ESG criteria”
and “incorporating ESG principles into the whole project investment process
from decision-making to post-investment management” (SAFE 2021: 54).235 In
parallel, through a multilateral forum known as the International Platform on
Sustainable Finance (IPSF), the EU and China launched a joint effort to assess
the commonalities and differences in their respective taxonomies for
environmentally sustainable investments (Moody's Investor Service 2022; IPSF
2022; HKGFA and Guangdong Green Finance Committee 2022).236 This
collaboration resulted in the 2021 publication of a so-called Common Ground
Taxonomy (CGT), which in turn was “incorporated into domestic regulation [in
China]” and “directly used by Chinese banks as standards for issuing green
bonds in the international market” (Cheng and Zhang 2023: 10).

233
The MOU that established the MCDF can be accessed in its entirety via
https://www.ndb.int/wp-content/uploads/2022/11/MCDF-MOU-for-disclosure.pdf.
234
The MCDF, which is administered by the AIIB, describes itself as “a multilateral initiative to increase
high-quality infrastructure and connectivity investments in developing countries in compliance with
International Financial Institution (IFI) standards, including by encouraging other investors and financial
institutions to adopt such standards” (AIIB 2021).
235
In September 2021, Xi Jinping announced at the UN General Assembly that China would no longer
finance new coal-fired power projects overseas. Then, in March 2022, NDRC, the Ministry of Foreign Affairs,
the Ministry of Commerce, and the Ministry of Ecology and Environment published “Opinions on Jointly
Promoting Green Development of the Belt and Road,” clarifying that China would “stop building new
coal-fired power projects abroad and prudently proceed with existing ones that are under construction”
(National Development and Reform Commission, Ministry of Foreign Affairs, Ministry of Ecology and
Environment, and Ministry of Commerce of the People’s Republic of China 2022).
236
IPSF is a multilateral forum that aims to enable the exchange of practices and increase international
cooperation on sustainable finance related matters. Its members include the EU, China, Singapore, Japan,
and India.
124
All of these actions and rhetorical commitments suggest that Beijing has some
level of interest in more effectively managing the ESG risks in its overseas
infrastructure project portfolio—and potentially even harmonizing its policies
and practices with prevailing international development finance rules and
standards. However, interest does not necessarily translate into implementation,
so our aim in this chapter is to determine if China has learned from past mistakes
and recalibrated the ways that it finances, designs, and implements
infrastructure projects in the Global South.237 More specifically, we will use the
3.0 version of AidData’s GCDF dataset to (a) document the scope and severity of
the ESG risks in China’s overseas infrastructure project portfolio; (b) identify
whether, when, and how it has sought to mitigate these project implementation
risks; and (c) determine whether its infrastructure projects with and without
strong ESG safeguards have fared differently during implementation.

Our findings demonstrate that, although the ESG risk profile of China’s overseas
infrastructure project portfolio deteriorated during the pre-BRI period and early
BRI period, there are signs of improvement during the late BRI (“BRI 2.0”)
period. Chinese lenders and donors have responded to rising levels of ESG risk
by putting in place increasingly stringent safeguards that may ultimately
undermine G7 and MDB efforts to outcompete Beijing on “quality” and “safety”
grounds. Chinese grant- and loan-financed infrastructure projects that are
subjected to strong ESG safeguards present fewer environmental, social, and
governance risks during implementation. They are also less likely to be
suspended or canceled. Perhaps most importantly, Chinese grant- and
loan-financed infrastructure projects that are subjected to strong ESG safeguards
do not face substantially longer implementation delays than those subjected to
weak ESG safeguards. Our findings therefore suggest Beijing enjoys a stronger
position in the global infrastructure financing market than its bilateral and
multilateral competitors realize. Developing countries have made their
preferences very clear: they want to work with lenders and donors that are
willing and able to quickly design and implement big-ticket, high-impact

237
In February 2022, Yunnan Chen of the Overseas Development Institute (ODI) told Euromoney magazine
that “China seems to have a more specific and targeted approach. Its financial institutions are learning,
recognising past mistakes and errors, and taking a more risk-averse approach to what projects they finance,
and how they go about financing and due diligence” (Wilson 2022).
125
infrastructure projects without unreasonably high levels of ESG risk.238 Beijing is
taking active measures to meet this challenge. Whether its competitors will do
the same is an open question.

Section 2: Measuring the scope and severity of ESG risk

exposure in China's infrastructure project portfolio
We begin by measuring the nature and extent of ESG risk exposure in China’s
overseas infrastructure project portfolio—and how it has changed over time.239
We do so in five ways. First, we identify whether China is locating large-scale
infrastructure projects in environmentally sensitive areas. Second, we analyze
whether China is placing such projects in socially sensitive areas—specifically, in
areas where indigenous populations are often denied free, prior, and informed
consent (FPIC). Third, we assess whether China is locating large-scale
infrastructure projects in geographical areas that are vulnerable to political
capture and manipulation by governing elites in host countries. Fourth, we
evaluate the extent to which China is relying on contractors sanctioned for
fraudulent and corrupt behavior for the implementation of its overseas
infrastructure projects. Fifth, based upon narrative evidence from the 3.0 version
of AidData’s GCDF dataset, we flag all infrastructure projects for which it is
known that a significant environmental, social, or governance challenge arose
before, during, or after implementation.

Figure A39 presents the cumulative number of Chinese grant- and loan-financed
infrastructure projects located in environmentally sensitive areas within LICs and
MICs between 2000 and 2021. We determine if a given infrastructure project is
located in one or more environmentally sensitive areas by first merging precisely
geocoded data on Chinese ODA- and OOF-financed infrastructure project sites
with two separate datasets: (1) the boundaries of designated terrestrial and
marine protected areas from the World Database on Protected Areas (WDPA),
which we convert into a 1 km x 1 km grid cell raster; and (2) the 1 km x 1 km grid
cell raster of terrestrial and marine critical habitats (as defined by the
238
On this point, see Humphrey 2015; Dollar 2016; Swedlund 2017; Humphrey and Michaelowa 2019; Zeitz
2021; Horigoshi et al. 2022; and Blair et al. 2022b.
239
Our analysis not only includes active and completed infrastructure projects, but also suspended and
canceled projects. We include suspended and canceled projects to avoid sample selection bias, since such
projects are more likely to present significant ESG risks (Lu et al. 2023b).
126
International Finance Corporation’s Performance Standard 6).240 We then identify
the subset of projects with one or more sites that overlap with a terrestrial
protected area, a marine protected area, a terrestrial critical habitat, and/or a
marine critical habitat.241 In total, we find 1,035 infrastructure projects in 108
countries supported by grants and loans from China worth $233 billion that are
located in environmentally sensitive areas.242

Figure A40 presents the cumulative number of Chinese grant- and loan-financed
infrastructure projects located in socially sensitive areas within LICs and MICs
between 2000 and 2021. We determine if a given infrastructure project is
located in one or more of these areas by first merging precisely geocoded data
on Chinese ODA- and OOF-financed infrastructure project sites with a 1 km x 1
km grid cell raster of indigenous lands.243 We then identify the subset of projects
with one or more sites that overlap with indigenous lands, which is a useful
measure of social risk because infrastructure projects can cause local harm by
encroaching upon the traditional territories of indigenous communities without
free, prior, and informed consent (FPIC).244 In total, we find 547 infrastructure
projects in 53 countries supported by grants and loans from China worth $112
billion that are located in socially sensitive areas.245

240
Martin et al. (2015); Brauneder et al. (2018); and UNEP-WCMC & IUCN (2023). The International Finance
Corporation’s Performance Standard 6 (PS6) is widely used by international lenders and donors to identify
“critical habitats,” which refer to areas of high biodiversity value (Narain et al. 2020, 2022).
241
More specifically, we identify all Chinese grant- and loan-financed infrastructure projects with locations
that physically overlap with areas that were designated as terrestrial or marine protected areas or “likely”
critical habitats (as defined by PS6) at any point between 2000 and 2021. We exclude all projects without
“precise” or “approximate” geocodes from the analysis. A project with “precise” geocodes is one for
which have highly precise boundaries of the project’s geofeature(s). A project with “approximate” geocode
is one identified within a 5 km radius of the precise boundaries of the project’s geofeature(s). As such, all
projects geocoded to the ADM8, ADM7, ADM6, ADM5, ADM4, ADM3, ADM2, ADM1, and ADM0 levels
are excluded.
242
Environmentalists have expressed particular concerns about the siting of Chinese government-financed
infrastructure projects in geographical areas that may facilitate legal and illegal logging, agricultural frontier
expansion, and human settlements in previously remote or pristine areas (Laurance et al. 2015; Yang et al.
2021; Baehr et al. 2022).
243
Garnett et al. 2018.
244
Free, prior, and informed consent (FPIC) refers to the right of Indigenous Peoples to provide or withhold
consent, at any point, for development projects affecting their territories. It is a right granted to Indigenous
Peoples in the UN Declaration on the Rights of Indigenous Peoples (UNDRIP) and it is based on the
principle that “all peoples have the right to self-determination.” UNDRIP requires states to “consult and
cooperate in good faith with the indigenous peoples concerned through their own representative
institutions in order to obtain their free, prior and informed consent before adopting and implementing
legislative or administrative measures that may affect them.”
245
The decision to locate an infrastructure project within or outside the traditional territories of indigenous
communities is only one way of understanding the social risk profile of such a project. Consistent with the
existing literature, we focus on this dimension of social risk because it can be consistently measured over
geographic space and time for nearly all infrastructure projects (e.g., Yang et al. 2021).
127
Figure A41 presents the cumulative number of Chinese grant- and loan-financed
infrastructure projects located in geographical areas within LICs and MICs that
are vulnerable to political capture and manipulation between 2000 and 2021.
We determine if a given infrastructure project is located in one or more of these
areas by first merging precisely geocoded data on Chinese ODA- and
OOF-financed infrastructure project sites with the Political Leaders’ Affiliation
Database (PLAD), which identifies the home (birth) districts (ADM2s) of political
leaders in LICs and MICs.246 Previous research has shown that Chinese aid and
credit is disproportionately allocated to the home provinces and districts of
political leaders in host countries and that Chinese lenders and donors lack
institutional safeguards to reduce the likelihood that politically motivated
projects will be approved (Dreher et al. 2019, 2022; Anaxagorou et al. 2020).247
In total, we find 216 infrastructure projects in 69 countries supported by grants
and loans from China worth $37 billion that are located in geographical areas
that are vulnerable to political capture and manipulation.248

Figure A42 presents 2000-2021 data on the cumulative number of Chinese

grant- and loan-financed infrastructure projects that relied on contractors
sanctioned by other international financiers for fraudulent and corrupt behavior.
We determine if a given infrastructure project relied on contractors sanctioned
by other international financiers for fraudulent and corrupt behavior with a
three-step process. First, we compile a list of firms historically or currently
debarred by the World Bank and five other multilateral development banks
(MDBs)—the African Development Bank, the Asian Development Bank, the
European Bank for Reconstruction and Development, the Inter-American

246
PLAD provides information on the birthplaces of 1,109 effective political leaders from 177 countries
between 1989 and 2021 (Bomprezzi et al. 2023). Birthplaces are geocoded to the ADM2 (district) level. In
order to identify projects that are vulnerable to political capture and manipulation, we identify all projects
with locations in the home (birth) districts (ADM2s) of effective political leaders that secured Chinese grant
or loan commitments during the periods of time when the leaders in question held office.
247
The World Bank uses ex ante, cost-benefit analysis to screen candidate projects. It employs a simple
project acceptability rule—“the expected present value of the project's net benefits must be higher than or
equal to the expected net present value of mutually exclusive project alternatives”—as “a safeguard
against project choices being captured by narrow political or sectional interests” (Warner 2010: 2). By
contrast, the Chinese grant-giving and lending institutions do not have analogous institutional safeguards
in place (Dreher et al. 2019, 2022).
248
An important caveat is that we only identify projects as being located within geographical areas that are
vulnerable to political capture and manipulation if they fall within home districts (ADM2s) of political
leaders. Therefore, projects that fall within the home regions (ADM1s), but not the home districts (ADM2s)
of political leaders, are excluded. Nor do we consider the presence of non-infrastructure projects in the
home districts (ADM2s) of political leaders.
128
Development Bank, and the Asian Infrastructure Investment Bank—as well as the
dates of each firm’s formal debarment period. Then, we compare the list of
debarred firm names to the firm names of implementing agencies and
contractors involved in the Chinese grant- and loan-financed infrastructure
projects. Finally, we identify the subset of projects that relied upon debarred
firms while they were still within their debarment periods by identifying all cases
in which there was calendar day overlap between the start and end dates of an
organization’s debarment period and the commitment, implementation, or
completion dates of the project(s) it supported. Projects that relied upon a
debarred firm are identified as posing a significant governance risk (see Section
A-6 in the Appendix for more details). In total, we find 296 infrastructure projects
in 81 countries supported by grants and loans from China worth $88.8 billion
that rely on such firms.

Figure A43 presents the cumulative number of Chinese grant- and loan-financed
infrastructure projects for which there is evidence that a significant
environmental, social, or governance challenge arose before, during, or after
implementation. A key feature of AidData’s 3.0 dataset—that sets it apart from
other publicly available Chinese development finance datasets—is the inclusion
of “cradle to grave” narratives that provide detailed information about how
projects were designed and implemented in practice and why they failed,
faltered, or succeeded. These narratives consist of 3.48 million words (roughly
the same number of words one would find in 34 full-length books) across 20,985
project records. They capture, among other details, project design and
implementation challenges related to land acquisition; preservation of cultural
heritage and archaeological sites; resettlement and compensation of indigenous
communities; pollution of air, water, and soil; and adherence to anti-corruption
standards. To make use of this vast trove of qualitative information, we apply a
set of systematic search and categorization procedures (described in Section A-7
in the Appendix) to identify the subset of infrastructure projects for which there
is evidence that a significant environmental, social, or governance challenge
arose before, during, or after implementation. In total, we find that at least 356
infrastructure projects in 131 countries supported by grants and loans from

129
China worth $250 billion encountered a significant environmental, social, or
governance challenge before, during, or after implementation.249

Finally, to gain a bird’s eye view of the scope and severity of ESG risk in China's
infrastructure project portfolio in the developing world, we use all five of these
measures in combination. To determine if any given Chinese grant- or
loan-financed infrastructure project presented a significant environmental, social,
or governance risk between 2000 and 2021, we identify if it was located in an
area that is environmentally sensitive, socially sensitive, or vulnerable to political
capture and manipulation; relied on contractors sanctioned by other
international financiers for fraudulent and corrupt behavior; and/or encountered
a significant environmental, social, or governance challenge before, during, or
after implementation.250 Across 125 LICs and MICs between 2000 and 2021, we
find that 1,693 infrastructure projects supported by grants and loans from China
worth $470 billion had significant environmental, social, or governance risk
exposure (see Figure 1.13). Over the same 22-year time period, we find that
$265 billion in Chinese grant and loan commitments for 1,101 infrastructure
projects had significant environmental risk exposure, while $192 billion in
Chinese grant and loan commitments for 701 infrastructure projects had
significant social risk exposure and $211 billion in Chinese grant and loan
commitments for 405 infrastructure projects had significant governance risk
exposure (see Figure 3.1).251

249
The “at least” qualifier is important because of our inability to address “false negatives” that may affect
our keyword-search based measures (see Section A-7 in the Appendix).
250
We restricted our searches to infrastructure projects supported by grant and loan commitments worth
$20 million (in constant 2021 USD) or more. Projects supported by larger financial commitments generally
have more detailed project descriptions, which provide a stronger basis for the identification of
environmental, social, and governance risks. They also present a lower risk of generating “false negatives.”
251
Chinese grant- and loan-financed infrastructure projects can—and often do—face more than one type of
ESG risk.

130
Figure 3.1: Infrastructure projects with significant environmental, social, and governance risk exposure

Figure 3.1

Notes: The presence of significant environmental, social, and governance risk (ESG) exposure is based on a
project-level composite measure that is described in Section 2 of Chapter 3.

The ESG risk prevalence rate—defined as the annual percentage of China’s

grant- and loan-financed infrastructure project portfolio (measured in constant
2021 USD) with significant environmental, social, or governance risk
exposure—has fluctuated over time (see Figure 3.2).252 During the pre-BRI
period, it sharply increased from 12% in 2000 to 65% in 2013. It then fell to 54%
(on average) during the early BRI period and 47% (on average) during the late
BRI period.253 By 2021, it fell to 33%.

252
A similar pattern is observable when one tracks the sheer number of Chinese grant- and loan-financed
infrastructure projects facing such risks. There is an apparent reduction in the number of infrastructure
projects affected by ESG risk in 2020 and 2021 (see Figure A44), but given that the probability of ESG risks
materializing and being detected increases as a project progresses from the financial commitment phase to
the implementation phase and the completion phase, we think the apparent reduction in 2020 and 2021
should be interpreted with caution.
253
If the ESG risk prevalence rate is redefined as the annual percentage of China’s grant- and loan-financed
infrastructure projects facing a significant environmental, social, or governance risk, it rose from 25% in
2000 to 40% in 2013. It then fell to 36% (on average) during the early BRI period and 31% (on average)
during the late BRI period (see Figure A44).
131
Figure 3.2: ESG risk prevalence in overseas infrastructure portfolio from China to LICs and MICs

Figure 3.2

Notes: Projects are recorded in the years when they secured financial commitments from China, although
the ESG risks that they encountered may have materialized after the financial commitment year. The
presence of significant ESG risk exposure is based on a project-level composite measure that is described
in Section 2 of Chapter 3. Likewise, environmental risk exposure, social risk exposure, and governance risk
exposure are based on the project-level composite measures that are described in Section 2 of Chapter 3.

In Figures 3.3, 3.4, and 3.5, we separately track the environmental risk
prevalence rate, the social risk prevalence rate, and the governance risk
prevalence rate.254 On average, over the entire 22-year period of analysis
(2000-2021), the environmental risk prevalence rate was higher (27%) than the
social risk prevalence rate (20%) or the governance risk prevalence rate (18%).255
Across these three measures, one can see a generally consistent pattern over
time: risk prevalence rates mostly increased during the pre-BRI period
(2000-2013) and mostly decreased during the BRI period (2014-2021). Some of

254
In Figures 3.3, 3.4, and 3.5, we define the risk prevalence rate as the annual percentage of China’s grant-
and loan-financed infrastructure project portfolio (measured in constant 2021 USD) facing a given type of
risk. In Figures A45, A46, and A47, we redefine the risk prevalence rate as the annual percentage of China’s
grant- and loan-financed infrastructure projects facing a given type of risk.
255
If the ESG risk prevalence rate is redefined as the annual percentage of China’s grant- and loan-financed
infrastructure projects with significant environmental, social, or governance risk exposure, the same pattern
holds: the environmental risk prevalence rate is substantially higher (22%) than the social risk prevalence
rate (14%) or the governance risk prevalence rate (7%).
132
the largest declines are observable during the late BRI period (2018-2021).256
However, these declines should be interpreted with caution, as they could be
the result of (a) newly approved projects not having progressed to phases of the
project lifecycle when ESG risks typically materialize, (b) actual improvements in
the ESG risk profile of China’s overseas infrastructure project portfolio, or (c)
some combination of these factors.257

Figure 3.3: Environmental risk prevalence in overseas infrastructure portfolio from China to LICs and MICs

Figure 3.3

Notes: Projects are recorded in the years when they secured financial commitments from China, although
the ESG risks that they encountered may have materialized after the financial commitment year. The
presence of significant environmental risk exposure is based on a project-level composite measure that is
described in Section 2 of Chapter 3.

256
During the late BRI period, the total number of Chinese grant- and loan-financed infrastructure projects
with significant environmental risk exposure, social risk exposure, and governance risk exposure also
apparently declined (see Figures 3.3, 3.4, and 3.5).
257
Given that the probability of ESG risk detection increases as an infrastructure project progresses from the
financial commitment phase to the implementation phase and the completion phase, still another
possibility is that the apparent declines in ESG risk prevalence during the late BRI period reflect
measurement imprecision.
133
Figure 3.4: Social risk prevalence in overseas infrastructure portfolio from China to LICs and MICs

Figure 3.4

Notes: Projects are recorded in the years when they secured financial commitments from China, although
the ESG risks that they encountered may have materialized after the financial commitment year. The
presence of significant social risk exposure is based on a project-level composite measure that is described
in Section 2 of Chapter 3.

134
Figure 3.5: Governance risk prevalence in overseas infrastructure portfolio from China to LICs and MICs

Figure 3.5

Notes: Projects are recorded in the years when they secured financial commitments from China, although
the ESG risks that they encountered may have materialized after the financial commitment year. The
presence of significant governance risk exposure is based on a project-level composite measure that is
described in Section 2 of Chapter 3.

Table A12 provides country-level summary statistics on the number and

monetary value of Chinese grant- and loan-financed infrastructure projects
between 2000 and 2021 with significant environmental, social, and governance
risk exposure. It also provides a country-by-country breakdown of the overall
ESG risk prevalence rate over the same time period. Analysis of the country-level
data from Table A12 demonstrates that the ESG risk in China’s overseas
infrastructure project portfolio is disproportionately concentrated in certain
regions (see Table A10). For example, only 16.9% of China’s grant- and
loan-financed infrastructure project portfolio (measured in constant 2021 USD)
was located in South and Central America between 2000 and 2021, but 38% of
its portfolio with significant governance risk exposure was concentrated in the

135
same region during the same time period.258 ESG risk in Beijing’s overseas
infrastructure project portfolio is also unevenly distributed across countries with
different per capita income levels (see Table A11). Governance risk is again a
case in point. 37.1% of China’s grant- and loan-financed infrastructure project
portfolio (measured in constant 2021 USD) was located in upper-middle income
countries (UMICs) between 2000 and 2021.259 Yet a staggering 52.5% of its
portfolio with significant governance risk exposure was concentrated in such
countries.260 Table A12 provides evidence that a small subset of large aid and
credit recipients—including Venezuela, Malaysia, and Argentina—contributed to
the disproportionate concentration of governance risk exposure in UMICs.

In Figure 3.6, we identify global hotspots by fusing data on the environmental,

social, and governance risk exposure of Chinese grant- and loan-financed
infrastructure projects with point, polygon, and line vector data (described in
Chapter 1) that capture the geographic footprints of these projects. To do so,
we first create a 200 km x 200 km grid covering every LIC and MIC in the 3.0
version of AidData’s GCDF dataset. We then use the point, polygon, and line
vector data to assign every geocoded infrastructure project to one of more of
these grid cells.261 We subsequently assign each grid cell a gradation of
color—along a “heat” spectrum—based on the cumulative monetary value of
Chinese grant- and loan-financed infrastructure projects with environmental,
social, or governance risk exposure in that geographical area between 2000 and

258
Table A10 demonstrates that environmental risk exposure was disproportionately concentrated in
Central and Eastern Europe: whereas 13% of China’s grant- and loan-financed infrastructure project
portfolio (measured in constant 2021 USD) was located in the region between 2000 and 2021, 20.5% of its
portfolio with significant environmental risk exposure was concentrated in Central and Eastern Europe
during the same 22-year period. By contrast, social risk exposure was disproportionately concentrated in
Asia: whereas 36.4% of China’s grant- and loan-financed infrastructure project portfolio (measured in
constant 2021 USD) was located in the region between 2000 and 2021, 42.3% of its portfolio with
significant social risk exposure was concentrated in Asia during that 22-year period (see Table A10).
259
According to Table A11, lower-middle income countries (LMICs) and low-income countries (LICs)
received 26.8% and 25.3%, respectively, of China’s grant- and loan-financed infrastructure project portfolio
(measured in constant 2021 USD) during the 2000-2021 period. However, 35.8% of its portfolio with
significant social risk exposure was concentrated in LMICs—and 37.8% of its portfolio with significant social
risk exposure was concentrated in LICs—between 2000 and 2021.
260
Only 15.8% of China’s grant- and loan-financed infrastructure project portfolio (measured in constant
2021 USD) with significant social risk exposure—and 25.6% with significant environmental risk
exposure—was located in UMICs (see Table A11).
261
If a project falls across multiple grid cells, we assume the monetary value of the commitment for the
project is evenly distributed within the project’s line or polygon. Thus, the total financial commitment value
for the project is split up among grid cells based upon the percentage of the project’s area that falls within
each grid cell.
136
2021.262 Whereas light pink grid cells represent areas where China has a
relatively low level of risk exposure in its infrastructure project portfolio, dark
purple grid cells represent areas where China has a relatively high level of risk
exposure in its infrastructure project portfolio.

The map in the upper-left hand corner of Figure 3.6 demonstrates that Beijing
has a particularly high level of environmental risk exposure in the Tropical Andes
(including Venezuela, Ecuador, and Peru), the Southern Cone (including
Argentina), East Africa (including Ethiopia, Kenya, and Uganda), West Africa
(including Ghana, Togo, Benin, Nigeria, and Cameroon), Central Asia (including
Kazakhstan, Kyrgyzstan, and Tajikistan), and Southeast Asia (including Laos,
Cambodia, Vietnam, and Indonesia). The geographical distribution of social risk
exposure, as depicted in the map in the upper-right hand corner of Figure 3.6, is
broadly similar, although the hotspots are less concentrated in Central Asia and
more concentrated in Ethiopia, Kenya, Pakistan, and Southeast Asia. The map in
the bottom-left hand corner of Figure 3.6 also demonstrates that Beijing has a
particularly high level of governance risk exposure in the Tropical Andes, East
Africa, and South Asia—including Zambia, Bangladesh, and Argentina (three
countries for which we provide in-depth case study evidence in Chapter 4).
Finally, in the bottom-right hand corner of Figure 3.6, we collapse all three
categories of risk exposure into a single map, such that each grid cell captures
the extent to which Chinese grant- and loan-financed infrastructure projects in
that area encountered significant environmental, social, or government risks.

262
In Figure A62, we replicate Figure 3.6 but scale the level of risk exposure in a given grid cell according
to the cumulative count of Chinese grant- and loan-financed infrastructure projects rather than the
cumulative monetary value of Chinese grant and loan commitments for the same projects.
137
Figure 3.6 A global map of China’s infrastructure project portfolio in LICs and MICs with significant environmental, social, governance (ESG) risk exposure

Figure 3.6
A global map of China’s infrastructure project portfolio in
LICs and MICs with significant environmental, social,
governance (ESG) risk exposure

Notes: This map presents the geographical areas where China’s grant- and loan-financed infrastructure
project portfolio (measured in constant 2021 USD) has significant environmental, social, or governance
(ESG) risk exposure. Darker (purple) colors represent areas where the portfolio has high levels of risk
exposure and lighter (pink) colors represent areas where the portfolio has lower levels of risk exposure.
Environmental risk exposure, social risk exposure, and governance risk exposure are based on the
project-level composite measures that are described in Section 2 of Chapter 3.

138
Section 3: Measuring the stringency of ESG safeguards in
China’s infrastructure project portfolio with new sources of
contractual evidence
Although Beijing clearly faces a wide array of ESG risks in its overseas
infrastructure project portfolio, little is known about the safeguards that it has
put in place to manage and mitigate these risks. Another blind spot is whether
and how Chinese state-owned lenders have strengthened or weakened their
ESG safeguards over time.

The 3.0 version of AidData’s GCDF dataset provides a unique opportunity to fill
this evidentiary gap.263 As part of the primary source identification work that was
undertaken to support the construction of the dataset, AidData obtained a large
cache of unredacted infrastructure financing agreements via official sources in
LICs and MICs, including government registers and gazettes, aid and debt
information management systems, and parliamentary oversight institutions.
These grant and loan agreements represent “high-value sources,” in that they
provide detailed information about whether financiers, at the time that they
signed the agreements with their host country counterparts, identified
behavioral expectations related to ESG risk management and mechanisms to
monitor and enforce compliance with those expectations.

Another important feature of the 3.0 version of AidData’s GCDF dataset—and

an improvement over the 2.0 version—is that it makes these unredacted
agreements available for the full range of financial instruments that Beijing uses
to bankroll infrastructure projects in the developing world, including:

1. Bilateral grants and interest-free loans issued by China’s Ministry of

Commerce (MOFCOM)

2. Bilateral loans issued by China Eximbank

263
The 3.0 version of the GCDF dataset provides stable URLs to hundreds of unredacted grant, loan, debt
forgiveness, debt rescheduling, and escrow account agreements. AidData published a subset of these
financing agreements in March 2021 when the How China Lends report was first published (Gelpern et al.
2021, 2022). However, the 3.0 dataset provides the full set of agreements retrieved by AidData.
139
 3. Bilateral loans issued by China Development Bank (CDB)

4. Bilateral loans issued by Chinese state-owned commercial banks, such as

ICBC, China Construction Bank, and Bank of China

5. Syndicated loans issued by China’s policy banks (China Eximbank and

CDB) and state-owned commercial banks

6. Syndicated loans issued by Chinese state-owned banks and multilateral

institutions

7. Grants and loans that China has channeled via multilateral institutions

8. Supplier’s credits issued by Chinese state-owned companies

These eight types of financing agreements, which account for 90% of China’s
grant- and loan-financed infrastructure project portfolio in the developing world
between 2000 and 2021, include widely divergent ESG terms and conditions
(see Tables A5 and A8).264 However, variation in de jure ESG safeguard
stringency has never been systematically documented across agreement types.
Nor has previous research demonstrated how Beijing’s use of these different
types of agreements—with varying levels of de jure ESG safeguard
stringency—has changed with the passage of time.

In order to overcome these obstacles, we developed a standardized set of

coding criteria related to ESG risk management that can be applied to any type
of Chinese loan contract or grant agreement that supports an overseas
infrastructure project.265 These 26 criteria, which are described in Section A-8
and Table A3 in the Appendix, include 8 focused on environmental safeguards,

264
These 8 financial instrument types were used by China to support 90.2% of its grant- and loan-financed
infrastructure project portfolio in LICs and MICs between 2000 and 2021. The remaining 9.8% of the
portfolio consisted of projects supported by more “exotic” financial instrument types (e.g., Engineering,
Procurement, Construction and Financing (ECPDF) agreements). The 3.0 version of the GCDF dataset does
not include many unredacted financing agreements for these projects, so we exclude them from our
analysis.
265
Environmental and social safeguards are typically inapplicable to projects that do not involve the
construction, rehabilitation, or expansion of infrastructure, although there are some exceptions to this
general rule (most notably, projects that involve natural resource extraction without infrastructure
components).
140
7 focused on social safeguards, and 11 focused on governance safeguards. They
are broadly aligned with the OECD Council Recommendation on Common
Approaches for Officially Supported Export Credits and Environmental and
Social Due Diligence, the IFC's Performance Standards on Environmental and
Social Sustainability, the Uniform Framework for Preventing and Combating
Fraud and Corruption, the OECD Council Recommendation on Bribery and
Officially Supported Export Credits, and the OECD Council Recommendation on
Public Procurement.266 The criteria are organized into three groups: those that
identify the presence or absence of (1) rules or standards to establish behavioral
expectations related to ESG risk management and mitigation, (2) oversight
mechanisms for monitoring compliance with those behavioral expectations;
and/or (3) enforcement mechanisms for sanctioning noncompliance with those
behavioral expectations (e.g., indemnification, withholding disbursements).

To construct our coding sample, we first identify all of the records (nearly 300) in
the 3.0 version of the GCDF dataset that include unredacted loan contracts and
grant agreements. We then remove all of the loan contracts and grant
agreements that do not support infrastructure projects. We subsequently
eliminate all loan contracts and grant agreements that do not correspond to one
of the 8 primary infrastructure financing agreement types. As shown in Section
A-10, we then prune the remaining sample of loan contracts and grant
agreements to identify 3 agreements for each of the 8 financial instrument
categories267 that provide broad geographical coverage (across Africa, Latin
America and the Caribbean, Asia and the Pacific, Central and Eastern Europe,
and the Middle East) and income bracket coverage (across upper-middle income
countries, lower-middle income countries, low-income countries, and least
developed countries), and temporal coverage (over our 22-year period of study).
For each financial instrument category, we also seek to identify agreements

266
The OECD Council Recommendation on Common Approaches for Officially Supported
Export Credits and Environmental and Social Due Diligence was previously known as the OECD
Revised Council Recommendation on Common Approaches on the Environment and Officially
Supported Export Credits.
267
For one of the eight financial instrument categories (“syndicated loans issued by Chinese
state-owned banks and multilateral institutions”), we were only able to identify two infrastructure
project financing agreements.
141
issued before and after the late BRI period,268 given that Beijing has made many
rhetorical commitments to strengthen ESG protections since late 2017.

Although we do not select infrastructure financing agreements for coding

purposes through a random sampling procedure and our coding sample
represents a small part of China's grant- and loan-financed infrastructure project
portfolio in LICs and MICs, our findings demonstrate that ESG terms and
conditions are highly standardized by infrastructure financing instrument, which
gives us confidence that we are capturing meaningful differences in de jure ESG
safeguard stringency across the main financial instruments that Beijing uses to
bankroll overseas infrastructure projects.269 We also find relatively little variation
in ESG terms and conditions across countries in different regions and income
brackets (see Table A4 in the Appendix).270 In this respect, our findings are
consistent with the first study to ever systematically evaluate the terms and
conditions governing China’s loan contracts with overseas borrowers. Gelpern et
al. (2022: 16) conclude that “our analysis of [100] contracts shows that Chinese
lending terms are highly standardized by lender and instrument, and do not
exhibit significant variation by [...] region or income bracket.”

The contract-level data from our coding sample are provided in Table A8 of the
Appendix. In order to convert the contract-level data into categorical measures
of safeguard stringency for each financial instrument type, we first make binary
determinations of whether there is any evidence that each financial instrument
type (before or after the late BRI period) established any (a) rules or standards
that create behavioral expectations related to ESG risk management and
mitigation, (b) oversight mechanisms for monitoring compliance with those

268
For two of the eight financial instrument categories (“supplier’s credits issued by Chinese state-owned
companies” and “syndicated loans issued by China’s policy banks and state-owned commercial banks”), we
relied on infrastructure financing agreements that were issued in 2022 (in lieu of agreements issued
between 2018 and 2021) to ensure adequate coverage during the late BRI period (see Tables A5, A6, and
A7 in the Appendix for more details).
269
The agreement-level ratings that are reported in Table A8 demonstrate that most of the observed
heterogeneity in ESG safeguard stringency is across financial instrument types rather than across
agreements within a given financial instrument type. See also Tables A6 and A7.
270
Our coding sample underrepresents China’s infrastructure financing to some regions and income
brackets and overrepresents its infrastructure financing to other regions and income brackets (see Section
A-10). The external validity of our sample would be a concern if China’s infrastructure financing agreements
varied systematically by region or income bracket. However, we do not find much evidence that China’s
infrastructure financing agreements differ significantly by region or income bracket (see Table A4 in the
Appendix).
142
behavioral expectations, or (c) enforcement mechanisms for sanctioning
noncompliance with those behavioral expectations. Based upon these
determinations, which are reported in Table A8, we assign high, medium, or low
environmental, social, and governance safeguard ratings to each financial
instrument type using the following criteria:

● Low: No rules and standards exist and there are no mechanisms for
monitoring compliance or sanctioning noncompliance.

● Medium: Rules and standards exist, but there are no mechanisms for
monitoring compliance or sanctioning noncompliance.

● High: There is a mechanism for monitoring compliance and/or a

mechanism for sanctioning noncompliance.

Our application of the standardized coding criteria to the sample of grant and
loan agreements produces a set of summary ESG ratings for the 8 financial
instrument categories over two time periods: the pre-BRI and early BRI period
(2000-2017) and the late BRI period (2018-2021).271 These summary ratings,
which measure the strength of ESG safeguards in a de jure rather than a de facto
sense, are provided in Table 3.1 and they call attention to several important
patterns and trends.272 First, among the infrastructure financing instruments at
Beijing’s disposal, policy bank (China Eximbank and CDB) loan agreements offer
the weakest ESG safeguards.273 This was certainly true before the BRI was

271
Beyond the fact that some Chinese financiers have published environmental policies and standards on
their websites and others have not, Narain et al. (2020) does not document any safeguard variation across
Chinese state-owned creditors that finance overseas infrastructure projects. However, Narain et al. (2020)
does not systematically evaluate the safeguard provisions contained in the financing agreements of
Chinese state-owned creditors. Nor does the study capture any of the changes that took place during the
late BRI period (2018-2021).
272
The “High,” “Medium,” and “Low” designations are not comprehensive measurements of ESG
safeguard stringency vis-à-vis international standards, such as PS6. They only provide measurements of
whether ESG rules and standards exist and whether there are mechanisms in place for monitoring
compliance or sanctioning noncompliance. A potentially productive avenue for future research would be to
construct “distance-to-frontier” safeguard stringency measures that are based on PS6 or an analogous set
of international standards that are broadly encompassing.
273
China Eximbank’s infrastructure loan agreements received low environmental, social, and governance
safeguard ratings. CDB’s infrastructure loan agreements received low environmental and social safeguard
ratings, but a medium governance safeguard rating—due to the fact that two out of the three CDB
contracts in the coding sample included anti-corruption and anti-money laundering requirements as well as
requirements to prepare and submit financial statements in accordance with International Financial
Reporting Standards (IFRS) or the Generally Accepted Accounting Principles (GAAP) of the Financial
Accounting Standards Board (FASB).
143
launched, and it remained largely true during the early and late BRI periods.274
Second, China’s state-owned commercial banks have strong de jure ESG
safeguards in their overseas loan agreements. They not only apply such
safeguards when they issue bilateral loan agreements, but also when they
participate in syndicated loan agreements. Third, strong ESG safeguards
consistently apply to the grants and loans that the PBOC and China’s Ministry of
Finance channel to LICs and MICs via multilateral institutions. They also apply to
syndicated loans that involve multilateral institutions, which highlights a fourth
(broader) finding from Table 3.1: the fact that syndicated loans have consistently
stronger de jure ESG safeguards than bilateral loans. Given that all participants
in a syndicated loan agreement for an infrastructure project must agree to a
common set of contractual terms and conditions, including applicable ESG
safeguards, one might think that a “least common denominator” dynamic could
go into effect. But Table 3.1 indicates that the opposite is true: syndicate
participants seem to defer to the lending institution(s) with the strongest
preference(s) for ESG risk mitigation.275

Table 3.1: De jure ESG safeguard stringency in China’s overseas infrastructure portfolio by type of financing instrument

Table 3.1
De jure ESG safeguard stringency in China’s overseas
infrastructure portfolio by type of financing instrument
Environmental Social Governance

Financing Instrument Safeguards Safeguards Safeguards

Pre/Early Pre/Early Pre/Early
Late BRI Late BRI Late BRI
BRI BRI BRI
Bilateral China Eximbank loan Low Low Low Low Low Low

Bilateral CDB loan Low Low Low Low Medium Medium

Bilateral MOFCOM loan or grant Low Medium Low Medium Low High

Bilateral Chinese state-owned

High High High Low High High
commercial bank loan

Syndicated loan with Chinese and

High High High High High High
multilateral bank participants

274
These findings are consistent with those of Narain et al. (2020, 2022).
275
The apparent benefit of including a multilateral institution or a state-owned commercial bank
in a lending syndicate is that it can lead every other member of the syndicate to adopt their
(stronger) safeguards.
144
 Environmental Social Governance

Financing Instrument Safeguards Safeguards Safeguards

Pre/Early Pre/Early Pre/Early
Late BRI Late BRI Late BRI
BRI BRI BRI
Syndicated loan with Chinese
state-owned commercial banks High High High Low High High
and/or policy banks

PBOC/MOF grant or loan channeled

High High High High High High
through multilateral institutions

Supplier’s credit from Chinese SOE Low High Low Low Low High
Notes: The safeguard stringency scores for each type of grant-giving and lending instrument are based on
the analysis described in Section A-9 of the Appendix.

Several changes that took place during our period of study (2000-2021) also
merit discussion. MOFCOM’s grant and interest-free loan agreements had weak
ESG safeguards prior to the late BRI period. However, we see evidence of
MOFCOM shifting toward stronger de jure ESG protections between 2018 and
2021. The same pattern is evident in supplier’s credit agreements issued by
Chinese state-owned enterprises: formal ESG safeguard stringency increased
with the passage of time.276 Table 3.1 also provides evidence that, during the
late BRI period, China’s state-owned commercial banks watered down their
social safeguards.277 During the pre-BRI and early BRI periods, these financial
institutions had mechanisms in place to monitor compliance and/or sanction
noncompliance with domestic and international social laws and standards.278
Their bilateral loan agreements and syndicated loan agreements made financial
disbursements conditional upon certification of compliance with social laws and
standards, or required borrowers to financially compensate (indemnify) lenders
for any losses or liabilities resulting from actual or alleged violations of social

276
Although the environmental and governance safeguards that apply to supplier’s credits strengthened
during the late BRI period, the social safeguards that apply to supplier’s credits did not.
277
Table 3.1 provides evidence that China’s state-owned commercial banks weakened the social safeguards
that apply to their bilateral loans and their syndicated loans during the late BRI period.
278
In the sample of financing agreements that we evaluated, social laws and standards were given
expansive definitions, including (a) laws, rules, and regulations in borrower countries related to work, social
security, industrial relations, occupational health and safety, public participation, property ownership
(formal and traditional), and the protection and empowerment of indigenous peoples and ethnic groups;
the projection, restoration, and promotion of cultural heritage and archaeological artifacts; and the
resettlement or economic displacement of persons; (b) the OECD Revised Council Recommendation on
Common Approaches on the Environment and Officially Supported Export Credits; (c) the Equator
Principles; (d) UN treaties and conventions on human rights; and (e) international labor agreements.
145
laws and standards. Yet, for reasons that we do not yet understand, these
safeguards vanished during the late BRI period.

The next step in our analysis is to apply the ESG safeguard stringency ratings
from Table 3.1 to China’s entire grant- and loan-financed infrastructure project
portfolio in the developing world. We do so by first mapping all loans and
grants for active, completed, suspended, or canceled infrastructure projects in
the 3.0 version of the GCDF dataset to one of the 8 financial instrument
categories (whenever possible). Then, we assign the aggregate ESG safeguard
stringency ratings—reported in Table 3.1—to the infrastructure loans and grants
in the 3.0 version of GCDF dataset that use the same loan or grant instrument.279

279
To map individual grants and loans for infrastructure projects to our taxonomy of infrastructure financing
instruments (consisting of 8 loan and grant-giving instruments), we use a combination of the funding
agency, implementing agency, co-financing agency, receiving agency, number of lenders, flow type, and
supplier’s credits fields in the 3.0 version of the GCDF dataset. These 8 financial instrument types cover
90.2% of China’s grant- and loan-financed infrastructure project portfolio in LICs and MICs between 2000
and 2021. We do not assign de jure ESG safeguard stringency ratings to the remaining 9.8% of the
portfolio, which represents infrastructure projects financed with other types of financial instruments. As
such, whenever we report portfolio-level summary statistics related to the application of de jure ESG
safeguards, we disregard projects for which de jure ESG safeguard stringency ratings could not be reliably
assigned (i.e., 9.8% of the LIC and MIC portfolio).
146
Figure 3.7: De jure ESG safeguard stringency in China’s overseas infrastructure portfolio

Figure 3.7

Notes: The safeguard stringency ratings for each grant-giving and lending instrument are based
on Table 3.1 and explained in Section A-9 of the Appendix.

Figure 3.7 presents the estimated percentage of China’s grant- and

loan-financed infrastructure project portfolio in LIC and MICs with strong de jure
environmental, social, and governance safeguards in place between 2000 and
2021. One can see a marked shift toward stronger ESG protections during the
late BRI period (2018-2021). By the eighth full year of BRI project
implementation (2021), approximately 57% of China’s grant- and loan-financed
infrastructure project portfolio in LIC and MICs had strong de jure
environmental, social, and governance safeguards in place.280 This represents a
major departure from past practice: at the turn of the century, China’s entire
grant- and loan-financed infrastructure project portfolio in LIC and MICs had
weak de jure environmental, social, and governance safeguards in place.

280
Table A12 provides country-level summary statistics on the percentage of China’s grant- and
loan-financed infrastructure project portfolio with strong de jure ESG safeguards between 2000 and 2021.

147
Figure 3.8: De jure safeguard stringency in China’s overseas infrastructure portfolio by safeguard type

Figure 3.8.1: Environment

Figure 3.8.2: Social

148
Figure 3.8.3: Governance

Notes: The safeguard stringency ratings for each grant-giving and lending instrument are based on Table
3.1 and explained in Section A-9 of the Appendix.

However, Beijing has not demonstrated comparable levels of enthusiasm for all
types of ESG safeguards during the late BRI era. Figure 3.8 demonstrates that
37% of the infrastructure project portfolio was subjected to strong de jure
environmental safeguards from 2018 to 2021, as compared to 20% during the
previous eighteen-year period (2000-2017). Similarly, 40% of the infrastructure
project portfolio was subjected to strong de jure governance safeguards from
2018 to 2021, as compared to 20% during the previous eighteen-year period
(2000-2017).

Yet Beijing demonstrated far less interest in applying stringent social safeguards
to its overseas infrastructure project portfolio during the late BRI era. Between
2018 and 2021, it shielded an increasing proportion of its grant- and
loan-financed infrastructure project portfolio from these types of safeguards (see
Figure 3.8). While the proximate explanation for this change during the late BRI
era was the removal of social safeguard enforcement mechanisms from the loan
contracts of China’s state-owned commercial banks (see Table 3.1), the

149
underlying reason why it took place is a mystery. One possibility—potentially
deserving attention in future research—is that China’s aversion to strong de jure
social safeguards is related to its own disconcerting experience with the World
Bank’s social safeguards during the late 1990s and early 2000s (see Box 3a).281

Box 3a: China’s experience with the application of World Bank social safeguards
to the Western Poverty Reduction Project in Qinghai

In 1997, the World Bank started working with China’s provincial government in Qinghai on the
design of a $40-million loan for the Western Poverty Reduction Project. The purpose of the
project was to resettle approximately 60,000 poor farmers to a new irrigation tract. The
resettlement area was located in central Qinghai, more than 500 kilometers from the border of
the Tibet Autonomous Region. However, in 1999, a transnational advocacy network—consisting
of Tibet NGOs (including the Tibet Information Network and the International Campaign for
Tibet) and multilateral development bank monitors (such as the Bank Information Center and the
Center for International Environmental Law)—launched a campaign to prevent the World Bank’s
Board of Directors from green-lighting the project. They claimed, with support from the U.S.
Congress and U.S. Treasury, that approval of the project would be tantamount to bankrolling
genocide (by diluting Tibet’s culture with 60,000 ethnic Chinese). They also claimed that the
World Bank had failed to comply with its own social safeguards policy—by classifying a project
as “Category B” when it should have been classified as “Category A.”282

When the Board of Directors voted to conditionally approve the project in June 1999, a group of
campaigners hung a “World Bank Approves China’s Genocide in Tibet” banner outside World
Bank headquarters. Robert Wade, who investigated claims about the project on behalf of the
World Bank’s Inspection Panel, recounts that "[t]he NGOs put together a formidably effective
campaign network. They established websites to share information and provide sample protest
letters to the Bank which could be emailed directly from the site or printed out and faxed. The
Tibet lobby sponsored rock concerts in cities around the world, with prepared postcards, fax
machines and email facilities on hand. The result was a deluge of letters, postcards, emails and
faxes the like of which the Bank had never seen, mainly from the U.S. and Europe. The Western
media, both press and TV, lined up behind the critics. Reports in leading newspapers like The
Financial Times, The New York Times and The Washington Post read as though taken straight
from NGO handouts. They repeated the NGOs’ portrait of the project in the same language,
often not distinguishing between what the NGOs claimed and what they, the journalists,
reported as fact. Many reported as fact, for instance, that the move-in area was the birthplace of
the Dalai Lama, which is simply false […]. Yet for all their claims to speak for Tibetans and for all

281
It is, however, worth noting that there was never much support in Beijing for strong social safeguards
across our entire 22-year period of study (2000-2021).
282
Category A projects pose the most severe environmental and social risks; they often involve large-scale
infrastructure, industrial-scale chemical manufacturing, or natural resource extraction activities. The World
Bank subjects these projects to its most stringent ESG safeguards, but it also acknowledges that the risks
these projects pose can be difficult or impossible to fully mitigate. Category B projects also pose significant
environmental and social risks, but the World Bank expects that it can reasonably and readily mitigate all or
most of these risks during implementation (Buchanan et al. 2018).
150
their denunciation of the consultation process, the NGOs never produced evidence that local
people did not want the project beyond a few very brief and anonymous letters sent to the Tibet
NGOs by people claiming to live near the move-in area” (Wade 2009: 32).

In July 2000, the World Bank’s Board of Directors convened to decide if it would approve the
project. The discussions dragged on for multiple days, with developing country representatives
advocating for project approval and certain developed country representatives calling for project
cancellation. The issue was ultimately resolved when China’s Executive Director withdrew the
project proposal from consideration. Beijing announced that the project would proceed with an
alternative source of funding.

Of course, Beijing’s critics and rivals might question whether any of the “fine
print” in its overseas infrastructure financing agreements even matters if ESG
safeguards are not put into practice. To gauge whether China’s de facto
application of ESG safeguards matches the de jure ESG safeguards in its
financing agreements, we leverage the detailed qualitative information that
AidData has collected about how projects were designed and implemented in
practice. The “cradle to grave” narratives in the 3.0 version of the GCDF dataset
include detailed descriptions of efforts to mitigate ESG risks before, during, and
after project implementation—for example, by adopting environmental
management plans (EMPs) that respond to the findings and recommendations of
an environmental impact assessment or by providing financial compensation to
project-affected persons (PAPs).

To make effective use of this qualitative information, we use a set of systematic

search and categorization procedures (described in Section A-7 in the Appendix)
to identify the subset of infrastructure projects for which there is evidence of
efforts being undertaken by Chinese financiers or implementing agencies to
mitigate environmental, social, or governance risks before, during, or after
project’s implementation. Between 2000 and 2021, we find evidence that de
facto ESG risk mitigation efforts were undertaken to support at least 210
infrastructure projects in 66 LICs and MICs supported by grants and loans from
China.283 The estimated cumulative value of China’s grant- and loan-financed
infrastructure project portfolio supported by de facto ESG risk mitigation efforts
increased from $55 million in 2000 to $86 billion in 2021 (see Figure A48).

283
The “at least” qualifier is important because of our inability to address “false negatives” that may affect
our keyword-search based measures (see Section A-7 in the Appendix).
151
According to Figure 3.9, the percentage of Chinese grant- and loan-financed
infrastructure projects in the developing world supported by a de facto ESG risk
mitigation effort steadily increased from 2% in 2000 to 18% in 2021, which is
broadly consistent with China’s increasing use of strong de jure ESG safeguards
over the same twenty-two year period.284 However, Figure 3.9 also highlights an
important shift that took place over time: although de jure and de facto risk
mitigation efforts mostly moved in tandem during the pre-BRI era (2000-2013),
the “delta” between de jure and de facto risk mitigation efforts widened during
the BRI era (2014-2021).285 By 2021, the gap between how Beijing applied ESG
safeguards in principle and in practice was substantial: 57% of its infrastructure
project portfolio in LICs and MICs benefited from strong de jure ESG
safeguards, yet there was evidence of de facto ESG risk mitigation efforts being
undertaken in only 18% of the portfolio (see also Box 3b).286

284
In the Appendix, we present a different version of this graph (Figure A50) that measures the annual
percentage of China’s grant- and loan-financed infrastructure projects in LICs and MICs supported by (a)
one or more de facto ESG risk mitigation efforts and (b) strong de jure ESG safeguards. It too shows that
Beijing’s de jure risk mitigation efforts generally outpaced its de facto risk mitigation efforts.
285
Figure 3.9 treats an infrastructure project’s financial commitment year as the year in which ESG risk
mitigation efforts were undertaken. However, given that the probability of ESG risk mitigation measures
being undertaken and detected increases as a project progresses from the financial commitment phase to
the implementation phase (and the completion phase), it may also be useful to treat an infrastructure
project’s commencement (implementation start) year or its completion (implementation end) year as the
year in which ESG risk mitigation efforts were undertaken. We do so in Figures A51 and Figure A52.
However, these two figures do not show substantially smaller (or larger) gaps between de jure and de facto
ESG risk mitigation efforts. In Figure A51, the average annual percentage point difference between
infrastructure projects with strong de jure ESG safeguards and infrastructure projects that involved de facto
ESG risk mitigation efforts based on the completion year is 10%. In Figure A52, it is 13.2% based on the
commencement year. In Figure 3.9, it is 12%.
286
An important caveat is that our measure of whether any effort was undertaken to mitigate ESG risks
before, during, or after project’s implementation almost certainly underestimates the true level of risk
mitigation effort (due to the previously-mentioned “false negative” challenge).
152
Figure 3.9: Infrastructure project portfolio with de jure vs. de facto ESG risk mitigation

Figure 3.9

Notes: De facto ESG risk mitigation efforts are measured using the methodology that is
described in Section A-8. Strong de jure ESG safeguards are defined in Section A-9 of the
Appendix.

Box 3b: De jure versus de facto application of ESG safeguards to the Lahore
Orange Line Metro Train Project
The Lahore Orange Line is Pakistan’s first-ever urban mass rail transit project. Since its
inauguration in October 2020, the average level of daily ridership (178,714) on the 27-km metro
line has remained below capacity, but transformed the megacity’s public transport landscape
(Hasnain 2023). During President Xi Jinping’s April 2015 visit to Pakistan, the project was
grandfathered into the China-Pakistan Economic Corridor (CPEC) as a “gift from China” (Khan
2018). But it was ultimately financed by China Eximbank with a mix of concessional and
non-concessional loans, including a $1.2 billion preferential buyer’s credit with a 2% interest rate,
an RMB 1.2 billion government concessional loan with a 2% interest rate, and a $203 million
buyer’s credit loan with a 5.2% interest rate.287 Pakistan’s government used the loan proceeds to
partially finance a $1.63 billion commercial contract between CR-NORINCO—a joint venture of
China State Railway Group Co. Ltd. (CR) and China North Industries Corporation
(NORINCO)—and Punjab Mass Transit Authority. CR-NORINCO, in turn, hired local contractors
to assist with a variety of activities, including the project's environmental impact assessment

287
For more details, see Project ID#54420, 53820, 37280 in the 3.0 version of AidData’s GCDF dataset.
153
(EIA), which was conducted by a local state-owned engineering services company (NESPAK)
prior to commencement of construction in August 2015 (NESPAK 2015b).

NESPAK’s “comprehensive and complete” studies, which included the EIA and a 37-page
environmental management plan (EMP), were deemed by third-party evaluators to be
“compliant with international codes and standards” (NESPAK v. Mumtaz 2017). The EMP
identified—and suggested corrective measures for—a series of risks related to land acquisition
and resettlement, flora and fauna, air quality and noise level, public utilities, seismic hazard, and
the health and safety of workers (NESPAK 2015a). After recognizing that several heritage sites, as
defined by the Antiquities Act of 1975 (“Act”), would be affected by construction, NESPAK
affirmed the need “to avoid any interference with cultural heritage site(s) and public property as
far as possible” (NESPAK 2015b). Noting that heightened noise levels could affect the structural
integrity of cultural heritage sites, it called upon contractors to employ “noise barriers during
construction” (NESPAK 2015a). Even though the Act prohibits construction activity within 200
feet of heritage sites, based on these plans, the Director General (DG) of Archeology issued a
No Objection Certificate (NOC) in November 2015, “giving permission to carry on construction
within prohibited limits of 200 feet of protected antiquities” (Mumtaz v. Punjab 2016).

Lahore’s iconic 17th-century monument, Chauburji, was built by Mughal Emperor Shahjehan for his beloved
daughter Jahanara Begum and served as an entrance to a royal garden, is shown here with the Orange Line
in the background. It is one of 11 heritage sites affected by the project’s construction activities.
Photo Credit: Anam Hussain/AlJazeera

At the time, all environmental and social requirements under local laws appeared to have been
met, giving CR-NORINCO and its local subcontractors the go-ahead to proceed with
implementation. However, when construction crews began marking sites for demolition and

154
earthworks in October 2015, it became apparent to local communities and civil society groups
that “construction work [would] be carried out within 95 feet of Shalimar Gardens” and several
other heritage sites (Ghani 2015). Almost immediately, a group of prominent environmental
lawyers, urbanists, and rights advocates petitioned the Lahore High Court (LHC), arguing that
the issuance of the NOC was “not only arbitrary, malafide, patently illegal, without lawful
authority but also without application of independent mind” (Mumtaz v. Punjab 2016). Before
issuing the NOC, the DG of Archeology allegedly did not consider Pakistan’s commitments to
international conventions for heritage conservation and was pressured by the government “to
issue NOC within two days time without consulting any independent experts” (Mumtaz v. Punjab
2016).

After the government failed to provide satisfactory responses to these concerns, in January
2016, the LHC ordered an immediate suspension of project activities near 11 heritage sites. It
also asked the authorities to report on their adherence to all de jure requirements related to land
acquisition, noise levels, and solid waste management (Shaukat and Tanveer 2016). The court
order threw the provincial government into a frenzy, as it anticipated long implementation delays
that could prevent the project from reaching completion ahead of the July 2018 election. It
immediately engaged experts to conduct separate Structural and Heritage Impact Assessments
(SIA and HIA) and re-issued the NOC in July 2016—before the LHC issued its full verdict the
following month. The matter was finally settled by the Supreme Court of Pakistan (SCP) when it
rejected the government’s revised NOC on the same grounds, questioning the integrity of the
government’s actions that clearly sought to remove this roadblock.

After several additional hearings and engagements with international experts to ascertain the
true dangers from vibrations to the integrity of historic buildings, the SCP finally authorized the
project’s resumption in December 2017 on the condition that its 31-item strong list of
requirements would be implemented (NESPAK v. Mumtaz 2017). Within days, “Shehbaz speed”
was on full display, as construction around these sites resumed after a delay of nearly two
years.288 Notwithstanding these efforts, the Sharif administration was unable to complete the
project by the end of its term, ultimately allowing its chief political rivals from the Pakistan
Movement for Justice party to cut the red ribbon in October 2020.

The saga of this project during the early BRI period demonstrates that even when strong de jure
ESG safeguards are in place, the de facto implementation of such safeguards can falter or fail for
a wide variety of reasons. In some cases, local officials may be incentivized to prioritize speed
over safety. In other cases, they may lack technical knowhow to enforce standards or may not
fear penalties for non-compliance.

288
Lahore is the capital of the Punjab province and the political power base of the then-incumbent Pakistan
Muslim League party. Its leadership, including then-Prime Minister Nawaz Sharif and his younger brother,
Punjab Chief Minister Shehbaz Sharif, belong to the city’s business elite. Since first coming to power in the
mid-1980s, the Sharifs’ political strategy has hinged on flagship infrastructure projects, such as major new
international airports and inter-city motorways. During his first tenure in office after returning from exile in
2008, the younger Sharif delivered the Lahore Bus Rapid Transit (BRT) at “Shehbaz speed” within 10
months (Majid et al. 2018; Express Tribune 2016). His party was rewarded with a resounding electoral
victory in the 2013 elections, which it attributed to the BRT.
155
What then can we conclude based upon the available evidence? First, it is
increasingly common for Chinese donors and lenders to include ESG safeguard
provisions in their infrastructure financing agreements with LICs and MICs. These
provisions are broadly compatible with international ESG safeguards, such as the
OECD Council Recommendation on Common Approaches for Officially
Supported Export Credits and Environmental and Social Due Diligence, the IFC’s
Performance Standards on Environmental and Social Sustainability, the Uniform
Framework for Preventing and Combating Fraud and Corruption, the OECD
Council Recommendation on Bribery and Officially Supported Export Credits,
and the OECD Council Recommendation on Public Procurement. Second, many
of these de jure provisions go far beyond identifying rules and standards related
to ESG risk management; a rapidly expanding percentage of China’s overseas
infrastructure project portfolio is underpinned by financing agreements that
include mechanisms for monitoring compliance and/or sanctioning
noncompliance with those ESG rules and standards. Third, China’s de facto risk
mitigation efforts are on the rise. Fourth, there is a growing gap between how
ESG safeguards are applied to China’s overseas infrastructure projects in
principle (de jure) and in practice (de facto), which is not unexpected given that
ESG risk mitigation only recently became a priority for Beijing.

Section 4: Do ESG safeguards produce a project

performance payoff or penalty?
Given that Beijing has recently taken significant measures to mitigate ESG risks
in its overseas infrastructure project portfolio, an important question is whether
these course corrections are compatible with the country’s reputation for speed
and convenience. China did not become the Global South’s go-to banker for
big-ticket infrastructure because of a happy accident. It earned the position by
addressing a key source of unmet demand among LICs and MICs: financing for
infrastructure without overly complex and cumbersome ESG safeguard policies

156
and procedures (Dollar 2016; Swedlund 2017; Parks 2019; Humphrey and
Michaelowa 2019; Zeitz 2021; Horigoshi et al. 2022).289

In 2015, the G-24—a group of countries that work together to coordinate the
positions of developing countries on international monetary and financial
issues—gave voice to the frustrations of LICs and MICs in a report entitled
Infrastructure Finance in the Developing World:

"One aspect of the business practices of the World Bank and major
[regional multilateral development banks] that has a particularly
strong impact on infrastructure investment is environmental and
social safeguard policies. Safeguards comprise procedures and
restrictions on different types of lending operations meant to
‘safeguard’ the project from having negative impacts on the
environment and social groups. Safeguards were first instituted at the
World Bank in the 1990s, and the other major [regional multilateral
development banks] followed suit in subsequent years. The World
Bank’s safeguards are still considered the most comprehensive and
rigorous, but the safeguards of the AsDB, IADB, and AfDB have been
gradually tightened over the years such that the differences between
them are relatively small, particularly on the hot-button issues of
environmental assessment and resettlement. As a project undergoes
the initial screening process, MDB staff members determine whether
it triggers any of the MDB’s applicable safeguards. Should that be the
case, a separate series of special requirements must be followed
before the loan can be approved and disbursed. The most frequently
triggered safeguards in the case of the World Bank relate to
environmental assessment and involuntary resettlement, and most
frequently affect investment projects in the transportation, energy,

289
According to David Dollar, who served as the World Bank’s country director for China (2004-2009) and
the U.S. Treasury Department’s economic and financial emissary to China (2009-2013), “[the] procedures
developed by the World Bank are the gold standard of environmental and social safeguards in
infrastructure projects. However, they have had a number of unintended consequences. It has become
time-consuming and expensive to do infrastructure projects with the World Bank, and as a result,
developing countries have turned to other sources of funding. [...] Given this situation, the emergence of
China as a major funder of [...] infrastructure projects has been welcomed by most developing countries.
China is seen as more flexible and less bureaucratic. It completes infrastructure projects relatively quickly so
that the benefits are seen sooner” (Dollar 2016).
157
 and urban sectors. The required procedures are extraordinarily
detailed and specific, and in many cases [...] extremely difficult for
borrowers and even staff to fully understand. Requirements often
include time-consuming, lengthy studies to be undertaken by
third-party experts (usually at the government’s cost), lengthy
consultations with affected parties (sometimes including unelected
non-governmental organizations), extensive mitigations measures,
and lengthy mandatory prior public disclosure and comment periods
during which time the project cannot move ahead. These
requirements supersede whatever national laws may be in place in
the borrowing country—a particularly troubling point of principle for
many borrowing countries, beyond the practical impacts of
safeguards” (Humphrey 2015).

China, which is a member of the G-24 and the World Bank’s largest borrower,
appreciated these concerns (see Box 3a) and used them as a way to differentiate
its offering to the global infrastructure financing market. Under the banner of
“South-South cooperation,” it emphasized its solidarity with the Global South
and offered LICs and MICs an alternative model of development that prioritized
the rapid installation of “hardware” over “software” investments that focus on
policies and institutions.290 Beijing’s message resonated—so much so that it
became the developing world’s financier of first resort for highways, railroads,
dams, bridges, seaports, airports, power plants, and electricity grids, while the
MDBs downsized their infrastructure departments and programs due to a lack of
borrower demand.291 Several years ago, Chris Humphrey of ETH Zurich’s Center
for Development Cooperation and Katharina Michaelowa of University of Zurich
published interview evidence from three African countries on the changing
nature of borrower demand for infrastructure financing. They found that:

“[o]ne issue which officials in all three countries noted as limiting their
own demand for infrastructure lending from the World Bank and to a

290
At the Belt and Road Forum for International Cooperation in 2017, Xi Jinping described the BRI as “a
new option for other countries and nations who want to speed up their development while preserving their
independence” (Belt and Road Forum for International Cooperation 2017, emphasis added).
291
In 2010, the World Bank’s Independent Evaluation Group found that “[t]wo thirds of [World Bank]
managers interviewed reported that some clients had avoided or were dropping a [World] Bank project
because of safeguard policies” (IEG 2010: 73).
158
 lesser degree the AfDB is the ‘hassle factor’ implicit in these types of
projects from project design rules and environmental and social
safeguards. Officials from all governments concurred that the World
Bank is particularly difficult. ‘For hydroelectric and railroads, we don’t
even talk to them, we just go straight to the Chinese,’ said an
Ethiopian official. Discussing a major gas pipeline project, a Tanzanian
official said, ‘The Chinese are a bit more expensive, but they are a lot
easier and a lot faster for this kind of project. We didn’t even send a
request to the World Bank for support, we went straight to the
Chinese.’ Even in Malawi, with only small amounts of Chinese finance,
officials were experiencing these dynamics with a planned new
coal-fired power plant, to be funded by the Chinese at market-based
interest rates. ‘The World Bank and AfDB wouldn’t fund it because the
powerful shareholders would not agree to that kind of thing for
environmental reasons. So we went with the Chinese.’” (Humphrey
and Michaelowa 2019: 23)

In light of LIC and MIC demand for low levels of “hassle factor,” Beijing’s pivot
toward a more stringent ESG safeguard regime raises the question of whether it
will undermine the value proposition that it has traditionally used to differentiate
itself from competitors in the global infrastructure financing market. To better
understand the implications of Chinese lenders and donors adopting stronger
ESG safeguards, we use the 3.0 version of AidData’s GCDF dataset to compare
the performance of Chinese government-financed infrastructure projects with
and without strong ESG safeguards. We do so with six outcome measures: (1)
the percentage of projects that run behind schedule (2) the average length of
commencement delays, (3) the average length of completion delays, (4) the
average amount of time it takes to reach completion, (5) the frequency and
value of project suspensions and cancellations, and (6) ESG risk prevalence rates.

We begin by comparing the percentage of China’s infrastructure project

portfolio with and without strong de jure ESG safeguards that ran behind
schedule. We classify a project as “behind schedule” if its actual implementation
start date took place 3 months or more after its originally scheduled
implementation start date, and/or if its actual completion date took place 3

159
months (or more) after its originally scheduled completion date. Figure A58
demonstrates that China’s grant- and loan-financed infrastructure projects in LICs
and MICs are equally as likely to run behind schedule when strong de jure ESG
safeguards are in place as they are when strong ESG safeguards are not in place:
74% of infrastructure projects with strong de jure ESG safeguards ran behind
schedule, and 75% without strong de jure ESG safeguards ran behind schedule.

Next, we examine the average length of commencement delays for China’s

overseas infrastructure projects that are subject to ESG safeguards with varying
levels of stringency. Figure A59 provides evidence that China’s overseas
infrastructure projects encounter slightly shorter commencement delays when
they are undertaken with strong ESG safeguards (47 days shorter, on average,
than China’s overseas infrastructure projects without strong safeguards). The fact
that infrastructure projects with strong ESG safeguards take slightly less time to
launch than those without such protections is consistent with the old adage that
“an ounce of prevention is worth a pound of cure.” It also implies that Beijing
may be able to implement strong ESG safeguards without losing its competitive
edge. As a general rule, MDB-financed infrastructure projects with strong ESG
safeguards face substantially longer commencement delays. Charles Kenny of
the Center for Global Development estimates that ‘‘Category A’’ World Bank
projects—environmentally and socially sensitive projects subjected to the
organization’s most stringent safeguards—take 7.4 years (2,689 days), on
average, to move from the proposal stage to the disbursement (project
commencement) stage.292

The principle of “an ounce of prevention is worth a pound of cure” also

evidently applies to completion delays in Chinese grant- and loan-financed
infrastructure projects. Figure A60 provides evidence that China’s overseas
infrastructure projects face slightly shorter completion delays when strong ESG
safeguards are in place (91 days less, on average, than China’s overseas
infrastructure projects without strong ESG safeguards).293 A separate, but closely

292
The findings reported in Kenny (2023) are specific to the 2010-2017 time period. In 2010, a study by the
World Bank’s Independent Evaluation Group (IEG) also revealed that the average cost of safeguards for a
Category A project at the World Bank is $19 million (IEG 2010).
293
These findings are likely related to the findings on commencement delays. ESG safeguards often require
that contractors and their host country counterparts take a series of time-consuming actions—such as
conducting environmental impact assessments (EIAs) and preparing resettlement action plans
160
related, insight from the 3.0 version of AidData’s GCDF dataset is that it takes an
average of 3.2 years (1,163 days) to complete an infrastructure project without
strong ESG safeguards, and it takes 8 fewer days (1,155 days) to complete an
infrastructure project with strong ESG safeguards (see Figure A61).294 These
findings do not suggest that China’s reputation for speed is in jeopardy.

What then can we conclude about the “speed of implementation” differences

between infrastructure projects with and without strong ESG safeguards? The
most important point is that they are not large, which means that there is not
much evidence to support the idea that ESG safeguards impose a significant
project performance penalty. The conventional wisdom is that ESG risk
mitigation measures substantially impede infrastructure project implementation,
thereby undermining a key component of the value proposition (speed) that
China has traditionally used to differentiate itself from its competitors in the
global infrastructure financing market (Swedlund 2017; Parks 2019; Humphrey
and Michaelowa 2019; Zeitz 2021). However, our findings do not support this
argument. Quite the opposite: they suggest that China can reduce the ESG risk
profile of its overseas infrastructure portfolio if it is willing to accept slightly
longer project implementation timelines (measured in dozens of days rather than
hundreds or thousands of days).295

Another potential way that the performance of Beijing’s overseas infrastructure

project portfolio might vary based on ESG safeguard stringency is the likelihood
of project suspension or cancellation. The latest version of AidData’s GCDF
dataset demonstrates that infrastructure projects with strong de jure ESG
safeguards are substantially less vulnerable to suspension and cancellation after
the finalization of a Chinese grant or loan agreement. According to Figure 3.10,
while 74 Chinese grant- and loan-financed infrastructure projects (worth $43
billion) with weak de jure ESG safeguards have been suspended or canceled

(RAPs)—during the pre-implementation phase of a project, which can eliminate implementation obstacles
that would otherwise delay completion.
294
By way of comparison, it takes World Bank and Asian Development Bank projects, on average, 6 years to
move from the commencement stage to the completion stage (see Bulman et al. 2017: 362).
295
This finding is relevant to the loan repayment challenges that we document in Chapter 2 because the
speed of implementation can affect a project’s revenue generation potential and thus a borrower’s ability to
meet its loan repayment obligations. The CDB-financed Jakarta-Bandung High-Speed Railway Construction
Project, which is running over-budget and behind schedule, is a case in point. Since it was financed through
a limited recourse project finance transaction and the railway is not yet in operation, the borrower is unable
to make debt service payments via railway revenues (Malik and Parks 2021; Kuo 2021).
161
since 2000, only 7 Chinese government grant- and loan-financed infrastructure
projects (worth $11 billion) with strong de jure ESG safeguards have been
suspended or canceled since 2000.296 These findings suggest that the
application of more stringent ESG safeguards may help rather than hinder
Beijing’s efforts to de-risk its overseas infrastructure project portfolio in the
developing world.

Figure 3.10: Monetary value of project suspensions and cancellations by de jure ESG safeguard strength

Figure 3.10

Notes: Strong and weak de jure ESG safeguards are defined in Section A-9 of the Appendix.

Finally, we can use the 3.0 version of AidData’s GCDF dataset to determine
whether ESG risk prevalence rates in China’s overseas infrastructure project
portfolio vary according to ESG safeguard stringency. Figure 3.11 compares the
percentage of China’s grant- and loan-financed infrastructure project portfolio
with significant ESG risk exposure across two cohorts: projects with and without
strong de jure ESG safeguards. Whereas 82% of projects that lacked strong de
jure ESG safeguards faced significant ESG risks, only 18% of projects with such
safeguards encountered similar risks.297 Figures A53, A54, and A55 in the
Appendix demonstrate that these patterns are equally applicable to all three
types (environmental, social, and governance) of ESG safeguards.

296
Our findings are consistent with those of Lu et al. (2023b). They find that Chinese-financed power plant
projects posing higher levels of environmental risk are more likely to be suspended.
297
Figure A49 tracks the same two cohorts over time. Notwithstanding a sharp increase in the
percentage of the infrastructure project portfolio subjected to strong de jure ESG safeguards
during the late BRI period, it shows that the same empirical pattern is generally consistent across
the 2000-2021 period.
162
Figure 3.11: Proportion of infrastructure project portfolio facing significant ESG risk exposure by level of safeguard stringency

Figure 3.10

Notes: Strong and weak de jure ESG safeguards are defined in Section A-9 of the Appendix. The presence
of significant ESG risk exposure is based on a project-level composite measure that is described in Section
2 of Chapter 3.

Section 5: Decoding Beijing’s ESG risk mitigation strategy

In the remainder of this chapter, we will analyze the 3.0 version of AidData’s
GCDF dataset to better understand how Beijing is seeking to manage and

163
mitigate ESG risks in its infrastructure project portfolio. We see evidence of
Chinese state-owned financiers taking four ESG risk mitigation efforts to:

1. Defund the bilateral development finance institutions with the weakest

safeguards

2. Support bilateral lenders and donors with the strongest safeguards

3. Outsource risk management via syndication and multilateralization

4. Unwind relationships with high-risk countries and double down on

relationships with low-risk countries

Risk mitigation strategy #1: Defund the bilateral development

finance institutions with the weakest safeguards
Most of Beijing’s official statements and publications about de-risking its
overseas infrastructure project portfolio are anodyne and difficult to interpret.
The “Green Development Guidelines for Foreign Investment and Cooperation”
that China’s Ministry of Commerce and Ministry of Ecology and Environment
issued in July 2021 state that “[t]he greening of outbound investment and
cooperation must be guided by Xi Jinping Thought on Socialism with Chinese
Characteristics for a New Era. In the process, we must implement the spirit of
the 19th CPC National Congress and the 2nd, 3rd, 4th and 5th Plenary Sessions
of the 19th CPC Central Committee, Xi Jinping Thought on Ecological
Civilization and the decisions of the CPC Central Committee and the State
Council. We must stay committed to the new development concept, striving for
the strong awareness of green development, efficient use of resources, strict
protection of the environment and effective control of carbon emissions. We
should work to showcase China’s leadership in global endeavor toward green
transition and our commitment to building the world into a better and cleaner
place and laying the groundwork for a new development paradigm” (MOFCOM
and MEE 2021).

However, when you cut through the flowery rhetoric used by Chinese politicians
and bureaucrats by following the money, a stark reality emerges: Beijing is

164
entrusting a shrinking proportion of its overseas infrastructure project portfolio
to the country’s policy banks (CDB and China Eximbank), which have particularly
weak de jure ESG safeguards (see Table 3.1 and Table 3.2). Figure 3.12 plots two
trends over time: the percentage of Chinese grant- and loan-financed
infrastructure projects with weak de jure ESG safeguards and the percentage of
China’s infrastructure project portfolio financed via bilateral loans from the
country’s policy banks from 2000 to 2021. Beijing’s reliance upon policy bank
financing for infrastructure projects in LICs and MICs plummeted from 86% in
2013 to 41% in 2021.298 The year-on-year changes that took place during this
period also track very closely with year-on-year changes in the percentage of
China’s overseas infrastructure project portfolio bankrolled by institutions with
weak de jure ESG safeguards (see Figure 3.12).

298
This 45 percentage point decline obscures some differences across the two policy banks. Whereas the
percentage of China’s infrastructure project portfolio in LICs and MICs financed via CDB declined from
39.79% in the pre-BRI period (2000-2013) to 11.61% during the late BRI period (2018-2021), the
percentage financed via China Eximbank actually increased from 38.06% to 48.75% across these two
periods (see Table 3.2). However, upon closer inspection, one can see that the percentages of China’s
infrastructure project portfolio in LICs and MICs financed via CDB and China Eximbank declined (for the
most part) over the course of the late BRI period. Figures A56 and A57 demonstrate that the percentage of
the portfolio financed via China Eximbank fell from 58.7% in 2018 to 18.3% in 2021 and the percentage of
the portfolio financed via CDB fell from 11.9% in 2018 to 4.3% in 2020 before ticking back up to 22.5% in
2021.
165
 Figure 3.12: Composition of infrastructure project portfolio: reliance upon the policy banks and weak de jure ESG safeguards

Figure 3.12

Notes: Weak de jure ESG safeguards are defined in Section A-9 of the Appendix.

Table 3.2: China’s grant- and loan-financed infrastructure portfolio by type of financing instrument over time

Table 3.2
China’s grant- and loan-financed infrastructure portfolio by
type of financing instrument over time

Pre-BRI Early BRI Late BRI

Contract Category (2000-2013) (2014-2017) (2018-2021)

Bilateral China Eximbank loan 38.06% 43.84% 48.75%

Bilateral CDB loan 39.79% 23.93% 11.61%

Bilateral MOFCOM loan or grant 2.64% 2.68% 3.10%

Bilateral Chinese state-owned commercial bank loan 5.14% 9.38% 8.84%

Syndicated loan with Chinese and multilateral bank participants 0.80% 2.17% 1.45%

Syndicated loan with Chinese state-owned commercial banks 8.86% 16.91% 20.39%
and/or policy banks
PBOC/MOF grant or loan channeled through multilateral 0% 0.12% 0.89%
institution

166
 Pre-BRI Early BRI Late BRI
Contract Category (2000-2013) (2014-2017) (2018-2021)

Supplier's credit from Chinese SOE 4.71% 0.77% 4.97%

Total 100% 100% 100%

Notes: This table presents the shares of China’s grant- and loan-financed infrastructure project portfolio
(measured in constant 2021 USD) in LICs and MICs delivered via 8 financial instrument categories across
three different time periods: (1) the pre-BRI period from 2000 to 2013, (2) the early BRI period from 2014 to
2017, and (3) the late BRI period from 2018 to 2021.

Risk mitigation strategy #2: Support bilateral lenders and donors

with the strongest safeguards
Across the various bilateral instruments that Beijing has at its disposal to bankroll
infrastructure projects in the developing world, CDB and China Eximbank loans
offer the weakest de jure ESG safeguards. Rather than relying on these policy
banks, Beijing is increasingly turning to a different set of financial
institutions—with lending and grant-giving instruments that include a more
stringent set of de jure ESG safeguards—to finance infrastructure projects in
LICs and MICs.

In Section 3 of Chapter 3, we discovered that bilateral loans from the country’s

state-owned commercial banks, supplier’s credits from the country’s state-owned
enterprises, and MOFCOM grants and interest-free loans have stronger de jure
ESG safeguards than CDB and China Eximbank loans (see Table 2.1 in Chapter
2). This finding begs the question: has Beijing increased its use of these bilateral
infrastructure financing instruments? Table 3.2 above demonstrates that it has in
fact done so, albeit in an incremental way: whereas the proportion of China’s
infrastructure project portfolio in LICs and MICs financed through these
instruments amounted to 12.49% during the pre-BRI period (2000-2013) and
12.83% during the early BRI period (2014-2017), it jumped up to 16.91% during
the late BRI period.

Risk mitigation strategy #3: Outsource risk management via

syndication and multilateralization
Another strategy that Beijing could pursue to de-risk its overseas infrastructure
project portfolio is outsourcing risk management to multilateral institutions,
167
which are widely considered to be norm-setting, industry leaders in the design
and implementation of ESG safeguards (Park 2010; Martin et al. 2015; Buntaine
2016; Brauneder et al. 2018; Buchanan et al. 2018; Narain et al. 2020, 2022).

One path to multilateralization is to directly entrust the management of grants

and loans to an intergovernmental organization. For example, through its $2
billion trust fund at the African Development Bank (known as the Africa Growing
Together Fund) and its $2 billion trust fund at the Inter-American Bank (known as
the China Co-Financing Fund for Latin America and the Caribbean), the PBOC
has fully delegated project design, preparation, implementation, and
supervision responsibilities (including ESG safeguard application) to the
multilateral institutions.

Another multilateralization option is to participate in syndicated loan

agreements that are arranged by multilateral institutions. Many of these
agreements have an A/B structure, whereby an MDB serves as the
lender-of-record and keeps a part of the loan for its own account (the “A-loan”)
while selling participation in the remainder of the loan (the “B-loan”). A common
feature of A/B syndicated loan agreements with multilateral arrangers is that all
B-loan participants yield authority to the A-loan provider for risk mitigation
purposes (Esty and Megginson 2003; Bae and Goyal 2009; Hainz and Kleimer
2012; Broccolini et al. 2021; Lu et al. 2023a).

According to Table 3.2 above, Beijing has modestly increased its reliance upon
multilateral institutions over time. The proportion of China’s infrastructure project
portfolio in LICs and MICs financed via syndicated loans with multilateral
participants and PBOC/MOF loans and grants entrusted to multilateral
institutions rose from 0.8% during the pre-BRI period (2000-2013) to 2.34%
during the late BRI period (2018-2021). Despite the small size of this increase, it
is noteworthy because PBOC and MOF grants and loans entrusted to
multilateral institutions and syndicated loans with Chinese bank and multilateral
institution participants have the most stringent ESG safeguards in our sample of
infrastructure financing agreements (see Table 3.1 in Section 3).

168
Syndicated loans with Chinese policy bank and state-owned commercial bank
participants may be analogous to syndicated loans with multilateral participants
if all members of the syndicate generally defer to the lending institution(s) with
the strongest preference(s) for ESG risk mitigation. Consistent with this
expectation, Table 3.1 above provides evidence that the ESG safeguards of
state-owned commercial banks do indeed prevail over those of the policy banks
in syndicated loan arrangements.299

Beijing has also intensified its use of these bilateral infrastructure financing
instruments over time. Figure 3.13 presents the percentage of Chinese grant-
and loan-financed infrastructure projects with strong de jure ESG safeguards in
conjunction with the percentage of China’s infrastructure project portfolio
financed via syndicated loans with Chinese policy bank and state-owned
commercial bank participants from 2000 to 2021. Beijing’s use of these types of
syndicated loan arrangements for infrastructure projects in LICs and MICs has
increased dramatically—from 0% in 2000 to 41% in 2021—and in tandem with
the usage of strong de jure ESG safeguards.300 The year-on-year changes that
took place over this twenty-two year period track closely with year-on-year
changes in the percentage of China’s overseas infrastructure project portfolio
bankrolled by institutions with strong de jure ESG safeguards.

299
Sufi (2007) demonstrates that lead arrangers reduce the costs of due diligence for all other syndicate
participants.
300
According to Table 3.2 above, the percentage of China’s infrastructure project portfolio in LICs and MICs
financed via syndicated loans involving state-owned policy banks and commercial banks increased from
8.86% during the pre-BRI period (2000-2013) to 16.91% during the early BRI period (2014-2017) and
20.39% during the late BRI period (2018-2021).
169
Figure 3.13: Composition of infrastructure project portfolio: Use of syndicated loans and strong de jure ESG safeguards

Figure 3.13

Notes: Strong de jure ESG safeguards are defined in Section A-9 of the Appendix.

Risk mitigation strategy #4: Unwind relationships with high-risk

countries and double down on relationships with low-risk countries
One additional way to reduce the ESG risk profile of an infrastructure project
portfolio is to use information about the past performance of host countries to
guide future lending and grant-giving activities. That is to say, once a donor or
lender has identified the subset of infrastructure projects in its portfolio that
have presented significant ESG risks, it may seek to identify where these projects
are geographically concentrated so that it can identify host countries posing
especially high levels of ESG risk and redirect funding for future infrastructure
projects elsewhere. This reallocation—or “selectivity”—practice is followed by
several major MDBs, including the Asian Development Bank and the World Bank
(Buntaine 2011, 2015, 2016).

170
Figure 3.14: Proportion of infrastructure portfolio allocated to LICs/MICs with high ESG risk prevalence rates

Figure 3.14

Notes: This figure compares the overall percentage of China’s infrastructure financing to LICs and MICs that
was allocated to countries with high ESG risk prevalence rates in two time periods: (1) the pre-BRI and early
BRI period (2000-2017) and (2) the late BRI period (2018-2021). Countries with a high ESG risk prevalence
rate are defined as those where at least 75% of China’s grant- and loan-financed infrastructure project
portfolio between 2000 and 2017 faced significant ESG risks. The presence of significant ESG risk exposure
is based on a project-level composite measure that is described in Section 2 of Chapter 3.

In order to gauge whether Beijing moved in this direction during the late BRI
era, we first create two cohorts of host countries: countries where at least 75% of
China’s grant- and loan-financed infrastructure project portfolio had significant
ESG risk exposure between 2000 and 2017 and countries where less than 75%
of China’s grant- and loan-financed infrastructure projects had significant ESG
risk exposure between 2000 and 2017. We then compare Beijing’s provision of
infrastructure financing to these two cohorts between 2018 and 2021. Figure
3.14 demonstrates that 2.8% of infrastructure financing from Beijing during the
late BRI era was directed to 9 LICs and MICs where at least 75% of China’s grant-
and loan-financed infrastructure project portfolio had significant ESG risk
exposure between 2000 and 2017. By way of comparison, Beijing allocated a
substantially larger proportion (6.83%) of its grant- and loan-financed
infrastructure project portfolio to the same 9 countries between 2000 and 2017.
This pattern is consistent with the idea that Beijing has rebalanced the
cross-country allocation of aid and credit to reduce the ESG risk profile of its
overseas infrastructure project portfolio.

Given that China has scaled back infrastructure spending in countries where its
projects have faced particularly high levels of ESG risk exposure, another way
that it could seek to recalibrate its portfolio is by ramping up support for
infrastructure projects in countries where its projects have faced particularly low

171
levels of ESG risk exposure. Figure 3.15 provides evidence that Beijing has in
fact moved in this direction. More specifically, it demonstrates that during the
late BRI period (2018-2021) 7.6% of infrastructure financing from China was
directed to 6 LICs and MICs where less than 10% of Chinese grant- and
loan-financed infrastructure projects had significant ESG risk exposure between
2000 and 2017. This represented a significant increase in late BRI era spending
for low-risk countries, as Beijing allocated only 1.73% of its grant- and
loan-financed infrastructure project portfolio to the same 6 countries between
2000 and 2017.

Figure 3.15: Proportion of infrastructure portfolio allocated to LICs/MICs with low ESG risk prevalence rates

Figure 3.15

Notes: This figure compares the overall percentage of China’s infrastructure financing to LICs and MICs that
was allocated to countries with low ESG risk prevalence rates in two time periods: (1) the pre-BRI and early
BRI period (2000-2017) and (2) the late BRI period (2018-2021). Countries with a low ESG risk prevalence
rate are defined as those where less than 10% of China’s grant- and loan-financed infrastructure project
portfolio between 2000 and 2017 faced significant ESG risks. The presence of significant ESG risk exposure
is based on a project-level composite measure that is described in Section 2 of Chapter 3.

Section 6: Is Beijing course-correcting or virtue-signaling?

When Beijing first signaled interest in subjecting its overseas infrastructure
project portfolio to more stringent ESG safeguards, critics and rivals were quick
to question its sincerity. Jonathan E. Hillman of the Center for Strategic &
International Studies (CSIS) penned an op-ed, in which he argued that China was
engaged in a “greenwashing effort.” He wrote that “Xi knows that China is
vulnerable on environmental issues. At a gathering of world leaders in Beijing
last year, ‘green’ and ‘sustainable’ were the salt and pepper in his speech about
the next phase of BRI. To address criticism, China unveiled no less than 11 new,
green initiatives under BRI, all of which are voluntary, and none of which are

172
binding or transparent” (Hillman 2020b). Kelly Sims Gallagher and Qi Qi of Tufts
University wrote that “Chinese government rhetoric about greening the BRI is
laudable, but it has yet to make any substantive changes toward that goal” (Sims
Gallagher and Qi 2021). The U.S. and its allies also rejected the notion that
Beijing’s overseas infrastructure projects had robust ESG protections in place. In
November 2019, the U.S., Japan, and Australia announced that they were
joining forces to establish a “Blue Dot Network” that would “evaluate and
certify nominated infrastructure projects based upon adherence to commonly
accepted principles and standards” and “promote market-driven, transparent,
and financially sustainable infrastructure development in the Indo-Pacific region
and around the world.” More recently, the U.S., the U.K., and the other
members of the G7 have promoted a Partnership for Global Infrastructure and
Investment (PGII)—previously known as the Build Back Better World (B3W)
initiative—that they characterize as an alternative to the BRI and an option for
countries that want to undertake infrastructure projects in strict accordance with
internationally accepted ESG safeguards.301

However, this chapter demonstrates—with many new sources and types of

evidence—that Beijing is not simply engaging in an international virtue-signaling
exercise. It has taken meaningful steps to de-risk its overseas infrastructure
project portfolio by ramping down the international lending activities of banks
that lack strong ESG risk management guardrails, ramping up the provision of
infrastructure financing via institutions that have strong ESG safeguards in place,
gradually unwinding aid and credit relationships with LICs and MICs that present
high level of ESG risk, and redirecting new infrastructure financing to lower-risk
countries. At the same time, it is still at a relatively early stage in its journey from
ESG skeptic to advocate and it will likely take many years—potentially even
decades—to close the gap between the de jure and de facto application of ESG
safeguards.302

301
In September 2023, the U.S., France, Germany, Italy, the EU, India, the UAE, and Saudi Arabia also
announced plans to develop an India-Middle East-Europe Economic Corridor (IMEC).
302
It is also important to keep in mind that, as of 2021, 40% of Chinese infrastructure financing to LICs and
MICs was still being channeled via bilateral CDB and China Eximbank grants and loans. This is significant,
since this chapter provides evidence that CDB and China Eximbank have for the most part not
reformed/modernized their de jure ESG safeguards in a way that is comparable to the practices of
multilateral institutions, state-owned commercial banks, or the lead arrangers of syndicated loans.
173
We do not see evidence of Chinese development finance institutions uniformly
complying with international ESG safeguards. Rather, we see evidence that
some Chinese lenders and donors are gradually and selectively harmonizing
their ESG safeguard policies and practices with those of traditional donors and
lenders.303 These changes should give pause to Beijing’s competitors in the
global infrastructure financing market. The G7 and some MDBs are currently
trying to convince would-be partners in the developing world that (a) the BRI is a
low-quality infrastructure option (privileging speed and convenience over safety
and long-term sustainability) and (b) they can provide alternative, high-quality
financing options for countries that want to undertake infrastructure projects
based on strict adherence to “international best practice” ESG safeguards.
However, this black-and-white branding strategy may lack resonance with its
target audience, as LICs and MICs have already made it very clear that they have
low levels of appetite for “gold standard” ESG safeguards. They want financing
partners that can quickly design and implement big-ticket, high-impact
infrastructure projects without unreasonably high levels of ESG risk.304 The
evidence in this chapter suggests that Beijing may be better-positioned to
answer this call than its competitors realize. It is now delivering large-scale
infrastructure projects with increasingly robust ESG safeguards but without the
lengthy implementation delays that often hobble similar projects backed by G7
members and MDB

303
There are reasons to believe that this approach of gradual and selective harmonization will be a
longer-term process. On July 16, 2021, China’s Ministry of Commerce and the Ministry of Ecology and
Environment issued “Green Development Guidelines for Overseas Investment and Cooperation,” which
recommend that project sponsors and contractors comply with international standards or Chinese
standards when the laws and regulations of host countries are vague or weak.
304
On this point, see Humphrey 2015; Dollar 2016; Swedlund 2017; Humphrey and Michaelowa 2019; and
Zeitz 2021.
174
Chapter 4: Reputational Rehabilitation on the
Belt and Road

Section 1: China’s quest for soft power during the BRI era
Great powers use aid and credit to expand their global influence.305 These tools
of economic statecraft can be particularly useful for winning support from
developing countries. During the Cold War, the Eastern and Western blocs used
grant-giving and lending instruments to promote their competing ideologies
and strengthen alliances around the world (Heurlin 2020). China’s BRI stands out
as another striking example of this approach during the 21st century. Beijing has
attempted to bolster its global influence by offering demand-responsive
infrastructure financing and rapid project delivery to developing countries
(Dreher et al. 2022). Prior to the unveiling of the BRI, Beijing tried to keep a
relatively low profile as an international donor and lender, adhering to Deng
Xiaoping’s “hide your capabilities and bide your time” principle. However, the
BRI marked an important shift. China became more proactive about cultivating
its brand and broadcasting positive messages about its overseas activities. It did
so by placing signage at project sites, organizing high-profile ceremonies to
celebrate the commencement and completion of projects, cultivating
relationships with journalists to encourage positive media coverage of project
accomplishments, and forging content-sharing partnerships with radio stations,
television channels, and newspapers (Custer et al. 2018, 2019; Wellner et al.
2023).

Washington sees the BRI as Beijing’s attempt to “reshape the international order
[and] [...] the rules of the road” by “us[ing] its economic power to coerce
countries” (White House 2022). The 2022 U.S. National Security Strategy
emphasizes the importance of “out-competing China” and calls upon
government agencies to “partner with, support, and meet the economic and
development needs of partner countries” (White House 2022). However, given
the difficulty of competing dollar-for-dollar with Beijing, Washington is

305
On this point, see Goldsmith et al. (2014), Blair et al. (2022a), Wellner et al. (forthcoming), and Asmus et
al. (forthcoming).
175
increasingly aware of the fact that it will need to carefully select countries,
sectors, and delivery instruments that offer the greatest return on investment.

U.S. policymakers have also demonstrated interest in understanding the causes

and consequences of BRI backlash—and Beijing’s response to it. On its 10th
anniversary, the BRI faces an array of reputational challenges in host countries
where enthusiasm for the initiative has waned. Debt repayment challenges
(discussed in Chapter 2) and ESG problems related to infrastructure project
implementation (discussed in Chapter 3) have led a growing number of LICs and
MICs to reevaluate the risks and rewards of continued participation in Beijing’s
“project of the century” (Horigoshi et al. 2022).

In this chapter, we seek to explain when, where, and how Chinese grant- and
loan-financed projects have become reputational assets or liabilities for China.
We also examine how Beijing has recalibrated its strategies and tactics in
countries where it has encountered public antipathy, unfavorable media
sentiment, or insufficient support among governing elites. With data on public
opinion, media sentiment, elite support, and Chinese grant and loan
commitments during the first eight years of the BRI era (2014-2021), we use an
“action-reaction” framework to examine how soft power gains and losses in LICs
and MICs during the early years of the initiative (2014-2017) influenced Beijing’s
responses in the so-called “BRI 2.0” era (2018-2021).

The pursuit of soft power for foreign policy advantage

States frequently use non-coercive tools—or so-called soft power
instruments—to win the support of foreign countries on issues of major national
importance (Nye 2004; Goldsmith and Horiuchi 2012; Goldsmith et al. 2014;
Guiso et al. 2009; Rose 2016, 2019).306 For China, these issues include Taiwan,
Tibet, Xinjiang, Falun Gong, human rights, and maritime claims in the South
China Sea (Flores-Macías and Kreps 2013; Kastner 2016; Strüver 2016, 2017;
Custer et al. 2018).

306
A popular definition of soft power is “the ability to achieve goals through attraction rather than
coercion” (Nye 2004: x).
176
Maintaining international support for the BRI is also a key priority. In October
2021, we observed Beijing’s readiness to manage reputational risk related to its
flagship, global infrastructure initiative. Within two days of the release of an
AidData report on public debt exposure to China, a leading English daily in
Pakistan, Dawn, wrote a hard-hitting editorial calling for greater transparency in
the planning and financing of the China-Pakistan Economic Corridor (CPEC).
Five days later, in response, two Cabinet-level ministers conducted a live press
conference on national television in which they presented a point-by-point
response to the editorial and the report (Malik et al. 2021; Dawn 2021). During
subsequent meetings with senior officials in Pakistan, we learned that the press
conference was organized in part due to pressure from the Chinese embassy,
which had expressed frustration at the “public relations impacts” resulting from
a rare instance of negative press coverage about the CPEC.307

Beijing uses a wide array of soft power investments—including scholarship

programs that allow foreign nationals to study in China, training programs that
expose foreign government officials to China’s policy positions, junkets for
foreign journalists to visit China and learn about its experiences and
perspectives, and political party outreach and capacity building activities—to
rally international support for its priorities and policy positions. According to
Dong and Chapman (2008: 162), China has provided thousands of university
scholarships to African students each year with the aim of “training future
leaders [...] who might serve as opinion leaders once back in their home
countries.” In an analysis of Beijing’s effort to cultivate relationships with African
political parties, Aiping (2015) finds that African countries frequently dispatch
delegations to China to learn more about its model of economic development
and system of governance, and that the Communist Party of China (CPC) uses
these exchanges to extol the virtues of the ruling party maintaining internal
stability to promote long-run economic growth and development.

307
Gelpern et al. (2022: 26, emphasis added) report a similar episode involving “a video obtained and
released by investigative journalists that revealed the terms of Ecuador’s multi-billion dollar oil-backed debt
to CDB. The release of the video shortly after the deal was signed prompted public debate about the new
borrowing [...]. In response, the head of CDB’s Resident Mission in Ecuador wrote to his counterpart in
Ecuador’s Ministry of Finance, complaining about the borrower’s apparent breach of [a] confidentiality letter,
called on the Ecuadorian government to launch a leak investigation and demanded that it take measures to
mitigate the reputational damage to CDB caused by the video. The CDB letter also implicitly threatened to
withhold future financing if the borrower did not adequately address the incident.”
177
Anecdotal evidence suggests that these soft power investments may be
effective. Consider, for example, the cohort of countries that agreed to abstain
on a key UNGA vote related to Russia’s invasion of Ukraine.308 On March 2, 2022,
UNGA passed a resolution condemning Russia’s aggression and calling on it to
“immediately, completely and unconditionally withdraw all of its military forces
from the territory of Ukraine.” 141 countries voted in favor of the resolution, 5
countries voted against it, and another 35 countries abstained. China abstained
and many of its largest aid and credit recipients—including Angola, Bangladesh,
Bolivia, Burundi, Central African Republic, Congo-Brazzaville, Cuba, El Salvador,
Ethiopia, Gabon, Guinea, Iran, Kazakhstan, Kyrgyzstan, Laos, Mongolia,
Mozambique, Namibia, Pakistan, South Africa, Sri Lanka, Sudan, Tajikistan, Togo,
Uganda, Uzbekistan, Vietnam, and Zimbabwe—followed suit (White and Holtz
2022).

Yet great powers do not use soft power instruments merely to secure
international support on discrete policy issues. They also strive for deeper
ideological alignment and widespread admiration of their economic
development and governance models (Repnikova 2022). Effective, long-term
soft power projection requires attractional influence: resonant ideas and values
that shape how governing elites in other countries diagnose problems, think
about cause-and-effect relationships, identify desirable policy outcomes, assign
priority among competing objectives, and determine how policies should be
formulated and implemented (Kroenig et al. 2010; Atkinson 2010).

For many of China’s target audiences in the Global South, its model of economic
development is attractive because of its apparent success: it successfully lifted
680 million people out of poverty in a generation (Ravallion 2009; Lin and Wang
2014).309 China’s model, which prioritizes the rapid installation of large-scale
physical infrastructure via state-led investment, is central to the way that it
308
Another interesting example is the policy position that many African governments take on China’s
treatment of Uyghurs living in Xinjiang Province. Rather than jumping on the Western bandwagon and
criticizing China for its alleged human rights abuses, African governments increasingly push back, arguing
that “Western forces [are] hyping up the so-called Xinjiang-related issues [to] serve their own ulterior
motives” (Olewe 2021).
309
According to Halper (2010), “China’s governing model is more appealing to the developing world and
some of the middle-sized powers than America’s market-democratic model. Given the choice between
market democracy and its freedoms and market authoritarianism and its high growth, stability, improved
living standards, and limits on expression—a majority in the developing world and in many middle-sized,
non-western powers prefer the authoritarian model.”
178
administers its overseas grant-giving and lending program (Ansar et al. 2016).310
This approach supposedly allows countries to leapfrog the process of
establishing liberal institutions and democratic values, which were prerequisites
for economic and social progress in Western countries (Naughton and Tsai
2015).

For political leaders seeking reelection within four- to five-year cycles and
autocrats vying for legitimacy on the basis of performance, China’s model of
economic development can be especially compelling. Arthur Mutambara,
Zimbabwe’s former Deputy Prime Minister, told the Wall Street Journal in 2011
that “China is my favorite country. [...] China's model is telling us you can be
successful without following the Western example” (Wonacott 2011). Several
years earlier, Abdoulaye Wade, the former President of Senegal, admonished
Western aid agencies and multilateral development banks for their insensitivity
to the conditions facing policymakers in developing countries and praised
China’s model: “China’s approach to our needs is simply better adapted than
the slow and sometimes patronizing post-colonial approach of European
investors, donor [agencies] and nongovernmental [organizations]. [...] I am a firm
believer in good governance and the rule of law. But when bureaucracy and
senseless red tape impede our ability to act—and when poverty persists while
international functionaries drag their feet—African leaders have an obligation to
opt for swifter solutions” (Wade 2008).

The effectiveness of China’s soft power instruments

But China’s soft power playbook goes beyond bankrolling and building
big-ticket infrastructure projects. Not unlike Western powers, it deploys a toolkit
that includes cultural diplomacy through Confucius Institutes, exchange
diplomacy using student scholarships and journalist visits, and elite-to-elite

310
An important feature of China’s model is its “portfolio approach.” Chin and Gallagher (2019: 256)
explain that “[w]hereas Western-backed [development finance institutions] and [multilateral development
banks] conduct individual project financing, China’s policy banks, at home and abroad, take a [...] portfolio
approach and finance what they refer to as ‘strategic credit spaces’ where bundles of loans or lines of credit
are issued for an array of coordinated and corresponding projects.” Coordinated public investment
strategies have a rich intellectual history related to “big push” theory (Rosenstein-Rodan 1943) and “growth
pole” theory (Perroux 1950; Hirschman 1958). According to Chin and Gallagher (2019: 251), “[s]ome in the
senior ranks of the Chinese state policy banks have drawn inspiration from [“big push” theory], including
the former chief economist at the China Development Bank [CDB], Lixing Zou, who saw CDB as having
played such a coordinating role within the Chinese growth miracle.”
179
diplomacy in the form of leadership visits (Benabdallah 2021; Custer et al. 2018).
Beijing has also put in place a well-resourced strategic communications strategy
with “distinctive” Chinese characteristics (Shambaugh 2015; Snow 2022). It has
dramatically expanded the global reach of official media organizations, such as
Xinhua News Agency and China Radio International.311 China’s messaging
operations are, in turn, buttressed by content-sharing partnerships with local
media, which republish stories carrying official Chinese narratives (Custer et al.
2019).

A growing body of statistical evidence suggests that China has effectively

utilized soft power instruments to influence public opinion, media sentiment,
and the foreign policy priorities of governing elites, albeit with varying degrees
of success (Brazys and Dukalskis 2019; Eichenauer et al. 2021; Blair et al. 2022a;
Dreher et al. 2022; Wellner et al. forthcoming, 2023; Asmus et al. forthcoming).
Wellner et al. (forthcoming) find that, on average, the completion of each
additional Chinese development project leads to a 3 percentage point increase
in public support for the Chinese government in the short term, although this
effect erodes over time.312 Brazys and Dukalskis (2019: 557) provide evidence
that geographical proximity to an active Confucius Institute “significantly and
substantively improves the tone of media reporting about events relevant to
China.”

The authorities in Beijing reportedly value more favorable media sentiment and
public sentiment because they believe it can “filter up and influence elite policy
to be more amenable to [their own] interests” (Brazys and Dukalskis 2019: 567).
However, Beijing also seeks to directly influence the foreign policy priorities of
governing elites. Dreher et al. (2022) provide evidence that a 10% increase in
voting alignment with China in the UN General Assembly yields a 276% increase

311
In Africa, its primary export, CGTN Africa, has been less successful in bringing China’s message to the
world than BBC Africa due to a lack of quality staffing (Zhang and Ong’ong’a 2021).
312
Building upon these model results, Wellner et al. (2023) estimate the country-specific magnitudes of
these public opinion effects by assuming a counterfactual scenario in which the Chinese government
equally distributed development projects across all countries. If Cambodia received 30 Chinese
development projects (the average number of projects over the 2006-2017 period of study) rather than the
91 that it actually received, the authors estimate that China would have suffered a 12.55 percentage point
loss of public support in that country.
180
in aid and credit from China, on average.313 These results suggest that Beijing,
like some of its rivals, uses its foreign aid and credit to encourage and/or reward
countries to support its foreign policy positions.314

Yet much remains unknown about China’s soft power strategy in the Global
South. Is Beijing responsive to soft power gains or losses vis-à-vis the U.S.? What
types of reputational liabilities or assets lead to competitive responses? Where,
when, and how is Beijing using its development finance instruments to make soft
power gains?

Leveraging the broad temporal and geographical coverage of the 3.0 version of
AidData’s GCDF dataset, we examine these questions with an “action-reaction”
framework. Based on China’s prior responses to changes in public opinion,
media sentiment, and elite support in LICs and MICs, we seek to anticipate
where and how China will mount competitive responses in the future.315 We
place special emphasis on understanding how China allocates limited
competitive resources across “safe bet,” “toss-up,” and “moonshot” countries.

Section 2: Vying for soft power: China’s play for public

opinion, media sentiment, and elite support
Our action-reaction framework evaluates Beijing’s aid and credit allocation
decisions during the late BRI period (2018-2021) after it observed various
changes on three measures of soft power during the early BRI period
(2014-2017): public support for the Chinese government, as measured via the

313
We thank Axel Dreher, Andreas Fuchs, Austin Strange, and Mike Tierney for generating and sharing
supplementary evidence derived from a statistical model in the fifth chapter of their book Banking on
Beijing.
314
Previous research demonstrates that foreign powers other than China use aid and credit to influence the
foreign policy positions of developing countries (Alesina and Dollar 2000; Kuziemko and Werker 2006;
Vreeland and Dreher 2014; Rose 2018).
315
China’s system of governance is often characterized as “fragmented authoritarianism,” where multilevel
power centers exist, each with their own listening posts, interests, and decision-making processes, which
can make it difficult to generate unified positions and approaches (Mertha 2009). Analysts have also argued
that “fragmented authoritarianism” affects the execution of China’s overseas development program, where
multiple stakeholders associated with the state (companies, banks, diplomatic missions, etc.) are advancing
their own interests without necessarily following a coherent strategy organized by a central coordinating
agency (Ye 2021, Lee 2020). A key objective of this chapter is to probe the plausibility of the hypothesis
that Beijing is capable of formulating coherent responses to the competitive (soft power) challenges that it
faces in the Global South.

181
Gallup World Poll (GWP); the favorability of media coverage related to the
Chinese government, as measured via the Global Database of Events,
Language, and Tone (GDELT); and elite support for China’s foreign policy
positions, as measured via UNGA voting alignment with China.

In Chapter 1, we documented that, in the Global South, these three soft power
metrics moved in varying directions over the first eight years of the BRI
(2014-2021). During the early BRI period, the contest for international public
opinion between Washington and Beijing remained neck and neck. However,
public support for China fell sharply by 16 percentage points between 2019 and
2021. In terms of media favorability, after years of maintaining a competitive
edge, Beijing lost significant ground to Washington during the late BRI period;
its favorability advantage over Washington shrank to a razor-thin margin. China
did relatively well at maintaining elite support for its foreign policy positions, as
measured by the extent to which LICs and MICs aligned their UNGA votes with
China. However, we also identified some grounds for concern in BRI participant
countries, as governing elites in this cohort took foreign policy positions (in
UNGA) that were increasingly out of alignment with those of China during the
late BRI period.

Our analysis in Chapter 1 also revealed that, while the battle for hearts and
minds between China and the U.S. was a “toss-up” during the early BRI period,
it became significantly more competitive during the late BRI period. Washington
made more reputational gains at Beijing’s expense—than vice-versa—during this
period. This was especially true in the contest for international public opinion,
where almost 85% of the country-level changes in public opinion represented
relative losses for China.

We now extend the analysis by focusing on the relative gains and losses
experienced by both powers on an annual basis across three soft power
measures during the early BRI period. We designed our analytic approach in
recognition of two realities. First, great power competition for soft power is
generally considered to be a zero-sum game by the participants.316 Second,

316
For example, in an empirical examination of competition between China and India in 2,333 provinces
across 123 countries between 2007 and 2014, one study finds that an “increase in the probability of a new
Indian Exim Bank loan in response to new Chinese development projects is more pronounced when
182
events (“shocks”) at local, regional, and global levels can simultaneously affect
overseas support for all great powers.317 To address these realities, we first
measure absolute year-on-year (YoY) changes to our three measures of soft
power at the country level (single delta) for each great power (China and the
U.S.). We then measure the difference (double delta) between the changes that
China experienced and the changes that the U.S. experienced (see Box 4a).
These double delta measures allow us to identify when and where China
experienced (a) larger losses (or smaller gains) than the U.S., and (b) larger gains
(or smaller losses) than the U.S.318

This methodological approach also makes it easier to think about how Beijing
allocates scarce resources across different categories of countries to advance its
soft power objectives. When a great power allocates scarce resources in pursuit
of grassroots, media, or elite support, it must make risk-adjusted reward
calculations—by balancing the magnitude of a potential soft power gain in a
given jurisdiction against the likelihood of success in that setting (Asmus et al.
forthcoming). As a general rule, it can make smaller gains in places where the
probability of success is high (safe bet territory), or it can venture into more
challenging places where the probability of success is low (moonshot territory)
but the opportunity to make a large gain (at the expense of a rival) is high.319 In
recognition of this tradeoff, we organize our analysis around four cohorts: safe
bet countries, toss-up countries that lean toward China, toss-up countries that
lean toward the U.S., and moonshot countries (based on the data sources and
methods described in Box 4a).

popular opinion in the recipient country is relatively more favorable about India than about China” (Asmus
et al. forthcoming: 26, emphasis added). The same study provides evidence that this effect “is driven by
the difference in public sentiment toward India and China rather than by the absolute levels of public
support for India in these countries” (Asmus et al. (forthcoming: 26-27).
317
For instance, when a country suffers from a major natural disaster induced by global climate change and
levels of foreign assistance are lacking, public resentment can increase toward all outside forces, as well as
domestic institutions.
318
Table A13 provides the country-specific double delta scores for each soft power measure.
319
Vadlamannati et al. (2023: 16) provide evidence that the “BRI prompts the largest positive U.S. response
when China is engaged with the target country but not yet dominant.” The existing empirical literature
does not provide conclusive evidence about whether Beijing follows a strategy of balancing,
bandwagoning, or hedging.
183
Box 4a: How AidData measures relative gains and losses in Chinese soft power

For all three measures of soft power in LICs and MICs (public opinion, media sentiment, and elite
support), we adopt an identical approach to measure relative gains and losses by China vis-à-vis
the U.S. at the host country level. In order to explain our method of measurement, we describe
how it is applied to one soft power indicator (public opinion) and one host country (Bangladesh).
We take the following steps:

Step 1: Use Bangladesh’s annual public approval ratings (from GWP) to calculate the
year-on-year change (single delta: t0 - t1) for China and the U.S. for each year of the early BRI
period.

Step 2: Calculate the difference between the “single delta” in China’s public approval rating and
the “single delta” in the U.S. public approval rating for each year. The resulting “double delta”
(China-U.S.) measure provides an indication of whether and to what extent China gained or lost
ground at the expense of the U.S. The larger the size of a (positive) double delta, the greater the
size of the gain that China made at the expense of the U.S.—or vice versa. Figure 4.1 provides a
visual representation of how this calculation works for one soft power indicator (GWP) in
Bangladesh between 2016 and 2017.

Figure 4.1: Bangladesh example: Is China losing or gaining ground vis-à-vis the U.S. in 2017?

Figure 4.1

Step 3: Calculate the average annual relative change (i.e., average annual double delta) in public
approval over the early BRI period (2014-2017). Averaging over the 2014-2017 period effectively
measures whether China gained more ground overall than the U.S.—or vice versa—during the

184
early BRI period. Table 4.1 provides the full calculation for Bangladesh, where 1.6 is the final
score for Bangladesh during this time period.
Table 4.1: Bangladesh example: Calculating an average double-delta score for soft power measures

Table 4.1
Bangladesh example: Calculating an average double-delta score for soft power measures

China Approval
China Single U.S. Approval U.S. Single China-U.S.
Year
Delta Rate Delta Double Delta
Rate

2013 69.6 - 66.2 - -

2014 47.5 -22.1 44.9 -21.3 -0.817

2015 55.6 +8.1 55.5 +10.6 -2.52

2016 70.4 +14.8 70.5 +15.0 -0.164

2017 62.0 -8.4 52.1 -18.4 +9.99

Average
58.8 -1.9 55.7 -3.5 1.6
(2014-17)

Toss-Ups - Leaning
Safe Bets Toss-Ups Moonshots
China
(75-100th percentile) (25-49.9th percentile) (0-24.9th percentile)
(50-74.9th percentile)

Step 4: Repeat this process for all country-year pairings, rank countries based on their average
annual double delta values during the early BRI period, and divide the global distribution of LICs
and MICs into four categories:

● Safe bets: The top quartile (75th to 100th percentile) of countries where China made the
largest gains at the expense of the U.S. during the early BRI period.

185
● Toss-ups (leaning China): Countries that fall between the 50th to 74.9th percentile represent
competitive jurisdictions where China made relatively small gains at the expense of the U.S.
during the early BRI period.320

● Toss-ups: Countries that fall between the 25th to 49.9th percentile represent competitive
jurisdictions where China made negligible gains at the expense of the U.S. during the early
BRI period—or the U.S. made relatively small gains at the expense of China during the early
BRI period.

● Moonshots: The bottom quartile (0 to 24.9th percentile) of countries where the U.S. made
the largest gains at the expense of China during the early BRI period.

Public opinion
To understand China’s public opinion gains and losses vis-à-vis the U.S. during
the early BRI years, we first rank-order 104 LICs and MICs (for which GWP data
are available) according to our double delta measure (see Figure 4.2). The
observed values on this measure range from -18 pp in Malaysia to +60 pp for
Libya. We then calculate a percentile ranking for each country in the global
distribution. These rankings are subsequently used to assign each country to
one of four categories (described in Box 4a): safe bets (75th to 100 the
percentile), toss-up leaning China (50th to 74.9th percentile), toss-up (25th to
49.9th percentile), and moonshots (0 to 24.9th percentile).

320
The “Toss-ups (leaning China)” category description is admittedly imperfect when applied to the elite
(UNGA voting) alignment measure, since the U.S. made gains at the expense of China in a substantial
number of the countries that fall within the category.

186
Figure 4.2: China’s public approval gains and losses vis-à-vis the U.S.

Figure 4.2

Notes: The relative change calculation is based on the “double delta” measure that is described in Box 4a.
Based on its position in the global distribution, each country is assigned to one of four categories: safe bet,
toss-up (leaning China), toss-up, or moonshot (as described in Box 4a).

As Figure 4.2 demonstrates, during the early BRI years, China achieved gains at
the expense of the U.S. in two-thirds of all host countries with GWP public
opinion data. The average double delta in this cohort was +3 pp. There are
several reasons why the U.S. may have suffered soft power losses during this
period. One is President Trump’s election and the “America First” agenda, which
brought about a period of global retrenchment. Another is President Xi’s
contrasting approach, which sought to establish “win-win” partnerships and new
“connectivities” with the Global South through the BRI (Rolland 2017).321

321
Before and after his 2016 electoral victory, President Trump’s fiery “America First” rhetoric and divisive
personal views raised worldwide concerns about his country’s commitment to maintaining global
leadership. His electoral agenda included proposals to impose tariffs on Chinese imports into the U.S. for
“protecting American jobs” which raised concerns about a new “trade war” that would ultimately hurt
developing countries (Kucik 2017). At the same time, Xi was seeking to position China as a leader on the
world stage. When he first introduced the BRI, he focused on creating “win-win” partnerships with host
countries through big-ticket infrastructure projects.
187
Notwithstanding Beijing’s overall success in making public opinion gains at the
expense of Washington, there were some notable exceptions. Negative double
deltas in Malaysia (-18 pp), Vietnam (-10 pp) and Niger (-2 pp) indicate
significant relative losses for China. During the same period of time, China
registered large, positive double deltas in Jamaica (+16 pp), Namibia (+10 pp),
and Egypt (+8 pp). In at least two-thirds of countries, we see evidence of robust
competition between Washington and Beijing, with average double delta values
in the -5 to +5 pp range.

While every country context is unique, public approval levels appear to be

sensitive to certain factors, such as changes of government from pro-China
incumbents to new political leaders who view Beijing with greater skepticism
(e.g., Malaysia), ongoing or worsening territorial disputes with China (e.g.,
Vietnam), and major changes in levels of aid or lending from a great power (e.g.,
Jamaica). The case studies in Section 4 demonstrate that the extent to which
policymakers in Beijing use development finance and public diplomacy tools to
respond to such changes depends on their perceptions of the strategic value of
a given country.

Media sentiment
We follow the same double delta approach to measure the media sentiment
gains and losses that China achieved vis-à-vis the U.S during the early BRI
period.322 According to Figure 4.3, Beijing achieved gains at the expense of
Washington in 79.4% of (104 out of 131) host countries for which GDELT data
are available. This pattern is consistent with previous research that suggests the
launch of the BRI, together with Beijing’s strategy of ramping up public
diplomacy efforts, including grassroots media management, generated more
positive media coverage about China’s government (Custer et al. 2019; Brazys
and Dukalskis 2019).

322
As we describe in Box 1b in Chapter 1, we restrict our analysis to media coverage about Chinese and
American governments, rather than the countries overall, to ensure that we focus on the activities of the
official sector. In most host countries, an array of Chinese and American private companies and nonprofits
engage in activities that are wholly or partially independent of their home country’s official policies and
approaches.
188
Figure 4.3 also demonstrates that China achieved the biggest relative gains in
the Democratic Republic of the Congo (+2.8 pp), Vanuatu (+1.8 pp), and
Tajikistan (+1.5 pp), while its largest relative losses took place in Comoros (-2.6
pp), Tunisia (-1.4 pp), and Dominica (-1 pp). Across the entire global distribution,
the average double delta size during the early BRI period was +0.54.

Figure 4.3: China’s media sentiment gains and losses vis-à-vis the U.S.

Figure 4.3

Notes: The relative change calculation is based on the “double delta” measure that is described in Box 4a.
Based on its position in the global distribution, each country is assigned to one of four categories: safe bet,
toss-up (leaning China), toss-up, or moonshot (as described in Box 4a).

Elite support
As a member of the Global South, China’s voting patterns in the UNGA are
closely aligned overall with those of developing countries—especially on issues
related to human rights and national sovereignty (Fung and Lam 2022).323
However, as Figure 4.4 demonstrates, unlike public approval and media

323
More often than not, we find that China and the U.S. vote differently at the UNGA. Since 2000, they
have only taken the same position on 14.5% of all resolutions. This misalignment becomes worse in certain
issue areas, such as a 7.8% alignment on human rights resolutions and a 4.6% alignment on the Palestinian
conflict. There is much closer alignment, 19.5%, between China and the U.S. on resolutions concerning
nuclear weapons and materials and arms control and disarmament, consistent with their nuclear statuses.
189
sentiment where Beijing achieved significant gains at Washington’s expense
during early BRI years, UNGA voting alignment is a measure of soft power for
which we observe the opposite trend: governing elites in LICs and MICs moved
their foreign policy positions into closer alignment with China in only 28% of the
107 countries (for which UNGA voting alignment data are available).324 China
experienced soft power losses vis-à-vis the U.S in India (-0.14 pp), Pakistan (-0.12
pp), Brazil (-0.09 pp), Venezuela (-0.05 pp), Mali (-0.16 pp) and Rwanda (-0.14
pp). The countries that handed relative soft power gains to China included the
Pacific island nations of Kiribati (0.4) and Micronesia (0.34), and several mid-sized
countries in Latin America including Panama (0.18) and Paraguay (0.12).

At the same time, these changes are important to understand in their correct
context. The vast majority of LICs and MICs remained far more strongly aligned
with China than the U.S. in the UNGA in absolute terms. Across all UNGA votes
cast between 2000 and 2021, we find that LICs and MICs on average aligned
with China 75.6% of the time, as compared to 23.1% with the U.S.

Figure 4.4: China’s elite alignment gains and losses vis-à-vis the U.S.

Figure 4.4

324
Across the global distribution, the average double delta “idealpoint” distance was -0.02.
190
Notes: The relative change calculation is based on the “double delta” measure that is described in Box 4a.
Based on its position in the global distribution, each country is assigned to one of four categories: safe bet,
toss-up (leaning China), toss-up, or moonshot (as described in Box 4a).

Section 3: From observations to reactions

Having measured the soft power gains and losses that China experienced during
the early BRI period, we now turn our attention to the second part of our
action-reaction framework: Beijing’s late BRI responses to soft power changes
during the early BRI period. We do so by analyzing a new feature of the 3.0
version of AidData’s GCDF dataset: development finance commitment data for
all LICs and MICs between 2018 and 2021.

Quantifying Beijing’s strategic prioritization: The expectation versus

reality framework
The next step in our analysis is to determine if there is evidence of Beijing
strategically allocating aid and credit during the late BRI period to pursue its soft
power objectives. We do so by analyzing whether Beijing favored or disfavored
any specific soft power country cohorts between 2018 and 2021, relative to a
reference point of aid and credit allocations that are based on the cross-country
distribution of economic need.325 In cases where we observe large differences
between the expected size of Chinese development finance commitments
(based on economic need) and the observed size of Chinese development
finance commitments, we argue that such differences may reflect strategic
decisions by Beijing to prioritize or de-prioritize a country for soft power
purposes.

325
The cross-country aid allocation literature generally uses two variables to measure economic need:
income and population size (e.g., Alesina and Dollar 2000; Dreher et al. 2022). However, income is used to
account for varying levels of economic need across HICs and LICs/MICs. Given that our analysis is
exclusively focused on LICs/MICs (i.e., only countries with relatively low levels of income), we rely on
population size to capture cross-country differences in economic need.

191
 Table 4.2: Expected versus observed allocation of Chinese development finance by soft power cohort

Table 4.2
Expected versus observed allocation of Chinese
development finance by soft power cohort
Public Opinion Media Sentiment Elite Support
Category
Expected Observed Difference Expected Observed Difference Expected Observed Difference
(%) (%) (pp) (%) (%) (pp) (%) (%) (pp)

Safe Bets 15 17 +2 10 9 -1 7 30 +22

Toss-Up -
18 48 +29 19 48 +29 17 23 +6
Leaning China

Toss-Up 22 15 -7 36 35 -1 28 16 -12

Moonshot 43 16 -27 34 7 -27 47 31 -16

Notes: This table illustrates which soft power cohorts (safe bet, toss-up leaning China, toss-up, and
moonshot) received relatively more or less development finance from China during the late BRI period
(2018-2021) compared to an expected allocation. The expected percentages for each cohort and soft
power measure (public opinion, media sentiment, and elite support) are based on a hypothetical scenario
in which China allocated development finance to each cohort on a non-strategic basis (i.e., based only on
the population size of each cohort). The observed percentages represent the actual percentages of China’s
ODA and OOF portfolio allocated to each soft power cohort during the 2018-2021 time period. The
differences between expected and observed allocations therefore provide an indication of how much China
may have prioritized or deprioritized each cohort based on strategic considerations.

To put this framework into practice, we follow a three-step process. First, we

assign each LIC and MIC (with available data) to a safe bet, toss-up (leaning
China), toss-up, or moonshot category based on its position within the global
distribution of relative gains and losses experienced by China during the early
BRI period. We follow this procedure for each of the three measures of soft
power, which results in each country being assigned to a different category for
each measure of soft power. For example, Morocco falls within the safe bet
category for the measure of public opinion, but it falls within the moonshot
category for the media sentiment measure and the toss-up (leaning China)
category for the elite support measure.

Second, we create measures of the expected level of development finance from

China based on the size of each host country’s population. We first average the

192
population size of each host country over the 2018-2021 time period, and then
estimate each host country’s share of the total population across all LICs and
MICs that received development finance commitments from China over the
same time period.326 We then multiply these country-specific shares by the total
size of China’s international development finance portfolio (measured in
constant 2021 USD) during the late BRI period. For example, host countries that
fall within the safe bets category (on the public opinion metric) are home to only
14.9% of the total population across all LICs and MICs, so we expect the same
percentage of China’s portfolio of international development finance
commitments during the late BRI period (14.9% of $368.8 billion, or $54.9
billion) to be allocated to the 26 countries in the safe bets category. (See Table
A14 in the Appendix for more details).

Third, for each of the four categories across every soft power metric, we
measure the size of the gap between the expected size and actual size of
development finance commitments from China. Consider again the global
distribution of public opinion gains and losses that China experienced during
the early BRI period and the countries that fall within the safe bets category.
With the 3.0 version of AidData’s GCDF dataset, we calculate total development
finance commitments from China to this cohort of 26 countries during the
2018-2021 period. This figure amounts to $62.2 billion, which represents 17% of
China’s portfolio of international development finance commitments during the
late BRI period. The $7.2 billion delta between the actual size and expected size
of Chinese development finance commitments represents a +2 pp difference for
countries in the safe bets category (see Table 4.2), which suggests that Beijing
may have prioritized this group of countries for soft power purposes. However,
one can also see in Table 4.2 that there is substantially stronger evidence of
Beijing prioritizing the toss-up (leaning China) category (+29 pp difference) and
de-prioritizing the moonshots category (-27 pp difference) for soft power
purposes.

326
Between 2018 and 2021, the 3.0 version of AidData’s dataset captures development finance
commitments to 138 countries and territories.
193
Decoding Beijing’s soft power playbook
Table 4.2 provides five key insights regarding how Beijing allocates aid and
credit in response to soft power gains and losses. Together, these insights help
decode Beijing’s playbook for competing with Washington for expanded
influence in the Global South.

First, across all three measures of soft power, China devoted nearly two-thirds of
its entire international development finance portfolio during the late BRI period
to countries belonging to the two toss-up categories. Table 4.2 also
demonstrates that Beijing prioritized toss-up countries that lean toward it. This is
true across all three measures of soft power, which suggests that Beijing’s foreign
policy braintrust is monitoring soft power gains and losses and redirecting aid
and credit at pivotal moments to battleground countries where it has a modest
advantage.327

Second, across all three measures of soft power, we find large negative deltas
between expected and observed development finance allocations to countries
in the moonshot category (see Table 4.2 and Figure 4.5). This pattern suggests
that Beijing is relatively risk-averse when it comes to the pursuit of soft power.
Rather than prioritizing audacious attempts to lure countries out of Washington’s
orbit, Beijing devoted few resources between 2018 and 2021 to countries that
moved in Washington’s direction during the early BRI period. The fact that China
deprioritized countries where it had recently suffered soft power losses vis-à-vis
its principal rival reveals how key decision-makers in Beijing approach
risk-reward calculations more generally. Moonshot countries are “high risk, high
reward” opportunities, in that they represent jurisdictions where the magnitudes
of soft power gains can be large (due to relatively low baseline levels of support
for China) but the probabilities of success are low. Figure 4.5 suggests that
Beijing’s soft power investment strategy is not focused on such countries but
rather on countries where the odds of success are higher—even if the size of the
gains that can be realized are less substantial.

327
As we will soon explain, there is some evidence that, in order to safeguard its geopolitical interests in
Bangladesh (where it forged a robust partnership with Prime Minister Sheikh Hasina), Chinese state-owned
financiers and implementation agencies accelerated the pace of project commencement and completion
ceremonies during election season.

194
Figure 4.5: Beijing’s aid and credit prioritization strategy

Figure 4.5

Notes: This figure provides a visual representation of the “Difference (pp)” columns from Table 4.2 by
focusing on the categories of countries which Beijing prioritized or deprioritized during the late BRI period,
in apparent response to soft power gains and losses vis-à-vis the U.S. The line labeled 0 represents
situations where there is no difference between the expectation and reality of Beijing's aid and credit
allocation. The positive or negative values represent categories where spending exceeds and fails to meet
expectations, respectively.

Third, in the contest for international public opinion, China has shown a low
tolerance level for reputational risk. During the late BRI period, it devoted
two-thirds of its international development finance portfolio to toss-up countries
and assigned special priority to the subset of toss-up countries that lean toward

195
China (see Table 4.2).328 By contrast, it directed only 16% of its international
development finance portfolio during the late BRI period to moonshot
countries.329 These resource allocation patterns suggest that China is less
interested in competing in countries where its principal rival has momentum on
its side and more interested in shoring up public support in countries with
favorable baseline conditions. Zambia is an example of a country that Beijing
showered with aid and credit when it was relatively pro-China. However, as we
explain in Section 4 of this chapter, after the government defaulted on its
external debt obligations in November 2020, China’s public approval rating
plummeted (by -9.2 pp) and the U.S. public approval rating remained mostly
unchanged (-0.3 pp). Beijing’s provision of aid and credit to Zambia all but
halted, completing a dramatic full-circle turn since the heyday of Beijing’s
engagement in the country during the early BRI period.

Fourth, in the quest for favorable media coverage, China strongly disfavors
countries in the safe bet and moonshot categories. It devoted only 16% of its
international development finance portfolio during the late BRI period to such
countries. Instead, it focused 83% of its international development finance
portfolio during the late BRI period on countries in the two toss-up categories.
These spending patterns suggest that Beijing is neither interested in high-risk,
high-reward opportunities nor low-risk, low-reward opportunities when it comes
to the pursuit of favorable media coverage. Instead, it appears to be doubling
down in competitive jurisdictions where it “has the wind at its back.” Figure 4.5
provides evidence of a large positive delta (+29 pp) between expected and
observed Chinese development finance allocations to countries in the toss-up
(leaning China) category. In the case of Bangladesh, where China experienced
media sentiment gains at the expense of the U.S., average annual commitments
tripled from $994 million during the early BRI period to over $3.3 billion in the
late BRI period, despite challenges associated with COVID-19. In the context of
growing strategic competition in the Indo-Pacific, Beijing’s partnership with

328
For context, the same set of countries that were classified as toss-up (leaning China) or toss-up in our
analysis received 21% of China’s total development finance portfolio in the pre-BRI period (2000-2013), and
28% during the early BRI period (2014-2017).
329
For context, the same set of countries that were classified as moonshot in our analysis received 31% of
China’s total development finance portfolio in the pre-BRI period (2000-2013), and 12% during the early BRI
period (2014-2017).
196
Prime Minister Sheikh Hasina’s 15-year old government strengthened
significantly.330

Fifth, in the international contest for elite support, China shows relatively little
interest in toss-up countries. Table 4.2} demonstrates that Beijing allocated only
39% of its international development finance portfolio during the late BRI period
to these highly competitive jurisdictions. It directed the remaining 61% of the
portfolio to countries in the moonshot category (i.e., jurisdictions where the U.S.
made soft power gains at its expense) and safe bets category (i.e., jurisdictions
where China made soft power gains at the expense of the U.S.).331 These
spending patterns suggest that Beijing is confident its tried-and-true strategy of
trading cash for foreign policy concessions can work in any number of settings.
As we explain in Section 2, the case of Argentina is consistent with the notion
that Beijing is willing to work with friendly governments, regardless of their
ideological commitments or political viewpoints.

Section 4: Country case studies

In Section 3, we learned that there is no one-size-fits-all approach in China’s soft
power playbook. Beijing tailors its strategy to local conditions in host countries,
while accounting for recent competitive gains and losses. However, policymakers
and analysts still lack a clear understanding of how China adapts its aid and
credit allocation based on local contextual factors. To gain greater insight, we
turn our attention to three case studies: Zambia (a toss-up country, leaning
toward China during the early BRI period on the public opinion measure),
Bangladesh (a toss-up country, leaning toward China during the early BRI period
on the media sentiment measure), and Argentina (a safe bet country during the
early BRI period on the elite support measure).332

330
During this time, Prime Minister Sheikh Hasina has presided over both stellar economic development
performance and growing authoritarianism. Most of the late-BRI financing has focused on large-scale
infrastructure projects related to transportation and energy.
331
According to Table 4.2, the largest positive delta (+22 pp) between expected and observed Chinese
development finance allocations to countries is observed in the safe bets category.
332
For each soft power measure, we selected a country that fell within a category (safe bet, toss-up, and
moonshot) that China prioritized during the late BRI period. To maximize generalizability, we also selected
countries from different regions and with varying levels of economic development.
197
Zambia case study: From reputational assets to reputational
liabilities
In terms of public opinion changes during the early BRI period, Zambia falls into
the toss-up (learning China) category of countries that moved closer to China
than the U.S., but not enough to be classified as a safe bet. During the early BRI
period (2014-2017), China’s average annual development commitments to
Zambia jumped to $1.93 billion, as compared to only $230 million during the
pre-BRI period (2000-2013). We investigate the factors that shaped China’s
engagement with Zambia during the early and late BRI periods in this case
study.

Vying for influence: The limits of Chinese soft power in Zambia

On her March 2023 trip to Zambia as U.S. Vice President (VP), where she last
visited as a child to see her Indian grandfather, Kamala Harris declared her
administration’s sincere and steadfast commitment to Africa. “Our presence here
is not about China,” she said. “It’s about an independent understanding of the
intertwined histories of our nations, and our mutual commitment to democratic
principles, and a recognition and understanding of what it means to engage in
smart investments and the potential for the future of the entire globe.” As proof,
she announced over $7 billion in commitments from the U.S. private sector to
various African countries. Most of the investments focused on advanced
technologies, such as artificial intelligence (AI) for improved weather prediction,
electric vehicle assembly plants, and biofertilizer facilities (White House 2023).

In the early days of the Biden administration, an effort was set in motion to
launch a major counter-initiative to China’s BRI (Sanger and Landler 2021). As the
White House was ironing out the details of PGII (rebranded from Build Back
Better World), Africa’s lukewarm condemnation of the Russian invasion of
Ukraine accelerated U.S. efforts to secure more support from African leaders. In
response to a March 2022 UN General Assembly Resolution (GA/12407) that
condemned Russian aggression, only 51% of African countries voted with
Washington (as compared to 81% for the rest of the world). A significant number

198
of African countries abstained, as China did, or decided not to vote at all (White
and Holtz 2022).

Keen to reset foreign relations after his predecessor's Chinese-backed

infrastructure spending spree, Zambia’s Hichilema administration decided to
break away from its neighbors—Zimbabwe and South Africa—and vote in favor
of the UNGA resolution (Obe and Vandome 2022). Weeks later, Hichilema was
rewarded with an invitation to the White House to meet VP Harris, the first by a
Zambian leader in nearly three decades. At this meeting and the December
2022 U.S.-Africa Leaders Summit, U.S. officials did not put China on the agenda,
focusing instead on messaging around their offerings to Africa. By contrast,
during the previous (Trump) administration, senior officials directly criticized
China’s activities in Zambia. In December 2018, national security advisor John
Bolton claimed that “China is now poised to take over Zambia’s national power
and utility company in order to collect on Zambia’s financial obligations” (Bolton
2019). A year later, speaking at a UN meeting for African leaders, Secretary
Pompeo advised that “[c]ountries should be wary of authoritarian regimes and
their empty promises,” because “[t]hey breed corruption, dependency and
instability, not prosperity, sovereignty and progress” (Paravicini 2020).333

Though the Biden administration moved away from its precedessor’s rhetoric,
China’s growing influence in Africa, particularly in mineral-rich regions such as
Zambia’s copperbelt, remained a point of concern among U.S. government
officials, and it loomed over VP Harris’ trip. At the news conference with VP
Harris in Lusaka, President Hichilema clarified his position on the U.S.-China
strategic competition. “When I'm in Washington, I'm not against Beijing,” he
said. “And when I'm in Beijing, I'm not against Washington” (White House
2023). Soon after his August 2021 landslide victory over incumbent Zambian
President Edgar Lungu, Hichilema struck a cordial tone with both powers,
seeking improved ties with each (Obe and Vandome 2022). But he later
discovered—during months of contentious restructuring negotiations with
creditors from the Paris Club, China, and private bondholders following the

333
Many African leaders complained about American neglect, indifference, or even disdain during the
Trump administration.
199
country’s November 2020 sovereign default—that resolving Zambia’s debt issues
would require delicate management of U.S.-China tensions.

From a soft power perspective, Beijing encountered the limitations of its

state-led overseas development financing model before Hichilema came to
power. Despite warnings from the International Monetary Fund that “the pace of
borrowing needs to be slowed down significantly,” the Patriotic Front (PF)
presidencies of Michael Sata and Edward Lungu oversaw a period of “profligacy
and theft of the ruling elite” that increased public debt from 32% of GDP in
2012 to 120% by 2020 (The Economist 2020). Chinese state-owned creditors
played a major role in bankrolling the country’s borrowing binge for big-ticket
infrastructure projects. From 2000 to 2011, China’s average annual lending
commitments to Zambia amounted to only $178.3 million. However, between
2012 and 2018 (during the Sata and Lungu administrations), its average annual
lending commitments soared to $932 million.

Efforts to construct, expand, and rehabilitate large-scale

infrastructure—including roads, airports, power plants, transmission lines, and
water and sanitation systems—with Chinese credit continued until late 2018,
when the country’s debt-to-GDP ratio surpassed 90%. This development
triggered alarm around the world and particularly in Beijing because of its high
level of exposure to non-performing loans. China also witnessed a sharp decline
in its public approval rating once it turned off the credit tap: whereas 79.5% of
Zambians approved of the Chinese government in 2017, this figure dropped by
12.9 percentage points (to 66.6%) in 2021.334

Therefore, an important lesson from this case study is that Beijing’s strategy of
buying soft power through the supply of credit-driven development projects has
limitations. This is likely true of any BRI participant country, but the risks are
especially high in poorly-governed countries with long histories of corruption
and financial mismanagement. Beijing’s experience in Zambia also calls attention
to the fact that its rivals are well-positioned to make inroads with governing
elites and the general public in such countries.

334
Approval ratings are taken from the Gallup World Poll. China’s lending commitments to Zambia in 2019
and 2020 amounted to only $143.7 million.
200
Why is Zambia relevant to great power competition?

China’s development finance portfolio in Zambia is among its most well-known

globally, due to the country’s widely-reported debt crisis. But there is much
more to the story of China in Zambia than the latter’s debt repayment struggles.
With the world’s 7th largest copper reserve, a mineral that is needed for
advanced technologies to fuel the green energy transition, Zambia is a
high-value target in the great power competition between China and the U.S.
(S&P Global 2022).

Within years of declaring independence from Britain, Zambia became home to

China’s first-ever foreign assistance megaproject, the iconic TAZARA Railway that
links Zambia’s copperbelt heartland to Tanzania’s port of Dar es Salaam (Dreher
et al. 2022). Today, China is Zambia’s largest trading partner, responsible for
buying over 75% of its main export, copper (The Growth Lab at Harvard
University 2023). China is also one of Zambia’s largest foreign creditors and
donors. Between 2000 and 2021, it issued grants and loans to Zambia worth
$13.7 billion. Beijing financed $10.03 billion in infrastructure projects during the
BRI era (2014-2021), including major intercity and urban highways, hydropower
stations and accompanying transmission lines, international airports,
communications systems, hospitals, universities, coal-fired power plants, and
water supply systems.335

335
China’s footprint in Zambia’s healthcare system goes beyond hospitals and equipment, with
medical teams from Hunan province serving on 22 tours in Zambia since 1978.
201
Figure 4.6: Official financial flows from China to Zambia, 2014-2021

Figure 4.6

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (as
described in Section A-2 of the Appendix).

202
Figure 4.7: China versus the U.S. in Zambia: Public approval rates

Figure 4.7

Given China’s extensive engagements with Zambia for more than seven
decades, it is not surprising that 72% of Zambians approved of the Chinese
government during the BRI era (see Figure 4.7 above). However, Beijing enjoyed
only a two percentage point advantage over its primary competitor
(Washington) during this period, despite the fact that Chinese development
finance commitments ($17.47 billion) dwarfed U.S. development finance
commitments ($3.40 billion). During the BRI era, U.S. development finance
remained relatively stable336—both in volume and composition—but Washington
saw its public approval rating in Zambia soar throughout early and late BRI
periods, from 63.8% in 2014 to 84.7% in 2021 (see Figure 4.7}).337 China, on the
other hand, saw its public approval rating steadily decline during the late BRI
period.

336
With an average annual allocation of only $423 million, U.S. development finance ranged from a low of
$399 million in 2019 to a peak of $548 million in 2017.
337
Given the relative stability of aid flows from Washington, Zambian public opinion appears to be primarily
driven by changes in the White House. After peaking at 92.5% during the first year of the Obama
administration, it subsequently fell to just 66.7% in 2017 after Trump took office, in line with the “Trump
effect” that hurt U.S. popularity globally.
203
Supplementary data from Afrobarometer suggest that, during the late BRI
period, Zambians re-evaluated the wisdom of following China’s state-centric and
infrastructure-focused development model as an alternative to the West’s rights,
freedoms, and market institutions-based approach. When Afrobarometer
surveyed Zambians in 2014-2015 about the country they considered to be the
“best model for development,” 32% chose China and 23% picked the U.S.
However, by 2019-2021, when the link between poorly-executed Chinese
megaprojects and Zambians’ economic woes had been laid bare, Beijing’s
nine-point advantage had flipped in favor of Washington, with only 22%
respondents expressing a preference for China’s model and 31% choosing the
U.S. model. This represented a reversal of fortune for China at the end of its
decade-long partnership with the ruling party (the Patriotic Fund, or PF) , which
started with President Sata’s May 2013 Beijing visit.

The Patriotic Fund’s infrastructure binge

The PF’s relationship with Beijing began in outright hostility during the
unsuccessful 2006 presidential campaign of a fiery former trade unionist,
Michael Sata, whose abrasive politics earned him the nickname “King Cobra”
(Kimenyi and Copley 2014). By orienting his campaign around a new agenda of
Sino-skepticism, he rallied support against the 15-year rule of the Movement for
Multi-Party Democracy (MMD), arguing that Chinese investors were “coming just
to invade and exploit Africa” (French 2011). At one point, he even suggested
that, if elected, he would recognize the Republic of China (Taiwan) as a
“sovereign state” (Mupuchi 2006). China’s Ambassador in Lusaka responded by
threatening that Chinese investment would be “put on hold [...] until the
uncertainty surrounding our bilateral relationship with Zambia is cleared”—that
is, until Sata’s electoral loss was confirmed (Shacinda 2006). After two
unsuccessful presidential bids, Sata relaxed his rhetoric during a successful 2011
presidential campaign by stating his appreciation for the benefits of Chinese
investment in Zambia while insisting that foreign mining companies abide by
local laws to protect workers (Mfula 2011; Shukla 2021).

Sata’s April 2013 visit to China reset Sino-Zambian relations. According to the
3.0 version of AidData’s GCDF dataset, Beijing issued grant and loan

204
commitments worth $810.2 million for nine new projects in 2013. In the same
year, it started delivering five additional projects worth $111.6 million, and it
completed delivery of a $7.4-million equipment supply project for the upkeep of
the TAZARA railway. Unlike the MMD-era, when China’s largest financial
commitments were earmarked for sports stadiums, the focus shifted toward
infrastructure projects ostensibly designed to generate economic growth. This
type of engagement continued until Sata’s death in October 2014. Over the
course of the Sata presidency, Beijing issued grant and loan commitments worth
$1.6 billion for high-profile projects, including upgrades to the Lusaka airport,
improvements to the Lusiwasi hydroelectric power plant, the construction of new
power transmission lines, and road rehabilitation activities. For a country where
less than half the population has electricity access and paved road density is
among the world’s lowest, China’s willingness to bankroll infrastructure projects
with significant economic growth potential was a welcome departure from the
OECD-DAC’s emphasis on health, education, and governance projects.338

After Sata’s death, his Minister of Justice and Defence Edward Lungu won a
highly contested election in January 2015 to complete the remainder of his
term. Anticipating an election the following year, Lungu visited China in March
2015 to accelerate the momentum behind new infrastructure projects. In the run
up to Lungu’s August 2016 re-election, Chinese grant and loan commitments
tripled from $910 million in 2015 to $2.9 billion in 2016. As in prior years, the
bulk of the money supported physical infrastructure, including more airports,
roads, and water supply systems. However, this time around, China also
committed $468 million for ICT (information and communications technology)
sector activities, including the installation of CCTV cameras and smart city
projects. These types of “AI-enabling” tools are reserved for Beijing’s closest
allies, especially in countries where political systems are turning more
authoritarian, such as in Kenya and Pakistan where they have been deployed
against political opponents (Bouey et al. forthcoming).

338
According to the CIA’s World Factbook, with an area of 752,618 square kilometers, Zambia is
the world’s 40th largest county, but its road network of 67,671 kilometers is the 70th most
extensive in the world. However, only 14,888 kilometers, or 28%, of Zambia’s roads are paved.
This means that its paved road density is only 0.02 kilometers per square kilometer, significantly
lower than other developing countries like India (0.7) and Brazil (0.6).
205
Beijing doubled down on its support for Zambia during the Lungu
administration—perhaps because it thought that a stable and pro-China
government could serve as a de facto insurer for its project and investments.
China’s popularity soared during this period. Public approval for China, as
measured by the Gallup World Poll, rose from 66.4% in 2015 to its highest ever
level of 79.5% in 2017.

After narrowly winning re-election with 50.4% of the vote in August 2016, Lungu
was quick to announce that “President Xi Jinping has expressed confidence in
my leadership.” He reiterated his resolve to further deepen his engagements
with China to “empower businessmen and benefit every Zambian” (Shaban
2016). In the following year, Chinese grant and loan commitments amounted to
$3.1 billion, half of which was a loan for the 750 MW Kafue Gorge hydropower
project.

Then, in September 2018, Lungu visited China to attend the Forum on

China-Africa Cooperation (FOCAC) summit where he formally signed up his
country to participate in the BRI. Beijing responded by providing an additional
$2.9 billion of aid and credit for roads ($1.4 billion), power transmission lines
($499 million), cement plants ($445 million), public housing ($229 million), and
university construction ($208 million).339

When reputational assets become liabilities

As Chinese state-owned entities lent record amounts to Zambia, the IMF

sounded the alarm about the composition and sustainability of Zambia’s external
debt burden (International Monetary Fund 2017). Between 2011 and 2016,
public and publicly guaranteed (PPG) external debt skyrocketed from 8.4% of
GDP to 36.5%. The share of central government debt owed to private
creditors—which was provided on mostly non-concessional terms—increased to
50%. The share of central government debt owed to MDBs—which was
provided on mostly concessional terms—shrank from 60% to 20.5%. The IMF
warned that the “pace of borrowing needs to be slowed significantly” to align

339
Throughout this period of high infrastructure activity, China’s signature public health engagements
continued through medical team visits, equipment and medical supplies donations, and interventions to
combat infectious diseases (Dolan et al. 2023).
206
with the country’s “absorptive capacity” through more stringent standards for
“the selection, procurement, and monitoring of infrastructure projects.”
(International Monetary Fund 2017). Despite these warnings and Zambia’s past
experiences with debt distress, the Zambian government and its external
creditors failed to pump the brakes (International Monetary Fund 2023).

Reports emerged that Zambian government agencies pressured the Ministry of

Finance to sign-off on politically advantageous but commercially non-viable
projects (Mutati 2016). In one instance, a senior government official dealing with
the Mbala-Nakonde road alleged that “the [Chinese] contractor had already
secured the agreement with the Ministry of Finance when they came to us. We
just negotiated about the details of design, not the amount of the loan” (Lee
2018). Zambia’s rush to launch as many high visibility projects as possible was a
function of political economy considerations. Thin electoral victory margins in
the country’s presidential elections “create powerful incentives to use public
works to reward constituencies” (Brautigam 2021).

A review of China’s ODA and OOF portfolio in Zambia reveals that 19 out of 21
projects identified as having ESG challenges faced governance-related
concerns, such as overpricing, corruption, and financial mismanagement (see
Table A12).340 In July 2019, then-opposition leader and future president
Hakainde Hichilma claimed that “the debt that was acquired for just one project,
Lusaka-Ndola (Dual Carriageway) road, which should cost US$400 million[...] is
costing US$1.2 billion.” He argued that Zambians “are giving [projects] to China
in the corrupt way, which is costing us too much of taxpayers’ money” (Ncube
2019).341 In 2022, Dr. Mbita Chitala, the former chairman of the Board of
Directors of ZESCO (Zambia’s state-owned power utility), published a book
entitled Corporate Capture: The Political Economy of Electricity Management in
Zambia 2014-2021 (How Not to Manage a State Enterprise). In it, he claimed
that large segments of the China Eximbank-financed Smart Zambia National ICT
Development Project were completely unnecessary and designed to enrich
senior government officials. He recalled that “the whole project was a

340
Between 2000 and 2021, the ESG risk prevalence rate in China’s grant- and loan-financed infrastructure
project portfolio was 56% in Zambia (see Table A12).
341
The Sata government also launched an investigation of the 750MW Kafue Gorge hydropower project to
determine if adequate transparency and financial controls were in place to monitor performance.
207
conspiracy to defraud Zambia” because Huawei was proposing solutions that
were “inferior to ZESCO existing network” and located in towns that were
already served (Chitala 2022).342

By late 2018, the Zambian government was cash-strapped and Beijing’s

development project portfolio began to suffer. The Zambian government was
increasingly unable to meet its financial commitments to Chinese contractors
and creditors. For example, after CDB issued a $469-million loan in December
2015, the Copperbelt Urban Roads Project faced a series of implementation
delays until the state-owned engineering, procurement, and construction (EPC)
contractor, China Henan International Cooperation Group Company (CHICO),
abandoned the project due to insufficient payments by the Zambian
government. The 750 MW Kafue Gorge hydropower project, the country’s single
largest energy sector project, faced severe implementation challenges when the
borrower—a wholly state-owned special purpose vehicle known as Kafue Gorge
Lower Power Development Corporation Limited—defaulted on its loan
repayment obligations to ICBC and China Eximbank.343 In both cases, senior
government officials claimed that Lusaka would come up with the cash needed
to complete the projects. The 750 MW Kafue Gorge hydropower project was
not completed until April 2023—approximately 13 years after Sinohydro, the
EPC contractor, signed an MOU to complete the project on a
342
Dr. Chitala claims in his book that Zambia’s Secretary to the Cabinet Dr. Roland Msiska and later Dr.
Simon Miri and his deputy, former Deputy Finance Minister and Bank of Zambia Governor Patrick Mvunga,
placed extraordinary pressure on ZESCO (the country’s state-owned power utility) to accept an on-lending
agreement for Phase II of the Smart Zambia National ICT Development Project. He writes that “it was
obvious that some people had either already benefited from the deal or were being prevented from
benefitting. Some of the members of the [ZESCO] management expressed genuine fear of the situation
and requested for my protection.” He also writes that “[t]he President [of Zambia] advised me to cancel [an
upcoming ZESCO Board] meeting which I gratefully did as I had planned to resign from the Board if the
Board members resolved in the affirmative to accept the on-lent loan. […] I had earlier made up my mind
that if the board accepted that [ZESCO] receive the USD 392 million as on lent to [ZESCO] by the Ministry
of Finance knowing that [ZESCO Limited could not accommodate the loan on its balance sheet and further
knowing that the loan was essentially a way of defrauding Zambia, I would respectfully resign as chairman
and board member of ZESCO Limited. […] It was common knowledge that the more than US$ 60 million
that the Ministry of Finance released as its 15% counterpart contribution funding for the loan was not only
wrongly released but was also grossly abused and that many people may have received bribes on that deal
and hence the pressure on ZESCO to accept the USD 392 million on-loan let instructions. […] It was
apparent that since government money had been expended without any work to show for, this would add
to the other scandals that the Auditor General used to publish as examples of abuse of public office of trust
or common theft of public resources by politicians and public servants” (Chitala 2022).
343
Given that the loan was insured by Sinosure, the EPC contractor (Sinohydro Corporation Ltd.) sought
compensation through an indemnity payment. In January 2021, Sinohydro Corporation Ltd. filed an
insurance claim with Sinosure, which in turn made an indemnity payment (worth $57.23 million) to
Sinohydro Corporation Ltd. in May 2021. See Project ID#92289 and #57536 in the 3.0 version of AidData’s
GCDF dataset.
208
build-operate-transfer basis. The Copperbelt Urban Roads Project was
indefinitely suspended. The Lusaka-Ndola (Dual Carriageway) Road Project
suffered a similar fate.

Between 2019 and 2021, new grant and loan commitments from China were
virtually non-existent. After providing average commitments worth $3 billion
between 2016 and 2018, Beijing sharply reduced its average annual
commitments to $66 million from 2019 to 2021. If China’s bid to accumulate soft
power requires demonstrating that its economic development model is effective
and sustainable, Zambia represents a spectacular failure. Projects that were
supposed to be reputational assets—and spur Chinese-led economic growth for
Zambians—became reputational liabilities.

The limitations of infrastructure financing for soft power accumulation

The Zambian case illustrates the limitations of China’s strategy of using

large-scale infrastructure projects to expand its influence overseas. This outcome
was not completely of China’s making. Nor was it inevitable. Rather, it was due
to overborrowing for projects whose feasibility was based on either optimistic or
“graft ridden” projections (The Economist 2018). For instance, the new terminal
at Lusaka’s Kenneth Kaunda Airport was designed to accommodate “an
improbable ten-fold increase in passenger traffic,” and at $360,000 per
kilometer, the average price tag of roads built between 2011 and 2018 was
“more than twice the African average” (The Economist 2018). Had the Zambian
government or Chinese creditors followed more robust due diligence
requirements (see relevant discussion in Chapters 2 and 3), these pitfalls could
have been minimized or otherwise avoided. However, these factors ultimately
led to the downfall of the entire model, resulting in massive losses for the Lungu
administration, China’s reputation, and above all else, the people of Zambia.

Going forward, Beijing might be able to reverse the reputational losses that it
has suffered if it follows through on a June 2023 debt relief commitment and if
local economic conditions improve. In addition, several previously stalled
infrastructure projects—those that were delayed due to COVID-19 and financial
distress—will soon reach completion and potentially increase public support for

209
China. Beijing could also consider sending high-ranking CPC officials or even
organize a state visit by President Xi to reset and deepen ties with Lusaka.
However, as compared to his predecessor, President Hichilema appears to be
playing a careful balancing act in his relations with China and the U.S., and he
will likely tread carefully.

210
Bangladesh case study: Stability in authoritarianism
Bangladesh, like Zambia, is a competitive jurisdiction that we classified as a
toss-up (leaning China) country. During the early BRI period, China made
modest media sentiment gains there at the expense of the U.S. It also
dramatically increased the provision of aid and credit to Bangladesh during the
late BRI period. Whereas average annual ODA and OOF commitments from
China to Bangladesh were only $994 million between 2014 and 2017, this figure
soared to $3.4 billion between 2018 and 2021. This unprecedented spending
increase coincided with a period of political stability, particularly after Prime
Minister Sheikh Hasina won her second consecutive electoral victory in 2018. It
also coincided with a period of rising authoritarianism.

China’s late BRI era foray into Bangladesh

In May 2021, Chinese Ambassador Li Jiming issued a stern warning to

Bangladesh’s elites. In response to reports that Dhaka might join the
Quadrilateral Security Dialogue (the Quad)—an informal coalition between the
U.S., Japan, Australia, and India—he said that such a move would inflict
“substantial damage” on Sino-Bangla relations. This uncharacteristically blunt
remark provoked strong pushback from Bangladesh’s Foreign Minister A.K.
Abdul Momen. “We are an independent and sovereign state. We decide our
foreign policy,” he said. He also clarified that Dhaka was neither approached by
the members of the Quad nor was its participation in the Quad a possibility
under active consideration (Islam 2021).

This diplomatic spat highlights Bangladesh’s desire to delicately balance

relations with the U.S. and its closest regional partner (India), while bolstering its
“strategic partnership” with China (Ministry of Foreign Affairs of the People’s
Republic of China 2016). Bangladesh is located in the strategically vital
Indo-Pacific region, serving as a gateway between South Asia and Southeast
Asia. For China, it is an entry point to the Bay of Bengal, and its proximity to
Myanmar increases its strategic value to the U.S. and its Quad allies.

Since Prime Minister Sheikh Hasina’s assumption of power in 2009, Bangladesh

has made tremendous economic and social development progress: extreme
211
poverty has been reduced by half; life expectancy has risen by 6 years; the
average economic growth rate has exceeded 7%; and per capita income has
risen by 3.5 times (United Nations 2022; Trotsenburg 2023). At the same time,
Sheikh Hasina has effectively turned Bangladesh into a single-party system,
ushering in a period of authoritarianism.344 Between 2009 and 2022, the
country’s position on the Reporters Without Borders’ Media Freedom Index fell
from 121 to 162 out of 180 countries (Reporters Without Borders 2023). Its
overall performance on Freedom House’s Freedom in the World Index tumbled
by a quarter, with political rights eroding by nearly one-third (Freedom House
2022, 2023).

For the U.S. and its allies, Bangladesh’s positive economic development
trajectory, democratic backsliding, and geostrategic position presents a policy
conundrum. Bangladesh’s concentration of power in one person, lack of
tolerance for political dissent, and singular focus on infrastructure-driven,
export-led economic growth is strikingly similar to Beijing’s own. Yet, the Quad is
seeking to expand its footprint in the Indo-Pacific and has reportedly reached
out to Bangladesh about the possibility of a regional “economic partnership”
(Kishida 2023).

Beijing is evidently aware that Bangladesh is a competitive jurisdiction where it

cannot afford to rest on its laurels. In a recently published op-ed, the Chinese
Ambassador to Bangladesh announced that “China is set to bring more
development opportunities to Bangladesh,” predicting that the country’s
exports to China would grow significantly after Beijing announced duty-free
market access to most Bangladeshi products (Wen 2023). In other statements,
he has accused the U.S. of pressuring Dhaka to deviate from its longstanding
position of neutrality in great power competitions. In April 2023, the
Bangladeshi Ministry of Foreign Affairs released its first-ever “Indo-Pacific
Outlook,” restating the “political dictum” of the country’s founding father Sheikh

344
The rise of authoritarianism, which began during Sheikh Hasina’s second term (2009-2014) and was
reinforced during her third term (2014-2019) with the Digital Security Act of 2018, has continued during her
fourth term (2019-present). The opposition party (BNP), which accuses the government of using the judicial
system to silence its critics and hatching plans to rig the January 2024 election, claims that the future of
multi-party democracy is in jeopardy (Amnesty International 2023; Paul 2023).

212
Mujib ur Rehman (‘Friendship toward all, malice toward none”) as a guiding
principle (Ministry of Foreign Affairs of Bangladesh 2023).

As we previously noted, Bangladesh was selected as a case study country

because it fell in the toss-up (leaning China) category during the early BRI period
(according to the media sentiment analysis described in Section 2). However, on
multiple dimensions of soft power, Bangladesh remains a toss-up country,
leaning toward neither China nor the U.S.. Both powers—as well as India—are
actively vying for influence there (Asmus et al. forthcoming). Beijing has
responded to its recent soft power gains in Bangladesh by doubling down on
the provision of aid and credit. This outcome was likely made possible by several
factors related to Bangladesh’s domestic political and economic environment: a
dominant single party government in-charge of a centralized governance
system, a strong focus on rapid delivery of highly visible physical infrastructure
projects, and proactive debt management.

Sheikh Hasina’s rise and the end of political instability

Within months of independence from Pakistan in 1971, the ruling Awami League
(AL) party started moving Bangladesh’s nascent political system toward a
single-party system. But a combination of factors—including economic
mismanagement, rising political repression, and growing concern about the
country’s future—led to multiple military coups and dictatorships from 1975 to
1990 (Riaz 2016). After a return to democracy, the period from 1991 to 2008 was
marred by political crises, dysfunctional governance, and frequent violent
clashes between supporters of Sheikh Hasina’s AL and Khaleda Zia’s Bangladesh
Nationalist Party (BNP) (Mannan 2018).

A key source of infighting was the bitter history of animosity between the two
leaders, originating in the 1975 coup d’etat against the government of Sheikh
Hasina’s father, Sheikh Mujib ur Rehman (the country’s founder). Military officers
not only assassinated Sheikh Mujib, but also all three of Sheikh Hasina’s brothers,
two of her sisters-in-law, and her mother. After the coup, Khaleda Zia’s husband
General Zia ur Rahman emerged as a key military and political leader, eventually
becoming president two years later and forming the BNP, which he led until his

213
assassination in 1981. The tremors from these events still reverberate in
Bangladesh’s politics, as AL seeks to fulfill Sheikh Mujib’s desire to run the
country as a single-party democracy.

Between 1991 and 2006, Zia served two terms in office and Hasina served one
term in office. This era represented an electorally competitive period, albeit one
marked by frequent political and constitutional crises (Vaughn 2008). Then, prior
to the elections scheduled for early 2007, the military intervened and appointed
a technocratic interim government to oversee fresh elections. These were
eventually held in December 2008, leading to Sheikh Hasina’s AL return to
power, which she has held onto ever since. She has steadily solidified her grip
on the political system, stifling opposition forces and retaining power through
heavy-handed authoritarian measures to control the media and judiciary (The
Economist 2023).

The soft power benefits of development finance

Between March 1971 (when the country gained its independence) and
December 2008, Bangladesh’s largest foreign aid and credit providers were
Japan, the U.S., and the U.K. China did not maintain an especially large
development finance program during this period. However, during the BRI era,
Beijing issued grant and loan commitments worth $17.5 billion, making it
Dhaka’s single largest development partner. In 2006, China also surpassed India
and the U.S. to become Bangladesh’s largest trading partner, with $4.7 billion in
total flows. One might assume that the combination of high trade volumes and
unprecedented levels of aid and credit for highly visible infrastructure projects
led to rapid soft power accumulation for China. However, in reality, the
relationship between development finance and soft power has not been
straightforward for several reasons.

214
Figure 4.8: Official financial flows from China to Bangladesh, 2014-2021

Figure 4.8

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (as
described in Section A-2 of the Appendix).

According to the 3.0 version of AidData’s GCDF dataset, China issued grant and
loan commitments worth $20.8 billion for 138 projects in Bangladesh between
2000 and 2021 (see Figure 4.8).345 That made Bangladesh one of China’s 20
largest aid and credit recipients in the developing world and its 7th largest
recipient in Asia. China’s ODA and OOF commitments to the country soared
from $3.3 billion during the pre-BRI period (2000-2013) to $17.5 billion during
the early and late BRI period (2014-2021). During the BRI era, the project
portfolio was largely focused on the construction and rehabilitation of power
plants, transmission lines, highways, and bridges.

345
China’s ODA and OOF commitments to Bangladesh during this 22-year period focused on four sectors:
energy ($10.3 billion), transport ($6.3 billion), industrial development ($1.9 billion), and ICT infrastructure
($1 billion). Beijing also supported a large number of education, health, and emergency response projects
(with relatively small financial commitments), particularly in crisis years such as the COVID-19 period.

215
Figure 4.9: China versus the U.S. in Bangladesh: Public approval rates

Figure 4.9

Given the unprecedented scale of these financial commitments and the highly
visible nature of the infrastructure projects undertaken during the BRI era, it is
not surprising that Beijing has made some public opinion and media sentiment
gains since 2014. The average level of public approval for the Chinese
government was 49% between 2006 and 2013 (see Figure 4.9). Then, during
Sheikh Hasina’s second and third consecutive terms in office (from 2014 to 2018
and from 2019 to 2022), Beijing saw its average public approval rating increase
to 60% and 56%, respectively.

The U.S. and China generally enjoyed similar levels of public support in
Bangladesh during the BRI era (see Figure 4.9). However, shortly after Donald
Trump came to power, the U.S. suffered an 18 percentage point decline in its
public approval rating (from 70% in 2016 to 52% in 2017), which gave China a
10% percentage point advantage over the U.S. in 2017. Beijing’s advantage was
nevertheless short-lived. By 2022, Washington saw its public approval rating rise
to 62%, thereby opening up an 11 percentage point advantage over Beijing in
2022.

216
Figure 4.10: China versus the U.S. in Bangladesh: Media sentiment scores

Figure 4.10

Notes: The data are drawn and processed from the GDELT 1.0 Event Database (related to government
actors from mainland China or the U.S.). See Box 1b in Chapter 1 for more details.

In terms of the competition for favorable media coverage, the GDELT data
indicate that, during Sheikh Hasina’s 2014-2018 term, the average tone of media
sentiment about China (+1.88) was significantly more positive than the average
tone of media sentiment (-0.33) about the U.S. (see Figure 4.10). However,
between 2019 and 2022, China’s average media sentiment score declined to
+0.48.346 One potential reason why media coverage about China may have
become less favorable during the late BRI period is that $10 billion (or nearly
38%) of Beijing’s development finance portfolio in Bangladesh encountered
significant ESG problems (see Table A12), including social and governance
challenges associated with the design and implementation of multiple coal-fired

346
With scores of +4.7 and +3, China’s largest media sentiment advantage over the U.S. came in
2016 and 2017. Since then (2018-2022), the average tone of media sentiment about China has
fallen to +1.1. This finding is consistent with empirical evidence that suggests the soft power
benefits of Chinese development projects are short-lived, with a significant fading away of
reputational benefits in the medium- to long-term (Wellner et al. forthcoming).

217
power generation plants.347 The U.S. also saw its average media sentiment score
deteriorate (to -0.6) during the 2019-2022 period, though the decline that it
experienced was smaller than the one experienced by China. As a result, the
media sentiment gains that Beijing made at the expense of Washington during
the early BRI period were partially reversed during the late BRI period (see
Figure 4.10).

The foundations of Beijing’s partnership with Sheikh Hasina

The rapid rise of Chinese development finance during the BRI era was facilitated
by Beijing’s strong partnership with Sheikh Hasina’s AL government, which grew
as she strengthened her grip on power. Since December 2008, the centralization
of political power has resulted in higher levels of administrative certainty and
created an enabling environment for Beijing to bankroll and build big-ticket
infrastructure projects.

Table 4.3: Development finance commitments from China to Bangladesh by executive administration

Table 4.3
Development finance commitments from China to
Bangladesh by executive administration
Annual average development
finance (ODA and OOF) from China
Political Administration ODA % from China
during each administration
(USD 2021 millions)

BNP’s Khaleda Zia (2002 - 2006) 175 72%

Military-appointed Fakhruddin
71 100%
Ahmed (2007-2008)

Sheikh Hasina I (2009-2013) 381 47%

Sheikh Hasina II
2,191 20%
(2014-2018)

Sheikh Hasina III

2,290 7%
(2019-present)
Notes: This figure provides a summary of Chinese ODA and OOF commitments to Bangladesh during each
executive administration in Bangladesh since 2002. Exact commitment dates in the 3.0 version of AidData’s

347
Between 2000 and 2021, the ESG risk prevalence rate in China’s grant- and loan-financed
infrastructure project portfolio was 59% in Bangladesh (see Table A12).

218
GCDF 3.0 were compared to election dates and political transition dates to categorize each
ODA/OOF-financed project according to the chief executive that was in power at the time of the
commitment. “ODA % from China” displays the proportion of China’s development portfolio that is
classified as ODA and that was committed during each executive administration. AidData relies on
OECD-DAC measurement criteria to make ODA and OOF determinations (as described in Section A-2 of
the Appendix).

As Table 4.3 demonstrates, the volume and composition of Chinese

development finance in Bangladesh have changed significantly over the course
of the last five administrations. Financial commitment amounts have soared
since the military-backed interim government of 2007-2008, but the proportion
of Chinese development finance provided via ODA (grants and highly
concessional loans) has fallen sharply, which reflects Beijing’s growing focus on
larger-scale, revenue-generating projects, such as power plants and tolled
bridges.

Between 2000 and 2008, under the government of Bangladesh National Party
(BNP) leader Khaleda Zia and the military-backed technocratic government of
Fakhruddin Ahmed, average annual Chinese development finance commitments
amounted to only $175 million and $71 million, respectively (see Table 4.3).
During this period, 82% of total commitments, or $1.2 billion, supported just
seven large projects, including a transmission line, fertilizer plant, and bridge.
Beijing also supported smaller projects to promote public goodwill, such as
cycle relief activities, scholarships to study in China, and the China-Bangladesh
convention center as a monument of friendship.

The nature of Beijing’s engagement in Bangladesh began to shift during the late
years of Sheikh Hasina’s second term (2009-2013), with average annual
commitments ($381 million) more than doubling those issued to the two
previous two governments (see Table 4.3). With a greater focus on “bankable”
infrastructure, the share of Chinese development finance provided via ODA
dropped to 47%. Project sizes also increased. Beijing issued loans worth $697
million for a fertilizer factory and $595 million for 3G mobile and national
internet broadband network upgrades, power plants, and an urban water
treatment plant.

219
But the most dramatic changes took place after Sheikh Hasina’s 2013 visit to
Beijing, at which time a major new focus on power generation and transport
connectivity was agreed upon at the highest levels. With greater political
stability and confidence in her political longevity, Beijing agreed to green-light a
set of transformative investments that could help Sheikh Hasina meet the
ambitious goals she set for export-led economic growth as a means of lifting
millions out of poverty. China’s own development model aligned with her vision,
which may explain why its average annual development finance commitment
grew by a factor of seven (to $2.2 billion) between 2014 and 2018 (see Table
4.3).

The conventional wisdom is that China dramatically scaled back its overseas
lending commitments during the late BRI period due to the COVID-19
pandemic and rising concerns about debt sustainability (see Chapter 2).
However, Beijing behavior in Bangladesh belies this claim. On average, between
2019 and 2021, Sheikh Hasina’s government accepted $2.3 billion per year in
new ODA and OOF commitments from China, which enabled the rapid progress
(or completion) of high-profile projects such as the Padma Bridge, the Dhaka
elevated expressway, and several power plants (see Table 4.3 and Table 4.4).
From Sheikh Hasina’s perspective, these are arguably the most politically
advantageous undertakings supported by China: Padma Bridge sits within her
home district and political constituency; the Dhaka expressway improves urban
mobility for the country’s political and business elite; and new power plants
could eliminate crippling power shortages affecting all Bangladeshis.348

The best of all worlds: Single-party governance, proactive debt

management, and insatiable infrastructure demand

Despite widespread international condemnation of growing authoritarianism and

alleged human rights abuses during her 15-year term in office, Asia’s “iron lady”
is poised to fulfill her late father’s vision of turning Bangladesh into a one-party
state (The Economist 2023). During recent media interactions, Sheikh Hasina has
claimed that her AL party is the only legitimate political force in the country that

348
In Chapter 3, we discuss the governance risks associated with siting development projects in
the home districts of political leaders. Also see Dreher et al. (2019, 2022).
220
should be allowed to contest elections because the main opposition BNP was
created by erstwhile military dictators. Her critics argue that, under her watch,
Bangladesh’s economic progress has become solely reliant on nepotistic textile
industrialists, and her team is out of fresh ideas to tackle deep rooted structural
problems, such as corruption.349 Irrespective of these challenges, with major
opposition parties in disarray and dissenting voices in civil society largely silent,
today she is predicted to win the January 2024 election.350

Table 4.4: Chinese ODA and OOF infrastructure project milestones during the second half of 2018 in Bangladesh

Table 4.4
Chinese ODA and OOF infrastructure project milestones
during the second half of 2018 in Bangladesh
Commitment Amount
Project Activity Type Month
(USD 2021 millions)

Padma Bridge Rail Link 2,900 Implementation Started July 2018

Dasherkandi Sewage
318 Implementation Started August 2018
Treatment Plant

Payra Coal-Fired Power Plant 2,200 Implementation Started October 2018

Bangladesh-China Friendship
94 Implementation Started November 2018
Bridge

Banskhali Coal-Fired Power

1,300 Financial Commitment December 2018
Plant
Notes: This table identifies Chinese ODA- and OOF-financed infrastructure project milestones in the
6-month period preceding the December 2018 national elections in Bangladesh. The financial commitment
amounts only represent those from official sector institutions in China. AidData relies on OECD-DAC
measurement criteria to make ODA and OOF determinations (as described in Section A-2 of the
Appendix).

Previous research has demonstrated that elections provide powerful incentives

for politicians to showcase their performance, and highly visible physical
infrastructure projects are ideal for achieving this goal (Marx 2018; Anaxagorou

349
Bangladesh’s perceived level of corruption is among the worst in South Asia (Transparency
International 2021).
350
According to a poll undertaken by the International Republican Institute’s (IRI) Center for
Insights in Survey Research (CISR) between March 1, 2023 and April 6, 2023, Sheikh Hasina
enjoys a 70% public approval rating. However, approval of the opposition increased from 36% in
September 2019 to 63% in March/April 2023 (CISR 2023).

221
et al. 2020). In Zambia, Brautigam (2022) argues that the competitiveness of
presidential elections heightens this desire, but in Bangladesh the stakes appear
to be high even in relatively uncompetitive elections, such as in 2018 when the
main opposition party BNP boycotted them. In the run-up to the December
2018 election, Beijing also seemed to make special efforts to ensure that its
favored candidate, Sheikh Hasina, was in the best possible position to win
reelection.351

In the six-month period leading up to the December 2018 election, Beijing

approved, started implementation on, or completed projects backed by grant
and loan commitments worth $6.8 billion (see Table 4.4). One of these projects
included the 1,320 MW Banskhali coal-fired power plant, which was financed by
five Chinese creditors (Agricultural Bank of China, China Development Bank,
China Construction Bank, and Bank of China, China Minsheng Bank) and one
local creditor (Rupali Bank Limited) through a $1.78 billion syndicated loan.
During the same six-month period, Chinese state-owned companies began the
implementation of six additional projects backed by financial commitments
worth $6.5 billion, including the 1,320 MW Payra coal-fired power plant and the
high-profile Padma bridge rail link project. Beijing kept up the momentum after
Sheikh Hasina won re-election. During the first six months of her new term in
office, Chinese donors and lenders green-lit new projects worth $1 billion;
Chinese firms started implementing new projects backed by official financial
commitments worth $1.2 billion; and two projects (backed by Chinese loans
worth $300 million) were completed (see Table 4.4). As in prior years, the
projects were infrastructural in nature (for electricity production and national
internet connectivity).

Similar to its relationship with Edward Lungu in Zambia and the Kirchners in
Argentina, Beijing’s partnership with Sheikh Hasina appears to reflect a
preference for working with incumbents who have good reelection
prospects—or political longevity for other reasons. From a return-on-investment
standpoint, this preference may reflect the fact that Chinese infrastructure

351
China has previously demonstrated that it is willing and able to fast-track infrastructure projects to help
friendly political incumbents (e.g., Jansson 2013).

222
financiers want their loans repaid on time and with interest.352 However, given
that infrastructure project completion increases public support for China (Wellner
et al. forthcoming 2023), Beijing’s diplomats likely also have incentives to
prioritize collaboration with stable governments that have put in place policies
and institutions which facilitate the delivery of large-scale infrastructure
projects.353

The gateway to South and Southeast Asia

Given its unique locational position and recent geopolitical trends—including

the rise of India and growing Sino-U.S. tensions in the South China
Sea—Bangladesh will likely remain a key battleground for great power
competition in the coming years. From Dhaka’s vantage point, growing
competition between China and the U.S. in the Indo-Pacific region only offers
more opportunities for improving economic growth. Its economy still needs
diversification—e.g., through new competitive industries adjacent to its
dominant textile industry. This will require investment, skills transfer, and further
integration of existing industries into global supply chains. A deepening trade
relationship with China, potentially through the planned rail link via Myanmar,
could increase economic prosperity through job creation.

From the U.S. perspective, its robust trade ties with Bangladesh and strong
alliance with India (Bangladesh’s most influential regional partner) may provide
opportunities to make inroads. Yet recent events—including vote rigging, voter
intimidation, the use of violence, and the targeting of Nobel Peace Prize
laureate Muhammad Yunus through the judicial system—have put Washington in
a tough position (Miller 2023). As a defender of liberal democracies around the
world, Washington may feel compelled to condemn actions that would push
Bangladesh further down the path of authoritarianism. On the other hand, it
cannot ignore the realpolitik consideration that China’s influence is expanding in
South and Southeast Asia.

352
Dreher et al. (2022) provide statistical evidence that when Beijing issues loans at or near market rates, it
favors countries with high levels of political stability.
353
Effective public debt management is one such example. Bangladesh currently has a manageable
debt-to-GDP ratio of under 20% (World Bank 2023a).
223
Argentina case study: ‘A family affair’—China’s generosity and
pragmatism
In terms of changes in elite support for China during the early BRI period,
Argentina falls into the safe bet category of countries. Its UNGA voting patterns
demonstrate that it moved into closer alignment with China than with the U.S.
between 2014 and 2017. We also chose to conduct a case study of Argentina
because it was richly rewarded by Beijing during the late BRI period—mostly in
the form of serial emergency rescue lending that has helped pull the country
back from the brink of economic catastrophe.

It is surprising that Argentina—an upper middle-income country with close, if at

times strained, relations with the United States—is one of the largest recipients
of Chinese aid and credit in the 3.0 version of AidData’s GCDF dataset. This was
not always the case. Beijing provided loans and grants worth only $4 billion
during the pre-BRI era (2000 to 2013). Then, during the BRI era (2014 to 2021),
China’s official financial flows to Argentina skyrocketed (see Figure 4.11).
Average annual financial commitments during this period amounted to $16.8
billion. Beijing bankrolled infrastructure projects such as solar and hydroelectric
power plants, irrigation systems, highways, railway lines and locomotives, and
even a space monitoring station. However, the vast majority of China’s official
sector financial flows to Argentina came in the form of currency swap debt from
the People’s Bank of China (PBOC).354

354
A bilateral currency swap (BCS) agreement is an agreement between the central banks of two countries
to exchange cash flows in different currencies at predetermined rates over a specified period of time. The
party to the BCS agreement that draws down on the swap line becomes the borrower and its counterparty
becomes lender. The currency of the borrower is held as collateral while the lender receives interest on the
amount drawn down by the borrower until repayment is made. In principle, swap lines with the PBOC are
designed to promote the use of RMB for trade and investment settlement purposes. However, in practice,
they are mostly used to provide balance of payments support to borrowers with high levels of outstanding
debt to China during periods of financial distress (Horn et al. 2023a, 2023b). Nearly all PBOC swap
borrowings carry de jure maturities of less than one year. However, PBOC swap debt is frequently rolled
over, resulting in average de facto maturities of 3.5 years (see Box 3a in Chapter 3).
224
Figure 4.11: Official financial flows from China to Argentina, 2014-2021

Figure 4.11

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF
determinations (as described in Section A-2 of the Appendix).

At the same time, Argentina is geographically and economically proximate to

the West. It has been a major non-NATO ally to the United States since 1998
(Center for Latin American and Latino Studies 2022). The U.S. is its third largest
export and import market (The Growth Lab at Harvard University 2023).
Argentina has close, if complicated, relations with the IMF, which has provided
tens of billions of dollars of credit to support the country. Argentina is also a
vibrant democracy and one with a highly educated populace.

But Argentina has a history of pursuing strategic ties with

ideologically-confounding partners in pursuit of its national agenda. During the
late Cold War, the Argentine right-wing military junta had good relations with
the Soviet Union to support its inflation-ridden economy (Schumacher 1981).
During the modern era of great power competition, Argentina’s status as a
regional leader in Latin America and a G-20 economy has made it particularly
important to both the United States and China, and it has pursued ties with both

225
nations to benefit itself. Nonetheless, our UNGA-based measure of elite support
suggests that Argentina moved into closer alignment with China than the U.S.
during the early BRI period. But the story begins a decade before the launch of
the BRI.

The Kirchners and the BRI

During the 2003-2007 and 2007-2015 Peronist presidencies of Néstor Kirchner

and his wife Cristina Fernández de Kirchner (CFK), U.S.-Argentina relations were
shaky, inconsistent, and at times tense and hostile, although not
cooperation-free (Sullivan and Nelson 2017). The Kirchners’ governments
deepened ties with American adversaries, including Russia, Venezuela, Iran, and
Cuba (Filkins 2015).

While these tensions with the United States grew and U.S. foreign policy
focused on the Global War on Terror, the Kirchners deepened Argentina’s
relationship with China, especially during CFK’s second term (Sullivan and
Nelson 2016). China became the second largest export and import market for
Argentina during the Kirchners’ presidencies (The Growth Lab at Harvard
University 2023). During CFK’s first term (December 2007 to December 2011),
official financial commitments from China to Argentina amounted to only $1
billion. However, during her second term (December 2011 to December 2015),
official financial commitments from China to Argentina skyrocketed to $23.9
billion.355

355
These figures, as well as other figures that rely on precise dates, only include projects where the
commitment dates within a given year are entirely unknown if the relevant measure [presidential term,
months before an election, etc.] makes up more than half of the year; so for 2015, since CFK served as
president for all but 21 days of the year, AidData includes commitments with entirely unknown dates as
under her presidency.
226
Figure 4.12: Argentina’s U.N. voting alignment with China and the U.S.

Figure 4.12

Notes: See Box 1b for more detail on this measure.

Figure 4.12 documents trends in Argentina’s level of UNGA voting alignment

with China and the U.S between 2000 and 2021. The influence of Kirchnerist
presidencies is evident. Argentina was never further away from China—from a
foreign policy alignment perspective—than in 2003. Then, under the Kirchners’
presidencies, Argentina steadily moved into closer alignment with China, a trend
that intensified during the early BRI period.

Argentina’s ties to China strengthened during CFK’s second term, which

coincided with the beginning of the BRI. In late 2014, several Chinese
state-owned banks green-lit a $4.71-billion syndicated loan for the 1,740 MW
Néstor Kirchner and Jorge Cepernic Hydroelectric Power Plant Construction
Project (KCHP Project), a twin dam of great personal significance to CFK, as it
was named in part after her late husband who died in 2010 (Lucci 2019).356 CFK

356
The unredacted loan contract was included in the How China Lends Dataset, Version 1.0
(Gelpern et al. 2022) and it is accessible in its entirety via
https://www.documentcloud.org/documents/20484849-arg_2014_435.
227
described the project as “the most important hydroelectric project” in
Argentina’s history and served as its major patron (Watts 2015). The EPC
contract was awarded to China Gezhouba Group as part of a wider consortium
with two Argentine companies.

The dam project combined Peronist political goals (infrastructure-driven

nation-building) with CFK’s personal political goals (developing the Santa Cruz
Province, the familial and political core of Kirchnerism) and the principles of the
BRI (infrastructure to spur global development and trade). Argentina’s leadership
was well-positioned to leverage large-scale financing from China to support its
agenda. For Beijing, the project represented an opportunity to promote the BRI
and pursue its soft power objectives.

On July 30, 2014, two days before signing of the loan agreement for the KCHP
Project, the Government of Argentina defaulted on its foreign bond repayment
obligations (von Hoffman 2014). At the time, Argentina was experiencing a
recession, a currency devaluation, and high levels of inflation. Earlier that month,
on July 18, 2014, during Xi Jinping’s visit to Buenos Aires, the People’s Bank of
China (PBOC) and the Central Bank of Argentina (BCRA) extended an RMB 70
billion (around $11 billion) bilateral swap agreement for the promotion of trade,
use of the renminbi, and the bolstering of Argentina’s foreign exchange reserves
(Arnold 2023). Later that year, the BCRA activated the swap arrangement,
drawing down RMB 14.2 billion between October and December. Then, in 2015,
it again drew down RMB 70 billion. The PBOC swap line helped rescue
Argentina without any significant involvement from Western powers. This was
significant because the default originated from an Argentine refusal to meet an
U.S. court-mandated repayment to a “vulture fund” (Shortell 2014).

The fact that China was willing to provide a large amount of credit during a
period of crisis convinced some local elites that China could be a viable
alternative to the West. After its 2001 default, Argentina had been isolated from
international capital markets, so the emergence of China as a major international
lender was a major boon, one that would allow the country to reduce its
dependence upon the IMF and Western sources of funding. The availability of

228
Chinese credit could also be leveraged to secure more favorable offers from the
West.

Partnership with China also offered domestic benefits that could help CFK’s
Justicialist Party remain in power. In exchange, China benefited from an
administration in Buenos Aires that was more willing to adopt foreign policy
positions that it favored. By 2015, the UNGA voting alignment data shows that
Argentina’s ideal distance from China reached its closest point across the entire
22-year period (at 0.0968).

Given the pro-China orientation of CFK’s Justicialist Party, Beijing was willing to
support the incumbent by authorizing multiple drawdowns under the PBOC
swap line to stabilize the economy.357 Nevertheless, the Justicialist Party
candidate, Daniel Scioli, who had been expected to win by a large margin, lost
the 2015 election by 3 percentage points (BBC News 2015). The winner was a
conservative businessman, Mauricio Macri. During his campaign, Macri
expressed his desire to improve ties with the U.S. and European Union. Macri
never directly voiced an anti-China position. In fact, he publicly stated the
importance of maintaining good relations. But he did signal that contracts
signed by CFK’s administration with Chinese companies required review for
corruption and insufficient “technical details.” CFK’s personal involvement in the
KCHP Project had been a source of domestic political grievance that the new
administration wished to rectify (Center for Latin American and Latino Studies
2022).

After years of heavy financing during CFK’s years, from Beijing’s perspective, the
prospect of an Argentinian repositioning toward the West would be a policy
failure. The fact that Macri’s accession came in the midst of a wave of new
center-right governments across Latin America likely amplified this concern
(Center for Latin American and Latino Studies 2022).

China’s suspicions about this matter were not ill-founded. Under the Macri
administration, U.S.-Argentina relations were much closer than when his

357
With the country’s economic problems a dominant issue during the campaign, inflation running around
25%, and currency reserves declining, Beijing’s financial rescue package was useful to governing elites
(GBH 2015).
229
predecessors were in power.358 Furthermore, in 2018, the Macri administration
negotiated the largest IMF loan ever—$57.1 billion—to support the country’s
economy (Sen 2018).

Relations with China were not straightforward at first. In late December 2015,
Macri suspended the twin dam (KCHP) project on environmental grounds (Lucci
2019).359 This triggered concerns from the Chinese lenders, which had already
disbursed $950 million. On March 10, 2016, CDB sent a letter on behalf of the
other members of the loan syndicate to Argentina’s finance ministry and warned
of legal and political consequences of project suspension: "[the KCHP Project
and the Belgrano Railway Modernization Project are] major projects promoted
by the Chinese parties in the same period of time in Argentina and each... [of
the loan agreements for the two projects] contain 'cross default' provisions." In
effect, CDB told the Argentine authorities that they could not cancel the KCHP
project without canceling the Belgrano railway project (Gelpern et al. 2022).360

In April 2016, after meeting with Xi in Washington D.C., Macri announced that
the KCHP Project would proceed, although it would be modified to minimize
negative environmental impacts (Koop 2016). Then, in December 2016, the
Supreme Court of Argentina suspended construction until an environmental
impact assessment was completed and a public hearing was held. Construction
ultimately resumed in March 2018. Macri, after learning of China’s determination
to see a project personally endorsed by Xi succeed, decided that canceling the
project was not worth the potential consequences of alienating or antagonizing
a major creditor (Patey 2017). He also took steps to develop a stronger
relationship with China while maintaining good relations with the West. During
the early BRI years of his presidency, official financial commitments from China
to Argentina remained substantial: $13.2 billion in 2016 and $14.4 billion in
2017.

358
This was bolstered by the ideological synergy between Macri and Donald Trump. Macri’s Argentina
supported the Trump administration’s campaign against Nicolás Maduro’s Venezuela (Wilkinson 2019).
359
Between 2000 and 2021, the ESG risk prevalence rate in China’s grant- and loan-financed infrastructure
project portfolio was 44% in Argentina (see Table A12).
360
The letter can be accessed in its entirety via
https://www.dropbox.com/s/q6s26ninx4ldnes/Cross-Default%20Letter%20from%20China%20Development
%20Bank%20to%20the%20Government%20of%20Argentina%2010%20March%202016.pdf?dl=0.
230
The late BRI period under Macri and Fernández

Argentina saw further growth in financial support from Beijing during the late BRI
period (2018-2021). Average annual financial commitments from China during
this period amounted to $21.4 billion. Most of this funding was provided via
PBOC swap line drawdowns for balance of payments support.361 By the time
Macri exited office in December 2019, Argentina’s RMB swap debt represented
50.5% of the country’s total foreign exchange reserves (see Table 4.5).

Table 4.5: Estimated percentage of Argentina’s foreign currency reserves derived from PBOC swap facility

Table 4.5
Estimated percentage of Argentina’s foreign currency
reserves derived from PBOC swap facility
Amounts outstanding under PBOC
Total reserves Estimated percentage of reserves
Year swap facility
(USD billions) originating from PBOC swap facility
(USD billions)

2009 $48.01 $0 0%

2010 $52.21 $0 0%

2011 $46.27 $0 0%

2012 $43.22 $0 0%

2013 $30.53 $0 0%

2014 $31.41 $2.32 7.39%

2015 $25.52 $10.78 42.25%

2016 $38.41 $10.07 26.22%

2017 $55.31 $10.75 19.44%

2018 $66.22 $18.96 28.65%

2019 $44.88 $18.60 41.46%

361
In July 2017, BCRA and PBOC extended the swap line by another three years; then, in December 2018,
during Xi’s state visit to Argentina, BCRA and PBOC signed a deal to increase the swap line’s limit from
RMB 70 billion to RMB 130 billion (around $19 billion) (Horn et al. 2023a). This agreement included a
stipulation that PBOC could reject currency swap drawdowns if Argentina’s IMF standby agreement was
suspended or canceled.
231
 Amounts outstanding under PBOC
Total reserves Estimated percentage of reserves
Year swap facility
(USD billions) originating from PBOC swap facility
(USD billions)

2020 $39.40 $19.89 50.49%

2021 $39.65 $20.40 51.47%

Notes: This table shows the estimated percentage of Argentina’s foreign currency reserves that originated
from the PBOC swap facility between 2009 and 2021. The data on total reserves are from the World Bank
and include gold reserves. Amounts outstanding under the PBOC swap facility are from Horn et al (2023a).
All amounts are reported in nominal USD.

Despite the initial challenges, Macri eventually became a valued partner to

China, as demonstrated by the increase in the swap line, Xi’s visits, and various
trade agreements, including the opening up of China to Argentina’s soymeal
livestock feed exports in September 2019 and lower barriers to beef and sheep
exports (Koop 2018, Bronstein and Heath 2019). Macri took measures to join the
China-backed Asian Infrastructure Investment Bank (AIIB) in 2018, which only
became official in 2020. Also, while Argentina did not officially join the BRI until
the next administration, Macri was one of only two heads of state from Latin
America to participate in the first Belt and Road Initiative Forum in 2017 (Center
for Latin American and Latino Studies 2022).

Beyond PBOC swap line borrowings, significant collaborative activities with

China during Macri’s presidency included the General Roca Railway Equipment
Acquisition Project, which received a $236-million buyer’s credit from CDB and
supported a commercial contract with state-owned CRRC Qingdao Sifang; the
National Road Line B (Corredor B Toll Road) Project, which was supported by a
$1.18-billion syndicated loan from three banks, including the Bank of China and
ICBC, to a Argentine-Chinese joint venture responsible for the construction and
operation of the road; and the 312 MW Caucharí Photovoltaic Power Plant
Project, financed by a $331.5-million China Eximbank loan. Macri’s personal
pragmatism and the prospect of what China could still offer helped foster a
good relationship.

China’s experience with Macri left Beijing with an optimistic outlook for the
future of Argentine-Chinese relations. Despite Macri’s pro-Western views and
actions and initial caution toward China, under his leadership Argentina grew
even closer to China. If China could thrive in Argentina under a conservative,
232
pro-Western president, then it could thrive under almost any conceivable
Argentine president—especially if the Kirchnerists returned to power, who had
deepened the relationship in the first place. China had demonstrated during
both CFK and Macri’s presidencies that it was a reliable, critical partner to the
Argentine government for the country’s economic stability and willing to
negotiate if necessary, leaving little willpower among Argentine elites to alienate
or antagonize a valuable partner; thus, China had turned Argentina into a “safe
bet” for the foreseeable future.

Macri’s center-right government ultimately proved insufficiently popular by the

time of the 2019 Argentine general elections, due to economic troubles. The
Kirchnerists returned to power with president Alberto Fernández and former
president, now vice president, Cristina Fernández de Kirchner (no relation).
During his campaign, Fernández often discussed China, arguing that Macri did
not appreciate China enough. Additionally, Fernández met with Zou Xiaoli, the
Chinese Ambassador in Argentina, and sent several of his foreign policy advisors
to the embassy in the weeks prior to the election. Observers expected that
under the Fernández administration, relations would be close, with China’s
projects in the country expected to be boosted (Koop 2019). As a
demonstration of the importance of relations, Fernández’s government
appointed Sabino Vaca Narvaja, son of one of the founders of the Montoneros
guerillas and the brother-in-law of CFK’s daughter, as ambassador to China and
assigned a general to serve as the defense attache, something previously only
done for the U.S. (Ellis 2021).

Examining the data on financial commitments from the 3.0 version of AidData’s
GCDF dataset during the latter half of the late BRI period, there is little evidence
of differential treatment across the Macri and Fernández administrations, with
Argentina receiving $21.2 billion in 2020 and $20.2 billion in 2021. However,
even though relations seemed poised to grow, the Fernández administration’s
relations with China reportedly stagnated because of issues on the Argentine
side (Giusto and Harán 2023). Numerous pledged or committed projects with
China never reached implementation because of currency controls,
protectionism, bureaucratic bungling, and inconsistent policymaking from
Argentina’s national leadership (Economist 2023).

233
An example is the Atucha III Nuclear Plant Project. First pledged in 2014 during
CFK’s second term, it went through multiple rounds of negotiations with the
Macri administration but never made solid progress. Soon after his victory,
Fernández announced that the project would proceed. In February 2022,
Argentina signed an EPC contract with China National Nuclear Corporation, only
for Argentina to ask China to 100% finance the project in April 2022 (as opposed
to the standard 85% maximum). More negotiations and requests for
modifications followed throughout the year, leaving the project on shaky
grounds (Bernhard 2022). Atucha is not an outlier. Between 2018 and 2021,
China pledged $664.6 million of additional financing, but none of these pledges
had become formal commitments by 2023, and an additional set of projects
worth $146 million were suspended.362

Even if Fernández was not as interested in strengthening Sino-Argentine ties,

Fernández, being a close Kirchner ally, was politically aligned with China. In
February 2022, he visited Beijing and signed an agreement to join the BRI
(Giusto and Harán 2023). China also issued a statement in support of Argentine
claims to the Falkland Islands (known as Las Islas Malvinas in Argentina) (Rey
2022). At the UNGA, he continued the overall pattern established by his
predecessors, bringing Argentina into closer voting alignment with China.363

The large volume of official financing from China to Argentina—especially in the

form of government-sustaining currency swap drawdowns, and in addition to
foreign direct investment and trade ties—has made China a critical financial
partner for Argentina. During the early BRI period, which saw the last part of
CFK’s rule and the initially difficult, but soon productive, beginning of the Macri
administration, Beijing realized how strong its foothold was, to the point that it
could feel confident that Argentine elites would not endanger relations, turning
it into a “safe bet” amongst all recipients for friendly elite alignment with China.

362
In addition to projects that secured financial commitments in the early BRI era—which do not
have the excuse of COVID-19 as a delaying factor and had more opportunities to be formally
committed—over $2.8 billion of financing that had been pledged never reached implementation
and projects worth $11 billion were suspended.
363
In 2021, he achieved one of the closest ideal distances (0.1452) from China during our entire
22-year period of study.
234
Based on the close relations under the Macri presidency and political alignment
during the Fernández presidency, Beijing appears to have correctly predicted
the course of bilateral relations. The overall trend has been one of increasing
UNGA voting alignment between China and Argentina and an expanding
envelope of Chinese aid and credit.

Entrenchment as a goal of soft power

During the BRI era, in light of good bilateral relations, China has provided
meaningful economic support to Argentina. Because the economy is the single
most important electoral issue and Argentina’s dependency on China is high,
major political parties have strong incentives to maintain good relations with
Beijing. Macri experienced the downside of being seen as uncooperative toward
Beijing, forcing him to course correct toward a friendlier path. But as great
power competition between China and the West intensifies, Argentina’s ability
to maintain good ties with both sides is becoming limited. A case in point is the
ongoing U.S. push to sell its own F-16 fighter jets to Argentina, in order to
prevent it from purchasing the cheaper Chinese-designed and
Pakistani-manufactured JF-17 jets (Buenos Aires Times 2023).

However, a candidate in the Presidential run-off election who received 30% of

the popular vote during the first round is far-right populist and libertarian
economist Javier Milei. In August 2023, Milei announced his intentions to freeze
relations with China, calling its government “an assassin” and describing his
proposed foreign policy as a “fight against socialists and statists” (Brandimarte
and Tobias 2023). Analysts have compared Milei to Donald Trump and former
Brazilian president Jair Bolsonaro, who both used anti-China rhetoric to bolster
their electoral campaigns (Shi 2023). Should Milei win the election, Argentina’s
foreign policy would likely shift into substantially closer alignment with the
United States. Beijing may hope that as president Milei would soften his rhetoric
on China once in office like Macri did, but regardless, China would likely prefer
another candidate.364

364
That being said, in October 2023, center-right candidate Patricia Bullrich announced that she
would reverse President Alberto Fernández’s decision to join the BRICs bloc if elected. She also
told the Financial Times that “[w]e believe that in some of the latest [Chinese] loans there are
235
In the run up to the first-round elections in October 2023, Chinese official sector
entities provided additional resources to Argentina, presumably to bolster the
campaigns of candidates it deems least problematic. In June 2023, Sergio
Massa, the incumbent economy minister and a leading candidate, visited
Beijing, where he struck deals on Argentine exports, $3.05 billion in financing for
various projects, and, most importantly, an extension and expansion of the
PBOC swap line (Alcoba 2023). Based on China’s past decision making in
Argentina and an empirical pattern of commiting more funds in election years
around the world, more agreements and commitments are likely forthcoming
(Dreher et al. 2019).

But the same logic of China being too big to push in Argentina may also apply
to its relationship with the U.S. Argentina lacks the leverage, or willingness, to
become a true American adversary akin to Cuba or Venezuela. Factors like
geographic proximity, democratic affinity, and cultural synergy all advantage the
U.S. over China. In 2021, 59.48% of Argentinians approved of the U.S.
government, according to Gallup, as compared to only 32.65% for China. The
economic dimension is also noteworthy. While not nearly as large as China, the
U.S. is still one of the major trade partners of Argentina. It is also the single
largest shareholder at the IMF, where Argentina is the single largest debtor.
Under its new government in 2024, Argentina’s foreign policy will have to
carefully tread between ties with the U.S. and China, both of which are critical
for its future economic prospects.

Section 5: Beijing’s approach to reputational risk

management
In this chapter, we have provided statistical and case study evidence that
China—like other foreign powers—makes adjustments to its international
development finance portfolio in response to soft power gains and losses on the
ground. It is taking reactive and proactive measures to manage reputational risk.

clauses which we don’t know about and we are ready to re-examine them” (Stott and Nugent
2023).

236
There are five key takeaways from our analysis. First, on three different measures
of soft power (that capture gains and losses in public opinion, media sentiment,
and elite support), Beijing devoted nearly two-thirds of its entire international
development finance portfolio during the late BRI period to “toss up” countries.
These countries represent competitive jurisdictions where neither China nor the
U.S. opened up an insurmountable lead vis-à-vis its principal rival.

Second, in settings where China recently made reputational gains at the

expense of its principal competitor, it doubled down with additional aid and
credit—i.e., it sought to maintain and build upon momentum.365 The fact that
Beijing is redirecting aid and credit to battleground countries where it has
recently gained advantage challenges a popular belief among Sinologists: that
the bureaucratic machinery responsible for Beijing’s overseas development
program is fragmented, uncoordinated, and inattentive to changing reputational
dynamics on the ground.366

Third, China does not have much of an appetite for reputational risk. It
consistently allocated a lower-than-expected share of its international
development finance portfolio to “moonshot” countries (dedicating between
16% and 27% less than expected for each soft power measure, see Figure 4.5).
These are countries where its principal rival has momentum on its side. China
rarely seeks to woo indifferent or antagonistic countries with aid and credit, but
instead prioritizes countries that are already moving in its direction. A separate,
but related, point is that, when reputational assets turn into reputational
liabilities (as we saw in the Zambia case study), Beijing mostly disengages from
discussions about new projects and financial commitments and refocuses on
managing risks within its existing portfolio of grant- and loan-financed projects.

Fourth, as case studies of Bangladesh and Argentina demonstrate, political

transitions in host countries are critical moments when the nature, level, and
pace of China’s engagement can change significantly. Beijing understands that it

365
More specifically, we find that Beiijing prioritized the provision of aid and credit to countries where it had
experienced public opinion and media sentiment gains at the expense of the U.S. during the early BRI
period, while it deprioritized toss-up countries where the momentum shifted in favor of the U.S. during the
early BRI period.

366
We thank Marina Rudyak for calling our attention to this point.
237
can make major reputational gains or suffer major reputational losses during
these windows, and it has demonstrated a willingness and ability to use
instruments of state power to protect its interests when such junctures arise.
During the transition from the Kirchner administration to the Macri
administration, a consortium of Chinese state-owned policy and commercial
banks invoked a cross-default clause in a loan agreement with the Argentine
Ministry of Finance to block the newly-elected president from following through
on his electoral pledge to suspend environmentally risky projects, which could
have jeopardized a $5.5 billion dam construction contract that was previously
issued to China Gezhouba Group Company Limited. Conversely, when new
leaders come to power and take a less adversarial posture (like Bangladesh’s
Sheikh Hasina and Zambia’s Edward Lungu), our findings suggest that Beijing
often seeks to cement bilateral relations by helping incumbents take credit for
high-profile projects.367 It does so by, among other things, approving new
financial commitments for projects that were previously under consideration,
organizing groundbreaking ceremonies for previously approved projects that
had not yet commenced, and providing state-sponsored media coverage of
recently completed projects.368 This strategy is especially relevant in democratic
countries where elections are more competitive, as in Zambia where incumbent
performance at the ballot box is linked to perceived effectiveness at delivering
big-ticket infrastructure projects that can create jobs and stimulate short-term
economic growth.

Finally, for those who make and shape policy in Western capitals, a key insight
from this chapter is that Beijing tends to disengage rather than double down in
countries where there are strong indications of BRI buyer’s remorse. These are
jurisdictions where Beijing’s competitors may be able to lure countries back into
the West’s orbit by focusing on their own areas of comparative advantage.
However, doing so would require that Western powers act quickly when these
windows of opportunities arise and adapt their programming to address the
unmet needs of partner countries.

367
On this point, also see Holslag 2011; Jansson 2013; DiLorenzo and Cheng 2019; Dreher et al. 2019;
Anaxagorou et al. 2020; Strange 2023; Tang 2021; Kern et al. 2022.
368
Wellner et al. (forthcoming) demonstrate that the completion of Chinese grant- and loan-financed
projects increases public support for the Chinese government and the host government.
238
References
Aamir, Adnan, Marwaan Macan-Markar, Shaun Turton, Cissy Zhou and Grace Li. 2022. Road to
nowhere: China’s Belt and Road Initiative at tipping point. 10 August 2022. Nikkei Asia.
Accessed on 13 August 2023 at
https://asia.nikkei.com/Spotlight/The-Big-Story/Road-to-nowhere-China-s-Belt-and-Road-Initiativ
e-at-tipping-point

Abi-Habib, Maria. 2018. How China Got Sri Lanka to Cough Up a Port. New York Times. 25 June
2018. Accessed on 24 July 2023 at
https://www.nytimes.com/2018/06/25/world/asia/china-sri-lanka-port.html.

Acker, Kevin, Deborah Brautigam, and Yufan Huang. 2020. Debt Relief with Chinese
Characteristics. SAIS-CARI Working Paper No. 2020/39. Washington DC: Johns Hopkins
University.

Aiping, Zeng. 2015. China-Africa Governance Exchanges and Experiences. Beijing: China
Institute of International Studies.

Akhtar, Shayerah I. and Nick M. Brown. 2021. U.S. International Development Finance
Corporation: Overview and Issues. Washington DC: Congressional Research Service. Accessed
on 16 July 2023 at https://sgp.fas.org/crs/row/R47006.pdf

Al Jazeera. 2018. China and Argentina Sign Currency Swap Deal after G20 Summit. Al Jazeera. 2
December 2018. Accessed on 1 October 2023 at
https://www.aljazeera.com/economy/2018/12/2/china-and-argentina-sign-currency-swap-deal-aft
er-g20-summit.

Alcoba, Natalie. 2023. Argentina’s Yuan Lifeline Sign of Brinksmanship between China, US. Al
Jazeera. 17 July 2023. Accessed on 1 October 2023 at
https://www.aljazeera.com/economy/2023/7/17/argentinas-yuan-lifeline-sign-of-brinksmanship-b
etween-china-us.

Alesina, Alberto, and David Dollar. 2000. Who Gives Foreign Aid to Whom and Why? Journal of
Economic Growth 5 (1): 33–63.

Amnesty International. 2023. Bangladesh: Unlawful Use of Force against Protesters Must End
Immediately. Amnesty International. 4 August 2023. Accessed on 1 October 2023 at
https://www.amnesty.org/en/latest/news/2023/08/bangladesh-unlawful-use-of-force-against-prot
esters-must-end-immediately/.

Anaxagorou, Christiana, Georgios Efthyvoulou, and Vassilis Sarantides. 2020. Electoral Motives
and the Subnational Allocation of Foreign Aid in Sub-Saharan Africa. European Economic Review
127: 103430.

Anderlini, Jamil. 2008. Beijing uses forex reserves to target Taiwan. 12 September 2008.
Financial Times. Accessed on 1 August 2023 at
https://www.ft.com/content/22fe798e-802c-11dd-99a9-000077b07658

Ang, Yuen Yuen. 2019. Demystifying Belt and Road The Struggle to Define China’s Project of the
Century. Foreign Affairs. 22 May 2019.

239
ANI. 2022. Finance Minister Kamal Warns Developing Nations of Chinese Loans Strapped with
Debt-Trap. ANI. 10 August 2022. Accessed on 1 August 2023 at
https://en.prothomalo.com/bangladesh/bangladesh-in-world-media/sysadnqpr3.

Ansar, Atif, Bent Flyvbjerg, Alexander Budzier and Daniel Lunn. 2016. Does infrastructure
investment lead to economic growth or economic fragility? Evidence from China. Oxford Review
of Economic Policy 32 (3): 360–390.

Arnold, Vincient. 2023. China: Central Bank Swaps to Argentina, 2014. Journal of Financial
Crises 5 (1): 158–88.

Asian Infrastructure Investment Bank (AIIB). 2021. AIIB, MCDF Partnership to Foster High-Quality
Infrastructure Investments. 25 August 2021. Beijing: AIIB. Accessed on 15 August 2023 at
https://www.aiib.org/en/news-events/news/2021/AIIB-MCDF-Partnership-to-Foster-High-Quality-
Infrastructure-Investments.html

Asmus, Gerda, Vera Z. Eichenauer, Andreas Fuchs and Bradley C. Parks. Forthcoming. Does India
Use Development Finance to Compete with China? A Subnational Analysis. Journal of Conflict
Resolution.

Atkinson, Carol. 2010. Does soft power matter? A comparative analysis of student exchange
programs. Foreign Policy Analysis 6: 1-22.

Bae, Kee‐Hong and Vidhan K. Goyal. 2009. Creditor Rights, Enforcement, and Bank Loans.
Journal of Finance 64(2): 823-860.

Baehr, Christian, Ariel BenYishay, and Bradley C. Parks. Forthcoming. Highway to the Forest?
Land Governance and the Siting and Environmental Impacts of Chinese Government-Funded
Road Building in Cambodia. Journal of Environmental Economics and Management.

Bahaj, Saleem, and Ricardo Reis. 2022. Jumpstarting an International Currency. CEPR Discussion
Paper DP14793. London: Centre for Economic Policy Research (CEPR).

Bailey, Michael A., Anton Strezhnev, and Erik Voeten. 2017. Estimating dynamic state
preferences from United Nations voting data. Journal of Conflict Resolution 61 (2): 430-56.

Banco Central de Bolivia. 2016. China Exim Bank: The Export-Import Bank of China [Slide deck].
Retrieved from
https://www.dropbox.com/s/ryfvumv7fl85gg3/banco_central_bol_EXIMBANK%20
CHINA%202016.pdf?dl=0.

Bank for International Settlements (BIS). 1995. Guide to the BIS statistics on international
banking. Basel, Switzerland: BIS.

Bank for International Settlements (BIS). 2018. Statistical release: BIS international banking
statistics at end-June 2018. Basel, Switzerland: BIS. Accessed on 18 July 2023 at
https://www.bis.org/statistics/rppb1810.pdf

Bank for International Settlements (BIS). 2022. BIS international banking statistics and global
liquidity indicators at end-December 2021. Basel, Switzerland: BIS. Accessed on 19 July 2023 at
https://www.bis.org/statistics/rppb2204.htm

Bank for International Settlements (BIS). 2023. Statistical release: BIS international banking
statistics and global liquidity indicators at end-December 2022. Basel, Switzerland: BIS.
Accessed on 19 July 2023 at https://www.bis.org/statistics/rppb2304.htm

240
Barbaro, Michael and Abdi Latif Dahir. 2023. America Has a Problem in Africa: China. New York
Times. 6 April 2023. Accessed on 1 August 2023 at
https://www.nytimes.com/2023/04/06/podcasts/the-daily/kamala-harris-us-africa-china.html.

Batson, Andrew. 2008. China Used Reserves To Sway Costa Rica. 13 September 2008. Wall
Street Journal. Accessed on 2 August 2023 at
https://www.wsj.com/articles/SB122121919505927749

BBC News. 2015. Conservative Mauricio Macri Wins Argentina Presidency. BBC News. 23
November 2015. Accessed on 1 August 2023 at
https://www.bbc.com/news/world-latin-america-34897150.

Belt and Road Forum for International Cooperation. 2017. Speech by Xi Jinping at the Opening
Ceremony of the Belt and Road Forum for International Cooperation on May 14, 2017 in Beijing.
Accessed on 15 August 2023 at
https://web.archive.org/web/20230514064500/http://www.beltandroadforum.org/n100/2017/05
14/c24-407.html

Benabdallah, Lina. 2020. Shaping the Future of Power: Knowledge Production and
Network-Building in China-Africa Relations. Ann Arbor, MI: University of Michigan Press.

Bernhard, Isabel. 2022. Why Argentina’s Nuclear Power Project With China Has Stalled. The
Diplomat. 14 December 2022. Accessed on 1 August 2023 at
https://thediplomat.com/2022/12/why-argentinas-nuclear-power-project-with-china-has-stalled/.

Bertaut, Carol, Bastian von Beschwitz, Stephanie Curcuru. 2023. ‘The International Role of the
U.S. Dollar’ Post-COVID Edition. FEDS Notes. Washington: Board of Governors of the Federal
Reserve System.

Blair, Robert A., Robert Marty, and Philip Roessler. 2022a. Foreign Aid and Soft Power: Great
Power Competition in Africa in the Early Twenty-first Century. British Journal of Political Science
52 (3): 1355-1376.

Blair, Robert A., Samantha Custer, Philip Roessler. 2022b. Dueling Aid Regimes: A Conjoint
Survey Experiment on Elites’ Development Finance Preferences in 141 Low- and Middle-Income
Countries. Working Paper #119. Williamsburg, VA: AidData at William & Mary.

Bluhm, Robert, Axel Fuchs, Bradley C. Parks, Austin M. Strange, and Michael J. Tierney. 2020.
Connective Financing: Chinese Infrastructure Projects and the Diffusion of Economic Activity in
Developing Countries. CEPR Discussion Paper 14818. London, UK: CEPR.

Bolton, John R. 2019. A New Africa Strategy: Expanding Economic and Security Ties on the
Basis of Mutual Respect. Speech at the Heritage Foundation on 20 February 2019. Accessed on
1 August 2023 at https://www.heritage.org/sites/default/files/2019-02/HL1306.pdf.

Bomprezzi, Pietro, Axel Dreher, Andreas Fuchs, Teresa Hailer. Andreas Kammerlander, Lennart
Kaplan, Silvia Marchesi, Tania Masi, Charlotte Robert and Kerstin Unfried. 2023. Light of Their
Lives: Country Leaders’ Spouses and Regional Favoritism. Mimeo.

Bon, Gatien and Gong Cheng. 2021. Understanding China's role in recent debt relief operations:
A case study analysis. International Economics 166 (C): 23-41.

Bouey, Jennifer, Lynn Hu, Keller Scholl, William Marcellino, Rafiq Dossani, Ammar A. Malik, Kyra
Solomon, and Sheng Zhang. Forthcoming. China’s AI Exports at the Dawn of an AI Age:

241
Technology, Development Finance, and Data Safety. Washington D.C.: RAND Corporation and
AidData.

Bourguignon, François and Jean-Philippe Platteau. 2015. The Hard Challenge of Aid
Coordination. World Development 69 (C): 86-97.

Brandimarte, Walter, and Manuela Tobias. 2023. Argentina’s Milei Says He’d Reject ‘Assassin’
China, Leave Mercosur. Bloomberg. 16 August 2023. Accessed on 16 August 2023 at
https://www.bloomberg.com/news/articles/2023-08-16/milei-s-foreign-policy-reject-assassin-chin
a-leave-mercosur.

Brauneder, Kerstin M., Chloe Monte, Simon Blyth, Leon Bennun, Stuart H M Butchart, Michael
Hoffmann, Neil D Burgess, Annabelle Cuttelod, Matt I. Jones, Val Kapos, John Pilgrim, and
Melissa J Toll. 2018.Global screening for Critical Habitat in the terrestrial realm. PLOS ONE 13
(3): 1-16.

Bräutigam, Deborah and Kevin Acker. 2020. Is China Hiding its Overseas Lending? Horn,
Reinhart and Trebesch's Hidden Loans and Hidden Data. SAIS-CARI Blog. Accessed on 19 July
2023 at www.chinaafricarealstory.com/2020/04/is-china-hiding-its-overseas-lending.html

Bräutigam, Deborah. 2019. Misdiagnosing the Chinese Infrastructure Push. The American
Interest. 4 April 2019. Accessed on 1 August 2023 at www.the-american-interest.com/2019/
04/04/misdiagnosing-the-chinese-infrastructure-push

Brautigam, Deborah. 2021. How Zambia and China Co-Created a Debt ‘Tragedy of the
Commons.’ Working Paper 2021/51. Washington DC: SAIS-CARI.

Brazys, Samuel. and Alexander Dukalskis. 2019. Rising powers and grassroots image
management: Confucius Institutes and China in the media. Chinese Journal of International
Politics 12 (4): 557–584.

Broccolini, Chiara, Giulia Lotti, Alessandro Maffioli, Andrea F. Presbitero and Rodolfo Stucchi.
2021. Mobilization Effects of Multilateral Development Banks. World Bank Economic Review 35
(2): 521–543.

Bronstein, Hugh, and Maximilian Heath. 2019. China Opens Soymeal Market to No. 1 Exporter
Argentina in ‘historic’ Deal. Reuters. 10 September 2019. Accessed on 1 August 2023 at
https://www.reuters.com/article/us-argentina-soyproducts-idUSKCN1VV2JP.

Brooker, Matthew. 2023. Zambia’s Debt Deal With China Is a Landmark Moment. Bloomberg. 25
June 2023. Accessed on 1 August 2023 at
https://www.bloomberg.com/opinion/articles/2023-06-25/zambia-debt-restructuring-with-china-i
s-a-landmark-moment.

Brooks, Sarah M., Raphael Cunha, and Layna Mosley. 2015. Categories, Creditworthiness, and
Contagion: How Investors' Shortcuts Affect Sovereign Debt Markets. International Studies
Quarterly 59 (3): 587–601.

Bruni, Jo. 2021. Ecuador strikes funding agreement with US's DFC. LatinFinance. January 15.
2021. Accessed on 16 July 2023 at
https://www.latinfinance.com/daily-briefs/2021/1/15/ecuador-strikes-funding-agreement-with-uss
-dfc

Buchanan, Graeme M., Paul F. Donald, Bradley Parks, Brian O’Donnell, John Swaddle, Lukasz
Tracewski, Daniel Runfola, and Stuart H. M. Butchart. 2018. The Local Impacts of World Bank

242
Development Projects Near Sites of Conservation Significance. Journal of Environment and
Development 27 (3): 299–322.

Buchheit, Lee C. 1985. Is Syndicated Lending a Joint Venture? International Financial Law Review
4: 12-14.

Buchheit, Lee C. 1991. Making Amends for Amendments. International Financial Law Review 10:
11.

Buchheit, Lee C. and Mitu Gulati. 2023. Enforcing comparable treatment in sovereign debt
workouts. Capital Markets Law Journal 18 (1): 71-77.

Buenos Aires Times. 2023. United States Steps up Pressure on Argentina over Fighter Jet
Choice. Buenos Aires Times. 26 July 2023. Accessed on 1 August 2023 at
https://www.batimes.com.ar/news/argentina/the-us-steps-up-pressure-on-argentina-to-choose-d
anish-f-16s-over-chinese-or-indian-fighter-jets.phtml.

Bulman, David, Walter Kolkma, and Aart Kraay. 2017. Good countries or good projects?
Comparing macro and micro correlates of World Bank and Asian Development Bank project
performance. The Review of International Organizations 12(3): 335-363.

Buntaine, Mark T. 2015. Accountability in Global Governance: Civil Society Claims for
Environmental Performance at the World Bank. International Studies Quarterly 59 (1): 99–111.

Buntaine, Mark T. 2011. Does the Asian Development Bank Respond to Past Environmental
Performance when Allocating Environmentally Risky Financing? World Development 39 (3):
336-350.

Buntaine, Mark. 2016. Giving Aid Effectively: The Politics of Environmental Performance and
Selectivity at Multilateral Development Banks. Oxford: Oxford University Press.

Center for Latin American and Latino Studies. 2022. Changes and Continuities through
Sino-Argentinean Recent History. Center for Latin American and Latino Studies.
https://www.american.edu/centers/latin-american-latino-studies/upload/china-argentina-english-f
inal.pdf.

Center for Insights in Survey Research (CISR). 2023. National Survey of Bangladesh: March-April
2023. Accessed at
https://www.iri.org/wp-content/uploads/2023/08/April-2023-National-Survey-Public-Release.pdf

Central Bank of Sudan (CBOS). 2023. Sudan External Debt Position First Quarter as at 31st of
March 2022. Khartoum: CBOS. Accessed on 19 July 2023 at
https://web.archive.org/web/20230208195131/https:/cbos.gov.sd/sites/default/files/Quarter%20
1%20Report%20%202022%20.pdf

Central Intelligence Agency. 2023. The World Factbook. In CIA World Factbook.
https://www.cia.gov/the-world-factbook/countries/zambia/.

Cerutti, Eugenio, Catherine Casanova, and Swapan-Kumar Pradhan. 2023. Banking Across
Borders: Are Chinese Banks Different? Journal of Banking & Finance 154: 106920.

Cerutti, Eugenio, Catherine Koch, and Swapan-Kumar Pradhan. 2018. The growing footprint of
EME banks in the international banking system. BIS Quarterly Review. Accessed on 19 July 2023
at https://www.bis.org/publ/qtrpdf/r_qt1812e.htm

243
Chen, Xiaoyi. 2014. 揭开外储委托贷款的神秘面纱 (Unveil the Mysteries of the Entrust Loans of
the Foreign Exchange Reserves in China). 21 August 2014. Accessed on 15 September 2023 at
https://www.dropbox.com/scl/fi/r5idp65hfybugsmc5v7b3/.pdf?rlkey=xr8jqe079moeb50pybrjh90
bx&dl=0

Chen, Muyang. 2020a. Beyond Donation: China’s Policy Banks and the Reshaping of
Development Finance. Studies in Comparative International Development 55: 436–459.

Chen, Muyang. 2020b. State Actors, Market Games: Credit Guarantees and the Funding of
China Development Bank. New Political Economy 25 (3): 453-468.

Chen, Muyang. 2023. China’s rise and the reshaping of sovereign debt relief. International Affairs
99 (4): 1755-1775.

Chen, Yunnan and David Landry. 2018. Capturing the rains: Comparing Chinese and World Bank
hydropower projects in Cameroon and pathways for South-South and North South technology
transfer. Energy Policy 115: 561–571.

Cheng, Wenting, and Kai Zhang. 2023. Setting ‘green’ boundaries for Chinese green finance:
Multi-level governance and regulatory stringency. Environmental Policy and Governance 1-13.

Chengjun, Zhou. 2021. A new perspective on the international application of RMB Bonds. China
Bond 11: 19. Accessed on 1 September 2023 at
https://finance.sina.cn/bond/zsyw/2021-12-03/detail-ikyamrmy6550370.d.html?vt=4&cid=76633
&node_id=76633

Chin, Gregory T., and Kevin P. Gallagher. 2019. Coordinated Credit Spaces: The Globalization of
Chinese Development Finance. Development and Change 50 (1): 245–274.

China Banking Regulatory Commission (CBRC). 2017a. Measures for the Supervision and
Administration of China Development Bank. 15 November 2017. Beijing: CBRC. Accessed on 15
August 2023 at https://www.gov.cn/gongbao/content/2018/content_5260809.htm

China Banking Regulatory Commission (CBRC). 2017b. Measures for the Supervision and
Administration of the Export-Import Bank of China. 15 November 2017. Beijing: CBRC.
Accessed on 15 August 2023 at
https://www.gov.cn/gongbao/content/2018/content_5264999.htm

China Daily. 2009. Huijin to fund Exim, Sinosure. 17 September 2009. China Daily Accessed on 1
August 2023 at www.chinadaily.com.cn/business/2009-09/17/content_8701521.htm

China Development Bank (CDB). 2020. Pricing Supplement. 19 November 2020. Accessed at
https:// secure.fundsupermart.com/fsm/bond/relatedBondDocument/XS2263032786/
China%20Development%20Bank_Pricing%20Supplement%20_%20Programme%20OC_Issue%2
0of%20CNY2,800,000, 000%203.03%20per%20cent.%20Notes%20due%202023.pdf

Chitala, Mbita. 2022. Corporate Capture; The Political Economy of Electricity Management in
Zambia 2014-2021: How Not to Manage a State Enterprise. Lusaka: Zambia Research
Foundation.

Chuku, Chuku, Prateek Samal, Joyce Saito, Dalia S. Hakura, Marcos d Chamon, Martin D.
Cerisola, Guillaume Chabert, and Jeromin Zettelmeyer. 2023. Are We Heading for Another Debt
Crisis in Low-Income Countries? Debt Vulnerabilities: Today vs the Pre-HIPC Era. IMF Working
Papers 2023 (079).

244
Cordella, Tito and Powell, Andrew. 2021. Preferred and non-preferred creditors. Journal of
International Economics 132(C): 103491

Cruces, Juan J, and Christoph Trebesch. 2013. Sovereign Defaults: The Price of Haircuts.
American Economic Journal: Macroeconomics 5 (3): 85–117.

Custer, Samantha, Brooke Russell, Matthew DiLorenzo, Mengfan Cheng, Sid Ghose, Jake Sims,
Jennifer Turner, and Harsh Desai. 2018. Ties that Bind: Quantifying China’s Public Diplomacy and
its Good Neighbor Effect. Williamsburg, VA: AidData at William & Mary.

Custer, Samantha, Tanya Sethi, Jonathan A. Solis, Joyce Jiahui Lin, Siddhartha Ghose, Anubhav
Gupta, Rodney Knight, and Austin Baehr. 2019. Silk Road Diplomacy: Deconstructing Beijing’s
Toolkit to Influence South and Central Asia. Williamsburg, VA: AidData at William & Mary.

Custer, Samantha, Tanya Sethi, Rodney Knight, Amber Hutchinson, Vera Choo, Mengfan Cheng.
2021. Listening to Leaders 2021: A report card for development partners in an era of contested
cooperation. Williamsburg, VA: AidData at the College of William & Mary.

Custer, Samantha. 2022. (Re)investing in Our Reputational Security: Alternative Models and
Options Strengthen U.S. Strategic Communications. Williamsburg, VA: AidData at William &
Mary.

Custer, Samantha, Axel Dreher, Thai-Binh Elston, Brooke Escobar, Rory Fedorochko, Andreas
Fuchs, Siddhartha Ghose, Joyce Jiahui Lin, Ammar Malik, Bradley C. Parks, Kyra Solomon, Austin
Strange, Michael J. Tierney, Lydia Vlasto, Katherine Walsh, Fei Wang, Lincoln Zaleski, and Sheng
Zhang. 2023. Tracking Chinese Development Finance: An Application of AidData’s TUFF 3.0
Methodology. Williamsburg, VA: AidData at William & Mary.

Dawn. 2021. Pakistan does not have a 'China debt' problem: Asad Umar on US report critical of
CPEC. Dawn. 6 October 2021. Accessed on 20 September 2023 at
https://www.dawn.com/news/1650453

Dennis, Steven A. and Donald J. Mullineaux. 2000. Syndicated loans. Journal of Financial
Intermediation 9(4): 404-426.

DiLorenzo, Matthew and Mengfan. 2019. Political Turnover and Chinese Development
Cooperation. Chinese Journal of International Politics 12 (1): 123-151.

Do Rosario, Jorgelina. 2023a. Exclusive: Argentina to use IMF money to pay back part of a
currency swap with China, sources say. Reuters. 24 August 2023. Accessed on 24 August 2023 at
https://www.reuters.com/world/americas/argentina-use-imf-money-pay-back-part-currency-swap-
with-china-sources-2023-08-23/

Do Rosario, Jorgelina. 2023b. Argentina used China yuan swap to pay the IMF in October
-source. Reuters. 31 October 2023. Accessed on 31 October 2023 at
https://www.nasdaq.com/articles/argentina-used-china-yuan-swap-to-pay-the-imf-in-october-sou
rce

Do Rosario, Jorgelina and Jorge Otaola. 2023. UPDATE 3-Argentina, short of dollars, to use
yuan in $2.7 bln IMF payment. Reuters. 29 June 2023. Accessed on 29 June 2023 at
https://www.reuters.com/article/argentina-imf/update-3-argentina-short-of-dollars-to-use-yuan-in
-2-7-bln-imf-payment-idINL1N38L21F

Do Rosario, Jorgelina and Karin Strohecker. 2023. Analysis: China holds the key to avoiding
Argentina's IMF default. The price tag is unknown. Reuters. 2 August 2023 Accessed on 2

245
August 2023 at
https://www.reuters.com/markets/china-holds-key-avoiding-argentinas-imf-default-price-tag-is-u
nknown-2023-08-02/

Do Rosario, Jorgelina and Rachel Savage. 2023. China development loans to emerging
economies hit 13-year low in 2021. Reuters. 24 January 2023, Accessed on 17 July 2023 at
https://www.reuters.com/markets/asia/china-development-loans-emerging-economies-hit-13-yea
r-low-2021-study-2023-01-24/

Dolan, Carrie B., Ammar A. Malik, Sheng Zhang, Wenhui Mao, Kaci Kennedy McDade, Eli
Svoboda, and Julius N. Odhiambo. 2023. Chinese Health Funding in Africa: The Untold Story.
PLOS Global Public Health 3 (6): e0001637.

Dollar, David. 2016. China As a Global Investor. Asia Working Group Paper 4. Washington DC:
Brookings Institution

Dong, Lili and David W. Chapman. 2008. The Chinese Government Scholarship Program: An
Effective Form of Foreign Assistance? International Review of Education 54 (2): 155-173.

Douglas, Jason. 2022. China Props Up Belt-and-Road Borrowers Via Unusual Channel. 11
December 2022. Wall Street Journal. Accessed on 1 August 2023 at
https://www.wsj.com/articles/china-props-up-belt-and-road-borrowers-via-unusual-channel-1167
0727153

Dreher, Axel, Andreas Fuchs, Bradley Parks, Austin M. Strange, and Michael J. Tierney. 2018.
Apples and Dragon Fruits: The Determinants of Aid and Other Forms of State Financing from
China to Africa. International Studies Quarterly 62 (1): 182-194.

Dreher, Axel, Andreas Fuchs, Bradley Parks, Austin Strange, and Michael J. Tierney. 2021. Aid,
China, and Growth: Evidence from a New Global Development Finance Dataset. American
Economic Journal: Economic Policy 13 (2): 135-174.

Dreher, Axel, Andreas Fuchs, Bradley Parks, Austin Strange, and Michael J. Tierney. 2022.
Banking on Beijing: The Aims and Impacts of China’s Overseas Development Program.
Cambridge, UK: Cambridge University Press.

Dreher, Axel, Andreas Fuchs, Roland Hodler, Bradley C. Parks, Paul A. Raschky, and Michael J.
Tierney. 2019. African Leaders and the Geography of China’s Foreign Assistance. Journal of
Development Economics 140: 44–71.

Duggan, Julian, Scott Morris, Justin Sandefur, and George Yang. 2020. Is the World Bank’s
COVID-19 crisis lending big enough, fast enough? New evidence on loan disbursements. CGD
Working Paper 554. Washington, DC: Center for Global Development.

Dyer, Geoff, Jamil Anderlini, and Henny Sender. 2011. China’s lending hits new heights. 17
January 2011. Financial Times. Accessed on 3 August 2023 at
https://www.ft.com/content/488c60f4-2281-11e0-b6a2-00144feab49a

Economic Relations Division of the Government of the People's Republic of Bangladesh. 2023.
Wing-8:Asia & JEC. Accessed on 14 June 2023 at
https://erd.gov.bd/site/page/e0ed373a-4243-4dd4-aa10-f402dea4a3cf/e0ed373a-4243-4dd4-aa
10-f402dea4a3cf

246
Eichenauer, Vera Z. Andreas Fuchs, and Lutz Brückner. 2021. The effects of trade, aid, and
investment on China's image in Latin America. Journal of Comparative Economics 49(2):
483-498.

Ellis, Evan. 2021. New Directions in the Deepening Chinese-Argentine Engagement. Global
Americans (blog). 11 February 2021. Accessed on 20 September 2023 at
https://theglobalamericans.org/2021/02/new-directions-in-the-deepening-chinese-argentine-eng
agement/.

Embassy of the People's Republic of China in the Republic of the Philippines. 2019. Belt and
Road Primer. Accessed on 1 July 2023 at
ph.china-embassy.gov.cn/eng/sgdt/201904/t20190411_1182334.htm

Esty, Benjamin C., and William L. Megginson. 2003. Creditor rights, enforcement, and debt
ownership structure: Evidence from the global syndicate loan market. Journal of Financial and
Quantitative Analysis 38: 37–59.

Euractiv. 2023. The Belt and Road Initiative -- Where are we 10 years on? Euractiv Hybrid
Conference on 19 September 2023. Accessed on 19 September 2023 at
https://events.euractiv.com/event/info/the-belt-and-road-initiative-where-are-we-10-years-on?_g
a=2.41858210.1856855957.1695222202-589043501.1692974998

Express Tribune. 2016. Chinese Diplomat Calls CM ‘Shahbaz Speed.’ Express Tribune. 20
December 2016. Accessed on 20 September 2023 at
https://tribune.com.pk/story/1268314/unprecedented-development-chinesediplomat-calls-cm-s
hahbaz-speed.

Ferry, Lauren and Alexandra Zeitz. 2023. China, the IMF, and Sovereign Debt Crisis. Paper
presented at the 6th Interdisciplinary Sovereign Debt Research and Management Conference
(DebtCon6) at Princeton University from 27-29 April 2023. Accessed on 1 August 2023 at
https://debtcon6.princeton.edu/sites/g/files/toruqf3611/files/documents/ferry.pdf

Filkins, Dexter. 2015. A Deadly Conspiracy in Buenos Aires? The New Yorker. 13 July 2015.
Accessed on 15 August 2023 at
https://www.newyorker.com/magazine/2015/07/20/death-of-a-prosecutor.

Financial Times. 2016. China Rethinks Developing World Largesse as Deals Sour. Financial
Times. 13 October 2016. Accessed on 15 August 2023 at
www.ft.com/content/5bf4d6d8-9073-11e6- a72e-b428cb934b78.

Flores-Macías, Gustavo A., and Sarah Kreps. 2013. The Foreign Policy Consequences of Trade:
China’s Commercial Relations with Africa and Latin America, 1992–2006. Journal of Politics 75
(2): 357–71.

Flyvbjerg, Bent, Mette Skamris Holm, and Søren Buhl. 2002. Underestimating Costs in Public
Works Projects: Error or Lie? Journal of the American Planning Association 68(3): 279–295.

Foysal, Quazi Omar. 2014. The China-Bangladesh Relations: From Eternity to Modern Era.
Society & Change VIII (4).

Freedom House 2022. Freedom in the World 2022: The Global Expansion of Authoritarian Rule.
New York: Freedom House.

Freedom House. 2023. Freedom in the World 2023.

https://freedomhouse.org/country/bangladesh/freedom-world/2023.

247
French, Howard W. 2011. In Africa, an Election Reveals Skepticism of Chinese Involvement. The
Atlantic. 29 September 2011. Accessed on 15 August 2023 at
https://www.theatlantic.com/international/archive/2011/09/in-africa-an-election-reveals-skepticis
m-of-chinese-involvement/245832/.

Fung, Courtney, and Shing-hon Lam. 2022. Mixed Report Card: China’s Influence at the United
Nations. Lowy Institute.
https://www.lowyinstitute.org/publications/mixed-report-card-china-s-influence-united-nations#h
eading-7004.

Furukawa, Mitsuaki. 2018. Management of the International Development Aid System: The Case
of Tanzania. Development Policy Review 36(1): 270–284.

Gallagher, Kevin P. and Rebecca Ray. 2020. Scope and Findings: China’s Overseas Development
Finance Database. BU Global Development Policy Center. Accessed on 17 July 2023 at
https://www.bu.edu/gdp/2020/12/13/scope-and-findings-chinas-overseas-development-finance-
database/

Gardner, Alysha, Joyce Lin, Scott Morris, and Bradley Parks. 2021. Bargaining with Beijing: A Tale
of Two Borrowers. Center for Global Development and AidData at William & Mary.

Garnett, Stephen T., Neil D. Burgess, John E. Fa, Alvaro Fernandez-Llamazares, Zsolt Molnar,
Cathy J. Robinson, James E. M. Watson, Kerstin K. Zander, Beau Austin, Eduardo S. Brondizio,
Neil French Collier, Tom Duncan, Erle Ellis, Hayley Geyle, Micha V. Jackson, Harry Jonas, Pernilla
Malmer, Ben McGowan, Amphone Sivongxay, and Ian Leiper. 2018. A spatial overview of the
global importance of Indigenous lands for conservation. Nature Sustainability 1 (7): 369-374.

GBH. 2015. Argentine Voters Focused On Economy In Presidential Election. GBH. 24 October
2015. Accessed on 15 August 2023 at
https://www.wgbh.org/news/2015-10-24/argentine-voters-focused-on-economy-in-presidential-e
lection.

Gelpern, Anna, Sebastian Horn, Scott Morris, Brad Parks, Christoph Trebesch. 2021. How China
Lends: A Rare Look into 100 Debt Contracts with Foreign Governments. Peterson Institute for
International Economics, Kiel Institute for the World Economy, Center for Global Development,
and AidData at William & Mary.

Gelpern, Anna, Sebastian Horn, Scott Morris, Brad Parks, Christoph Trebesch. 2022. How China
Lends: A Rare Look into 100 Debt Contracts with Foreign Governments. Economic Policy
eiac054.

Gelpern, Anna. 2016. Sovereign debt: now what? Yale Journal of International Law 41 (2): 45-96.

Geo News. 2018. PTI Govt Decides to Investigate Mass-Transit Projects. Geo News. 24 August
2018. Accessed on 15 August 2023 at
https://www.geo.tv/latest/208466-pti-govt-decides-to-investigate-mass-transit-projects.

Ghani, Amel. 2015. Divisive Project: LDA Hopes to Secure Archaeology NOC in a Week. The
Express Tribune. 4 November 2015. Accessed on 15 August 2023 at
https://tribune.com.pk/story/985371/divisive-project-lda-hopes-to-secure-archaeology-noc-in-a-
week.

Gilchrist, Simon, Bin Wei, Vivian Z. Yue, and Egon Zakrajšek. 2022. Sovereign risk and financial
risk. Journal of International Economics 136 (C): 103603.

248
Giusto, Patricio, and Juan Manuel Harán. 2023. Argentina’s Failed China Policy. The Diplomat.
28 February 2023. Accessed on 15 August 2023 at
https://thediplomat.com/2023/02/argentinas-failed-china-policy/

Goldsmith, Benjamin E., Yusaku Horiuchi, and Terence Wood. 2014. Doing Well by Doing Good:
The Impact of Foreign Aid on Foreign Public Opinion. Quarterly Journal of Political Science 9 (1):
87–114.

Goldsmith, Benjamin E. and Yusaku Horiuchi. 2012. In Search of Soft Power: Does Foreign Public
Opinion Matter for US Foreign Policy? World Politics 64 (3): 555–585.

Gong, Xue. 2023. The Belt and Road Initiative Is Still China’s ‘Gala’ but Without as Much Luster.
Carnegie China. Washington D.C.: Carnegie Endowment for International Peace.
https://carnegieendowment.org/2023/03/03/belt-and-road-initiative-is-still-china-s-gala-but-with
out-as-much-luster-pub-89207.

Goodman, Seth, Sheng Zhang, Ammar Malik, Bradley C. Parks, and Jacob Hall. 2023. AidData's
Geospatial Global Chinese Development Finance Dataset. Mimeo.

Gormley, Todd A. 2010. The Impact of Foreign Bank Entry in Emerging Markets: Evidence from
India. Journal of Financial Intermediation 19 (1): 26–51.

Government of Costa Rica. 2007. Establecimiento de relaciones diplomáticas con la República

Popular China ha resultado beneficioso para el país. 10 September 2007. San Jose: Ministerio
de Relaciones Exteriores y Culto Dirección de Comunicación Institucional. Accessed on 2 August
2023 at
https://www.dropbox.com/scl/fi/g8kkmicps7o8z9vdxcn5d/1243-RPCH-beneficios.pdf?rlkey=3u5
afpnqeeg6myy358wxj2kn9&dl=0

Government of Costa Rica. 2008. Letter from Costa Rican Minister of Finance Guillermo E.
Zúñiga to SAFE Deputy Administrator Fang Shangpu. 7 January 2008. Accessed on 13 August
2023 at
https://www.dropbox.com/scl/fi/48qdeslzwim9t6dqhyk0m/7-January-2008-Letter-from-Minister-o
f-Finance-of-Costa-Rica-to-SAFE.pdf?rlkey=pr3p4advi1wfzzwjr5t8gva5m&dl=0

Guiso, Luigi, Paola Sapienza, and Luigi Zingales. 2009. Cultural Biases in Economic Exchange?
Quarterly Journal of Economics 124 (3): 1095–1131.

Gurara, Daniel, Andrea Presbitero, and Miguel Sarmiento. 2020. Borrowing costs and the role of
multilateral development banks: Evidence from cross-border bank syndicated lending. Journal of
International Money and Finance 100: 102090.

Hainz, C. and S. Kleimeier. 2012. Political Risk, Project Finance, and the Participation of
Development Banks in Syndicated Lending. Journal of Financial Intermediation 21 (2): 287–314.

Halper, Stefan. 2010. The Beijing Consensus: How China’s Authoritarian Model Will Dominate
the Twenty-First Century. Philadelphia: Basic Books.

Hancock, Tom and Matthew Hill. 2022. Debt Defaults Are a Stress Test for China’s Soft Power
Strategy. Bloomberg Businessweek. 26 September 2022. Accessed on 1 October 2023 at
https://www.bloomberg.com/news/articles/2022-09-26/debt-crisis-puts-scrutiny-on-china-soft-po
wer#xj4y7vzkg

Hasnain, Khalid. 2023. Orange Line Touches 100m Passengers Count. Dawn. 6 August 2023.
Accessed on 15 August 2023 at https://www.dawn.com/news/1768572.

249
Heurlin, Christopher. 2020. Authoritarian Aid and Regime Durability: Soviet Aid to the
Developing World and Donor–Recipient Institutional Complementarity and Capacity.
International Studies Quarterly 64 (4): 968–79.

Hillman, Jonathan E. 2020. A ‘China Model?’ Beijing’s Promotion of Alternative Global Norms
and Standards. 13 March 2020. Congressional Testimony. Washington DC: CSIS. Accessed on 1
October 2023 at
https://www.csis.org/analysis/china-model-beijings-promotion-alternative-global-norms-and-stan
dards.

Hillman, Jonathan E. 2020. Don’t Let China Greenwash Its Belt and Road. Nikkei Asia. 6 October
2020. Accessed on 22 August 2023 at
https://www.csis.org/analysis/dont-let-china-greenwash-its-belt-and-road

Hirschman, Albert O. 1958. The Strategy of Economic Development. New Haven, CT: Yale
University Press.

Hoffman, Constantine von. 2014. How Argentina Wound up in Another Default Mess. CBS News.
31 July 2014. Accessed on 15 August 2023 at
https://www.cbsnews.com/news/q-a-the-argentina-bond-default-explained/.

Holslag, Jonathan. 2011. China and the Coups: Coping with Political Instability in Africa. African
Affairs 110 (440): 367–386.

Hong Kong Green Finance Association (HKGFA) and Guangdong Green Finance Committee.
2022. CGT Research Series Phase 2: Understanding Use Cases of the Common Ground
Taxonomy. Accessed on 1 October 2023 at
https://www.hkgreenfinance.org/wp-content/uploads/2022/09/CGT_Phase2reportENG_final.pdf

Horigoshi, Ana, Samantha Custer, Bryan Burgess, Kelsey Marshall, Vera Choo, Katie
Andrzejewski, and Emily Dumont. 2022. Delivering the Belt and Road: Decoding the supply of
and demand for Chinese overseas development projects. Williamsburg, VA: AidData at William
& Mary.

Horn, Sebastian, Bradley C. Parks, Carmen M. Reinhart, and Christoph Trebesch. 2023a. China as
an International Lender of Last Resort. NBER Working Paper #31105. Cambridge, MA: NBER.

Horn, Sebastian, Bradley C. Parks, Carmen M. Reinhart, and Christoph Trebesch. 2023b. Debt
Distress on China’s Belt and Road. American Economic Association Papers & Proceedings 113:
131-34.

Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2019. China’s Overseas Lending.
NBER Working Paper No. 26050. Cambridge, MA: National Bureau of Economic Research.

Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2020a. China’s Overseas
Lending: A Response to Our Critics. CGD Note. Washington DC: Center for Global
Development (CGD). Accessed on 19 July 2023 at
https://www.cgdev.org/sites/default/files/chinas-overseas-lending-response-our-critics.pdf

Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2020b. Coping with Disasters:
Two Centuries of International Official Lending. NBER Working Paper No. 27343.

Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2021. China’s Overseas Lending.
Journal of International Economics 133: 103539.

250
Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2022a. Hidden Defaults. World
Bank Policy Research Working Paper Series 9925. Washington DC: World Bank.

Horn, Sebastian, Carmen M. Reinhart, and Christoph Trebesch. 2022b. Hidden Defaults. AEA
Papers and Proceedings 112: 531–35.

Human Rights Watch. 2021. Bangladesh: Events of 2021. In World Report 2022. New York, NY:
Human Rights Watch. Accessed on 15 August 2023 at
https://www.hrw.org/world-report/2022/country-chapters/bangladesh.

Humphrey, Chris, and Katharina Michaelowa. 2019. China in Africa: Competition for Traditional
Development Finance Institutions? World Development 120:15–28.

Humphrey, Christopher. 2015. Infrastructure Finance in the Developing World: Challenges and
Opportunities for Multilateral Development Bank. Washington DC: G-24.

Hwang, Jyhjong,, Oyintarelado Moses, Lucas Engel, and Sobia Shadbar. 2022. Chinese Loans to
Africa During the COVID-19 Pandemic. BU Global Development Policy Center. GCI Policy Brief
012. Accessed on 17 July 2023 at
https://www.bu.edu/gdp/files/2022/11/GCI_PB_012_EN_FIN.pdf

Iacoella, Francesco, Bruno Martorano, Laura Metzger, and Marco Sanfilippo. 2021. Chinese
Official Finance and Political Participation in Africa. European Economic Review 136: 103741.

Independent Evaluation Group of the World Bank (IEG). 2010. Safeguards and Sustainability
Policies in a Changing World. Washington DC: World Bank.

International Monetary Fund (IMF). 2020. IMF Policy Paper: The Evolution of Public Debt
Vulnerabilities in Lower Income Economies. Washington DC: IMF.

International Monetary Fund (IMF). 2021. Request for an Extended Arrangement Under the
Extended Fund Facility -- Press Release; Staff Report; Staff Statement; and Statement by the
Executive Director for Suriname. IMF Country Report No. 21/280. Washington DC: IMF.
Accessed at
https://www.imf.org/-/media/Files/Publications/CR/2021/English/1SUREA2021001.ashx

International Monetary Fund (IMF). 2022. First Review Under the Extended Arrangement under
the Extended Fund Facility, and Financing Assurances Review--Press Releease; Staff Report; Staff
Statement; and Statement by the Executive Director for Suriname. IMF Country Report No.
22/90. Washington DC: IMF. Accessed on 3 August 2023 at
https://www.imf.org/en/Publications/CR/Issues/2022/03/25/Suriname-First-Review-under-the-Ext
ended-Arrangement-under-the-Extended-Fund-Facility-and-515732

International Monetary Fund. 2017. Zambia: Staff Report on the 2017 Article IV Consultation.
Debt Sustainability Analysis. Washington D.C.: International Monetary Fund.
https://www.imf.org/en/Publications/Search#sort=%40imfdate%20descending&numberOfResult
s=50&f:country=[Zambia]&f:series=[COUNTRYREPS,DSAS]:~:text=Zambia%3A%20Debt%20Sust
ainability%20Analysis%3B%20IMF%20Country%20Report%20No.%2017/327%3B%20Septembe
r%2025%2C%202017.

International Monetary Fund. 2023. International Financial Statistics. IMF Data.

https://data.imf.org/?sk=4c514d48-b6ba-49ed-8ab9-52b0c1a0179b.

International Platform on Sustainable Finance (IPSF). 2022. Common Ground Taxonomy –

Climate Change Mitigation. Instruction Report of the IPSF Taxonomy Working Group Co-chaired

251
by the EU and China. 3 June 2022. Accessed on 1 October 2023 at
https://finance.ec.europa.eu/system/files/2022-06/220603-international-platform-sustainable-fin
ance-common-ground-taxonomy-instruction-report_en.pdf

Isaksson, Ann-Sofie and Andreas Kotsadam. 2018a. Chinese Aid and Local Corruption. Journal
of Public Economics 159: 146–159.

Isaksson, Ann-Sofie, and Andreas Kotsadam. 2018b. Racing to the Bottom? Chinese
Development Projects and Trade Union Involvement in Africa. World Development 106 (C):
284–298.

Isaksson, Ann-Sofie. 2020. Chinese Aid and Local Ethnic Identification. International
Organization 74 (4): 833–852.

Islam, Md. Safiqul, Huang Ailian, and Zhang Jie. 2018. Major Challenges and Remedies in
Building the Bangladesh-China-India-Myanmar Economic Corridor. China Quarterly of
International Strategic Studies 04 (04): 613–29.

Islam, Syful. 2021. Bangladesh Hits Back after China Envoy Warns against Joining Quad. Nikkei
Asia. 11 May 2021. Accessed on 15 August 2023 at
https://asia.nikkei.com/Politics/International-relations/Bangladesh-hits-back-after-China-envoy-w
arns-against-joining-Quad.

Ivashina, Victoria. 2009. Asymmetric information effects on loan spreads. Journal of Financial
Economics 92 (2): 300–319.

Jansson, Johanna. 2013. The Sicomines Agreement Revisited: Prudent Chinese Banks and
Risk-Taking Chinese Companies. Review of African Political Economy 40 (135): 152–62.

Jia, Chen. 2015. Policy banks get $90b cash infusion. 19 August 2015. China Daily. Accessed on
10 July 2023 at https://www.chinadaily.com.cn/business/2015-08/19/content_21643373.htm

Jinping, Xi. 2017. Full Text of President Xi’s Speech at Opening of Belt and Road Forum. Speech
presented at the The Belt and Road Forum for International Cooperation, Beijing, May 14.
https://www.fmprc.gov.cn/mfa_eng/wjdt_665385/zyjh_665391/201705/t20170527_678618.html.

Jones, Catherine. 2018. China’s Challenge to Liberal Norms: The Durability of International
Order. London: Palgrave Macmillan UK.

Jones, Lee. 2020. China’s Belt and Road Initiative Is a Mess, Not a Master Plan. Foreign Policy
(blog). October 9, 2020.
https://foreignpolicy.com/2020/10/09/china-belt-and-road-initiative-mess-not-master-plan/.

Joosse, Alexandra, Ammar A. Malik, Sheng Zhang, and Thai-Binh Elston. 2023. Networks of the
Belt & Road: The Hidden Role of Financial Brokers. Mimeo.

Justice Ejaz Afzal Khan. 2017. NESPAK, Punjab Mass Transit Authority, Lahore Development
Authority, Province of Punjab and Civil Society Network Versus Kamil Khan Mumtaz & others and
Province of Punjab. Supreme Court of Pakistan.
https://www.supremecourt.gov.pk/online-case-status/

Kaminsky, Graciela and Carmen M. Reinhart. 2000. On Crises, Contagion and Confusion. Journal
of International Economics 51 (1): 145-168.

252
Kastner, Scott A. 2016. Buying Influence? Assessing the Political Effects of China’s International
Trade. Journal of Conflict Resolution 60 (6): 980– 1007.

Kenny, Charles. 2023. World Bank Investment Projects Aren’t Designed for Crises. 15 March
2023. Washington DC: Center for Global Development. Accessed on 27 August 2023 at
https://www.cgdev.org/blog/world-bank-investment-projects-arent-designed-crises

Kern, Andreas, Bernhard Reinsberg, and Patrick Shea. 2022. IMF Programs, Chinese Lending,
and the Political Economy of Leader Survival. Working Paper #118. Williamsburg, VA: AidData at
William & Mary.

Khan, Yasir. 2018. Metro Train Lahore, a Gift from China. China Today. Accessed on 28
September 2023.
http://www.chinatoday.com.cn/ctenglish/2018/ii/202011/t20201105_800226035.html.

Kim, Jim Yong. 2017. Remarks of World Bank Group President Jim Yong Kim at the Belt and
Road Forum for International Cooperation on 14 May 2017. Accessed at 15 August 2023 at
https://www.worldbank.org/en/news/speech/2017/05/14/remarks-of-world-bank-group-presiden
t-jim-yong-kim

Kimenyi, Mwangi S., and Amy Copley. 2014. The Death of President Michael Sata and Issues of
the Health of African Leaders. Brookings (blog). 30 October 2014. Accessed at 15 August 2023
at
https://www.brookings.edu/articles/the-death-of-president-michael-sata-and-issues-of-the-health
-of-african-leaders/.

Kishida, Fumio. 2023. The Future of the Indo-Pacific: Japan’s New Plan for a Free and Open
Indo-Pacific. Ministry of Foreign Affairs of Japan. https://www.mofa.go.jp/files/100477791.pdf.

Kroenig, Matthew, Melissa McAdam, and Steven Weber. 2010. Taking Soft Power Seriously.
Comparative Strategy 29 (5): 412-431.

Kondo, Illenin O., Astghik Mkhitaryan, and Cesar Sosa-Padilla. 2022. Borrowing in the Shadow of
China. Mimeo.

Kong, Bo, and Kevin P. Gallagher. 2017. Globalizing Chinese Energy Finance: The Role of Policy
Banks. Journal of Contemporary China 26 (108): 834–851.

Koop, Fermín. 2015. New Argentina President to Put Deals with China under the Microscope.
Dialogo Chino. 27 November 2015. Accessed on 15 August 2023 at
https://dialogochino.net/en/trade-investment/4292-new-argentina-president-to-put-deals-with-c
hina-under-the-microscope/.

Koop, Fermín. 2016. Argentina Confirms Santa Cruz Dam Projects. Dialogo Chino (blog). 15
April 2016. Accessed on 15 August 2023 at
https://dialogochino.net/en/climate-energy/5888-argentina-confirms-santa-cruz-dam-projects/.

Koop, Fermín. 2018. Macri Extends Hand to China as Trade Deficit Grows. Buenos Aires Times. 5
May 2018. Accessed on 15 August 2023 at
https://www.batimes.com.ar/news/economy/macri-extends-hand-to-china-as-trade-deficit-grows.
phtml.

Koop, Fermín. 2019. New Argentina President Looks Set to Deepen China Ties. Dialogo Chino.
31 October 2019. Accessed on 15 August 2023 at

253
https://dialogochino.net/en/trade-investment/31387-new-argentina-president-looks-set-to-deep
en-china-ties/.

Kose, M. Ayhan, Sergio Kurlat, Franziska Ohnsorge, and Naotaka Sugawara. 2022. A
Cross-Country Database of Fiscal Space. Journal of International Money and Finance 128
(November): 102682.

Kucik, Jeffrey. 2017. Trump’s Trade War Begins. Foreign Affairs, October 10, 2017.
https://www.foreignaffairs.com/articles/united-states/2017-10-10/trumps-trade-war-begins.

Kuo, Mercy A. 2021. China’s BRI Lending: $385 Billion in ‘Hidden Debts’. The Diplomat. 29
November 2021. Acccesed on 27 August 2023 at
https://thediplomat.com/2021/11/chinas-bri-lending-385-billion-in-hidden-debts/

Kuziemko, Ilyana, and Eric Werker. 2006. How Much Is a Seat on the Security Council Worth?
Foreign Aid and Bribery at the United Nations. Journal of Political Economy 114 (5): 905–930.

Lagarda, Guillermo, Jennifer Linares, Arnoldo Lopez Marmolejo and Osmel Manzano. 2018.
Efficiency Before Effectiveness: The Case of the Inter-American Development Bank. GEGI
Working Paper 022. Boston, MA: Boston University Global Development Policy Center.

Lagarde, Christine. 2019. BRI 2.0: Stronger Frameworks in the New Phase of Belt and Road.
Speech delivered at Belt and Road Forum on 26 April 2019. Accessed on 1 August 2023 at
https://www.imf.org/en/News/Articles/2019/04/25/sp042619-stronger-frameworks-in-the-new-ph
ase-of-belt-and-road

Landers, Clemence, Nancy Lee and Scott Morris. 2021. Why Does DFC Want to Pay Off
Ecuador’s Chinese Creditors? 19 January 2021. Washington DC: Center for Global Development.
Accessed on 16 July 2023 at
https://www.cgdev.org/blog/why-does-dfc-want-pay-ecuadors-chinese-creditors

Laurance, William F., Sean Sloan, Lingfei Weng and Jeffrey A. Sayer. 2015. Estimating the
Environmental Costs of Africa’s Massive Development Corridors. Current Biology
25(24):3202–3208.

Lee, Ching Kwan. 2018. The Specter of Global China: Politics, Labor, and Foreign Investment in
Africa. Chicago, IL: University of Chicago Press.
https://press.uchicago.edu/ucp/books/book/chicago/S/bo22657847.html.

Lin, Justin Yifu and Yan Wang. 2014. China-Africa co-operation in structural transformation:
ideas, opportunities, and finances. UNU-WIDER Working Paper 2014/046. Helsinki: United
Nations University.

Liu, Zongyuan Zoe. 2023. Sovereign Funds: How the Communist Party of China Finances Its
Global Ambitions. Cambridge, MA: Harvard University Press.

Longstaff, Francis A., Pan, Jun, Pedersen, Lasse H. and Singleton, Kenneth J. 2011. How
Sovereign Is Sovereign Credit Risk? American Economic Journal: Macroeconomics 3 (2): 75-103.

Lu, Christina. 2023. China’s Belt and Road to Nowhere. Foreign Policy. 13 February 2023.
Accessed on 10 July 2023 at
https://foreignpolicy.com/2023/02/13/china-belt-and-road-initiative-infrastructure-development-
geopolitics/

254
Lu, Yangsiyu, Cecilia Springer, and Bjarne Steffen. 2023a. Cofinancing and Infrastructure Project
Outcomes in Chinese Lending and Overseas Development Finance. GCI Working Paper #27.
Boston, MA: Boston University Global Development Policy Center.

Lu, Yangsiyu, Xiao Yan Zhou, and B. Alexander Simmons. 2023b. Empirical analysis of Chinese
overseas power plant investments: Likelihood of suspensions and associated environmental risks.
iScience 26 (8): 107457.

Lucci, Juan. 2019. Is It Possible to Say No to China? The Case of the Kirchner-Cepernic Dams in
Argentine Patagonia. Leadership Academy for Development. Palo Alto, CA: Stanford University.

Ma, Guonan. 2006. Sharing China’s Bank Restructuring Bill. China & World Economy 14 (3):
19-37.

Majid, Hadia, Ammar Malik, and Kate Vyborny. 2018. Infrastructure Investments and Public
Transport Use: Evidence from Lahore, Pakistan. IGC Working Paper C-89231-PAK-1. London, UK:
IGC. Accessed on 1 October 2023 at
https://www.dropbox.com/s/2gbnajqo55d14jz/BRT_IGC_WP.pdf?dl=0.

Malik, Ammar A. and Bradley C. Parks. 2021. Hidden debt exposure to China: What is it, where
is it, and should we be concerned? Williamsburg, VA: AidData at William & Mary.

Malik, Ammar A., Bradley C. Parks, Brooke Russell, Joyce Lin, Katherine Walsh, Kyra Solomon,
Sheng Zhang, Thai-Binh Elston, and Seth Goodman. 2021. Banking on the Belt and Road:
Insights from a New Global Dataset of 13,427 Chinese Development Projects. Williamsburg, VA:
AidData at William & Mary.

Mandan, Tanvi. 2020. Emerging Global Issues: The China-India Boundary Crisis and Its
Implications. Washington D.C.: U.S. Brookings Institute. Accessed on 18 August 2023 via
https://www.brookings.edu/articles/emerging-global-issues-the-china-india-boundary-crisis-and-i
ts-implications/.

Mandon, Pierre and Martha Tesfaye Woldemichael. 2023. Has Chinese Aid Benefited Recipient
Countries? Evidence from a Meta-regression Analysis. World Development 166: 106211.

Mannan, Md. Abdul. 2018. Islamo-Nationalism, Domestic Politics, and Bangladesh’s Policy of
Balancing against India since the 1990s. International Area Studies Review 21 (4): 340–64.

Martin, C. S., M. J. Tolley, E. Farmer, C. J. Mcowen, J. L. Geffert, J. P. W. Scharlemann, H. L.

Thomas, J. H. van Bochove, D. Stanwell-Smith, J. M. Hutton, B. Lascelles, J. D. Pilgrim, J. M. M.
Ekstrom, D. P. Tittensor. 2015. A global map to aid the identification and screening of critical
habitat for marine industries. Marine Policy 53 (1): 45-53.

Martorano, Bruno, Laura Metzger, and Marco Sanfilippo. 2020. Chinese development assistance
and household welfare in Sub-Saharan Africa. World Development 129 (C): 104909.

Marty, Robert, Seth Goodman, Michael Le Few, Carrie B. Dolan, Ariel BenYishay, and Dan
Runfola. 2019. Assessing the Causal Impact of Chinese Aid on Vegetative Land Cover in Burundi
and Rwanda Under Conditions of Spatial Imprecision. Development Engineering 4: 100038.

Marx, Benjamin. 2018. Elections as Incentives: Project Completion and Visibility in African
Politics. Mimeo. Accessed on 1 October 2023 at
https://drive.google.com/file/d/1aSDEhq-ZGlUcX5QxWbb-0bAcAdSUXlKc/view

255
Mertha, Andrew. 2009. ‘Fragmented Authoritarianism 2.0’: Political Pluralization in the Chinese
Policy Process. The China Quarterly 200 (December): 995–1012.

Mfula, Chris. 2011. Interview - Zambia Opposition Leader Says to Keep China Ties. Reuters. 16
September 2011. Accessed on 1 October 2023 at
https://www.reuters.com/article/zambia-opposition-idAFL5E7KG14Y20110916.

Mfula, Chris. IMF's Georgieva sees broad deal with China on debt restructurings. Reuters. 24
January 2023. Accessed on 30 July 2023 at
https://www.reuters.com/world/africa/imf-chief-says-its-time-creditors-restructure-zambias-debt-2
023-01-24/

Miller, Matthew. 2023. U.S. Department of State Press Briefing by Department Spokesperson on
24 May 2023. Washington DC: U.S. Department of State. Accessed on 15 September 2023 at
https://www.state.gov/briefings/department-press-briefing-may-24-2023/

Mingey, Matthew and Agatha Kratz. 2021. China’s Belt and Road: Down but not Out. Rhodium
Group Note. Accessed on 21 July 2023 at https://rhg.com/research/bri-down-out/

Mingey, Matthew and Logan Wright. 2023 China’s External Debt Renegotiations After Zambia.
Rhodium Group Note. Accessed on 1 August 2023 at
https://rhg.com/research/chinas-external-debt-renegotiations-after-zambia/

Ministry of Commerce (MOFCOM) and Ministry of Ecology and Environment (MEE) of the
People's Republic of China. 2021. Guidlines for Green Development in Foreign Investment and
Cooperation. 15 July 2021. Beijing: MOFCOM and MEE.Accessed on 27 August 2023 at
https://www.followingthemoney.org/wp-content/uploads/2022/07/2021_MOFCOM-MEE_Guidel
ines-on-Green-Development-in-Foreign-Investment-and-Cooperation_E.pdf

Ministry of Commerce of the People's Republic of China (MOFCOM). 2014. MOFCOM Held a
Media Debriefing on the Measures for the Administration of Foreign Aid. Beijing, China: Ministry
of Commerce. Accessed on 1 August 2023 at
www.mofcom.gov.cn/article/fbhfn/fbh2014/201412/2014120082 4824.shtml.

Ministry of Finance and Economic Development (MOFED). 2022. Annual Public Debt Bulletin
2021 Financial Year. Harare, Zimbabwe: Ministry of Finance and Economic Development
(MOFED). Accessed on 20 July 2023 at
https://veritaszim.net/sites/veritas_d/files/2021%20Annual%20Public%20Debt%20Bulletin.pdf

Ministry of Finance of the People's Republic of China. 2020. Written Interview with Finance
Minister Liu Kun on G20 Debt Agenda. Beijing: Ministry of Finance of the People's Republic of
China. Accessed on 1 August 2023
atwww.mof.gov.cn/en/news/spe/202011/t20201120_3626593.htm

Ministry of Foreign Affairs of the People's Republic of China. 2019. Working Together to Deliver
a Brighter Future For Belt and Road Cooperation. Keynote Speech by H.E. Xi Jinping President
of the People's Republic of China At the Opening Ceremony of the Second Belt and Road
Forum for International Cooperation. 26 April 2019.Accessed on 14 August 2023 at
https://www.fmprc.gov.cn/mfa_eng/wjdt_665385/zyjh_665391/201904/t20190426_678729.html

Ministry of Foreign Affairs of the People’s Republic of China. 2014. Xi Jinping Holds Talks with
President Cristina Fernández de Kirchner of Argentina and Two Heads of State Announce the
Establishment of China-Argentina Comprehensive Strategic Partnership. Ministry of Foreign
Affairs of the People’s Republic of China. July 19, 2014.

256
https://www.fmprc.gov.cn/eng/gjhdq_665435/3447_665449/3453_664968/3455_664972/20140
7/t20140722_594264.html.

Ministry of Foreign Affairs of the People’s Republic of China. 2016. Xi Jinping Holds Talks with
Prime Minister Sheikh Hasina of Bangladesh Both Sides Agree to Upgrade China-Bangladesh
Relations to Strategic Partnership of Cooperation and Continuously Move Forward Bilateral
Relationship at Higher Levels, October 14, 2016.
https://www.fmprc.gov.cn/mfa_eng/topics_665678/2016zt/XJPDJPZMJLGJXGSFWBCXZYDGAJ
XDJZGJLDRDBCHW/201610/t20161018_704267.html.

Ministry of Foreign Affairs. 2023. Indo-Pacific Outlook of Bangladesh. Ministry of Foreign Affairs,
Dhaka. https://mofa.gov.bd/site/press_release/d8d7189a-7695-4ff5-9e2b-903fe0070ec9.

Moody's Investor Service. 2022. China Green Bond Principles will support the domestic
sustainable finance market’s development and align with international standards. 30 August
2022. Accessed on 1 October 2023 at
https://www.moodys.com/web/en/us/capabilities/esg/resources/insights/china-green-bond-princ
iples-will-support-the-domestic-sustainable-finance-markets-development-and-align-with-interna
tional-standards.html

Morris, Scott, Brad Parks, and Alysha Gardner. 2020. Chinese and World Bank Lending Terms: A
Systematic Comparison across 157 Countries and 15 Years. CGD Policy Paper 170. Washington,
DC: Center for Global Development.

Moses, Oyintarelado Jyhjong Hwang, Lucas Engel, and Victoria Yvonne Bien-Aimé. 2023. A New
State of Lending: Chinese Loans to Africa. GCI Policy Brief 019. Boston, MA: Boston University's
Global Development Policy Center.

Mupuchi, Speedwell. 2006. Zambia: Sata Is Misleading the Nation Over Taiwan - KK. The Post.
29 August 2006. Accessed on 1 October 2023 at
https://allafrica.com/stories/200608290697.html.

Mutati, Felix. 2016. Preparing for the Worst. Africa Confidential 57 (25). 16 December 2016.
Accessed on 20 September 2023 at
https://www.africa-confidential.com/article-preview/id/11858/Preparing_for_the_worst.

Myers, Margaret and Rebecca Ray. 2023. At a Crossroads: Chinese Development Finance to
Latin America and the Caribbean, 2022. InterAmerican Dialogue and BU Global Development
Policy Center. Accessed on 17 July 2023 at
https://www.thedialogue.org/wp-content/uploads/2023/03/Chinese-Development-Finance-2023
-FINAL.pdf

Narain D., Maron M., Teo H.C., Hussey K., Lechner A.M. 2020. Best-practice biodiversity
safeguards for Belt and Road Initiative’s financiers. Nature Sustainability 3: 650–657.

Narain D., Teo H.C., Lechner A.M., Watson J.E., Maron M. 2022. Biodiversity risks and
safeguards of China’s hydropower financing in Belt and Road Initiative (BRI) countries. One Earth
5: 1019–1029.

Nasrin, Sharmina. 2001. China-Bangladesh Relations: Need for Economic Partnership. Journal of
International Affairs 5 (3).

National Bank of Serbia (NBS). 2007. Financial Statements prepared in accordance with
International Financial Reporting Standards for the year ended 31 December 2006 with
Independent Auditor's Report. Belgrade, Serbia: NBS. Accessed on 1 July 2023 at

257
https://nbs.rs/export/sites/NBS_site/documents-eng/onbs-poslovanje/NBS_IAS_Report_2006.pd
f

National Development and Reform Commission, Ministry of Foreign Affairs, Ministry of Ecology
and Environment, and Ministry of Commerce of the People’s Republic of China. 2022. Opinions
on Jointly Promoting Green Development of the Belt and Road. 16 March 2022. Accessed on 25
August 2023 at https://www.ndrc.gov.cn/xxgk/zcfb/tz/202203/t20220328_1320629_ext.html

National Economic and Development Authority (NEDA). 2017. Official Development Assistance
Terms and Conditions of Loan Assistance (as of 30 June 2017). Manila, NEDA. Accessed on 13
June 2023 at
https://web.archive.org/web/20210627183145/https://www.neda.gov.ph/wp-content/uploads/2
017/07/Q2-2017-ODA-Terms-and-Conditions-Loans-Bilateral-Institutions.pdf

Naughton, Barry, and Kellee S. Tsai. 2015. State Capitalism, Institutional Adaptation, and the
Chinese Miracle. Cambridge, UK: Cambridge University Press.

Ncube, Sipilisiwe. 2019. China’s $22m Debt Cancellation a Mockery - HH. Zambia: News
Diggers! (blog). 3 July 2019. Accessed on 1 October 2023 at
https://diggers.news/local/2019/07/03/chinas-22m-debt-cancellation-a-mockery-hh/.

NESPAK. 2015a. Lahore Orange Line Metro Rail Project PC-1. Lahore, Pakistan: Punjab
Masstransit Authority. Accessed on 1 October 2023 at
https://pma.punjab.gov.pk/system/files/pcI.pdf.

NESPAK. 2015b. Environmental Impact Assessment (EIA) Lahore Orange Line Metro Train
Project. Lahore, Pakistan: Punjab Masstransit Authority. Accessed on 1 October 2023 at
https://epd.punjab.gov.pk/system/files/7%2901%20Tital%2C%20Table%20of%20Contant%2C%
20Abbreviations%2C%20Executive%20%20Summary.pdf.

North, Douglass C. 1991. Institutions. Journal of Economic Perspectives 5 (1): 97–112.

Nugent, Clara. 2023a. Argentina reaches last-minute deal with IMF to avoid default. Financial
Times. 28 July 2023. Accessed on 2 August 2023 at
https://www.ft.com/content/42c509a8-d75e-4350-b448-b6b8db13942a

Nugent, Ciara. 2023b. Argentina taps Chinese credit line to keep economy afloat during
elections. Financial Times. 18 October 2023. Accessed on 19 October 2023 at
https://www.ft.com/content/c92e56b7-e619-4cbb-b615-e401140c10cc

Nye, Joseph S. 2004. Soft Power: The Means To Success In World Politics. New York, NY:
PublicAffairs.

Obe, Stephen Chan, and Chrisopher Vandome. 2022. Zambia: The First Year of a Hichilema
Foreign Policy | Chatham House – International Affairs Think Tank. London: Chatham House.
Accessed on 1 October 2023 at
https://www.chathamhouse.org/2022/08/zambia-first-year-hichilema-foreign-policy.

Organisation for Economic Co-operation and Development (OECD). 2023a. Development

Assistance Committee: Total official flows. Accessed on Aug 10, 2023 via stats.oecd.org.

Organisation for Economic Co-operation and Development (OECD). 2023b. Converged

Statistical Reporting Directives for the Creditor Reporting System (CRS) and the Annual DAC
Questionnaire, Chapter 1-6. 27 April 2023. Paris: OECD. Accessed on 14 September 2023 at
https://one.oecd.org/document/DCD/DAC/STAT(2023)9/FINAL/en/pdf

258
Office of the Leading Group for Promoting the Belt and Road Initiative. 2019. The Belt and Road
Initiative Progress, Contributions and Prospects. Beijing: Office of the Leading Group for
Promoting the Belt and Road Initiative. Accessed on 1 July 2023 at
https://eng.yidaiyilu.gov.cn/p/86739.html

Olander, Eric. 2023. What’s Driving the Steady Decline in Chinese Overseas Development
Lending? China Global South Project. Accesed on 01 July 223 at
https://chinaglobalsouth.com/podcasts/whats-driving-the-steady-decline-in-chinese-overseas-de
velopment-lending/

Olewe, Dickens. 2021. Why African Countries Back China on Human Rights. BBC News, 1 May
2021. Accessed on 1 October 2023 at https://www.bbc.com/news/world-africa-56717986.

Organisation for Economic Co-operation and Development (OECD). 2021. Converged Statistical
Reporting Directives for the Creditor Reporting System (CRS) and the Annual DAC
Questionnaire. Paris: OECD. Accessed on 1 August 2023 at
https://one.oecd.org/document/DCD/DAC/STAT(2020)44/FINAL/en/pdf

Paravicini, Giulia. 2020. Pompeo Takes Veiled Swipe at China on Final Leg of Africa Trip. Reuters.
19 February 2020. Accessed on 1 October 2023 at
https://www.reuters.com/article/us-usa-pompeo-ethiopia-idUSKBN20D1PB.

Park, S. 2010. The World Bank’s global safeguard policy norm? in Owning Development:
Creating Policy Norms in the IMF and the World Bank, edited by Susan Park and Antje
Vetterlein. Cambridge, UK: Cambridge University Press.

Parks, Bradley. 2019. Chinese Leadership and the Future of BRI: What Key Decisions Lie Ahead?
CGD Note. Washington, DC: Center for Global Development.

Parsons, Nick. 2010. CIC bonds: The good, the bad and the ugly. The Asset. 23 April 2010.
Accessed on 15 July 2023 at
https://theasset.com/article/17866/cic-bonds-the-good-the-bad-and-the-ugly

Patey, Luke. 2017. China Made Mauricio Macri a Deal He Couldn’t Refuse. Foreign Policy. 24
January 24 2017. Accessed on 1 October 2023 at
https://foreignpolicy.com/2017/01/24/china-made-mauricio-macri-a-deal-he-couldnt-refuse/.

Paul, Ruma. 2023. Bangladesh promises free, fair elections in response to US visa curbs. 25 May
2023 Reuters. Accessed on 28 September 2023 at
https://www.reuters.com/world/asia-pacific/bangladesh-says-will-take-steps-address-any-interfer
ence-in-elections-2023-05-25/

People’s Daily. 2014. Xi Eyes More Enabling Intl’ Environment for China’s Peaceful Development
(3). People’s Daily. 30 November 2014. Accessed on 1 August 2023 at http://en.people.cn/n/2
014/1130/c90883-8815967-3.html.

Perroux, Francois. 1950. Economic Space: Theory and Applications. Quarterly Journal of
Economics 64 (1): 89–104.

Pilling, David, and Emily Feng. 2018. Chinese Investments in Africa Go Off the Rails. Financial
Times. 5 December 2018. Accessed on 1 August 2023 at
https://www.ft.com/content/82e77d8a-e716-11e8-8a85- 04b8afea6ea3.

Pogliani, Pamela and Philip Wooldridge. 2022. Cross-border financial centres. BIS Working Paper
1035. Basel, Switzerland: Bank for International Settlements (BIS).

259
Qing, Shi. 2023. Dollar Hegemony is the true cause of Africa’s debt problems, not China. 19
April 2023. Global Times. Accessed on 1 October 2023 at
https://www.globaltimes.cn/page/202304/1289429.shtml

Ravallion, Martin. 2009. Are there Lessons for Africa from China’s Success Against Poverty?
World Development 37 (2): 303-313.

Ray, Rebecca. 2023. Small is Beautiful: A New Era in China’s Overseas Development Finance?.
GCI Policy Brief 017. BU Global Development Policy Center. Accessed on 17 December 2023 at
https://www.bu.edu/gdp/files/2023/01/GCI_PB_017_CODF_EN_FIN.pdf

Reinhart, Carmen M. and Christoph Trebesch. 2016. Sovereign debt relief and its aftermath.
Journal of the European Economic Association 14 (1): 215–251.

Rendon, Moises, and Sarah Baumunk. 2018. When Investment Hurts: Chinese Influence in
Venezuela. Washington DC: CSIS. Accessed on 1 October 2023 at
https://www.csis.org/analysis/when-investment-hurts-chinese-influence-venezuela.

Repnikova, Maria. 2022. Chinese Soft Power. Cambridge, UK: Cambridge University Press.
https://www.cambridge.org/core/elements/chinese-soft-power/4E9C9E445C5F8A9F81F5AAEEA
D2DF21D.

Reporters Without Borders. 2023. World Press Freedom Index. https://rsf.org/en/index.

République du Congo. 2018. Annexe Opérationnelle N°2, Document Cadre de Politiques et de

Programmation Macroéconomiques et Budgétaires (DPPMB) 2018-2022, Plan National de
Développement (PND) 2018-2022. Brazzaville: République du Congo. Accessed on 31 July 2023
at
https://web.archive.org/web/20200722130921/https://pnd.plan.gouv.cg/wp-content/uploads/20
18/07/Parlement-CMEB-24-07-18.pdf

Reuters. 2009. China Huijin to recapitalise Exim Bank soon-media. 12 November 2009. Reuters.
Accessed on 15 July 2023 at
https://www.reuters.com/article/china-exim-huijin/china-huijin-to-recapitalise-exim-bank-soon-me
dia-idUSPEK13684620091112

Reuters. 2010a. SAFE intends to change policy-oriented foreign exchange loans to entrusted
banks to handle them--Media (外管局拟将政策性外汇贷款改为委托银行办理形式--媒体). 17 May
2010. Reuters. Accessed on 15 July 2023 https://www.reuters.com/article/idCNCHINA-230572

Reuters. 2010b. Huijin to invest 100 bln yuan in China banks -report. 19 August 2010. Reuters.
Accessed on 5 June 2023 at
https://www.reuters.com/article/china-economy-huijin/huijin-to-invest-100-bln-yuan-in-china-ban
ks-report-idINTOE67J02720100820

Reuters. 2016. China forex regulator buys $4.2 bln in stocks via new platform - media. 31 March
2016. Reuters. Accessed on 15 July 2023 at
https://www.reuters.com/article/china-markets-regulator-stocks/china-forex-regulator-buys-4-2-bl
n-in-stocks-via-new-platform-media-idUSL3N1731HH

Rey, Débora. 2022. UK Rebukes China for Supporting Argentina’s Falklands Claim. AP News. 7
February 2022. Accessed on 1 October 2023 at
https://apnews.com/article/europe-china-asia-beijing-xi-jinping-df1ca5eebfd3b988813f22bba0b
3a7f0.

260
Riaz, Ali. 2016. Bangladesh: A Political History since Independence. Bloomsbury Publishing.

Rolland, Nadège. 2017. China’s ‘Belt and Road Initiative’: Underwhelming or Game-Changer?
The Washington Quarterly 40 (1): 127–42.

Rose, Andrew K. 2016. Like Me, Buy Me: The Effect of Soft Power on Exports. Economics &
Politics 28 (2): 216–232.

Rose, Andrew K. 2019. Soft Power and Exports. Review of International Economics 27(5): 1573–
1590.

Rose, Sarah. 2018. Linking US Foreign Aid to UN Votes: What Are the Implications. CGD Note.
Washington DC: Center for Global Development. Accessed on 15 September 2023 at
https://www.cgdev.org/publication/linking-us-foreign-aid-un-votes-what-are-implications

Rosenstein-Rodan, Paul N. 1943. Problems of Industrialization of Eastern and South-Eastern

Europe. Economic Journal 53 (210/211): 202–211.

S&P Global. 2022. The Future of Copper: Will the Looming Supply Gap Short-Circuit the Energy
Transition?
https://cdn.ihsmarkit.com/www/pdf/0722/The-Future-of-Copper_Full-Report_14July2022.pdf.

Sahoo, Pravakar, and Abhirup Bhunia. 2014. BCIM Corridor a Game Changer for South Asian
Trade. East Asia Forum. 18 July 2014. Accessed on 1 October 2023 at
https://www.eastasiaforum.org/2014/07/18/bcim-corridor-a-game-changer-for-south-asian-trade/
.

Sanderson, Henry and Michael Forsythe. 2013. China’s Superbank: Debt, Oil and Influence - How
China Development Bank is Rewriting the Rules of Finance. New York, NY: Wiley.

Sanger, David E., and Mark Landler. 2021. Biden Tries to Rally G7 Nations to Counter China’s
Influence. The New York Times. 12 June 2021. Accessed on 1 October 2023 at
https://www.nytimes.com/2021/06/12/world/europe/biden-china-g7.html.

Schlegl, Matthias, Christoph Trebesch, and Mark L. J. Wright. 2019. The Seniority Structure of
Sovereign Debt. NBER Working Paper 25793. Cambridge, MA: National Bureau of Economic
Research (NBER).

Schumacher, Edward. 1981. Argentina and Soviet are no longer just business partners.. The New
York Times. 12 July 1981. Accessed on 1 October 2023 at
https://www.nytimes.com/1981/07/12/weekinreview/argentina-and-soviet-are-no-longer-just-busi
ness-partners.html.

Sen, Ashish Kumar. 2018. IMF Throws Argentina a $57 Billion Lifeline. Atlantic Council (blog).
September 27, 2018.
https://www.atlanticcouncil.org/blogs/new-atlanticist/imf-throws-argentina-a-57-billion-lifeline/.

Séri, Regina Stéphanie. 2021. How giant discoveries of natural resources impact sovereign debt
ratings in developing and emerging countries? Études et Documents n°10. Paris: CERDI.

Setser, Brad W. 2023a. Shadow reserves — how China hides trillions of dollars of hard currency.
The China Project. Accessed on 1 July 2023 at
https://thechinaproject.com/2023/06/29/shadow-reserves-how-china-hides-trillions-of-dollars-of-
hard-currency/

261
Setser, Brad W. 2023b. The Common Framework and Its Discontents. 26 March 2023. New York:
Council on Foreign Relations. Accessed on 1 August 2023 at
https://www.cfr.org/blog/common-framework-and-its-discontents

Setser, Brad W. 2023c. How to Hide Your Foreign Exchange Reserves—A User’s Guide. 29 June
2023. New York, NY: Council on Foreign Relations. Accessed on 1 September 2023 at
https://www.cfr.org/blog/how-hide-your-foreign-exchange-reserves-users-guide

Sevastopulo, Demetri and Gideon Long. 2021. U.S. Development Bank Strikes Deal to Help
Ecuador Pay China Loans. Financial Times. 14 January 2021. Accessed on 16 July 2023 at
https://www.ft.com/content/affcc432-03c4-459d-a6b8-922ca8346c14

Shaban, Abdur Rahman Alfa. 2016. Chinese President Has Confidence in My Leadership –
Zambia’s Edgar Lungu. Africanews. 6 September 2016. Accessed on 1 October 2023 at
https://www.africanews.com/2016/09/06/chinese-president-has-confidence-in-my-leadership-za
mbia-s-edgar-lungu/.

Shacinda, Shapi. 2006. China Puts Zambia Mine Investment on Hold over Taiwan. Reuters. 5
September 2006. Accessed on 1 October 2023 at
http://www.minesandcommunities.org/article.php?a=5019.

Shahid Karim. 2016. Kamil Khan Mumtaz etc Versus Government of Punjab etc. Lahore High
Court. https://sys.lhc.gov.pk/appjudgments/2016LHC2454.pdf

Shaukat, Aroosa, and Rana Tanveer. 2016. LHC Suspends Work on Orange Line Train Project.
The Express Tribune. 28 January 2016. Accessed on 1 October 2023 at
https://tribune.com.pk/story/1035831/lhc-suspends-work-on-orange-line-train-project.

Shambaugh, David. 2015. China’s Soft-Power Push. Foreign Affairs. 16 June 2015. Accessed on 1
October 2023 at
https://www.foreignaffairs.com/articles/china/2015-06-16/chinas-soft-power-push.

Shan, He. 2011. CIC injects 20 bln yuan into Sinosure. Accessed on 1 June 2023 at
www.china.org.cn/business/2011-11/17/content_23944979.htm

Shepherd, Christian, and Ben Blanchard. 2018. China’s Xi offers another $60 billion to Africa, but
says no to ‘vanity’ projects. Reuters. 3 September 2018. Accessed on 24 July 2023 at
https://www.reuters.com/article/china-africa/chinas-xi-says-funds-for-africa-not-forvanity-
projects-idUSL3N1VO018

Shi, Jiangtao. 2023. Sudden Rise of ‘Argentina’s Trump’ Puts Economic Ties with China at Risk.
South China Morning Post. 18 August 2023. Accessed on 1 October 2023 at
https://www.scmp.com/news/china/diplomacy/article/3231570/sudden-rise-argentinas-trump-jav
ier-milei-puts-economic-ties-china-risk.

Shortell, Paul. 2014. Will Default Dampen China-Argentina Ties? The Dialogue. 18 September
2014. Accessed on 1 October 2023 at
https://www.thedialogue.org/blogs/2014/09/will-default-dampen-china-argentina-ties/.

Shukla, Manish. 2021. How Anti-China Movement Launched by Late Zambian President Michael
Sata Is Influencing Zambian Elections. Zee News. 24 January 2021. Accessed on 1 October 2023
at
https://zeenews.india.com/world/how-anti-china-movement-launched-by-late-zambian-president
-michael-sata-is-influencing-zambian-elections-2337420.html.

262
Sims Gallagher, Kelly and Qi Qi. 2021. Climate Proofing China’s Belt and Road Initiative (BRI). 7
May 2021. Climate Policy Lab. Accessed on 25 August 2023 at
https://www.climatepolicylab.org/climatesmart/2021/3/4/climate-proofing-chinas-belt-and-road-i
nitiative-bri

Silver, Laura, Kat Devlin and Christine Huang. 2020. Unfavorable Views of China Reach Historic
Highs in Many Countries. 6 October 2020. Washington DC: Pew Research Center. Accessed on 1
September 2023 at
https://www.pewresearch.org/global/2020/10/06/unfavorable-views-of-china-reach-historic-highs
-in-many-countries/

Snow, Nancy. 2022. A Reliable Friend and Strategic Partner in the Indo-Pacific Region: Japan’s
Strategic Communications and Public Diplomacy. AidData Research Report.
https://docs.aiddata.org/reports/gf01/gf01-07/A-Reliable-Friend-and-Strategic-Partner-in-the-Ind
o-Pacific-Region-Japans-Strategic-Communications-and-Public-Diplomacy.html.

State Administration of Foreign Exchange (SAFE). 2004. Annual Report of the State
Administration of Foreign Exchange (2004). Beijing: SAFE. Accessed on 1 August 2023 at
https://www.safe.gov.cn/en/2004/0726/1272.html

State Administration of Foreign Exchange (SAFE). 2008. Letter from SAFE Deputy Administrator
Fang Shangpu to Costa Rican Minister of Finance Guillermo E. Zúñiga . 2 January 2008.
Accessed on 13 August 2023 at
https://www.dropbox.com/scl/fi/a20fdxb2lecowt8tp5lym/2-January-2008-SAFE-Letter-to-Ministe
r-of-Finance-of-Costa-Rica.pdf?rlkey=y9ywdrrhap4rdkt0cyqj4mj4y&dl=0

State Administration of Foreign Exchange (SAFE). 2017a. Pan Gongsheng: Foreign Exchange
Administration Supports the Belt and Road Initiative. Beijing: SAFE. Accessed on 1 July 2023 at
https://www.safe.gov.cn/en/2017/0505/1263.html

State Administration of Foreign Exchange (SAFE). 2017b. Annual Report of the State
Administration of Foreign Exchange 2017. Beijing: SAFE. Accessed on 1 July 2023 at
https://www.safe.gov.cn/en/file/file/20180822/2f5f632ccd964e02876b7455883e0449.pdf?n=An
nual%20Report%20of%20the%20State%20Administration%20of%20Foreign%20Exchange%20(2
017)

State Administration of Foreign Exchange (SAFE). 2021. State Administration of Foreign

Exchange 2021 Annual Report. Beijing: SAFE. Accessed on 1 August 2023 at
https://www.safe.gov.cn/en/file/file/20221226/9186b372d33e4a6b9d156ab154cc2979.pdf?n=A
nnual%20Report%20of%20the%20State%20Administration%20of%20Foreign%20Exchange%20(
2021)

State Council of the People's Republic of China. 2021. Party Secretary General Xi Jinping
attended the third 'Belt and Road' symposium. 21 November 2021. Accessed on 1 October
2023 at https://www.gov.cn/xinwen/2021-11/21/content_5652298.htm

Stott, Michael and Ciara Nugent. 2023. Argentina must rethink China relationship, says
presidential candidate Patricia Bullrich. Financial Times. 16 October 2023. Accessed on 16
October 2023 at https://www.ft.com/content/6aaa091d-8031-493a-9655-9b16bc7aee35

Strange, Austin. 2023. Symbols of State: Explaining Prestige Projects in the Global South.
Working Paper.

Strüver, Gregor. 2016. What Friends Are Made Of: Bilateral Linkages and Domestic Drivers of
Foreign Policy Alignment with China. Foreign Policy Analysis 12 (2): 170–91.

263
Strüver, Gregor. 2017. China's Partnership Diplomacy: International Alignment Based on Interests
or Ideology. The Chinese Journal of International Politics 10 (1): 31–65.

Sufi, Amir, 2007. Information asymmetry and financing arrangements: Evidence from syndicated
loans. The Journal of Finance 62: 629–668.

Sullivan, Mark P., and Rebecca M. Nelson. 2016. Argentina: Background and U.S. Relations.
Washington DC: Congressional Research Service. Accessed on 1 October 2023 at
https://www.hsdl.org/?view&did=791518.

Sullivan, Mark P., and Rebecca M. Nelson. 2017. Argentina: Background and U.S. Relations.
Washington DC: Congressional Research Service. Accessed on 1 October 2023 at
https://sgp.fas.org/crs/row/R43816.pdf.

Swedlund, Haley J. 2017. The Development Dance: How Donors and Recipients Negotiate the
Delivery of Foreign Aid. Ithaca, NY: Cornell University Press.

Tang, Keyi. 2021. Development Finance and Distributive Politics: Comparing Chinese and World
Bank Finance in sub-Saharan Africa. SAIS-CARI Working Paper No. 2021/45. Washington DC:
SAIS-CARI.

Tangjun, Song, Zhang Jianguang, and Wang Lin. 2016. Research on State-owned Assets
Management of Administrative Units (行政单位国有资产管理研究). Beijing: China Financial and
Economic Publishing House

Teixeira, João C.A., Francisco J.F. Silva, Manuel B.S. Ferreira, and José A.C. Vieira. 2018.
Sovereign credit rating determinants under financial crises. Global Finance Journal 36: 1-13.

The Controller and Auditor General of the National Audit Office of Tanzania. 2018. Report of the
Controller and Auditor General on the Consolidated and Separate Financial Statements of
Tanzania Petroleum Development Corporation for the Year Ended 30 June 2018. Dar es Salaam:
The Controller and Auditor General of the National Audit Office of Tanzania. Accessed at
https://www.dropbox.com/s/dyt43g53m2btrmz/TPDC%20Financial%20Statements%2030%20Ju
ne%202018.pdf?dl=0

The Economist. 2011. Trying to Pull Together. The Economist. 20 April 2011. Accessed on 1
October 2023 at https://www.economist.com/briefing/2011/04/20/trying-to-pull-together.

The Economist. 2018. Zambia Slumps toward Another Debt Crisis. The Economist. 13
September 2018. Accessed on 1 October 2023 at
https://www.economist.com/middle-east-and-africa/2018/09/13/zambia-slumps-toward-another-
debt-crisis.

The Economist. 2020. How to Stop Zambia from Turning into Zimbabwe. The Economist. 12
November 2020. Accessed on 1 October 2023 at
https://www.economist.com/leaders/2020/11/12/how-to-stop-zambia-from-turning-into-zimbab
we.

The Economist. 2023. Argentina Is Wasting the Vast Opportunities China Offers It. The
Economist. 7 March 2023. Accessed on 1 October 2023 at
https://www.economist.com/the-americas/2023/03/07/argentina-is-wasting-the-vast-opportunitie
s-china-offers-it.

The Economist. 2023. Sheikh Hasina’s Iron Grip on Bangladesh May yet Strangle Its Laudable
Progress. The Economist. 2 June 2023. Accessed on 1 October 2023 at

264
https://www.economist.com/podcasts/2023/06/02/sheikh-hasinas-iron-grip-on-bangladesh-may-
yet-strangle-its-laudable-progress.

The Export-Import Bank of China. 2017. Offering Memorandum. 7 March 2017. Accessed on 4
August 2023 at
https://links.sgx.com/FileOpen/China%20EximBank_USD%20Bonds%20Offering%20Memorand
um%20dated%207%20March%202017.ashx?App=Prospectus&FileID=30816

The Export-Import Bank of China. 2021. The Party Secretary and President of the Export-Import
Bank of China Hu Xiaolian attends events at the 2021 Boao Asia Forum. 20 April 2021. Accessed
on 1 October 2023 at www.eximbank.gov.cn/info/news/202104/t20210420_30419.html

The Growth Lab at Harvard University. 2023. The Atlas of Economic Complexity. Accessed on 1
August 2023 at http://www.atlas.cid.harvard.edu.

The People's Bank of China (PBOC). 2007. China Development Bank to Receive US$20 billion as
Capital Injection. 31 December 2007. Beijing: PBOC. Accessed on 2 June 2023 at
www.pbc.gov.cn/en/3688110/3688172/4048344/3715513/index.html

The People's Bank of China (PBOC). 2012. China Financial Stability Report 2012. eijing: PBOC.
Accessed on 1 June 2023 at
www.pbc.gov.cn/eportal/fileDir/image_public/UserFiles/english/upload/File/China%20Financial
%20%20Stability%20Report(1).pdf

The White House. 2022. National Security Strategy. Washington DC: The White House.
Accessed on 2 June 2023 at
https://www.whitehouse.gov/wp-content/uploads/2022/10/Biden-Harris-Administrations-Nationa
l-Security-Strategy-10.2022.pdf.

The White House. 2023. Fact Sheet: Vice President Harris Announces Over $7 Billion in Private
Sector and U.S. Government Commitments to Promote Climate Resilience, Adaptation, and
Mitigation across Africa. Washington DC: The White House. Accessed on 2 June 2023
athttps://www.whitehouse.gov/briefing-room/statements-releases/2023/03/31/fact-sheet-vice-pr
esident-harris-announces-over-7-billion-in-private-sector-and-u-s-government-commitments-to-p
romote-climate-resilience-adaptation-and-mitigation-across-africa/.

Transparency International. 2021. Corruption Perceptions Index.

https://www.transparency.org/en/cpi/2021.

Tripathi, Salil. 2021. Bangladesh’s Long Journey From ‘Basket Case’ to Rising Star. Foreign Policy
(blog). 10 April 2021. Accessed on 2 June 2023 at
https://foreignpolicy.com/2021/04/10/bangladesh-independence-anniversary-basket-case-rising-
star/.

Trotsenburg, Axel van. 2023. Defying the Odds: Bangladesh’s Journey of Transformation and
Resilience. 1 March 2023. Washington DC: World Bank. Accessed on 2 June 2023
athttps://www.worldbank.org/en/news/opinion/2023/03/01/defying-the-odds-bangladesh-s-jour
ney-of-transformation-and-resilience.

U.S. Department of Defense. 2018. US National Defense Strategy 2018: Sharpening the
American Military’s Competitive Edge. 11 August 2018. Washington, DC: U.S. Department of
Defense. Accessed on 1 October 2023 at
https://dod.defense.gov/Portals/1/Documents/pubs/2018-National-Defense-
Strategy-Summary.pdf

265
U.S. International Development Finance Corporation (DFC). 2021a. DFC Surpasses $6.7 Billion in
Commitments in 2021 During Record Fiscal Year. October 12, 2021. Washington DC: DFC.
Accessed on 16 July 2023 at
https://www.dfc.gov/media/press-releases/dfc-surpasses-67-billion-commitments-2021-during-re
cord-fiscal-year

U.S. International Development Finance Corporation (DFC). 2021b. DFC and Ecuador Sign
Framework Agreement to Support Development, Private Sector Investment. January 14, 2021.
Washington DC: DFC. Accessed on 16 July 2023 at
https://www.dfc.gov/media/press-releases/dfc-and-ecuador-sign-framework-agreement-support-
development-private-sector

U.S. International Development Finance Corporation (DFC). 2022. DFC Marks Largest Year with
More Than $7.4 Billion Committed in FY 2022. November 15, 2022. Washington DC: DFC.
Accessed on 16 July 2023 at
https://www.dfc.gov/media/press-releases/dfc-marks-largest-year-more-74-billion-committed-fy-
2022

UNEP-WCMC & IUCN. 2023. Protected Planet: The World Database on Protected Areas (WDPA).
August 2023. Cambridge, UK: UNEP-WCMC & IUCN.

United Nations. 2022. World Population Prospects 2022. Dataset.

https://population.un.org/wpp/.

Vadlamannati, Krishna Chaitanya, Samuel Brazys, Alexander Dukalskis, and Yuanxin Li. 2023.
Building Bridges or Breaking Bonds? The Belt and Road Initiative and Foreign Aid Competition.
Foreign Policy Analysis 19 (3): orad015.

Vaughn, Bruce. 2008. Bangladesh: Political Turmoil and Transition. CRS Report for Congress.
Washington DC: Congressional Research Service.
https://www.everycrsreport.com/files/20080530_RL34522_27d61b3aea89509df1d24e4d578959
04992d8120.pdf.

Vreeland, James Raymond, and Axel Dreher. 2014. The Political Economy of the United Nations
Security Council. Money and Influence. New York: Cambridge University Press.

Wade, Abdoulaye. 2008. Time for the West to Practice What It Preaches. Financial Times. 23
January 2008. Accessed on 15 August 2023 at
https://www.ft.com/content/5d347f88-c897-11dc-94a6-0000779fd2ac.

Wade, Robert H. 2009. Accountability Gone Wrong: The World Bank, Nongovernmental
Organisations and the US Government in a Fight over China. New Political Economy 14 (1):
25-48.

Warner, Andrew. 2010. Cost-Benefit Analysis in World Bank Projects. Washington, DC:
Independent Evaluation Group of the World Bank.

Watts, Jonathan. 2015. Argentina Leader Leaves Controversial Legacy with Patagonia Dams
Project. The Guardian. 1 December 2015. Accessed on 2 June 2023 at
https://www.theguardian.com/world/2015/dec/01/argentina-president-cristina-fernandez-de-kirc
hner-patagonia-hydroelectric-dam-project.

Wei, Lingling. 2013. China's Zhu Changhong Helps Steer Nation's Currency Reserves. 16 July
2013. Wall Street Journal. Accessed on 1 August 2023 at
https://www.wsj.com/articles/SB10001424127887323664204578606301739504368

266
Wellner, Lukas, Axel Dreher, Andreas Fuchs, Bradley C. Parks, and Austin Strange. 2023. Chinese
development finance and public opinion. 2 October 2023. VoxEU column. Accessed on 2
October 2023 at
https://cepr.org/voxeu/columns/chinese-development-finance-and-public-opinion

Wellner, Lukas, Axel Dreher, Andreas Fuchs, Bradley C. Parks, and Austin Strange. Forthcoming.
Can Aid Buy Foreign Public Support? Evidence from Chinese Development Finance. Economic
Development and Cultural Change.

Wen, Yao. 2023. BD an Important Strategic Partner of China. The Financial Express (blog). 2
August 2023. Accessed on 2 August 2023 at
https://thefinancialexpress.com.bd/views/views/bd-an-important-strategic-partner-of-china.

White, Abraham, and Leo Holtz. 2022. Figure of the Week: African Countries’ Votes on the UN
Resolution Condemning Russia’s Invasion of Ukraine. Washington DC: Brookings Institution.
Accessed on 2 October 2023 at
https://www.brookings.edu/articles/figure-of-the-week-african-countries-votes-on-the-un-resoluti
on-condemning-russias-invasion-of-ukraine/.

Wigglesworth, Robin and Sun Yu. 2023. How China changed the game for countries in default.
13 April 2023. Financial Times. Accessed on 14 August 2023 at
https://www.ft.com/content/19add278-aa83-45f8-a84f-12750f32258f

Wilkinson, Tracy. 2019. Argentina Election Raises Doubts about Trump’s Bet on Right-Wing
Leaders. Los Angeles Times. 10 September 2019. Accessed on 2 June 2023 at
https://www.latimes.com/world-nation/story/2019-09-09/argentina-election-trump-right-wing-lati
n-america-support.

Wilson, Elliot. 2022. BRI: Have we passed peak China? 16 February 2022. Euromoney. Accessed
on 3 August 2023 at
https://www.euromoney.com/article/29ppq3xrh5p5bxgoh9dz4/opinion/bri-have-we-passed-pea
k-china

Wonacott, Peter. 2011. In Africa, U.S. Watches China’s Rise. Wall Street Journal. 2 September
2011. Accessed on 2 June 2023 at
https://www.wsj.com/articles/SB10001424053111903392904576510271838147248

Wong, Chun Han and Keith Zhai. 2023. Court Papers Open Rare Window Into Role of Graft in
China’s Overseas Lending. 9 June 2023. Wall Street Journal. Accessed on 7 August 2023 at
https://www.wsj.com/articles/court-papers-open-rare-window-into-role-of-graft-in-chinas-oversea
s-lending-69f7ab81

Wooley, Alex. 2023. AidData’s China data wins 2023 Best Dataset Award from IPES. The First
Tranche. 29 August 2023. Accessed on 10 September 2023 at
https://www.aiddata.org/blog/aiddatas-china-data-wins-2023-best-dataset-award-from-ipes

World Bank Group. 2020. World Bank Group Approaches to Mobilize Private Capital for
Development: An Independent Evaluation. Washington DC: World Bank
https://ieg.worldbankgroup.org/sites/default/files/Data/Evaluation/files/PrivateCapitalMobilizatio
n.pdf

World Bank. 2000. World Bank Debtor Reporting System Manual. Washington DC: World Bank.
Accessed on 20 February 2023 via
https://web.archive.org/web/20220901043448/https://databankfiles.worldbank.org/data/downlo
ad/debt/DRS%20Manual%202013.pdf

267
World Bank. 2012. World Bank Sanctions Procedures. Washington DC: World Bank. Accessed on
20 August 2023 via
https://www.worldbank.org/content/dam/documents/sanctions/other-documents/osd/WBGSanc
tions_Procedures_April2012_Final.pdf

World Bank. 2020a. International Debt Statistics 2021. Washington DC: World Bank. Accessed
on 20 February 2023 via https://openknowledge.worldbank.org/handle/10986/34588

World Bank. 2020b. World Bank Debtor Reporting System – What it Measures. World Bank
Development Economics Data Group. Washington DC: World Bank. Accessed on 15 September
2023 at
https://pubdocs.worldbank.org/en/951771586884732835/pdf/DRS-What-it-Measures.pdf

World Bank. 2021. International Debt Statistics 2022. Washington DC: World Bank. Accessed on
20 February 2023 via https://openknowledge.worldbank.org/handle/10986/36289

World Bank. 2023a. World Development Indicators.

https://databank.worldbank.org/source/world-development-indicators.

World Bank. 2023b. Procurement - World Bank Listing of Ineligible Firms and Individuals.
Washington DC: World Bank. Accessed on June 1, 2023 via
https://www.worldbank.org/en/projects-operations/procurement/debarred-firms

World Bank. 2023c. Remarks by World Bank Group President Ajay Banga at the 2023 Annual
Meetings Plenary on 13 October 2023. Washington DC: World Bank. Accessed on 13 October
2023 at
https://www.worldbank.org/en/news/speech/2023/10/13/remarks-by-world-bank-group-presiden
t-ajay-banga-at-the-2023-annual-meetings-plenary

Xiao, Cai. 2016. SAFE buys into A-share market. 1 April 2016. China Daily. Accessed on 1 June
2023 at https://www.chinadaily.com.cn/cndy/2016-04/01/content_24224169.htm

Xie, Yifan and Mia Lamar. 2016. China Currency Regulator Invests in Bank Stocks. 4 April 2016.
Wall Street Journal. Accessed on 5 July 2023 at
https://www.wsj.com/articles/chinas-foreign-exchange-regulator-dips-into-stocks-1459523466

Xinhua. 2015 China raises capital base of development bank. 18 August 2015. Xinhua. Accessed
on 5 June 2023 at https://www.chinadaily.com.cn/business/2015-08/18/content_21640311.htm

Xining, Wang. 2021. Whole-Process People’s Democracy Is A High Quality Democracy. 2021.
Beijing: Ministry of Foreign Affairs of the People’s Republic of China. Accessed on 5 September
2023 at
https://www.fmprc.gov.cn/eng/wjb_663304/zwjg_665342/zwbd_665378/202112/t20211213_104
67431.html.

Xueqing, Jiang. 2017. CBRC Tells Policy Banks to Beef up Risk Management. 16 November
2017. China Daily. Accessed on 1 August 2023 at
https://www.chinadaily.com.cn/business/2017-11/16/content_34591397.htm

Xueqing, Jiang. 2022. Tsinghua report exposes Africa's bond crises. 5 August 2022. China Daily.
Accessed on 5 August 2022 at
https://www.chinadaily.com.cn/a/202208/05/WS62ec70d1a310fd2b29e706f6.html

268
Yang, Hongbo, B. Alexander Simmons, Rebecca Ray, Christoph Nolte, Suchi Gopal, Yaxiong Ma,
Xinyue Ma, Kevin P. Gallagher, 2021. Risks to global biodiversity and Indigenous lands from
China’s overseas development finance. Nature Ecology & Evolution 11:1520-1529.

Ye, Min. 2021. Fragmented Motives and Policies: The Belt and Road Initiative in China. Journal
of East Asian Studies 21 (2): 193–217.

Ying, Diao. 2008. Policy bank to gradually go commercial. 3 March 2008. China Daily. Accessed
on 15 June 2023 at https://www.chinadaily.com.cn/china/2008-03/03/content_6500728.htm

Yuan, Yao. 2014. Global Investment M&A Newsletter: Revealing the Mystery of Foreign
Exchange Reserve Entrusted Loans. Accessed on 1 July 2023 at
https://m.gaodun.com/caiwu/379542.html

Yuzhe, Zhang. 2010. China finds policy job for forex reserves. 23 May 2010. Caixin Online.
Accessed on 1 August 2023 at
https://web.archive.org/web/20100523082852/http://english.caing.com/2010-05-20/100145743
.html

Zeitz, Alexandra O. 2021. Emulate or differentiate? Chinese development finance, competition,

and World Bank infrastructure funding. The Review of International Organizations 16: 265–292.

Zhang, Yanqiu, and Daniel Oloo Ong’ong’a. 2022. Unveiling China’s Digital Diplomacy: A
Comparative Analysis of CGTN Africa and BBC News Africa on Facebook. Asian Journal of
Comparative Politics 7 (3): 661–83.

Zhang, Denghua, and Graeme Smith. 2017. Foreign Aid System: Structure, Agencies, and
Identities. Third World Quarterly 38 (10): 2330–2346.

Zhou, Haonan and Eugenio Cerutti 2018. The Chinese banking system: Much more than a
domestic giant. VoxEU Blog. Accessed on 19 July 2023 at
https://cepr.org/voxeu/columns/chinese-banking-system-much-more-domestic-giant

Zhu, Yanling. 2022. China’s ‘New Cultural Diplomacy’ in International

Broadcasting: Branding the Nation through CGTN Documentary. International
Journal of Cultural Policy 28 (6): 671–83.

269
 Appendix: Supplementary Material

Section A-1: Figures and tables referenced in the report

Table A1: AidData’s Global Chinese Development Finance Dataset, Version 3.0 at a glance

Table A.1

AidData’s Global Chinese Development Finance Dataset, Version 3.0 at a glance

Global Chinese Development Finance Global Chinese Development Finance
AidData Dataset Dataset, Version 2.0 Dataset, Version 3.0
(Published September 2021) (Published November 2023)

Scope & Sectors All All

Coverage
Country Coverage 165 countries globally 165 countries globally
(including 145 countries with projects (including 146 countries with projects
identified) identified)

Financiers 334 Chinese official sector donors and 791 Chinese official sector donors and
lenders lenders

Financial Instrument Loans, grants, scholarships, technical Loans (with categorization of 23 distinct
assistance, debt rescheduling, debt loan instruments), grants, scholarships,
forgiveness technical assistance, debt rescheduling,
debt forgiveness

Dataset Number of Records 13,427 20,985

Summary
Number of Fields 70 133

Sources Publicly Available 91,125 147,703

(including 62,750 unique sources) (including 99,393 unique sources)

Total Financial Value $851 billion (2017 prices) $1.34 trillion (2021 prices)
(excluding short-term “rollover” facilities)

Timeframe 2000-2017 (with implementation details 2000-2021 (with implementation details

through 2021) through 2023)

Financial Details Transaction amount, collateral, interest Transaction amount, collateral,

Project rate, grace period, maturity, commitment interest rate, default interest rate, grace
Details fee, management fee period, maturity, commitment fee,
management fee, insurance fee, first and
last loan repayment dates, level of public
liability

Participating Agencies Funding agencies, co-financing agencies, Funding agencies, co-financing agencies,
receiving agencies, implementing direct receiving agencies, indirect
agencies, accountable agencies receiving agencies, implementing
agencies, guarantor, insurance provider,
collateral provider, security
agent/collateral agent

Implementation Details Commitment year, status, planned and Commitment date, status, planned and
actual start and completion dates actual start and end dates, deviation from
planned start and completion dates,

270
 Global Chinese Development Finance Global Chinese Development Finance
AidData Dataset Dataset, Version 2.0 Dataset, Version 3.0
(Published September 2021) (Published November 2023)

infrastructure project flag

Description Average of 142 words per project Average of 166 words per project

OECD Classifications Sector, flow class Sector, flow class, recipient country
income classification, grant-equivalent
measure

Sub-national Details 3,285 physical locations 9,497 physical locations

Figure A1: Composition of official financial flows from China to the developing world, 2000-2021

Figure A.1

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (see
Section A-2 of the Appendix for details). The Vague (Official Finance) is a residual category for official
financial commitments from China that could not be reliably categorized as ODA or OOF because of
insufficiently detailed information.

271
Figure A2: Stock of official financial flows from China and the G7 to the developing world during the early and late BRI periods

Figure A.2

Notes: AidData relies on OECD-DAC measurement criteria to make ODA and OOF determinations (as
described in Section A-2 of the Appendix). The Vague (Official Finance) is a residual category for official
financial commitments from China that could not be reliably categorized as ODA or OOF because of
insufficiently detailed information. G7 ODA and OOF data represent gross disbursements from the

272
OECD-DAC. This figure excludes short-term “rollover” facilities from the tally of official financial
commitments (see Box 2c and Section A-3 in the Appendix).

Figure A3: Sectoral composition of official financial flows from China to developing world, 2014 vs. 2021

Figure A.3

AidData’s Global Chinese Development Finance Dataset, Version 3.0 at a

glance

Notes: This figure shows the sectoral composition of Chinese ODA and OOF commitments (measured in
constant 2021 USD) in 2014 (left panel) and 2021 (right panel).

273
Figure A4: Number of financially-distressed low- and middle-income countries with outstanding debt to China

Figure A.4

Notes: To determine if a country has “outstanding debt to China,” we use the 3.0 version of AidData’s
GCDF dataset to measure whether in a given year if it had at least one official sector loan from China within
its originally scheduled repayment period (i.e., after the expiration of the grace period but before the final
maturity date). To determine if a country was experiencing financial distress in a given year, we use the
measure that is described in Box 1a. Since 2000 is the first year in which we measure loan commitments in
the 3.0 version of the GCDF dataset, we do not capture any outstanding debt to China (loans within their
originally scheduled repayment periods) in that year.

274
Figure A5: Sovereign overdue repayments owed to official sector creditors in China

Figure A.5

Notes: This figure measures the average per country monetary value (in constant 2021 USD millions) of
sovereign arrears to a subset of official sector creditors in China (using data from the World Bank’s
International Debt Statistics). Sovereign arrears capture principal and interest arrears (i.e., overdue
repayments) on PPG debt to China Eximbank, China Development Bank, and China’s Ministry of Commerce
contracted by a subset of LICs and MICs that participate in the World Bank’s Debtor Reporting System
(DRS). Years in which a country maintained diplomatic relations with Taiwan are excluded. Each country-year
observation is given equal weight in a given year to generate global averages.

275
Figure A6: Project life cycle

Figure A.6

Project life cycle

276
Figure A7: Average length of completion delays across all Chinese ODA- and OOF-financed infrastructure projects

Figure A.7

Notes: This graph shows the average length of completion delays (in calendar days) across all Chinese
ODA- and OOF-financed infrastructure projects by year that were formally committed, undergoing
implementation, or completed. Completion delays are calculated by taking the difference (in calendar days)
between the originally scheduled project completion date and the actual project completion date. The
data are drawn from the 3.0 version of AidData’s GCDF dataset.

277
Figure A8: China’s soft power gains and losses vis-à-vis the U.S.

Figure A.8

Notes: Figure A8 presents the proportion of LICs and MICs in a given year (from 2014-2021) in which China
experienced relative gains or losses in public approval, media sentiment, and UNGA voting alignment

278
vis-à-vis the U.S. To measure the relative gains or losses in popular support, we follow a three-step
calculation for each country: (1) calculate the difference between the public approval rating for China in a
given year and the prior year; (2) calculate the difference between public approval rating for the U.S. in a
given year and the prior year; and (3) calculate the “double difference” between (1) and (2) to determine if
China experienced a greater gain or loss in public support than the U.S. in the same country-year. For
relative gains and losses in media sentiment and UNGA voting alignment, the same three-step calculation
was followed using the average media sentiment score for each country-year from the GDELT 1.0 Event
Database (related to government actors from mainland China or the U.S.) and the average
“idealpointdistance” estimate between each country and China (or the U.S.) in a given year.

Figure A9: China versus the U.S.: Public disapproval rates

Figure A.9

Notes: This graph shows the annual weighted average disapproval rate for China and the U.S. between
2014-2021 from Gallup World Poll. The average disapproval rate is weighted by population size for each
country. When surveying respondents, respondents can answer “approve”, “disapprove”, or “unsure”
when asked if they approve of China’s or the U.S’s leadership. This graph represents the average that
answered “disapprove” for each country’s leadership. The construction of this variable is described in
greater detail in Box 1b.

279
Figure A10: Composition of China’s overseas lending portfolio by level of public liability

Figure A.10

Notes: This graph shows the annual composition of China’s overseas lending portfolio (as measured in 2021
constant USD) in LICs and MICs according to the extent to which the host governments may eventually be
liable for debt repayment. Central government debt and other public sector debt represent loans where
the borrower is a government agency or a wholly- or majority-owned state entity. Central government debt
represents loans that have a sovereign guarantee from the host government. Potential public debt
represents loans to entities (including special purpose vehicles or joint ventures) where the host
government has a minority stake. Private debt captures loans to private entities.

280
Figure A11: Proportion of China’s overseas lending portfolio provided via buyer’s credits

Figure A.11

Figure A12: Proportion of China’s overseas lending portfolio provided via Government Concessional Loans (GCLs) and Preferential Buyer’s Credits (PBCs)

Figure A.12

281
Notes: This graph shows the annual proportion of official sector loan commitments from China (measured
in constant 2021 USD) to LICs and MICs provided via Government Concessional Loans (GCLs) and
Preferential Buyer’s Credits (PBCs from China Eximbank.

Figure A13: Proportion of China’s overseas lending portfolio provided via infrastructure project loans

Figure A.13

Notes: This graph shows the annual proportion of official sector loan commitments from China (measured
in constant 2021 USD) to LICs and MICs that were issued to build, maintain, or renovate infrastructure in
the borrowing country. Infrastructure project loans in the 3.0 version of AidData’s GCDF dataset are those
(a) categorized as investment project loans (IPLs), and (b) explicitly designated as supporting infrastructure
(through the “infrastructure” marker).

282
Figure A14: Composition of China’s overseas lending portfolio by emergency and non-emergency lending instrument

Figure A.14

Notes: This figure measures the percentages of China’s overseas lending (in constant 2021 USD) to LICs
and MICs that consists of emergency rescue loans (ERLs) that are rollovers, ERLs that are not rollovers, and
all other types of loans. The “rescue” variable in the 3.0 version of the GCDF dataset is used to identify
emergency rescue loans. Rollover ERL amounts are calculated by subtracting the values in the Adjusted
Amount (Constant USD 2021) field from the values in the Amount (Constant USD 2021) field. Non-rollover
ERL amounts are directly drawn from the Adjusted Amount (Constant USD 2021) field.

283
Figure A15: Proportion of RMB-denominated rescue lending to countries in and not in financial distress

Figure A.15

Notes: This graph presents the annual proportion of China’s RMB-denominated loan commitments (as
measured in 2021 constant USD) to two country cohorts between 2000 and 2021: (1) countries in financial
distress, and (2) countries not in financial distress. To determine if a country experienced financial distress in
a given year, we use the binary measure that is described in Box 1a in Chapter 1.

284
Figure A16: Composition of China’s overseas lending portfolio by currency of denomination for countries in financial distress

Figure A.16

Notes: This figure presents the composition of China’s lending portfolio in LICs and MICs (as measured in
2021 constant USD) by the currencies in which the loans were denominated for country-years that were
designated as “in financial distress.” To determine if a country experienced financial distress in a given year,
we use the binary measure that is described in Box 1a in Chapter 1.

285
Figure A17: Composition of China’s overseas loan portfolio by currency of denomination for countries not in financial distress

Figure A.17

Notes: This figure shows the composition of China’s lending portfolio in LICs and MICs (as measured in
2021 constant USD) by the currencies in which the loans were denominated for country-years that were not
designated as “in financial distress.” To determine if a country experienced financial distress in a given year,
we use the binary measure that is described in Box 1a in chapter 1.

Figure A18: Early versus late BRI: weighted average interest rates

Figure A.18

286
Figure A19: Early versus late BRI: weighted maturity lengths

Figure A.19

Figure A20: Early versus late BRI: weighted average grant element

Figure A.20

287
Figure A21: Weighted average grant element of overseas lending from China

Figure A.21

Notes: This graph shows the average grant element of official sector lending from China to LIC and MICs
between 2000 and 2021. This grant element is calculated using the Grant Element (OECD cash-flow)
variable in the 3.0 version of AidData’s GCDF 3.0, which uses a 10% discount rate for all borrowing
countries based on the OECD’s cash-flow grant element calculation. The annual averages are weighted by
the constant 2021 USD values of the loan commitments in each respective year.

288
Figure A22: Weighted average maturity lengths for countries in and not in financial distress

Figure A.22

Notes: This graph shows the average maturity length across all official sector loans from China to LICs and
MICs between 2000 and 2021 across two cohorts: (1) countries in financial distress and (2) countries not in
financial distress. The annual averages are weighted by the constant 2021 USD commitment values of the
loans in each respective year. To determine if a country experienced financial distress in a given year, we
use the binary measure that is described in Box 1a in Chapter 1.

289
Figure A23: Weighted average interest rates for countries in and not in financial distress

Figure A.23

Notes: This graph shows the average interest rate across all official sector loans from China to LICs and
MICs between 2000 and 2021 for two cohorts: (1) countries in financial distress and (2) countries not in
financial distress. The annual averages are weighted by the constant 2021 USD commitment values of the
loans in each respective year. To determine if a country experienced financial distress in a given year, we
use the binary measure that is described in Box 1a in chapter 1.

290
Figure A24: Weighted average grant elements for countries in and not in financial distress, 2014-2021

Figure A.24

Notes: This graph shows the average grant element across all official sector loans from China to LICs and
MICs between 2000 and 2021 for two cohorts: (1) countries in financial distress and (2) countries not in
financial distress. To determine if a country experienced financial distress in a given year, we use the binary
measure that is described in Box 1a in chapter 1. The grant element is calculated using the Grant Element
(OECD cash-flow) variable in the 3.0 version of AidData’s GCDF dataset, which uses a 10% discount rate for
all borrowing countries based on the OECD’s cash-flow grant element calculation. The annual grant
element averages are weighted by the constant 2021 USD commitment values of the loans in each
respective year.

Figure A25: Grant element across China’s portfolio to countries in and not in financial distress

Figure A.25

Notes: This graph shows the average grant element across China’s portfolio of official sector loans to LICs
and MICs across two cohorts: countries experiencing financial distress and countries not experiencing
financial distress. The grant element is calculated using the Grant Element (OECD cash-flow) variable in the

291
3.0 version of AidData’s GCDF dataset, which uses a 10% discount rate for all borrowing countries based
on the OECD’s cash-flow grant element calculation. The grant element average is weighted by constant
2021 USD commitment values.

Figure A26: Composition of China’s overseas lending portfolio by type of lending instrument

Figure A.26

Notes: This figure shows the annual percentage of official lending commitments (in constant 2021 USD)
from China to LICs and MICs that was provided via (1) investment project loans, (2) emergency rescue
loans, and (3) all other types of loans. Project loans are defined as those in the 3.0 version of AidData’s
GCDF dataset that are categorized as investment project loans (IPLs).

292
Figure A27: Composition of China’s overseas lending portfolio by type by creditor category

Figure A.27

Figure A28: Proportion of China’s overseas lending portfolio involving a multilateral institution

Figure A.28

293
Notes: This graph shows the annual percentage of project loans from official sector creditors in China to
borrowers in LICs and MICs that involve multilateral institutions as co-financiers, implementing agencies, or
a receiving agency between 2000 and 2021.

Figure A29: Composition of China’s overseas lending with and without credit enhancements to countries in and not in financial distress

Figure A.29

Notes: This graph shows the proportion of China’s overseas lending commitments (measured as constant
2021 USD) to LICs and MICs backed by a credit insurance policy, a third-party repayment guarantee,
and/or collateral to two cohorts: (1) countries in financial distress and (2) countries not in financial distress.
To determine if a country experienced financial distress in a given year, we use the binary measure that is
described in Box 1a in Chapter 1.

Figure A30: Composition of China’s non-emergency overseas loan portfolio by repayment risk

Figure A.30

Notes: This graph shows the percentage of China’s non-emergency loan commitments between 2000 and

294
2021 that supported two cohorts: (1) LICs and MICs with scores of 5 or less on the sovrate index; and (2)
LICs and MICs with scores above 5 on the sovrate index. The World Bank’s sovrate index is a measure of
repayment risk that varies from 0 to 21, with higher scores indicating lower levels of sovereign credit risk
(Kose et al. 2022). Countries with scores of 0-5 are in default or at a high risk of default (see Box 1a). Official
sector Chinese lending commitments are measured in constant 2021 USD. Country-year observations
without official sector Chinese lending commitments or sovrate scores are excluded.

295
Figure A31: Annual loan commitments to LICs and MICs by financial institution and capital injections from SAFE, 2000-2021

Figure A.31

Annual loan commitments to LICs and MICs by financial institution and capital
injections from SAFE, 2000-2021

Notes: This table presents annual lending commitments to LICs and MICs (in constant 2021 USD) from
selected Chinese state-owned policy banks, state-owned commercial banks, and state-owned funds. It also
presents cumulative lending commitments to LICs and MICs (in constant 2021 USD) that are backed by
credit insurance from Sinosure. The vertical dashed lines represent years in which a SAFE capital injection is
known to have taken place.

296
Figure A32: Proportion of China’s syndicated overseas lending that involves non-Chinese creditors

Figure A.32

Notes: This graph shows the annual percentage of syndicated loans involving Chinese state-owned
creditors (by the number of loans) that also involve non-Chinese creditors.

297
Figure A33: Proportion of Chinese overseas syndicated lending involving a multilateral institution

Figure A.33

Notes: This graph shows the annual percentage of syndicated loans involving Chinese state-owned
creditors (by the number of loans) that also involve multilateral institutions.

Figure A34: Average financial commitments from Chinese state-owned creditors to different types of syndicated loans

Figure A.34

Notes: This figure represents the average monetary commitment value size (in constant 2021 USD) from
official sector creditors in China that participated in syndicated loans to LICs and MICs across two cohorts:
(1) syndicated loans with Chinese and non-Chinese participants, and (2) syndicated loans with only Chinese
participants.

298
Figure A35: Chinese state-owned bank contributions to syndicated loans in LICs and MICs by participant cohort

Figure A.35

Notes: This figure presents all individual syndicated loans captured in the 3.0 version of AidData’s GCDF
dataset based on the size of the monetary commitments (in constant 2021 USD) from the official sector
creditors in China that contributed to the loans. The color codes correspond to two separate types of
syndicated loans: (1) those with Chinese and non-Chinese participants, and (2) those with only Chinese
participants.

Figure A36: Composition of China’s overseas project lending portfolio by channel of delivery

Figure A.36

299
Figure A37: Official lending commitments by lending institution type, 2000-2021

Figure A.37

300
Figure A38: Cumulative percentage of infrastructure portfolio with significant environmental, social, or governance risk exposure

Figure A.38

301
Figure A39: Infrastructure projects located within environmentally sensitive areas

Figure A.39

Notes: This figure shows the cumulative number and monetary value (in constant 2021 USD) of Chinese
grant- and loan-financed infrastructure projects in LICs and MICs between 2000 and 2021 that are located
in environmentally sensitive areas. Section 2 of Chapter 3 describes the methods that were used to identify
infrastructure projects in environmentally sensitive areas.

302
Figure A40: Infrastructure projects located within socially sensitive areas

Figure A.40

Notes: This figure shows the cumulative number and monetary value (in constant 2021 USD) of Chinese
grant- and loan-financed infrastructure projects in LICs and MICs between 2000 and 2021 that are located
in socially sensitive areas. Section 2 of Chapter 3 describes the methods that were used to identify
infrastructure projects in sociallly sensitive areas.

303
Figure A41: Infrastructure projects located in geographical areas vulnerable to political capture and manipulation

Figure A.41

Notes: This figure shows the cumulative number and monetary value (in constant 2021 USD) of Chinese
grant- and loan-financed infrastructure projects in LICs and MICs between 2000 and 2021 that are located
in areas that are vulnerable to political capture and manipulation. Section 2 of Chapter 3 describes the
methods that were used to identify infrastructure projects in geographical areas that are vulnerable to
political capture and manipulation.

304
Figure A42: Infrastructure projects involving contractors sanctioned for fraudulent and corrupt behavior

Figure A.42

Notes: This figure shows the cumulative number and monetary value (in constant 2021 USD) of Chinese
government-financed infrastructure projects in LICs and MICs that relied on contractors sanctioned by other
international financiers for fraudulent and corrupt behavior between 2000 and 2021. Section 2 of Chapter 3
describes the methods used to identify infrastructure projects that relied on contractors sanctioned by other
international financiers for fraudulent and corrupt behavior.

305
Figure A43: Infrastructure projects that encountered significant environmental, social, or governance challenges over time

Figure A.43

Notes: This figure shows the cumulative number and monetary value (in constant 2021 USD) of Chinese
government-financed infrastructure projects in LICs and MICs that encountered significant environmental,
social, or governance challenges over time. Whether a project experienced a significant environmental,
social, or governance challenge is measured with a binary indicator from AidData that uses the project’s
description field to determine if there was evidence of a significant environmental, social, or governance
challenge before, during, or after implementation. For more details, see Section 2 of Chapter 3.

306
Figure A44: ESG risk prevalence in overseas infrastructure portfolio from China to LICs and MICs

Figure A.44

Notes: This figure shows the annual number (bars) and (with a solid line) the annual percentage of the
Chinese grant- and loan-financed infrastructure projects (count) in LICs and MICs that encountered
significant environmental, social, or governance risks between 2000-2021. Projects are recorded in the
years when they secured financial commitments from China, although the ESG risks that they encountered
may have materialized after the financial commitment year. The presence of significant ESG risk exposure is
based on a project-level composite measure that is described in Section 2 of Chapter 3.

307
Figure A45: Environmental risk prevalence in overseas infrastructure

Figure A.45

Notes: This figure shows the annual number (bars) and percentage (of project count) of Chinese
government-financed infrastructure projects in LICs and MICs that encountered significant environmental
risks between 2000-2021. Projects are recorded in the years when they secured financial commitments from
China, although the environmental risks that they encountered may have materialized after the financial
commitment year. The presence of significant environmental risk exposure is based on a project-level
composite measure that is described in Section 2 of Chapter 3.

308
Figure A46: Social risk prevalence in overseas infrastructure portfolio

Figure A.46

Notes: This figure shows the annual number (bars) and percentage (of project count) of Chinese
government-financed infrastructure projects in LICs and MICs that encountered significant social risks
between 2000-2021. Projects are recorded in the years when they secured financial commitments from
China, although the social risks that they encountered may have materialized after the financial
commitment year. The presence of significant social risk exposure is based on a project-level composite
measure that is described in Section 2 of Chapter 3.

309
Figure A47: Governance risk prevalence in overseas infrastructure portfolio

Figure A.47

Notes: This figure shows the annual number (bars) and percentage (of project count) of Chinese
government-financed infrastructure projects in LICs and MICs that encountered significant governance risks
between 2000-2021. Projects are recorded in the years when they secured financial commitments from
China, although the governance risks that they encountered may have materialized after the financial
commitment year. The presence of significant governance risk exposure is based on a project-level
composite measure that is described in Section 2 of Chapter 3.

310
Figure A48: Infrastructure projects supported by one or more de facto environmental, social, or governance risk mitigation efforts

Figure A.48

Notes: This figure shows the cumulative monetary value of Chinese grant- and loan-financed infrastructure
projects (in constant 2021 USD) supported by one or more de facto environmental, social, or governance
risk mitigation efforts (as measured via the keyword search-based method) from 2000 to 2021.

311
Figure A49: De jure ESG safeguard stringency in China’s overseas infrastructure portfolio with ESG risk exposure

Figure A.49

Notes: This figure shows the annual percentage of China’s grant- and loan-financed infrastructure project
portfolio in LICs and MICs (in constant 2021 USD) that presented significant environmental, social, or
governance risks across two cohorts: those with strong de jure ESG safeguards and those weak de jure ESG
safeguards. Projects with “strong” de jure ESG safeguards are defined as those with at least two out of
three (environmental, social and governance) safeguard stringency metrics that are “high.” Projects that do
not meet this standard are classified as having “weak” de jure ESG safeguards. The presence of significant
ESG risk exposure is based on a project-level composite measure that is described in Section 2 of Chapter
3.

312
Figure A50: Infrastructure project portfolio with de jure vs. de facto ESG risk mitigation

Figure A.50

Notes: This figure presents (in bars) the annual percentage of Chinese grant- and loan-financed
infrastructure projects (by project count) in LICs and MICs supported by one or more de facto
environmental, social, or governance risk mitigation efforts (as measured via the keyword search-based
method) from 2000 to 2021. It also presents (with a line) the annual percentage of Chinese grant- and
loan-financed infrastructure projects (by project count) in LICs and MICs with strong de jure ESG
safeguards. Projects with “strong” de jure ESG safeguards are defined as those with at least two out of
three (environmental, social and governance) safeguard stringency metrics that are “high.”

313
Figure A51: Infrastructure project portfolio with de jure vs. de facto ESG risk mitigation, by year of project completion

Figure A.51

Notes: This figure presents (in bars) the annual percentage of Chinese grant- and loan-financed
infrastructure project portfolio (in constant 2021 USD) in LICs and MICs supported by one or more de facto
environmental, social, or governance risk mitigation efforts (as measured via the keyword search-based
method) from 2000 to 2021. It also presents (with a line) the annual percentage of Chinese grant- and
loan-financed infrastructure project portfolio (in constant 2021 USD) in LICs and MICs with strong de jure
ESG safeguards. Both measures are based on the year the project was completed (where known) instead of
the commitment year. Projects with “strong” de jure ESG safeguards are defined as those with at least two
out of three (environmental, social and governance) safeguard stringency metrics that are “high.”

314
Figure A52: Infrastructure project portfolio with de jure vs. de facto ESG risk mitigation, by year of project commencement

Figure A.52

Notes: This figure presents (in bars) the annual percentage of Chinese grant- and loan-financed
infrastructure project portfolio (in constant 2021 USD) in LICs and MICs supported by one or more de facto
environmental, social, or governance risk mitigation efforts (as measured via the keyword search-based
method) from 2000 to 2023. It also presents (with a line) the annual percentage of Chinese grant- and
loan-financed infrastructure project portfolio (in constant 2021 USD)) in LICs and MICs with strong de jure
ESG safeguards over the same time period. Both measures are based on the year the project began
implementation (where known) instead of the commitment year. Projects with “strong” de jure ESG
safeguards are defined as those with at least two out of three (environmental, social and governance)
safeguard stringency metrics that are “high.”

315
Figure A53: Proportion of infrastructure project financing facing significant environmental risks by whether the project financing had strong de jure environmental safeguards versus those with
weak de jure ESG safeguards

Figure A.53

Notes: This figure shows the percentage of Chinese grant- and loan-financed infrastructure project portfolio
(in constant 2021 USD) in LICs and MICs from 2000 to 2021 that presented significant environmental risk
across two cohorts: (1) projects with strong de jure environmental safeguards, and (2) projects without
strong de jure environmental safeguards. Projects with “strong” de jure environmental safeguards are
defined as those with a score of “high” on AidData’s safeguard stringency scale. Projects that present
significant environmental risk are measured with 1/0 project-level composite measure (based on two of the
1/0 input indicators described in Section 2 of Chapter 3: whether the project was located in an
environmentally sensitive area and/or whether the project’s description field provides evidence of a
significant environmental challenge before, during, or after implementation).

316
Figure A54: Proportion of infrastructure project financing facing significant social risks by whether the project financing had strong de jure environmental safeguards versus those with weak de
jure ESG safeguards

Figure A.54

Notes: This figure shows the percentage of Chinese grant- and loan-financed infrastructure project portfolio
(in constant 2021 USD) in LICs and MICs from 2000 to 2021 that presented significant social risk across two
cohorts: (1) projects with strong de jure social safeguards, and (2) projects without strong de jure social
safeguards. Projects with “strong” de jure social safeguards are defined as those with a score of “high” on
AidData’s safeguard stringency scale. Projects that present significant social risk are measured with 1/0
project-level composite measure (based on two of the 1/0 input indicators described in Section 2 of
Chapter 3: whether the project was located in a socially sensitive area and/or whether the project’s
description field provides evidence of a significant social challenge before, during, or after
implementation).

317
Figure A55: Proportion of infrastructure project financing facing significant governance risks by whether the project financing had strong de jure environmental safeguards versus those with
weak de jure ESG safeguards

Figure A.55

Notes: This figure shows the percentage of Chinese grant- and loan-financed infrastructure project portfolio
(in constant 2021 USD) in LICs and MICs from 2000 to 2021 that presented significant governance risk
across two cohorts: (1) projects with strong de jure governance safeguards, and (2) projects without strong
de jure governance safeguards. Projects with “strong” de jure governance safeguards are defined as those
with a score of “high” on AidData’s safeguard stringency scale. Projects that present significant governance
risk are measured with 1/0 project-level composite measure (based on three of the 1/0 input indicators
described in Section 2 of Chapter 3: whether the project was located in an area vulnerable to political
capture and manipulation, whether the project relied on contractors sanctioned by other international
financiers for fraudulent and corrupt behavior; and/or whether the project’s description field provides
evidence of a significant governance challenge before, during, or after implementation).
318
Figure A56: Composition of infrastructure project portfolio: Reliance on China Eximbank and weak de jure ESG safeguards

Figure A.56

Notes: This figure shows the percent of Chinese grant- and loan-financed infrastructure projects (measured
as constant 2021 USD) with weak de jure ESG safeguards as well as the percentage of Chinese grant- and
loan-financed infrastructure projects (measured by constant 2021 USD) financed via bilateral loans from
China Eximbank. Projects with “strong” de jure ESG safeguards are defined as those with at least two out
of three (environmental, social and governance) safeguard stringency metrics that are “high.” Projects that
do not meet this standard are classified as having “weak” de jure ESG safeguards.

319
Figure A57: Composition of infrastructure project portfolio: Reliance on China Development Bank and weak de jure ESG safeguards

Figure A.57

Notes: This figure shows the percentage of Chinese grant- and loan-financed infrastructure projects
(measured as constant 2021 USD) with weak de jure ESG safeguards as well as the percentage of Chinese
grant- and loan-financed infrastructure projects (measured by constant 2021 USD) financed via bilateral
loans from CDB. Projects with “strong” de jure ESG safeguards are defined as those with at least two out of
three (environmental, social and governance) safeguard stringency metrics that are “high.” Projects that do
not meet this standard are classified as having “weak” de jure ESG safeguards.

Figure A58: Proportion of China’s infrastructure project portfolio behind schedule by de jure ESG safeguard strength

Figure A.58

Notes: This figure presents the percentage of China’s grant- and loan-financed infrastructure project
portfolio (in constant 2021 USD) that ran behind schedule across two cohorts: projects with strong de jure
environmental, social, and governance safeguards and projects with weak de jure environmental, social,
and governance safeguards. Projects with “strong” de jure ESG safeguards are defined as those with at

320
least two out of three (environmental, social and governance) safeguard stringency metrics that are “high.”
Projects that do not meet this standard are classified as having “weak” de jure ESG safeguards. “Behind
schedule” is defined as projects where the actual implementation start date took place 3 months or more
after its originally scheduled implementation start date as well as projects where the actual completion date
took place 3 months (or more) after its originally scheduled completion date. Only active projects and
completed projects that secured official commitments from China are included in the analysis.

Figure A59: Average length of commencement delays in infrastructure projects by de jure ESG safeguard strength

Figure A.59

Notes: This figure compares the average length of commencement delays (in days) in China’s grant- and
loan-financed infrastructure project portfolio across two cohorts: projects with strong de jure environmental,
social, and governance safeguards and projects with weak de jure environmental, social, and governance
safeguards. Projects with “strong” de jure ESG safeguards are defined as those with at least two out of
three (environmental, social and governance) safeguard stringency metrics that are “high.” Projects that do
not meet this standard are classified as having “weak” de jure ESG safeguards. Commencement delays are
calculated by taking the difference (in calendar days) between the originally scheduled project
implementation start date and the actual project implementation start date. Only active projects and
completed projects that secured official commitments from China are included in the analysis.

Figure A60: Average length of completion delays in infrastructure projects by de jure ESG safeguard strength

Figure A.60

Notes: This figure compares the average length of completion delays (in days) in China’s grant- and
loan-financed infrastructure project portfolio across two cohorts: projects with strong de jure environmental,
social, and governance safeguards and projects with weak de jure environmental, social, and governance
safeguards. Projects with “strong” de jure ESG safeguards are defined as those with at least two out of
three (environmental, social and governance) safeguard stringency metrics that are “high.” Projects that do
not meet this standard are classified as having “weak” de jure ESG safeguards Completion delays are
calculated by taking the difference (in calendar days) between the originally scheduled project completion
321
date and the actual project completion date. Only active projects and completed projects that secured
official commitments from China are included in the analysis.

Figure A61: Average time to complete an infrastructure project by de jure ESG safeguard strength

Figure A.61

Notes: This figure compares the average number of calendar days that it has taken to complete Chinese
grant- and loan-financed infrastructure projects in LICs and MICs across two cohorts: projects with strong
de jure environmental, social, and governance safeguards and projects with weak de jure environmental,
social, and governance safeguards. Projects with “strong” de jure ESG safeguards are defined as those with
at least two out of three (environmental, social and governance) safeguard stringency metrics that are
“high.” Projects that do not meet this standard are classified as having “weak” de jure ESG safeguards. The
amount of time needed to complete a project is calculated by measuring the number of calendar days
between the actual project implementation start date and the actual project completion date. Only active
projects and completed projects that secured official commitments from China are included in the analysis.

322
Figure A62: ESG map - project count

Figure A.62

ESG map - project count

Notes: This map presents the geographical areas where China’s grant- and loan-financed infrastructure
project portfolio (measured in project counts) has significant environmental, social, or governance (ESG) risk
exposure. Darker (purple) colors represent areas where the portfolio has high levels of risk exposure and
lighter (pink) colors represent areas where the portfolio has lower levels of risk exposure. Environmental risk
exposure, social risk exposure, and governance risk exposure are based on the project-level composite
measures that are described in Section 2 of Chapter 3. If a project falls across multiple grid cells, it is
counted in every grid cell that it intersects. In other words, for every grid cell, we count the number of
project points, lines or polygons that intersect it.

Section A-2: How AidData measures concessionality and

intent
As part of its data collection and classification system, AidData designates each financial and in-kind
transfer (“flow”) from an official sector institution as Official Development Assistance (ODA) or Other
Official Flows (OOF). The OECD’s Development Assistance Committee (DAC) has used these designations
since 1972 to distinguish between flows from official sector institutions that (a) are provided on
concessional terms and that promote and specifically target the economic development and welfare of
developing countries (ODA), and (b) are provided on non-concessional terms or do not specifically target

323
the economic development and welfare of developing countries (OOF). The sum of ODA and OOF is
sometimes referred to as Official Financial Flows, Official Financing, or Overseas Development Finance.
Many DAC countries, non-DAC countries, and multilateral institutions report the volume and composition
of their official sector flows according to these categories and criteria. In alignment with the OECD-DAC’s
own definitions, AidData classifies each project in the 3.0 dataset as either “ODA-like” or “OOF-like.” This
unique feature of the 3.0 dataset sets it apart from other publicly available datasets that measure Chinese
development finance in that it allows analysts to make “apples-to-apples” comparisons of Chinese
development finance and other international sources of development finance (that report their ODA and
OOF flow data to the OECD-DAC).

The criteria for whether a flow qualifies as ODA or OOF is determined by the OECD-DAC. It is based on (1)
the intent of the flow (whether its primary intent was development or not), (2) the income classification of
the receiving country, and (3) the concessionality level of the flow.369 All grants and in-kind transfers are
treated as concessional. However, a “grant element” measure is used to calculate the concessionality level
of all loans. This measure, which varies from 0 percent to 100 percent, seeks to capture the generosity of a
loan—or the extent to which it is priced below market rates. In principle, any loan provided on entirely
non-concessional terms should have a grant element of 0 percent.

While the first two criteria have remained consistent since the concept of ODA was introduced more than
five decades ago, the OECD-DAC recently made changes to the third (concessionality) criterion. Until 2017,
a loan from an official sector institution to a low-income or middle-income country had to meet a
concessionality (grant element) threshold of 25% to qualify as ODA. However, in 2018, the OECD-DAC
introduced a tiered system of discount rates and concessionality thresholds based on the income
classifications of borrower countries and whether borrowing institutions are official sector or private sector
institutions. The 2018 definition of concessionality is based on the following criteria:

● For loans to official sector institutions, the following concessionality thresholds apply:
○ Least-developed countries and low-income countries: a minimum grant element of 45%
(calculated using a 9% discount rate).
○ Lower-middle income countries: a minimum grant element of 15% (calculated using a 7%
discount rate).
○ Upper-middle income countries: a minimum grant element of 10% (calculated using a
discount rate of 6%).
● For loans to private sector institutions, the OECD-DAC maintains the pre-2018 definition of
concessionality and requires a grant element of at least 25% (that is calculated using a 10%
discount rate).370

To ensure comparability between the flows documented in the 3.0 version of the GCDF dataset and the
flow data published by the OECD-DAC, AidData has applied these definitions in the following manner:

Intent: AidData codes the intent of each financial and in-kind transfer (“flow”). Flows with “development
intent” are those that are primarily oriented toward the promotion of economic development and welfare in
the recipient country. Flows with “commercial intent” are those that primarily seek to promote the
commercial interests of the country from which the financial transfer has originated (e.g., encouraging the

369
An additional criteria is that the flow must be provided by official agencies, including state and local
governments or their executive agencies. AidData’s GCDF 3.0 only tracks official Chinese agencies, so this
criteria is always met.
370
According to the OECD, the method for calculating the ODA grant equivalent for loans to private sector
institutions has not yet been formalized, and discussions to do so are currently ongoing at the OECD-DAC.
Until an agreement has been formalized, the pre-2018 concessionality definition still applies.

324
export of Chinese goods and services). Flows with “representational intent” are those that primarily seek to
promote a bilateral relationship with another country or otherwise promote the language, culture, or values
of the country from which the financial transfer has originated (e.g., the establishment of a Confucius
Institute or Chinese cultural center). Flows with “military intent” are those that seek to promote the security
interests of the country from which the financial transfer originates or strengthen the lethal force capabilities
of military institutions in the recipient country.

ODA Income Classification: AidData reports the income classification group of the borrowing country.
Flows to countries not eligible for ODA are automatically assigned to the “OOF-like” category.

Concessionality:

● For flows committed between 2000 and 2017, a flow is classified as “ODA-like” when it (1) has
development intent, (2) has a grant element of at least 25% (using a 10% discount rate), and (3)
supports a country that is ODA-eligible according to the OECD-DAC’s ODA income classification
list.
● For flows committed between 2018 and 2021, a flow is classified as “ODA-like” when it (1) has
development intent, (2) has a concessionality level that meets the new criteria (established in 2018
definition), and (3) supports country that is ODA-eligible according to the OECD-DAC’s ODA
income classification list.

By definition, any international official sector flows not classified as ODA-like are classified as OOF-like. The
OOF-like flows in the 3.0 version of AidData’s GCDF dataset largely consist of export credits and
non-concessional loans.

In some cases, we are not able to determine if an international official sector flow would qualify as ODA or
OOF because of insufficiently detailed information in source documentation. In such cases, the flow in
question is categorized as Vague (Official Finance).

Section A-3: How does AidData measure the cumulative

stock of official financial flows from China to LICs and MICs?
As we explain at greater length in Chapter 2, short-term emergency rescue loans represent an increasingly
important part of China’s overseas portfolio of loans to LICs and MICs. Nearly all of these borrowings, which
are typically used to refinance maturing debts, carry de jure maturities of one year or less (i.e., they are
initially scheduled for repayment in 12 months or less). However, it is not unusual for financially-distressed
LICs and MICs to receive short-term emergency rescue loans from the same Chinese creditor in a series of
consecutive years. So-called “rollover” emergency rescue loans come in two varieties: (1) those that reach
their original contractual maturity dates and secure final maturity date extensions; and (2) those that are
repaid on their original contractual maturity dates and reissued (with similar or different face values and
borrowing terms) and assigned new maturity dates. However, among serial recipients of short-term
emergency rescue loans, it is seldom possible—with publicly available sources of information—to
differentiate between those who had their final maturity dates extended and those who fully repaid on their
original contractual maturity dates but were reissued new loans.

This relatively new feature of China’s overseas lending program raises an important question about how to
accurately estimate the cumulative stock of official financial flows—or lending commitments—from China to

325
LICs and MICs. Neither the OECD's Creditor Reporting System (CRS) nor the World Bank’s Debtor
Reporting System (DRS) ask lenders or borrowers to disclose loans with maturities of one year or less.
However, most of China’s short-term emergency rescue loans have de facto maturities that substantially
exceed one year (Horn et al. 2023a), which makes it difficult to justify the exclusion of all emergency rescue
loans from stock- or flow-based measures of official financial commitments (or lending commitments) from
China to LICs and MICs (see Box 2c). At the same time, rollover debt presents an overcounting risk because
it straddles a fine line between new lending commitments and maturity extensions of existing lending
commitments. This risk is particularly relevant to estimations of the cumulative stock of official financial
flows (or lending commitments) from China.

In order to address this challenge, the 3.0 version of AidData’s GCDF dataset includes three new variables
(fields) that measure transaction amounts without including any rollover amounts from PBOC swap line
borrowings or emergency rescue loans from other creditors (with maturities of one year or less).371 These
amounts are reported in their original currencies of denomination, nominal USD, and constant 2021 USD
via the "Adjusted Amount (Original Currency),” "Adjusted Amount (Constant USD 2021)," and "Adjusted
Amount (Nominal USD)" variables.372

Figures 1.3, 1.4, 1.8, 2.8, and A14—as well as Tables 2.1, A15, and A16—use the “Adjusted Amount
(Constant USD 2021)” variable to represent the cumulative stock of official financial flows from China to
countries or borrowing institutions. The other figures and tables in this report use the “Amount (Constant
USD 2021)” field for financial calculations, unless otherwise noted.

Section A-4: Additional details on the BIS-based estimates

of China’s international lending portfolio
In the section, we provide a step-by-step description of how the BIS-based measures of total outstanding
credit from Chinese banks to overseas borrowers are derived. However, before we do so, there are five key
caveats and considerations that readers should keep in mind regarding the BIS cross-border lending
data.373 First, the data are represented as amounts outstanding, which is effectively equivalent to
cumulative disbursements minus cumulative repayments (i.e., credit stocks rather than credit flows). A
separate, but related, point is that cumulative lending commitments usually exceed amounts outstanding.

371
Whenever possible, for each emergency rescue loan (PBOC swap borrowing) of the rollover variety, we
calculate a transaction amount that excludes the rollover amount by taking the difference between the level
of outstanding debt in the current year and the previous year. This approach is consistent with the one
taken by Horn et al. (2023a) to derive net (new) PBOC swap borrowings. In cases when this approach
cannot be applied but there is evidence of the same lender providing a series of short-term emergency
rescue loans (with identical face values and de jure maturities of 1 year or less) to the same borrower that
are repaid on their original contractual maturity dates and subsequently reissued in consecutive years, we
record the face value of the original loan commitment in the first year but not the face values of the loan
commitments in subsequent years.
372
Users of the 3.0 version of AidData’s GCDF dataset can estimate “rollover” loan amounts (in their
original currencies of denomination) by subtracting the values in the Adjusted Amount (Original Currency)
field from the values in the Amount (Original Currency) field. Nominal USD “rollover” loan amounts can be
estimated by subtracting the values in the Adjusted Amount (Nominal USD) field from the values in the
Amount (Nominal USD) field. Constant 2021 USD “rollover” loan amounts can be estimated by subtracting
the values in the Adjusted Amount (Constant USD 2021) field from the values in the Amount (Constant USD
2021) field.
373
We are grateful to Haonan Zhou of Princeton University for his advice on how to utilize the LBS data
from the BIS.
326
Second, the BIS relies on self-reported data from “internationally active banks,” so unlike AidData’s GCDF
dataset it does not capture overseas credit extended by non-bank institutions (e.g., supplier credits from
Chinese companies, loans from China’s Ministry of Commerce). Third, the BIS data are reported by
state-owned and privately-owned banks, but China’s banking sector is dominated by state-owned banks,
which are responsible for nearly all of China’s overseas lending activities (Horn et al. 2021). Fourth, BIS data
can be accessed and analyzed according to the nationality of the reporting institution or the residence of
the counterparty, and since a significant proportion of Chinese bank lending to LICs and MICs is channeled
through offshore financial centers and foreign affiliates of Chinese banks (Cerutti et al. 2018), it is generally
advisable for analysts of China’s overseas lending activities to use the BIS data that are organized according
to the nationality of the reporting institution. Fifth, BIS data rely on the voluntary disclosure efforts of banks,
and Horn et al. (2021) provide evidence of some underreporting to the BIS.374

We now provide a detailed explanation of how we derived the BIS-based measures of total outstanding
credit from Chinese banks to overseas borrowers. BIS (2022) reports the following breakdown of total
outstanding cross-border credit during the second quarter of 2018: $20.27 trillion to HIC borrowers, $4.02
trillion to LIC and MIC borrowers, and $4.76 trillion to offshore financial centers (OFC) borrowers.375 The
LBS data from the BIS also indicate that total outstanding credit from Chinese banks to all overseas
borrowers—including those in LICs, MICs, HICs, and OFCs—was $2.15 trillion during the second quarter of
2018.376 Cerutti et al. (2023) provide complementary evidence from the LBS; they find that, during the
second quarter of 2018, Chinese banks were responsible for 2.4% of total outstanding cross-border credit
to HIC borrowers, 23.7% of total outstanding cross-border credit to LIC and MIC borrowers, and 13.7% of
total outstanding cross-border credit to OFC borrowers.

These figures imply that, during the second quarter of 2018, 22.6% of overseas credit from Chinese banks
was directed to HIC borrowers, while 44.3% was directed to LIC and MIC borrowers and 30.3% was
directed to OFC borrowers.377 Given that the total amounts outstanding under Chinese bank loans to
overseas borrowers are available from the LBS data, one can use these ratios (22.6%, 44.31%, and 30.33%)
to estimate the the total amounts outstanding under Chinese bank loans to HIC borrowers, LIC and MIC
borrowers, and OFC borrowers in previous years (2015-2017) and subsequent years (2019-2021).378 We do
so in the third, fifth, and eighth columns of Table 2.1.

However, something important happened in December 2022: the BIS removed OFCs as a country grouping
from the statistical tables on its public website and reassigned the countries that were previously assigned
to the OFC grouping to the HIC country grouping or the LIC/MIC country grouping.379 The latest vintage of

374
More specifically, they find that their own debt stock estimates (based on AidData and other sources)
“significantly exceed BIS implied debt stocks for some of the riskiest and most volatile debtor countries
worldwide, such as Angola, Equatorial Guinea, Venezuela or Zimbabwe” (Horn et al. 2021: 30).
375
See https://www.bis.org/statistics/rppb2207/intgraphs/ch1graphA2.htm.
376
See https://stats.bis.org/statx/srs/table/a7?c=CN&p=20182.
377
The estimate for AE (HIC) borrowers is based on the following calculation: ($20.27 trillion*0.024)/$2.15
trillion. The estimate for EMDE (LIC and MIC) borrowers is based on the following calculation: ($4.02
trillion*0.237)/$2.15 trillion. The estimate for OFC borrowers is based on the following calculation: ($4.76
trillion*0.137)/$2.15 trillion.
378
These income bracket-level ratios are very slow-moving over time. See
https://www.bis.org/statistics/rppb2304/intgraphs/ch1graphA2.htm.
379
OFCs are legal jurisdictions that serve as intermediaries of cross-border financial flows and specialize in
the provision of banking services to non-residents (Pogliani and Wooldridge 2022). The BIS defines OFCs
as “countries with banking sectors dealing with non-residents and/or in foreign currency on a scale out of
proportion relative to the size of the host economy” (BIS 1995). The BIS identified 21 OFCs as of 2011:
327
the LBS data on total outstanding cross-border credit, which relies upon these updated country groupings,
is therefore useful in that it has created an alternative way of estimating total outstanding cross-border
credit from Chinese banks to LIC/MIC and HIC borrowers.

BIS (2023) provides evidence that total cross-border lending to HIC borrowers was $20.5 trillion, total
cross-border lending to LIC and MIC borrowers was $5.48 trillion, and total cross-border lending to other
borrowers (that could not be assigned to either the HIC or LIC/MIC category) was $3.4 trillion during the
second quarter of 2018.380 These figures—in conjunction with the aforementioned Cerutti et al. (2023)
estimates—imply that during the second quarter of 2018: 22.8% of total cross-border Chinese bank lending
was directed to HIC borrowers, while 60.4% was directed to LIC and MIC borrowers, and 21.6% was
directed to “other” borrowers (i.e., overseas borrowers that cannot be allocated by counterparty
residence).381

Here again, since total amounts outstanding under cross-border Chinese bank loans are available from the
LBS data, one can use these (22.8%, 60.4%, and 21.6%) ratios to estimate the the total amounts
outstanding under Chinese bank loans to HIC borrowers, LIC and MIC borrowers, and “other” borrowers in
previous years (2015-2017) and subsequent years (2019-2021). Table 2.1 provides a lower bound estimate
($1.16 trillion) and an upper bound estimate ($1.58 trillion) of total outstanding cross-border credit from
Chinese banks to LIC and MIC borrowers as of 2021 as well as a lower bound estimate ($594.6 billion) and
an upper bound estimate ($599.9 billion) of total outstanding cross-border credit from Chinese banks to
HIC borrowers in 2021.382 Regardless of whether one uses lower bound or upper bound estimates, Table
2.1 demonstrates that total outstanding credit from Chinese banks to LIC and MIC borrowers effectively
doubled between 2015 and 2021 (either from $644 billion to $1.16 trillion or from $878 billion to $1.58
trillion). This topline pattern is remarkably consistent with the GCDF data on China’s cumulative overseas
lending commitments between 2015 and 2021, which increased (in nominal terms) from $620 billion in
2015 to $1.03 trillion in 2021.383

Aruba, Bahamas, Bahrain, Barbados, Bermuda, British Virgin Islands (VG) and West Indies UK, Cayman
Islands, Gibraltar, Guernsey, Hong Kong SAR, Isle of Man, Jersey, Lebanon, Liberia, Macao SAR, Mauritius,
Netherlands Antilles (Curaçao and Sint Maarten), Panama, Samoa, Singapore, and Vanuatu. The BIS did not
identify VG as a stand-alone jurisdiction; instead, it is included in a larger, BIS-defined country grouping
called “West Indies UK” that includes Anguilla, Antigua and Barbuda, Montserrat, and St Kitts and Nevis.
380
See https://www.bis.org/statistics/rppb2304/intgraphs/ch1graphA2.htm.
381
The estimate for AE (HIC) borrowers is based on the following calculation: ($20.5 trillion*0.024)/$2.15
trillion. The estimate for EMDE (LIC and MIC) borrowers is based on the following calculation: ($5.48
trillion*0.237)/$2.15 trillion. The estimate for “other” borrowers is based on the following calculation: ($3.4
trillion*0.137)/$2.15 trillion.
382
These estimates are reported in the second, third, fifth, and sixth columns of Table 2.1. The EMDE (LIC
and MIC) estimates are based on a lower bound assumption (that 44.31% of total cross-border Chinese
bank lending was directed to EMDE borrowers) and an upper bound assumption (that 60.4% of total
cross-border Chinese bank lending was directed to EMDE borrowers). The AE (HIC) estimates are based on
a lower bound assumption (that 22.6% of total cross-border Chinese bank lending was directed to AE
borrowers) and an upper bound assumption (that 22.8% of total cross-border Chinese bank lending was
directed to AE borrowers). Table 2.1 also indicates that total outstanding cross-border credit from Chinese
banks to OFC borrowers in 2021 was approximately $798 billion and total outstanding cross-border credit
from Chinese banks to “other” borrowers in 2021 was approximately $568.3 billion. All of these figures are
based on aggregate BIS data that measure total outstanding cross-border credit from Chinese banks in the
final quarter of each year between 2015 and 2021. See
https://stats.bis.org/statx/srs/table/a7?c=CN&p=20154
383
For comparability’s sake, these figures are reported in current (nominal) USD. However, the 3.0 version of
AidData’s GCDF dataset demonstrates that China’s cumulative overseas lending commitments increased
from $823 billion in 2015 to $1.3 trillion in 2021 in constant 2021 USD.
328
To be clear, the time-varying, BIS-based estimates rest upon an important assumption: that the geographic
allocation of outstanding credit from Chinese banks across income groups (22.8% to borrowers in HICs,
60.4% to borrowers in LICs/MICs and 21.6% to “other” borrowers) was stable between 2015 and 2021. If
this assumption is not true (or at least not for the most part true), an alternative explanation for the major
increase in China’s total outstanding credit to borrowers in LICs, MICs, HICs, and “other” overseas
jurisdictions (captured in the ninth column of Table 2.1)—from $1.45 trillion in 2015 to $2.63 trillion in
2021—could be that lending to borrowers in HICs and/or “other” overseas jurisdictions outpaced lending
to LICs/MICs between 2015 and 2021. However, we think that this alternative explanation is implausible.
AidData recently collected data on all official sector lending commitments from China to borrowers in 26
HICs and the preliminary summary statistics show that annual lending commitment to HICs declined from
2015 to 2021 (see Figure A63). The data also demonstrate that changes in official sector lending
commitments from China to HICs and LICs/MICs have generally moved in tandem over time. Between
2015 and 2021, China’s official sector lending commitments to LICs/MICs declined by 38.5%. China’s
official sector lending commitments to 26 HICs declined by 36.7% over the same six-year period.384

Figure A63: Official sector PRC lending to LICs and MICs versus Y1 HICs

Figure A.63

Notes: This figure presents annual lending commitments from official sector institutions in China to two
cohorts between 2000 and 2021: (1) the recipient countries captured in the 3.0 version of AidData’s GCDF
dataset, which covers 165 low-income and middle-income countries, and (2) twenty-seven high-income
countries.385 The data on high-income countries represents preliminary data collected by AidData as part of

384
Over the next two years, AidData expects to achieve comprehensive coverage of all official sector
lending and grant-giving commitments from China to HICs between 2000-2021. Given the important but
underappreciated role that OFCs play in China’s overseas lending program, AidData also intends to allow
users of its data to differentiate between official sector financial flows from China that travel to versus
through countries.
385
The 3.0 version of AidData’s GCDF dataset covers a total of 165 countries and territories. It specifically
covers all countries that are or have been classified as low- or middle-income countries during the
2000-2021 time period.

329
an ongoing data collection effort to capture all official sector lending commitments from China to all
high-income countries between 2000 and 2021.

Section A-5: How does AidData categorize Chinese lending

to different types of borrowers?
The “Level of Public Liability” field in the 3.0 version of AidData’s GCDF dataset captures the extent to
which the host government may eventually be liable for debt repayment. It is hierarchically and
automatically determined based on the following criteria:

1. The loan record is classified as "Central government debt" if it is an official sector loan to a central
government institution in the recipient country, measured by whether there is at least one
receiving agency (direct or indirect) from the recipient country that is classified as a government
agency;
2. If the loan record does not meet the first (1) criterion, it is classified as "Central
government-guaranteed debt" if it is an official sector loan to a state-owned entity (e.g.,
state-owned enterprise and state-owned bank) or privately-owned entity in the recipient country
that benefits from a sovereign (central government) repayment guarantee;
3. If the loan record does not meet the first (1) criterion or the second (2) criterion, it is classified as
"Other public sector debt" if (a) it is an official sector loan to a state-owned entity (such as a
city/municipal government, a state-owned bank, or a state-owned enterprise) in the recipient
country that does not benefit from a sovereign (central government) repayment guarantee; (b) it is
an official sector loan to a private entity or state-owned entity in the recipient country that is
backed by a repayment guarantee from a state-owned entity other than the central government in
the recipient country (such as a city/municipal government, a state-owned bank, or a state-owned
enterprise), OR (c) it is an official sector loan to a special purpose vehicle (SPV) or joint venture (JV)
that is majority-owned by one or more public sector institutions in the recipient country and that
does not benefit from a sovereign (central government) repayment guarantee or a repayment
guarantee from a state-owned entity other than the central government in the recipient country
(such as a city/municipal government, a state-owned bank, or a state-owned enterprise).
4. If the loan record does not meet the first (1) criterion, the second (2) criterion, or the third (3)
criterion, it is classified as "Potential public sector debt" if it is an official sector loan to a special
purpose vehicle (SPV) or joint venture (JV) borrower that is minority-owned by one or more public
sector institutions in the recipient country and that does not benefit from a sovereign (central
government) repayment guarantee or a repayment guarantee from a state-owned entity other
than the central government in the recipient country (such as a city/municipal government, a
state-owned bank, or a state-owned enterprise).
5. If the loan record does not meet the first (1) criterion, the second (2) criterion, the third (3)
criterion, and the fourth (4) criterion, it is classified as "Private debt" if it is an official sector loan to
a privately-owned entity that does not benefit from a repayment guarantee from a public sector
institution in the recipient country (this includes lending to a private entity, or lending to a Joint
Venture or Special Purpose Vehicle with no level of host government ownership (i.e., the "JV/SPV
Host Government Ownership" variable is set to "No Host Government Ownership";
6. If the loan record does not meet the first (1) criterion, the second (2) criterion, the third (3)
criterion, the fourth (4) criterion, or the fifth (5) criterion, then it is classified as "Unallocable" due
to a lack of information.

330
Section A-6: How did we identify projects that rely on
contractors sanctioned for fraudulent and corrupt behavior?
As part of our broader effort to measure the nature and extent of governance risk in China’s overseas
infrastructure project portfolio, we sought to identify the extent to which Chinese lenders and donors relied
on contractors formally debarred for fraudulent and corrupt behavior to implement their grant- and
loan-financed infrastructure projects in LICs and MICs. Debarment is a type of sanction imposed on firms or
individuals, which prohibits the entity in question from participating in future or current project preparation
or implementation. MDBs regularly investigate reports of wrongdoing—such as fraud, corruption, and
collusion—by entities involved in their projects. If the investigation confirms the firm or individual
participated in the alleged wrongdoing, a sanctions board or committee debars that firm or individual,
rendering them ineligible to “be awarded or otherwise benefit from a Bank-financed contract.”386 Each
debarment is usually enforced for a specific period of time, with the length of debarment proportional to
the severity of the wrongdoing in which the firm or individual participated. In extreme cases, a firm or
individual may be debarred indefinitely. While MDBs maintain their own list of debarred institutions, a
growing group of MDBs have agreed to uphold the debarments from each other’s lists (an approach that is
known as “cross-debarment”).387

To identify the subset of Chinese grant- and loan-financed infrastructure projects that pose elevated levels
of governance risk, we began by collecting current and historical information on firms that have been
debarred by MDBs since 2004 (when the earliest data is available). We collected the name of each firm
debarred and the length of each debarment from World Bank Sanctions Systems Annual Reports from
2004-2022. These annual reports provide information on World Bank historical debarment beginning in
2004 as well as any cross-debarments from the other MDBs (the African Development Bank, the Asian
Development Bank, the European Bank for Reconstruction and Development, the Inter-American
Development Bank, and the Asian Infrastructure Investment Bank) beginning in 2010. We then combined
this list with a list of current debarments and cross-debarments maintained by the World Bank on its
website (World Bank 2023).388 We subsequently matched the list of debarred firms with a list of all the firms
that are (or were) directly involved in Chinese grant- and loan-financed infrastructure projects—specifically,
any firm listed as an implementing agency or a receiving agency for a infrastructure project in the 3.0
version of AidData’s GCDF dataset. In total, we found 21 formally debarred organizations that were
involved in 453 Chinese grant- and loan-financed infrastructure projects.

After isolating the projects that involved debarred organizations, we sought to identify the subset of
projects that relied upon debarred organizations while they were still within their debarment periods. We
did so by identifying all cases in which there was calendar day overlap between the start and end dates of
an organization’s debarment period and the commitment, implementation, or completion dates of the
project(s) it supported. After eliminating all projects that involved the debarred agency before or after their
official debarment periods, we identified 324 Chinese grant- and loan-financed infrastructure projects that

386
According to the World Bank, debarment is defined as “[t]he sanctioned party is declared ineligible,
either indefinitely or for a stated period of time, (1) to be awarded or otherwise benefit from a
Bank-financed contract, financially or in any other manner; (2) to be a nominated sub-contractor, consultant,
manufacturer or supplier, or service provider of an otherwise eligible firm being awarded a Bank-financed
contract; and (3) to receive the proceeds of any loan made by the Bank or otherwise to participate further in
the preparation or implementation of any Bank-Financed Project” (World Bank 2012).
387
The World Bank, the African Development Bank, the Asian Development Bank, the European Bank for
Reconstruction and Development, the Inter-American Development Bank, and most recently the Asian
Infrastructure Investment Bank have signed the "Agreement on Mutual Enforcement of Debarment
Decisions.”
388
We compared similar online resources from the other MDBs participating in cross-debarment and found
the World Bank’s to be the most comprehensive.
331
involved the debarred agency during the active debarment period (either as an implementing agency or a
receiving agency). Table A2 provides a list of the debarred agencies and associated details.

Table A2: Contractors sanctioned for fraudulent and corrupt behavior that were involved in Chinese ODA- and OOF-financed infrastructure projects during their debarment periods

Table A.2

Annual loan commitments to LICs and MICs by financial institution and capital
injections from SAFE, 2000-2021
Beginning
Grounds for End of Debarring
Firm Name of
debarment Debarment Institution
Debarment
Asian
China CAMC Engineering Co., LTD. Fraudulent practices 03/09/2022 10/26/2026 Development
Bank
China Energy Engineering Group Hunan
Fraudulent practices 07/01/2019 05/10/2021 World Bank
Electric Power Design Institute Co., Ltd.
African
Fraudulent and
China First Highway Engineering Co. Ltd 07/01/2014 06/30/2018 Development
collusive practices
Bank
China First Metallurgical Construction
Fraudulent practices 09/28/2011 09/27/2014 World Bank
Corporation (CFMCC)

China Geo-Engineering Corporation Fraudulent practices 01/28/2009 1/27/2014 World Bank

China International Water & Electric Corp. Fraudulent practices 09/24/2014 09/24/2017 World Bank

China Jiangsu International Economic And

Fraudulent practices 02/14/2014 2/13/2017 World Bank
Technical Cooperation Group Ltd.

China Jiangxi Corporation For International

Fraudulent practices 04/15/2014 4/15/2015 World Bank
Economic And Technical Cooperation

China Machinery Industry Construction Group

Fraudulent practices 07/18/2018 07/17/2022 World Bank
Inc. (also known as SINOCONSTRUCTION)

China Nuclear Industry Fifth Construction Co.

Fraudulent practices 07/19/2018 07/18/2020 World Bank
Ltd.

China Railway 20 Bureau Group Co. Fraudulent practices 06/26/2017 12/31/2017 World Bank

China Railway Construction Corporation

Fraudulent practices 06/05/2019 02/04/2020 World Bank
(International) Limited

China Railway First Group Co. Ltd. Fraudulent practices 09/18/2019 08/18/2021 World Bank

China State Construction Engineering

Fraudulent practices 01/14/2009 1/13/2015 World Bank
Corporation (CSCEC)

Fraudulent and
China Wuyi Co. Ltd. 01/14/2009 1/13/2015 World Bank
collusive practices

Shandong Taikai Power Engineering Company

Fraudulent practices 08/18/2015 02/17/2017 World Bank
Limited

Fraudulent and
SNC-Lavalin Group Inc. 08/18/2015 02/17/2017 World Bank
corrupt practices

332
 Beginning
Grounds for End of Debarring
Firm Name of
debarment Debarment Institution
Debarment
Inter-American
Fraudulent and
Tractebel Engineering S.A. 12/23/2021 09/28/2025 Development
corrupt practices
Bank

Zhengtai Group Co., Ltd. Fraudulent practices 05/05/2017 08/05/2018 World Bank

Zhonghao Overseas Construction Eng. Co., Ltd. Fraudulent practices 06/02/2012 06/01/2014 World Bank

Norconsult Corrupt practices 07/01/2013 12/31/2014 World Bank

Notes: This table lists contractors that (1) were sanctioned by multilateral development banks for fraudulent
and corrupt behavior, and (2) during the time of the contractor’s debarment, they were identified as a
receiving or implementing agency on a Chinese-financed infrastructure project. See Section 2 of Chapter 3
for more details on how these firms were identified. The World Bank, the African Development Bank, the
Asian Development Bank, the European Bank for Reconstruction and Development, the Inter-American
Development Bank, and the Asian Infrastructure Investment Bank have a formal agreement to cross-debar
firms.

Section A-7: How did we identify ESG risks and risk

mitigation measures in AidData project descriptions?
In order to support the analysis in chapter 3, we constructed three binary variables that identify whether
there is any narrative evidence from the 3.0 version of AidData’s GCDF dataset that a given infrastructure
project encountered a significant environmental problem, social problem, or governance problem before,
during, or after implementation. To do so, we identified a set of keywords for each type of challenge
(environmental, social, or governance) that, if present in a project’s “description” field within the 3.0
dataset, may indicate the existence of such a problem. We applied these keyword searches to all projects
backed by official commitments (including those that were subsequently suspended or canceled) and all
implemented and completed projects.389 After identifying the subset of projects with “description” fields
that contain one or more of the pre-specified keywords, we manually reviewed each of the corresponding
project descriptions to eliminate “false positives”—e.g., a description that references the keyword “lawsuit”
to describe environmental litigation affecting an earlier phase of a project backed by a non-Chinese
financier but not the subsequent, Chinese government-financed phase of the project. More specifically, we
confirmed or disconfirmed that the projects in question encountered the types of problems that we
intended to identify with the prespecified set of keywords. However, at the manual review stage, we did not
seek to differentiate between problems that were effectively or ineffectively managed. Nor did we seek to
independently evaluate the veracity of any reported instances of harm, wrongdoing, or poor performance.
Our goal was to identify which types of problems arose during implementation (regardless of whether a
mitigation response was undertaken). However, we were not able to account for “false negatives” (i.e.,
cases in which a specific type of problem arose during project implementation, but the problem was not

389
We also restricted our searches to infrastructure projects supported by grant and loan commitments
worth $20 million (in constant 2021 USD) or more. Projects supported by larger financial commitments
generally have more detailed project descriptions, which provide a stronger basis for the identification of
environmental, social, and governance risks and risk mitigation efforts. They also present a lower risk of
generating “false negatives.”
333
captured in the “description” field within AidData’s 3.0 dataset), so the summary statistics that we report
should be treated as lower-bound estimates.

We identified the subset of infrastructure projects that encountered significant environmental

problems—before, during, or after implementation—by first applying the following keyword search terms
to the infrastructure project descriptions in the 3.0 dataset: disaster, destroy*, destruction, ruin*, despoil,
degrad*, pollut*, dispos*, displace*, danger*, hazard*, exploit*, involuntary, unsafe, crack*, substandard,
low-quality, deficien*, defect*, ecology*, ecosystem, habitat, biodiverse*, flora, fauna, species, non-native,
invasive, pesticide*, chemical*, cancer*, discharge*, emission*, emit*, subcritical, odor, smell. We then
reviewed each of the identified project descriptions “by hand” to remove false positives.

We followed an analogous set of procedures to identify those infrastructure projects that encountered
significant social challenges and governance problems, respectively. To construct a binary, project-level
measure of exposure to social problems, we applied the following keyword search terms: strike*, protest*,
riot*, violat*, noncomplian*, non-complian*, involuntary, forced, evict* injur*, noise, vibrat*, nuisance, dead,
death, died, kill*, harm*, unsafe, crack*, substandard, low-quality, defect*, deficien*, danger*, indigenous,
burial, sacred, spiritual, ritual*, religio*, heritage, overcrowd*, displace*, grievanc*, underserve*,
disadvantage*, minorit*, aboriginal, tribe, vulnerab*, marginalize*, ethnic, archaeol*, custom*, manipul*,
interfer*, coerc*, discriminat*, intimidat*, workplace, layoff, fired, underpaid, unpaid, wages, “working
conditions”, abuse*, resettle*, aggrieved, sexual. To construct a binary, project-level measure of exposure
to governance problems, we applied the following keyword search terms: misuse*, abuse*, bid-rig*,
misappropriat*, mismanage*, steal, theft, stole*, corrupt*, bribe*, graft, fraud*, kickback, siphon*,
embezzle*, illicit, illegal, inflat*, overprice*, over-price*, wrongdoing, collusion, collusive, collude*, loot*,
plunder*, abuse*, obstructive, defraud*, fictitious, launder*. We then manually reviewed each of the
identified project descriptions to remove false positives.

In addition to keywords that are specific to environmental, social, or governance challenges, we identified a
set of keywords that could indicate a challenge in any of the three categories. We used these cross-cutting
keywords to identify additional projects requiring manual review for a potential environmental, social, or
governance challenge. For this “cross-cutting” keyword search, we applied the following search terms:
allege*, allegation, criticiz*, alarm, criticism, complain*, controvers*, fiasco, turmoil, breach*, probe*,
irresponsib*, audit*, inquiry, scrutin*, uncover*, scandal, dispute*, fined, sued, lawsuit, arbitrat*,
adjudicate*, court, litigat*, rescope*, jeopard*, rescind*, revoke*, failure, delay*, threat*, renege*,
renegotiat*, “conditions precedent”, *indemnif*, adverse, “liable”.

We also constructed three binary variables that identify whether there is any narrative evidence from the 3.0
version of AidData’s GCDF dataset that efforts were undertaken by Chinese financiers or implementing
agencies to mitigate environmental, social, or governance risks before, during, or after project
implementation.390 These variables seek to measure whether ESG risk mitigation measures were undertaken
by Chinese financiers or implementing agencies, irrespective of whether or not such measures were
successful.391 Consistent with the approach that was used to identify infrastructure projects that

390
Here too we restricted our searches to infrastructure projects supported by grant and loan commitments
worth $20 million (in constant 2021 USD) or more.
391
We only sought to identify ESG risk mitigation measures that involved choice or consent from the
Chinese side. Therefore, if a Chinese financier or implementing agency was compelled by an entity outside
of China to take an ESG risk mitigation measure (i.e., a Chinese company does not implement an
334
encountered significant ESG challenges, we construct three binary, project-level measures of environmental
risk mitigation, social risk mitigation, and governance risk mitigation by applying the following keyword
search terms:

● Environmental risk mitigation: remedia*, compensat*; resettle*, reclaim*, reclamation, restor*,

restitution, certify*, permit*, feedback, disclosure, amend*, minimize,consent*, FPIC, safeguard,
mandat*, mitigat*, heritage, consult*, dialogue, “public hearing”, “Environmental Assessment”,
“EA”, “Environmental Impact Assessment”, “EIA”, “Environmental and Social Impact
Assessment”, “ESIA”, “Resettlement action plan”, “RAP”, “Environmental Management Plan”,
“EMP”, “Project Affected Persons”, “PAPs”, “Project-Affected”, “Environmental Categorization”,
“Full Resettlement Plan”, “FRP”, “compliance certificate”, “ECC”, “Environmental and Social
Policy”, “E&S”, “Stakeholder Engagement Plan”, “SEP”, “Due Diligence”, “ESDD”,
“Environmental, Health, Safety and Social Management System”, “EHSSMS”, “Environmental and
Social Action Plan”, “ESAP”, “ESCP, “protected area”, "Equator Principles", desulphurization,
desulphurization, “FGD”, super-critical, “Initial Environmental Examination”, “IEE”,
“Environmental indemnity”, “Environmental claim”
● Social risk mitigation: compensat*; resettle*, reclaim*, reclamation, restor*, restitution, consent*,
FPIC, safeguard, mandat*, mitigat*, certify*, permit*, feedback, disclosure, amend*, minimize,
heritage, consult*, dialogue, “public hearing”, “health and safety”, “Resettlement action plan”,
“RAP”, “Full Resettlement Plan”, “FRP”, “Environmental and Social Impact Assessment”, “ESIA”,
“compliance certificate”, “Environmental and Social Policy”, “E&S”, “Stakeholder Engagement
Plan”, “SEP”, “Due Diligence”, “ESDD”, “Environmental, Health, Safety and Social Management
System”, “EHSSMS”, “Environmental and Social Action Plan”, “ESAP”, “ESCP”, occupational
health, “OHS”, "Equator Principles", , “Gender Action Plan”, “collective bargaining”, “Safeguards
Monitoring Report”
● Government risk mitigation: debar*, cross-debarment, sanction*, blacklist*, prosecut*, arrest,
convict*, criminal, extradit*, imprison, anticorruption, anti-corruption, value-for-money, freeze,
froze*, halt*, suspend*, suspension, cancel*, withheld, withhold, abandon*, postpon*, abeyance,
mothball*, forensic, prohibit*, “Open Competitive Bidding”, OCB, “International Competitive
Bidding”, ICB, “Competitive Bidding”, IFRS, “International Financial Reporting Standards”,
“GAAP”, “know your customer”, “generally accepted accounting principles”, “anti-money
laundering”

We then manually reviewed each of the identified project descriptions to remove false positives.

Section A-8: How did AidData identify ESG risk mitigation

measures in infrastructure financing agreements?
In order to benchmark the ESG risk management provisions in the infrastructure financing agreements of
Chinese state-owned lenders and donors, we first developed a standardized set of 26 evaluation criteria (or

Environmental Impact Assessment until it is compelled to do so by a court order issued by a judge in the
recipient country), it does not count as an environmental risk mitigation measure because of the absence of
choice/consent by the Chinese company. Similarly, if a Chinese company employee is convicted on
corruption charges in the recipient country, this does not count as a governance risk mitigation measure
because of the absence of choice/consent on the Chinese side.
335
diagnostic questions). The criteria are organized into three groups: those that identify the presence or
absence of (1) rules or standards to establish behavioral expectations related to ESG risk management and
mitigation, (2) oversight mechanisms for monitoring compliance with those behavioral expectations; and/or
(3) enforcement mechanisms for sanctioning noncompliance with those behavioral expectations (e.g.,
indemnification, withholding disbursements). 8 of the criteria are specific to environmental safeguards,
while 7 are specific to social safeguards, and 12 are specific to governance safeguards.

The criteria that we used to evaluate the 23 contracts in the coding sample are provided below in Table A3
(and organized according to safeguard type and whether the evaluation criteria relates to standards,
monitoring, or enforcement).

Table A3: Criteria to evaluate environmental, social, and governance safeguards in Chinese grant and loan contracts

Table A.3

Criteria to evaluate environmental, social, and governance safeguards in

Chinese grant and loan contracts
Question Question Type Safeguard Type
Are any environmental clauses or conditions included in the Standards Environmental
agreement?
Is an Environmental Impact Assessment (EIA) identified as a Standards Environmental
requirement in the agreement?
Is an Environmental Management Plan (EMP) – or a functional Standards Environmental
equivalent of an EMP – identified as a requirement in the
agreement?
Is the borrower required to report to the lender on its Monitoring Environmental
implementation of the EIA recommendations or EMP?
Does the agreement include any environmental conditions Enforcement Environmental
precedent (for entry into force or initial or ongoing
disbursement)?
Does the agreement specify that noncompliance with Enforcement Environmental
environmental rules, standards or laws is a sufficient basis for the
lender to demand early repayment (“prepayment”) of the loan
or cancel the loan?
Does the agreement require compliance with international Standards Environmental
environmental standards, such as the Equator Principles or the
OECD Revised Council Recommendation on Common
Approaches on the Environment and Officially Supported Export
Credits?
Is the borrower required to indemnify the lenders, guarantors, or Enforcement Environmental
insurers ("Finance Parties") against any loss or liability incurred
by the Finance Parties as a result of any actual or alleged breach
of an environmental law or standard?
Are any social clauses or conditions included? Standards Social
Is an Environmental and Social Impact Assessment (ESIA) Standards Social
identified as a requirement?

336
Is an Environmental and Social Action Plan (ESAP), a Standards Social
Resettlement Action Plan (RAP), or the functional equivalent of
an ESAP or RAP identified as a requirement?
Is the borrower required to report to the lender on its Monitoring Social
implementation of the ESIA recommendations or RAP?
Does the agreement include any social conditions precedent (for Enforcement Social
entry into force or initial or ongoing disbursement)?
Does the agreement indicate the penalty of violating social Enforcement Social
standards/laws leads to demanding early prepayment to close
the loan?
Is the borrower required to indemnify the lenders, guarantors, or Enforcement Social
insurers ("Finance Parties") against any loss or liability incurred
by the Finance Parties as a result of any actual or alleged breach
of a social law or standard?
Are any governance (e.g., anti-corruption, competitive bidding, Standards Governance
audited financial statements, and/or anti-money laundering)
clauses or conditions present in the agreement?
Are independently audited financial statements required of the Standards Governance
borrower?
Must the borrower’s financial statements comply with Standards Governance
International Financial Reporting Standards (IFRS) standards?
Does the agreement include any competitive bidding Standards Governance
requirements?
Are bidding documents or bid evaluation reports subject to prior Monitoring Governance
approval by the lender?
Does the agreement include any anti-corruption or anti-money Standards Governance
laundering requirements?
Is the borrower required to indemnify the lenders, guarantors, or Enforcement Governance
insurers ("Finance Parties") against any loss or liability incurred
by the Finance Parties as a result of any actual or alleged breach
of an anti-corruption or anti-money laundering law or standard?
Does the lender explicitly reserve the right to prevent or Monitoring Governance
investigate anti-corruption or anti-money laundering crimes?
Does the borrower affirm that it will seek to ensure that the Monitoring Governance
proceeds from the loan or grant are not used to finance or
benefit any internationally sanctioned entity?
Does the agreement specify that noncompliance with Enforcement Governance
governance rules, standards, or laws is a sufficient basis for the
lender to demand early repayment (“prepayment”) of the loan
or cancel the loan?
Does the agreement include any competitive bidding, Enforcement Governance
anti-corruption, or anti-money laundering conditions precedent
(for entry into force, initial disbursement, or ongoing
disbursements)?

337
Section A-9: Assigning ESG Safeguard Stringency Ratings to
Infrastructure Financing Agreements and the 8 Infrastructure
Financing Instrument Categories
To assess the presence or absence of each safeguard type (standards, monitoring, and enforcement) within
each infrastructure financing agreement (“contract”), we examine the binary (yes/no) responses to 26
diagnostic questions that are categorized by safeguard type (see Table A8).392 For each contract, if we
identify “yes” responses to any questions related to standards in a given ESG domain (environmental,
social, or governance), then we determine rules and standards to be present in that particular ESG
domain-contract dyad. Conversely, if we identify “no” responses to all questions related to standards in a
given ESG domain (environmental, social, or governance), then we determine rules and standards to be
absent in that particular ESG domain-contract dyad. The same criteria are applied to the questions related
to monitoring and the questions related to enforcement in a given ESG domain. Table A4 records all of the
yes/no determinations at the level of ESG domain-contract dyads.

Then, for each ESG domain-contract dyad, we assign high, medium, or low ratings based upon the
following criteria:

● Low: Contracts lacking established standards, monitoring mechanisms, and enforcement

mechanisms.
● Medium: Contracts that stipulate standards, but lack monitoring mechanisms or enforcement
mechanisms.
● High: Contracts that stipulate standards and incorporate either monitoring mechanisms or
enforcement mechanisms.

Table A4 records all of the ESG domain-contract dyad ratings. Next, to account for changes in ESG
safeguard stringency within each financial instrument category over time, we segment our ESG
domain-contract dyad ratings into two distinct periods: 2000-2017 (pre- and early BRI period) and
2018-2022 (late BRI period). We generate summary (high/medium/low) ratings for the environmental
safeguards, social safeguards, and governance safeguards that applied to each financial instrument
category over each time period (2000-2017 and 2018-2021). We do so by following a "most frequent
designation" decision rule. If the majority of contracts within a financial instrument category share the same
rating for a given ESG domain, the most frequent rating designation is applied to the entire financial
instrument category. However, in cases where two contracts within the same time period and financial
instrument category are assigned different ratings for a given ESG domain, the higher rating is applied to
the financial instrument category as a whole.

For various types of analysis, we use the distinction between “strong” and “weak” de jure ESG safeguards.
Projects with “strong” de jure ESG safeguards are defined as those with at least two out of three

392
The “standards” measures seek to identify whether the contract identifies rules or standards that create
behavioral expectations related to ESG risk management and mitigation. The “monitoring” measures seek
to identify whether the contract identifies oversight mechanisms for monitoring compliance with those
behavioral expectations. The “enforcement” measures seek to identify whether the contract identifies
enforcement mechanisms for sanctioning noncompliance with those behavioral expectations.
338
 (environmental, social and governance) safeguard stringency ratings that are “high.” Projects that do not
meet this standard are classified as having “weak” de jure ESG safeguards.

Table A4: Safeguard stringency ratings for infrastructure financing agreements at the ESG domain-contract dyad level

Table A.4

Safeguard stringency ratings for infrastructure financing agreements at the ESG

domain-contract dyad level
Financial
Instrument Contract ESG Domain Standards Monitoring Enforcement Level
Category

Environmental No No No Low
Mombasa-Nairobi Standard Gauge Railway Project
Social No No No Low
(Kenya)
Governance No No No Low

Environmental No No No Low
Bilateral China
New Power Plant Project (Antigua and Barbuda) Social No No No Low
Eximbank loan
Governance No No No Low

Environmental No No No Low
New Centennial Water Source-Kaliwa Dam Project
Social No No No Low
(Philippines)
Governance No No No Low

Environmental No No No Low
Term Facility for Infrastructure and Social
Social No No No Low
Development Projects (Costa Rica)
Governance Yes No Yes High

Environmental Yes No Yes High

Bilateral CDB
Ship Construction Project (Marshall Islands) Social No No No Low
loan
Governance Yes No No Medium

Environmental No No No Low
Term Facility Agreement for Infrastructure Projects
Social No No No Low
(Ecuador)
Governance No No No Low

Environmental Yes No No Medium

Project for the Realization of the Bridge Renovation
Social Yes No No Medium
on the River Tara (Montenegro)
Governance Yes No No Medium
Bilateral
MOFCOM loan Environmental No No No Low
or grant National Emergency Operations Center (COEN)
Social No No No Low
Project (Peru)
Governance No No No Low

Port Infrastructure Construction Project (Mongolia) Environmental No No No Low

339
Financial
Instrument Contract ESG Domain Standards Monitoring Enforcement Level
Category

Social No No No Low

Governance Yes Yes No High

Environmental Yes No Yes High

Masindi-Biiso, Kibaale-Kiziranfumbi and
Hohwa-Nyairongo-Kyarusesa-Butoole Road Social No No No Low
Upgrading Project (Uganda)
Governance Yes Yes Yes High
Bilateral
Environmental Yes No No Medium
Chinese
state-owned Electrification Project (Ghana) Social No No No Low
commercial
Governance Yes No Yes High
bank loan
Environmental Yes No Yes High
Patuca III Hydroelectric Plant Project (Piedras
Social Yes No Yes High
Amarillas) Phase II (Honduras)
Governance Yes No No Medium

Environmental Yes Yes Yes High

Syndicated
Ituango Hydroelectric Project (Colombia) Social Yes Yes Yes High
loan with
Chinese and Governance Yes Yes Yes High
multilateral Environmental Yes Yes Yes High
bank
Castellana Wind Power Project (Argentina) Social Yes Yes Yes High
participants
Governance Yes Yes Yes High

Environmental Yes No Yes High

Works on Construction of the Municipal (Sewage)
Social No No No Low
Infrastructure Project (Serbia)
Syndicated Governance Yes Yes Yes High
loan with
Environmental Yes No Yes High
Chinese
President Dr Néstor Carlos Kirchner and Gobernador
state-owned Social No No No Low
Jorge Cepernic Hydropower Project (Argentina)
commercial
Governance Yes No No Medium
banks and/or
policy banks Environmental Yes No Yes High
Queen Elizabeth ll Quay at Freetown Port Project
Social Yes No Yes High
(Sierra Leone)
Governance Yes Yes Yes High

PBOC/MOF Environmental Yes Yes Yes High

grant or loan Infrastructure Transport Program (PIT) (Costa Rica) Social Yes Yes Yes High
channeled
Governance Yes Yes Yes High
through
multilateral Sustainable Water Supply and Sanitation Program Environmental Yes Yes No High
institutions (Rwanda) Social Yes Yes Yes High

340
Financial
Instrument Contract ESG Domain Standards Monitoring Enforcement Level
Category

Governance Yes Yes Yes High

Environmental Yes Yes Yes High

Water Supply Scheme for Tete Settlement Project
Social Yes No Yes High
(Papua New Guinea)
Governance Yes Yes Yes High

Environmental Yes No Yes High

Regional Hospitals Project – Lot 3 (Guyana) Social No No No Low

Governance Yes No Yes High

Environmental Yes No Yes High

Supplier's
Phase 1 Lot 5 Adenta-Dodowa Dual Carriageway
credit from Social No No No Low
Project (Ghana)
Chinese SOE
Governance Yes No Yes High

Environmental No No No Low
Energy Transmission Network Construction Project
Social No No No Low
Associated with the Imboulou Power Plant (Congo)
Governance No No No Low

Table A5: Sample of infrastructure financing agreements used to code the de jure stringency of ESG safeguards

Table A.5

Sample of infrastructure financing agreements used to code the de jure

stringency of ESG safeguards
Financial
Commitm
Instrument Project Title Lender/Donor Borrower/Recipient
ent Year
Category

Mombasa-Nairobi Standard Gauge Government of

2014 Export-Import Bank of China
Railway Project [hyperlink] Kenya

Ministry of Finance
and Economy of
2008 New Power Plant Project [hyperlink] Export-Import Bank of China
Bilateral China Antigua and
Eximbank loan Barbuda

Metropolitan
The New Centennial Water Waterworks and
2018 Export-Import Bank of China
Source-Kaliwa Dam Project [hyperlink] Sewerage System of
the Philippines

341
Financial
Commitm
Instrument Project Title Lender/Donor Borrower/Recipient
ent Year
Category

Term Facility for Infrastructure and Social National Bank of

2008 China Development Bank
Development Projects [hyperlink] Costa Rica

Nereus Navigation
Bilateral CDB loan 2013 Ship Construction Project [hyperlink] China Development Bank Ltd. and Irises
Shipping Ltd

Term Facility Agreement for Government of

2016 China Development Bank
Infrastructure Projects [hyperlink] Ecuador

Project for the Realization of the Bridge Government of

2020 China Ministry of Commerce
Renovation on the River Tara [hyperlink] Montenegro

Bilateral MOFCOM National Emergency Operations Center

2014 China Ministry of Commerce Government of Peru
loan or grant (COEN) Project [hyperlink]

Port Infrastructure Construction Project Government of

2019 China Ministry of Commerce
[hyperlink] Mongolia

Ministry of Finance,
Masindi-Biiso, Kibaale-Kiziranfumbi and Planning and
2021 Hohwa-Nyairongo-Kyarusesa-Butoole China Construction Bank Economic
Road Upgrading Project [hyperlink] Development of the
Bilateral Chinese
Republic of Uganda
state-owned
commercial bank Industrial and Commercial Bank Ministry of Finance
2016 Electrification Project [hyperlink]
loan of China of Ghana

Patuca III Hydroelectric Plant Project Empresa Nacional

Industrial and Commercial Bank
2013 (Piedras Amarillas) Phase II [hyperlink] de Energía Eléctrica
of China
[hyperlink] of Honduras

People's Bank of China via

Inter-American Development
Bank, Inter-American Empresas Públicas
2017 Ituango Hydroelectric Project [hyperlink]
Investment Corporation, de Medellín E.S.P.
Inter-American Development
Bank
Syndicated loan
State Administration on
with Chinese and
Foreign Exchange (SAFE) via
multilateral bank
International Finance
participants
Corporation, International
Castellana Wind Power Project CP La Castellana
2017 Finance Corporation,
[hyperlink] S.A.U.
Inter-American Investment
Corporation, Inter-American
Development Bank, Canadian
Climate Fund for the Private

342
Financial
Commitm
Instrument Project Title Lender/Donor Borrower/Recipient
ent Year
Category

Sector in the Americas

Works on Construction of the Municipal Bank of China Hungarian

Ministry of Finance
2022 (Sewage) Infrastructure Project Branch, Bank of China Srbija
of Serbia
[hyperlink] A.D. Beograd
Syndicated loan
with Chinese President Dr Néstor Carlos Kirchner and
China Development Bank, Ministry of Economy
state-owned Gobernador Jorge Cepernic
2014 Industrial and Commercial Bank and Public Finance
commercial banks Hydropower Project
of China, and Bank of China of Argentina
and/or policy [hyperlink]
banks
Queen Elizabeth ll Quay at Freetown Industrial and Commercial Bank National Port
2017 Port Project of China, Export-Import Bank Development (SL)
[hyperlink] of China Limited

People's Bank of
People's Bank of China via
Infrastructure Transport Program (PIT) China via
2014 Inter-American Development
PBOC/MOF grant [hyperlink] Inter-American
Bank
or loan channeled Development Bank
through
Sustainable Water Supply and Sanitation People's Bank of China via Government of
multilateral 2017
Program [hyperlink] [hyperlink] African Development Bank Rwanda
institutions
Water Supply Scheme for Tete PRC Poverty Reduction Fund Government of
2019
Settlement Project [hyperlink] via Asian Development Bank Papua New Guinea

Regional Hospitals Project – Lot 3 Ministry of Finance

2022 China CAMC Engineering
[hyperlink] of Guyana

Phase 1 Lot 5 Adenta-Dodowa Dual Government of

Supplier's credit 2018 Sinohydro Corporation Limited
Carriageway Project [hyperlink] Ghana
from Chinese SOE
Energy Transmission Network China National Machinery & Government of
2005 Construction Project Associated with the Equipment lmport & Export Republic of the
Imboulou Power Plant [hyperlink] Corporation (CMEC) Congo

343
 Table A6: Summary of environmental, social, and governance safeguard clauses in the contract sample for the pre-BRI and early BRI period (2000-2017)

Table A.6

Summary of environmental, social, and governance safeguard clauses in the contract sample for the pre-BRI and
early BRI period (2000-2017)

Environmental Social Governance

Standards Monitoring Enforcement Standards Monitoring Enforcement Standards Monitoring Enforcement

Bilateral China None None None None None None None None None
Eximbank loan

Bilateral CDB loan 1 of 3 contracts None 1 of 3 contracts None None None Most contracts None 1 of 3 contracts
includes includes include anti-money treat breaching
compliance and Environmental laundering and IFRS these requirements
no Indemnity clause requirements as event of default,
environmental thus requires
claims indemnification
requirement
Bilateral MOFCOM None None None None None None None None None
loan or grant
Bilateral Chinese Compliance with None 1 of 2 contracts 1 of 2 None 1 of 2 contracts Anti-corruption and None 1 of 2 contracts
state-owned Environmental includes issuance of contracts includes anti-money requires
commercial bank Laws and no the Certificate of includes Compliance with laundering cancellation and
loan Environmental Environmental Compliance Social Laws as requirements mandatory
Claims Compliance as a with Social conditions prepayment in the
requirements conditions Laws precedent vent of breach
precedent requirement
Syndicated loan Requirements to Requires Requires Requirement Requires Requires Social Anti-corruption and Reserve rights to Cancellation,
with Chinese and Comply with Implementatio environmental s to Comply Implementation conditions anti-money investigate mandatory
multilateral bank environmental n report of conditions with report of precedent; laundering potential prepayment, or
participants law and EIA/EMP precedent; requires Sociallaw; ESIA/ESAP requires requirements; breach; requires indemnification in
international cancelation, conduct cancelation, requires the proceedings event of breach
environmental mandatory ESIA and mandatory independent audit will not benefit
standards; pre-payment, or ESAP; pre-payment, or or compliance with sanctioned
conduct EIA and indemnification In indemnification in IFRS entities
EMP events of breach events of breach
Syndicated loan 2 of 2: EMP, None 1 of 2 1 of 2 no None 1 of 2 no social 1 of 2 requires 1 of 2 1 of 2 no

344
 Environmental Social Governance

Standards Monitoring Enforcement Standards Monitoring Enforcement Standards Monitoring Enforcement

with Chinese Environmental Indemnification and social enforcement providing representation governance
state-owned law compliance acceleration via standards information for made that enforcement
commercial banks required Event of Default if 1 of 2 know your customer neither borrower
and/or policy banks environmental laws 1 of 2 social indemnification checks nor associates 1 of 2 acceleration
1 of 2: broken law for breach of are via Event of Default
international compliance social laws or 1 of 2 requires anti internationally for
environmental 1 of 2 required standards corruption law sanctioned misrepresentation
standards indemnification if compliance, about compliance
(Equator environmental laws independently with anti corruption
Principles, broken audited financial laws or sanctions by
OECD) statements guarantor
compliant with IFRS
PBOC/MOF grant ESMP and EIA presentation 1 of 2 conditions RAP and presentation of 1 of 2 Competitive Bidding Acceleration,
or loan channeled (or equivalents) of EMAP (or precedent, ESMP (or RAP and ESAP acceleration/mand bidding documents mandatory
through required equivalent) to mandatory equivalents) (or equivalents) atory prepayment requirements, subject to prepayment for
multilateral lender, prepayment/acceler required, 1 to lender, for violations, 2 of anti-corruption laws lender review, 1 anti-corruption/anti-
institutions ongoing ation of 2 requires ongoing 2 conditions of 2 money laundering
reporting on ESIA reporting on precedent investigation by violations, 1 of 2
social social lender indemnification for
management management violations
Supplier's credit none none none none none none none none none
from Chinese SOE

345
 Table A7: Summary of environmental, social, and governance safeguard clauses in the contract sample for the late BRI period (2018-2021)

Table A.7

Summary of environmental, social, and governance safeguard clauses in the contract sample for the late BRI
period (2018-2021)

Environmental Social Governance

Standards Monitoring Enforcement Standards Monitoring Enforcement Standards Monitoring Enforcement

Bilateral None None None None None None None None None
China
Eximbank
loan
Bilateral 1 of 3 contracts None 1 of 3 contracts None None None Most contracts None 1 of 3 contracts treat
CDB loan includes includes Environmental include anti-money breaching these
compliance and no Indemnity clause laundering and IFRS requirements as event of
environmental requirements default, thus requires
claims requirement indemnification

Bilateral 1 of 2 contracts None None 1 of 2 None None 1 of 2 contracts Bidding None

MOFCOM requires contracts includes Competitive documents
loan or Environmental require equal Bidding subject to lender
grant Impact Assessment treatment requirements; approval
(EIA) principle Anti-corruption
requirement

Bilateral Environmental None Breach of None None None Anti-corruption and Reserve rights to Breach treated as an
Chinese Laws Compliance Environmental Laws is anti-money investigate event of default,
state-owned and no treated as event of laundering potential breach; condition precedent that
commercial Environmental default, thus requires requirements requires the the borrowers shall
bank loan Claims indemnification and proceedings will provide all requested
Requirement acceleration not benefit evidence to prove
sanctioned compliance
entities

Syndicated Requirements to Requires Requires Requirement Requires Requires Social Anti-corruption and Reserve rights to Requires cancellation,
loan with Comply with Implementation environmental s to Comply Implement conditions anti-money investigate mandatory prepayment,
Chinese and environmental law report of conditions precedent; with Social ation precedent; laundering potential breach; or indemnification in
multilateral and international EIA/EMP requires cancelation, Law; conduct report of requires requirements; requires the event of breach
bank environmental mandatory ESIA and ESIA/ESAP cancelation, requires proceedings will
participants standards; conduct prepayment, or ESAP; mandatory independent audit not benefit
EIA and EMP; indemnification in prepayment, or or compliance with sanctioned
events of breach indemnification in IFRS entities

346
 Environmental Social Governance

Standards Monitoring Enforcement Standards Monitoring Enforcement Standards Monitoring Enforcement

events of breach

Syndicated Compliance with None None None None None Compliance with Proceeds not to Mandatory prepayment
loan with environmental laws Anti-Corruption and benefit for business with
Chinese and permits Anti-Money sanctioned sanctioned entities,
state-owned Laundering Laws entities non-compliance with
commercial Anti-Corruption or
banks Anti-Money Laundering
and/or Laws
policy banks
PBOC/MOF EMP required Reporting on Withdrawal of grant Gender None Withdrawal of Compliance with Lender reserves Withdraw grant funds for
grant or EMP progress funds for Action Plan grant funds for anti-corruption, right to noncompliance
loan required noncompliance (GAP), noncompliance anti-money investigate
channeled prevention laundering, corrupt practices
through of anti-terrorism finance
multilateral involuntary laws; competitive
institutions resettlement, bidding required
labor
protections
in bidding
documents
and
contracts
required

Supplier's Compliance with None Indemnification and None None None Anti-money None Indemnification and
credit from environmental laws acceleration clauses laundering, acceleration clauses
Chinese and permits anti-corruption and
SOE counter terrorism
financing law
compliance

347
Table A8: Environmental Social Governance Safeguard Questions

Table A.8

Environmental Safeguard Questions

Page 1 of 6 (Environment)
Grant and Loan Instrument Category ① ① ① ② ② ② ③ ③ ③ ④ ④ ④
① Bilateral China Eximbank loan Kenya Antigua Philippines Costa Marshall Ecuador Montenegro Peru Mongolia Uganda Ghana Honduras
② Bilateral CDB loan (2014) (2008) (2018) Rica Islands (2016) (2020) (2014) (2019) (2021) (2016) (2013)
③ Bilateral MOFCOM loan or grant (2008) (2013)
④ Bilateral Chinese state-owned commercial bank loan
Are any environmental clauses or conditions included in
No No No No Yes No Yes No No Yes Yes Yes
the agreement?
Is an Environmental Impact Assessment (EIA) identified as a
No No No No NA No Yes No No No No No
requirement in the agreement?
Is an Environmental Management Plan (EMP) – or a
functional equivalent of an EMP – identified as a No No No No NA No No No No No No No
requirement in the agreement?
Is the borrower required to report to the lender on its
No No No No NA No No No No No No No
implementation of the EIA recommendations or EMP?
Does the agreement include any environmental conditions
precedent (for entry into force or initial or ongoing No No No No No No No No No Yes No Yes
disbursement)?
Does the agreement specify that noncompliance with
environmental rules, standards or laws is a sufficient basis
No No No No No No No No No Yes No No
for the lender to demand early repayment (“prepayment”)
of the loan or cancel the loan?
Does the agreement require compliance with international
environmental standards, such as the Equator Principles or
the OECD Revised Council Recommendation on Common No No No No No No No No No No No No
Approaches on the Environment and Officially Supported
Export Credits?
Is the borrower required to indemnify the lenders,
guarantors, or insurers ("Finance Parties") against any loss
or liability incurred by the Finance Parties as a result of any No No No No Yes No No No No Yes No No
actual or alleged breach of an environmental law or
standard?

348
Page 2 of 6 (Environment)
Grant and Loan Instrument Category ⑤ ⑤ ⑥ ⑥ ⑥ ⑦ ⑦ ⑦ ⑧ ⑧ ⑧
⑤ Syndicated loan with Chinese and multilateral bank participants Colombia Argentina Serbia Argentina Siena Costa Rwanda Papua Guyana Ghana Congo
⑥ Syndicated loan with Chinese state-owned commercial banks (2017) (2017) (2022) (2014) Leone Rica (2017) New (2022) (2018) (2005)
and/or policy banks (2017) (2014) Guinea
⑦ PBOC/MOF grant or loan channeled through multilateral (2019)
institutions
⑧ Supplier's credit from Chinese SOE
Are any environmental clauses or conditions included in the
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No
agreement?
Is an Environmental Impact Assessment (EIA) identified as a
Yes Yes No No Yes Yes Yes No No No No
requirement in the agreement?
Is an Environmental Management Plan (EMP) – or a functional
equivalent of an EMP – identified as a requirement in the Yes Yes No Yes Yes Yes No Yes No No No
agreement?
Is the borrower required to report to the lender on its
Yes Yes No No No Yes Yes Yes No No No
implementation of the EIA recommendations or EMP?
Does the agreement include any environmental conditions
Yes No No Yes No Yes No No No No No
precedent (for entry into force or initial or ongoing disbursement)?
Does the agreement specify that noncompliance with
environmental rules, standards or laws is a sufficient basis for the
Yes No Yes Yes No Yes No Yes Yes Yes No
lender to demand early repayment (“prepayment”) of the loan or
cancel the loan?
Does the agreement require compliance with international
environmental standards, such as the Equator Principles or the
No Yes No No Yes No No No No No No
OECD Revised Council Recommendation on Common Approaches
on the Environment and Officially Supported Export Credits?
Is the borrower required to indemnify the lenders, guarantors, or
insurers ("Finance Parties") against any loss or liability incurred by
No Yes No Yes Yes No No No Yes Yes No
the Finance Parties as a result of any actual or alleged breach of an
environmental law or standard?

349
Page 3 of 6 (Social)
Grant and Loan Instrument Category ① ① ① ② ② ② ③ ③ ③ ④ ④ ④
① Bilateral China Eximbank loan Kenya Antigua Philippines Costa Marshall Ecuador Montenegro Peru Mongolia Uganda Ghana Honduras
② Bilateral CDB loan (2014) (2008) (2018) Rica Islands (2016) (2020) (2014) (2019) (2021) (2016) (2013)
③ Bilateral MOFCOM loan or grant (2008) (2013)
④ Bilateral Chinese state-owned commercial bank loan
Are any social clauses or conditions included? No No No No No No Yes No No No No Yes
Is an Environmental and Social Impact Assessment (ESIA)
No No No No NA No No No No No No No
identified as a requirement?
Is an Environmental and Social Action Plan (ESAP), a
Resettlement Action Plan (RAP), or the functional No No No No NA No No No No No No No
equivalent of an ESAP or RAP identified as a requirement?
Is the borrower required to report to the lender on its
No No No No NA No No No No No No No
implementation of the ESIA recommendations or RAP?
Does the agreement include any social conditions
precedent (for entry into force or initial or ongoing No No No No No No No No No No No Yes
disbursement)?
Does the agreement indicate the penalty of violating Social
standards/laws leads to demanding early prepayment to No No No No No No No No No No No No
close the loan?
Is the borrower required to indemnify the lenders,
guarantors, or insurers ("Finance Parties") against any loss
No No No No No No No No No No No No
or liability incurred by the Finance Parties as a result of any
actual or alleged breach of an social law or standard?

350
Page 4 of 6 (Social)
Grant and Loan Instrument Category ⑤ ⑤ ⑥ ⑥ ⑥ ⑦ ⑦ ⑦ ⑧ ⑧ ⑧
⑤ Syndicated loan with Chinese and multilateral bank participants Colombia Argentina Serbia Argentina Siena Costa Rwanda Papua Guyana Ghana Congo
⑥ Syndicated loan with Chinese state-owned commercial banks (2017) (2017) (2022) (2014) Leone Rica (2017) New (2022) (2018) (2005)
and/or policy banks (2017) (2014) Guinea
⑦ PBOC/MOF grant or loan channeled through multilateral (2019)
institutions
⑧ Supplier's credit from Chinese SOE
Are any social clauses or conditions included? Yes Yes No No Yes Yes Yes Yes No No No
Is an Environmental and Social Impact Assessment (ESIA) identified
Yes Yes No No No No Yes No No No No
as a requirement?
Is an Environmental and Social Action Plan (ESAP), a Resettlement
Action Plan (RAP), or the functional equivalent of an ESAP or RAP Yes Yes No No No Yes Yes No No No No
identified as a requirement?
Is the borrower required to report to the lender on its
Yes Yes No No No Yes Yes No No No No
implementation of the ESIA recommendations or RAP?
Does the agreement include any social conditions precedent (for
Yes No No No No Yes Yes No No No No
entry into force or initial or ongoing disbursement)?
Does the agreement indicate the penalty of violating Social
standards/laws leads to demanding early prepayment to close the Yes No No No No Yes No Yes No No No
loan?
Is the borrower required to indemnify the lenders, guarantors, or
insurers ("Finance Parties") against any loss or liability incurred by
No Yes No No Yes No No No No No No
the Finance Parties as a result of any actual or alleged breach of an
social law or standard?

351
Page 5 of 6 (Governance)
Grant and Loan Instrument Category ① ① ① ② ② ② ③ ③ ③ ④ ④ ④
① Bilateral China Eximbank loan Kenya Antigua Philippines Costa Marshall Ecuador Montenegro Peru Mongolia Uganda Ghana Honduras
② Bilateral CDB loan (2014) (2008) (2018) Rica Islands (2016) (2020) (2014) (2019) (2021) (2016) (2013)
③ Bilateral MOFCOM loan or grant (2008) (2013)
④ Bilateral Chinese state-owned commercial bank loan
Are any governance (eg. anti-corruption, competitive
bidding, audited financial statements, and/or anti-money No No No Yes Yes No Yes No Yes Yes Yes Yes
laundering ) clauses or conditions present in the agreement?
Are independently audited financial statements required of
No No No No No No No No No No No No
the borrower?
Must the borrower’s financial statements comply with
No No No Yes Yes No No No No No No No
International Financial Reporting Standards (IFRS) standards?
Does the agreement include any competitive bidding
No No No No No No Yes No Yes No No No
requirements?
Are bidding documents or bid evaluation reports subject to
No No No No No No No No Yes No No No
prior approval by the lender?
Does the agreement include any anti-corruption or
No No No Yes Yes No Yes No No Yes Yes Yes
anti-money laundering requirements?
Is the borrower required to indemnify the lenders, guarantors,
or insurers ("Finance Parties") against any loss or liability
incurred by the Finance Parties as a result of any actual or No No No Yes No No No No No Yes No No
alleged breach of an anti-corruption or anti-money laundering
law or standard?
Does the lender explicitly reserve the right to prevent or
No No No No No No No No No Yes No No
investigate anti-corruption or anti-money laundering crimes?
Does the borrower affirm that it will seek to ensure that the
proceeds from the loan or grant are not used to finance or No No No No No No No No No Yes No No
benefit any internationally sanctioned entity?
Does the agreement specify that noncompliance with
governance rules, standards, or laws is a sufficient basis for
No No No Yes No No No No No Yes Yes No
the lender to demand early repayment (“prepayment”) of the
loan or cancel the loan?
Does the agreement include any competitive bidding,
anti-corruption, or anti-money laundering conditions
No No No No No No No No No Yes No No
precedent (for entry into force, initial disbursement, or
ongoing disbursements)?
Page 6 of 6 (Governance)

352
Grant and Loan Instrument Category ⑤ ⑤ ⑥ ⑥ ⑥ ⑦ ⑦ ⑦ ⑧ ⑧ ⑧
⑤ Syndicated loan with Chinese and multilateral bank participants Colombia Argentina Serbia Argentina Siena Costa Rwanda Papua Guyana Ghana Congo
⑥ Syndicated loan with Chinese state-owned commercial banks (2017) (2017) (2022) (2014) Leone Rica (2017) New (2022) (2018) (2005)
and/or policy banks (2017) (2014) Guinea
⑦ PBOC/MOF grant or loan channeled through multilateral (2019)
institutions
⑧ Supplier's credit from Chinese SOE
Are any governance (eg. anti-corruption, competitive bidding,
audited financial statements, and/or anti-money laundering ) Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No
clauses or conditions present in the agreement?
Are independently audited financial statements required of the
Yes Yes No No Yes Yes No No No No No
borrower?
Must the borrower’s financial statements comply with International
No Yes No No Yes No No No No No No
Financial Reporting Standards (IFRS) standards?
Does the agreement include any competitive bidding
No No No No No Yes Yes Yes No No No
requirements?
Are bidding documents or bid evaluation reports subject to prior
No No No No No Yes Yes No No No No
approval by the lender?
Does the agreement include any anti-corruption or anti-money
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No
laundering requirements?
Is the borrower required to indemnify the lenders, guarantors, or
insurers ("Finance Parties") against any loss or liability incurred by
No Yes No No No No Yes No Yes Yes No
the Finance Parties as a result of any actual or alleged breach of an
anti-corruption or anti-money laundering law or standard?
Does the lender explicitly reserve the right to prevent or
Yes Yes No No No Yes No Yes No No No
investigate anti-corruption or anti-money laundering crimes?
Does the borrower affirm that it will seek to ensure that the
proceeds from the loan or grant are not used to finance or benefit Yes Yes Yes No Yes No No No No No No
any internationally sanctioned entity?
Does the agreement specify that noncompliance with governance
rules, standards, or laws is a sufficient basis for the lender to
Yes Yes Yes No Yes Yes Yes Yes Yes Yes No
demand early repayment (“prepayment”) of the loan or cancel the
loan?
Does the agreement include any competitive bidding,
anti-corruption, or anti-money laundering conditions precedent (for No Yes No No No Yes No No No No No
entry into force, initial disbursement, or ongoing disbursements)?

353
Section A-10: How does the composition of the coding sample of
infrastructure financing agreements compare to the composition of China’s
entire grant- and loan-financed infrastructure project portfolio in LICs and
MICs?
In Chapter 3, we construct a coding sample of 23 infrastructure financing agreements to analyze trends in ESG safeguards in the full
grant- and loan-financed infrastructure projects portfolio from China to LICs and MICs. These 23 financing agreements were identified
out of a set of nearly 300 unredacted loan contracts and grant agreements included in the 3.0 version of the GCDF dataset. All 23
contracts represent infrastructure financing agreements that correspond to the 8 primary infrastructure agreement types listed in
Chapter 3 Section 3. These 8 financial instrument types were used by China to support 90.2% of its grant- and loan-financed
infrastructure project portfolio in LICs and MICs between 2000 and 2021. The remaining 9.8% of the portfolio consisted of projects
supported by more “exotic” financial instrument types (e.g., EPCF agreements). The 3.0 version of the GCDF dataset does not include
many unredacted financing agreements for these projects, so we exclude them from our analysis.

This contract coding sample has broad geographical coverage, income bracket coverage, and temporal coverage (for both the pre-
and early BRI period as well as the late BRI period). Table A9 describes the composition of the contract coding sample and the
composition of China’s entire grant- and loan-financed infrastructure project portfolio (as measured in the 3.0 version of AidData’s
GCDF dataset) on three dimensions: region, income bracket, and time period.

Table A9: Composition of coding sample of infrastructure financing agreements versus China’s entire grant- and loan-financed infrastructure project portfolio

Table A.9

Composition of coding sample of infrastructure financing agreements versus China’s entire grant-
and loan-financed infrastructure project portfolio)
Contract Coding Full infrastructure
Sample project portfolio
(count) (% of constant USD
2021 value)

Region

Africa 34.7% (8) 29%

Latin America and the Caribbean 43.4% (10) 18%

Asia and the Pacific 13% (3) 38%

Central and Eastern Europe 8.2% (2) 11%

Middle East 0% 1%

Income Bracket

Upper-middle income countries 52.2% (12) 34%

Lower-middle income countries 34.8% (8) 24%

354
 Contract Coding Full infrastructure
Sample project portfolio
(count) (% of constant USD
2021 value)

Low income & least developed countries 13% (3) 34%

Time Period

Pre- and early BRI (2000-2017) 65% (15) 81%

Late BRI (2018-2022) 35% (8) 19%

Section A-11: ESG risk exposure and de jure ESG safeguard protection in
China’s overseas infrastructure project portfolio tables
Table A10: Regional distribution of environmental, social, and governance risk exposure in China’s overseas infrastructure project portfolio

Table A.10

Regional distribution of environmental, social, and governance risk exposure in China’s overseas
infrastructure project portfolio
Infrastructure Infrastructure
Infrastructure projects
Infrastructure projects projects with projects with
with social risk Infrastructure
Region with ESG risk exposure environmental risk governance risk
exposure projects (%)
(%) exposure exposure
(%)
(%) (%)

Africa 24.66% 26.31% 31.24% 23.29% 27.61%

South & 20.90% 13.87% 12.45% 38.05% 16.92%

Central
America

Asia 34.26% 35.46% 42.39% 24.74% 36.40%

Europe 16.60% 20.54% 10.13% 11.01% 13.12%

Middle East 2.95% 2.98% 2.75% 2.58% 4.53%

Oceania 0.63% 0.84% 1.04% 0.33% 1.42%

Total 100.00% 100.00% 100.00% 100.00% 100.00%

Notes: This table provides a regional breakdown of China’s grant- and loan-financed infrastructure project portfolio in LICs and MICs
with environmental, social, or governance risk (ESG) exposure (columns 2-5). It also provides a regional breakdown of China’s grant-
and loan-financed infrastructure project portfolio in LICs and MICs (column 6). ESG risk exposure (column 2) is based on the
project-level composite measure that is described in Section 2 of Chapter 3. Likewise, environmental risk exposure, social risk
exposure, and governance risk exposure (column 3-5) are based on the project-level composite measures that are described in Section
2 of Chapter 3. All percentages represent percentages of China’s grant- and loan-financed infrastructure project portfolio (in constant
2021 USD) between 2000 and 2021 (exposed to one or more types of ESG risk). The regional groupings are defined by the OECD’s
regional classifications. The “Other” category captures grant and loan commitments for infrastructure projects in countries that were
classified as high-income countries as of 2021 and commitments that could not be assigned to one specific country.

355
Table A11: Distribution of environmental, social, and governance risk exposure in China’s overseas infrastructure project portfolio by host country income level

Table A.11

Distribution of environmental, social, and governance risk exposure in China’s overseas

infrastructure project portfolio by host country income level
Infrastructure
Infrastructure Infrastructure
Income level Infrastructure projects with projects with
projects with ESG projects with social Infrastructure projects
of host environmental risk exposure governance risk
risk exposure risk exposure (%)
country (%) exposure
(%) (%)
(%)

Upper-middle 34.30% 25.66% 15.88% 52.51% 37.17%

income
countries
(UMIC)

Lower-middle 23.54% 26.00% 35.89% 12.88% 26.85%

income
countries
(LMIC)

Low-income 26.61% 28.44% 37.87% 24.74% 25.37%

countries (LIC)

Other 15.55% 19.91% 10.36% 9.87% 10.61%

Grand Total 100.00% 100.00% 100.00% 100.00% 100.00%

Notes: This table provides an income bracket breakdown of China’s grant- and loan-financed infrastructure project portfolio in LICs
and MICs with environmental, social, or governance risk (ESG) exposure (columns 2-5). It also provides an income bracket breakdown
of China’s grant- and loan-financed infrastructure project portfolio in LICs and MICs (column 6). ESG risk exposure (column 2) is based
on the project-level composite measure that is described in Section 2 of Chapter 3. Likewise, environmental risk exposure, social risk
exposure, and governance risk exposure (column 3-5) are based on the project-level composite measures that are described in Section
2 of Chapter 3. All percentages represent percentages of China’s grant- and loan-financed infrastructure project portfolio (in constant
2021 USD) between 2000 and 2021 (exposed to one or more types of ESG risk). The income brackets are defined by the OECD’s ODA
income categories as of 2021. The “Other” category captures grant and loan commitments for infrastructure projects in countries that
were classified as high-income countries as of 2021 and commitments that could not be assigned to one specific country.

356
Table A12: Country-by-country distribution of ESG risk exposure and de jure ESG safeguard protection in China’s overseas infrastructure project portfolio, 2000-2021

Table A.12

Country-by-country distribution of ESG risk exposure and de jure ESG safeguard protection in China’s overseas infrastructure project
portfolio, 2000-2021
Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Albania 5 1% 3 33% 0 0 0 0 5 3 0% 0%

Angola 17,670 44% 48 20% 11,691 36 4,602 7 12,083 30 27% 8%

Antigua and 246 49% 9 33% 246 9 48 1 66 1 0% 0%

Barbuda

Argentina 10,046 44% 26 53% 9,020 13 9,180 22 8,090 7 90% 82%

Azerbaijan 274 34% 3 30% 0 0 156 1 118 2 98% 60%

Bahamas 3,596 100% 5 83% 3,536 3 3,412 1 61 2 0% 0%

Bangladesh 12,075 59% 27 45% 6,542 15 8,423 15 6,028 11 22% 24%

Barbados 3 2% 1 10% 3 1 0 0 0 0 0% 0%

Belarus 2,367 21% 15 23% 1,863 11 222 1 1,086 5 7% 3%

Benin 1,651 70% 24 56% 1,574 19 917 6 57 1 31% 13%

Bolivia 3,285 98% 11 69% 1,850 8 1,988 4 553 1 0% 9%

Bosnia and 1,381 64% 3 27% 1,136 2 798 1 245 1 0% 0%

Herzegovina

Botswana 169 11% 8 42% 169 8 0 0 2 1 83% 7%

Brazil 1,292 16% 11 35% 1,022 10 0 0 270 1 34% 63%

357
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Brunei 2,102 99% 6 86% 2,102 6 0 0 0 0 89% 83%

Bulgaria 162 31% 4 50% 162 2 0 0 0 2 0% 0%

Burkina Faso 118 31% 2 25% 118 2 118 2 0 1 24% 20%

Burundi 1 0% 3 13% 0 0 0 0 1 3 0% 0%

Cabo Verde 143 35% 6 14% 15 3 0 0 134 4 0% 0%

Cambodia 8,665 59% 50 36% 7,258 45 6,001 34 351 4 20% 13%

Cameroon 3,317 39% 15 26% 855 7 2,503 7 172 3 10% 19%

Central African 96 25% 7 29% 0 0 96 7 0 1 0% 0%

Republic

Chad 877 51% 9 32% 711 8 877 9 0 0 0% 0%

Chile 164 10% 5 28% 71 1 71 1 93 4 15% 60%

Colombia 1,102 74% 11 69% 557 7 579 5 244 2 42% 57%

Comoros 68 27% 7 35% 68 7 0 0 0 0 0% 0%

Congo (Republic) 3,874 52% 30 38% 3,183 23 864 8 523 6 0% 0%

Cook Islands 36 45% 4 50% 36 4 0 0 0 0 0% 0%

Costa Rica 1,772 99% 10 91% 757 8 0 0 1,028 4 3% 10%

Cote d'Ivoire 2,767 42% 18 28% 1,235 10 2,344 12 102 2 45% 33%

Cuba 218 8% 3 13% 218 3 0 0 184 2 0% 0%

North Korea 57 8% 6 38% 0 2 55 2 2 2 0% 0%

358
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Congo (DRC) 2,519 28% 31 30% 1,260 17 1,401 20 1,171 4 0% 0%

Djibouti 1,048 46% 4 11% 582 2 582 2 1,021 3 3% 5%

Dominica 111 40% 6 23% 81 5 30 1 0 0 0% 0%

Ecuador 6,447 65% 14 35% 5,627 11 2,543 4 3,700 4 18% 20%

Egypt 183 3% 3 7% 183 3 150 2 0 0 60% 27%

El Salvador 185 98% 3 38% 82 1 124 2 61 1 23% 20%

Equatorial Guinea 2,204 26% 4 8% 2,204 4 0 0 0 0 0% 0%

Eritrea 1,927 75% 7 30% 0 0 1,895 4 32 3 65% 13%

Ethiopia 17,845 86% 76 74% 9,510 37 14,507 73 7,482 13 19% 4%

Fiji 338 66% 21 68% 233 20 105 1 0 0 0% 0%

Gabon 2,035 67% 26 62% 1,180 10 1,060 21 0 1 22% 31%

Gambia 182 100% 4 100% 182 4 99 1 99 1 0% 0%

Georgia 49 17% 2 33% 49 2 0 0 0 0 0% 0%

Ghana 1,670 19% 21 27% 1,670 21 0 0 0 0 22% 36%

Grenada 305 67% 16 44% 305 16 0 0 0 0 0% 0%

Guinea 1,760 43% 15 54% 479 14 1,281 1 1,281 1 16% 28%

Guinea-Bissau 120 67% 12 55% 120 12 0 0 62 4 0% 0%

Guyana 358 32% 12 60% 75 2 210 2 269 10 0% 0%

359
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Honduras 336 90% 1 50% 336 1 336 1 336 1 100% 100%

India 352 14% 2 13% 352 2 0 0 0 0 0% 0%

Indonesia 20,912 50% 60 43% 14,423 49 16,586 39 1,285 3 33% 39%

Iran 9,309 37% 13 27% 5,133 5 3,462 7 4,243 4 6% 14%

Iraq 314 5% 3 17% 312 2 0 0 2 1 93% 79%

Israel 2,435 73% 9 90% 2,435 9 0 0 1,181 2 60% 75%

Jamaica 1,852 81% 13 57% 1,378 12 0 0 474 1 0% 0%

Jordan 1,808 85% 6 29% 13 1 1,808 5 0 1 91% 35%

Kazakhstan 12,196 25% 13 21% 10,400 12 2,770 1 1,796 1 45% 31%

Kenya 10,761 82% 41 47% 9,658 28 8,266 18 5,459 14 13% 7%

Kiribati 0 0% 2 29% 0 2 0 0 0 0 0% 0%

Kyrgyzstan 2,831 77% 21 42% 1,726 9 527 6 1,369 12 1% 3%

Laos 13,531 71% 49 35% 11,699 41 13,209 46 12 1 17% 16%

Lesotho 342 76% 8 32% 0 0 193 2 342 8 0% 0%

Liberia 225 59% 16 42% 66 2 0 0 159 14 0% 0%

Libya 0 0% 0 0% 0 0 0 0 0 0 100% 100%

Madagascar 322 44% 4 14% 322 4 0 0 32 1 0% 0%

Malawi 209 23% 5 12% 107 3 0 0 103 3 13% 10%

360
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Malaysia 24,105 94% 19 59% 2,733 10 1,357 5 20,014 4 10% 36%

Maldives 1,516 74% 18 62% 1,516 18 0 0 210 2 30% 29%

Mali 1,023 65% 21 58% 610 4 1,023 21 0 0 0% 0%

Marshall Islands 0 0% 0 0% 0 0 0 0 0 0 65% 22%

Mauritania 509 30% 11 30% 170 3 509 11 0 1 0% 0%

Mauritius 80 8% 1 3% 0 0 0 0 80 1 0% 0%

Mexico 289 32% 4 27% 28 1 261 3 0 0 17% 80%

Micronesia 21 21% 9 32% 21 9 0 0 0 0 0% 0%

Mongolia 1,095 41% 15 42% 933 11 90 2 100 4 3% 3%

Montenegro 1,090 90% 2 50% 0 0 45 1 1,090 2 0% 0%

Morocco 332 30% 1 10% 332 1 332 1 0 0 42% 38%

Mozambique 2,374 43% 9 25% 1,225 5 1,336 4 694 3 29% 12%

Myanmar 8,395 65% 21 23% 3,031 8 3,211 12 4,907 7 4% 11%

Namibia 73 15% 3 9% 73 3 55 1 0 0 0% 0%

Nauru 0 0% 1 25% 0 1 0 0 0 0 0% 0%

Nepal 1,388 89% 40 87% 832 18 1,388 40 256 4 0% 0%

Niger 1,988 51% 18 56% 337 2 1,988 18 1,342 1 0% 0%

Nigeria 6,326 52% 22 47% 3,022 10 4,492 16 787 3 28% 35%

361
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Niue 16 83% 1 25% 16 1 0 0 0 0 0% 0%

North Macedonia 1,346 99% 7 54% 1,032 6 0 0 1,086 3 17% 20%

Oman 0 0% 0 0% 0 0 0 0 0 0 100% 100%

Pakistan 20,448 52% 63 49% 6,373 16 17,887 55 7,380 16 25% 31%

Panama 207 81% 5 71% 123 3 0 0 140 4 100% 100%

Papua New Guinea 2,270 33% 11 22% 1,638 5 1,894 6 633 6 31% 6%

Paraguay 0 0% 0 0% 0 0 0 0 0 0 100% 100%

Peru 454 58% 8 73% 0 0 421 1 34 7 0% 0%

Philippines 2,335 37% 16 36% 1,978 14 491 7 736 4 55% 27%

Russia 63,128 81% 15 42% 42,875 5 15,939 7 19,197 8 13% 32%

Rwanda 26 2% 1 3% 0 0 0 0 26 1 7% 10%

St. Lucia 52 76% 1 33% 52 1 0 0 0 0 0% 0%

Samoa 148 34% 8 22% 148 8 0 0 61 1 0% 0%

Sao Tome and 0 0% 1 20% 0 1 0 0 0 0 0% 0%

Principe

Senegal 2,436 67% 16 36% 2,436 16 1,681 6 756 4 6% 6%

Serbia 3,443 73% 9 56% 3,443 9 59 1 0 0 5% 14%

Seychelles 74 40% 14 42% 74 14 0 0 0 0 0% 0%

Sierra Leone 1,227 33% 25 52% 1,227 25 0 0 954 3 21% 8%

362
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

South Africa 4,404 50% 10 45% 2,675 9 0 0 1,729 1 6% 33%

South Sudan 1,817 92% 14 41% 506 10 0 0 1,310 4 0% 0%

Sri Lanka 10,620 70% 50 54% 7,893 34 850 3 6,476 25 1% 3%

Sudan 4,230 28% 9 8% 1,727 4 2,008 5 1,976 1 0% 0%

Suriname 664 48% 7 24% 0 0 84 4 579 3 10% 19%

Syria 9 4% 3 30% 0 0 0 0 9 3 0% 0%

Tajikistan 2,057 36% 20 29% 1,612 15 0 0 988 6 10% 15%

Tanzania 1,857 64% 23 34% 1,753 18 159 6 207 3 8% 16%

Thailand 110 2% 2 10% 0 0 110 2 0 0 2% 17%

Timor-Leste 29 30% 4 36% 29 4 0 0 0 0 0% 0%

Togo 721 50% 21 47% 666 19 162 4 0 0 0% 0%

Tonga 70 23% 4 11% 70 4 0 0 0 0 0% 0%

Trinidad and 943 65% 6 55% 943 6 39 1 246 2 0% 0%

Tobago

Tunisia 0 0% 2 12% 0 1 0 0 0 1 26% 14%

Türkiye 4,590 43% 8 27% 3,768 6 2,339 4 34 1 83% 95%

Turkmenistan 11 0% 2 14% 0 0 0 0 11 2 0% 0%

Uganda 3,274 74% 12 32% 2,847 9 1,699 3 2,684 6 6% 5%

Ukraine 562 11% 4 40% 104 2 0 0 458 2 8% 43%

363
 Strong Strong
de jure de jure
ESG ESG ESG ESG
Risk ESG Risk Environmental Environmen Social Governance Governan safeguar safeguar
ESG Risk (% of risk (% of risk tal risk Social risk risk risk ce risk ds ds
Country ($) $) (#) #) ($) (#) ($) (#) ($) (#) (% of $) (% of #)

Uruguay 20 49% 1 33% 20 1 0 0 0 0 100% 100%

Uzbekistan 0 0% 0 0% 0 0 0 0 0 0 96% 36%

Vanuatu 63 23% 1 17% 63 1 0 0 0 0 0% 0%

Venezuela 64,321 77% 35 44% 10,390 25 4,529 3 63,743 13 6% 3%

Vietnam 15,967 62% 48 49% 12,418 31 8,086 32 88 3 51% 34%

Zambia 8,101 56% 45 41% 2,045 26 2,261 6 5,829 19 36% 29%

Zimbabwe 559 13% 7 14% 387 6 0 0 345 2 0% 0%

Africa, regional 407 69% 6 67% 407 6 379 5 28 1 0% 0%

Asia, regional 0 0% 0 0% 0 0 0 0 0 0 100% 100%

Notes: ESG risk exposure is based on a project-level composite measure that is described in Section 2 of Chapter 3. Likewise, environmental risk exposure, social risk exposure,
and governance risk exposure are based on project-level composite measures that are described in Section 2 of Chapter 3. Strong de jure ESG safeguards are defined in Section
A-9 of the Appendix. All monetary amounts are provided in constant 2021 USD millions. All values in the "% of $" columns represent percentages of China’s grant- and
loan-financed infrastructure project portfolio (in constant 2021 USD) between 2000 and 2021 in a specific recipient country. All values in the "% of #" columns represent
percentages of China’s grant- and loan-financed infrastructure projects between 2000 and 2021 in a specific recipient country.

364
Section A-12: Soft power gains and losses during early and
late BRI periods
Table A13 shows the relative change (double delta) scores for each soft power measure described in Box
4a, including steps 1-3 for both the early BRI period (2014-2017) and the late BRI period (2018-2021).
Public opinion is based on Gallup World Poll data. Media sentiment is based on the average tone of media
coverage related to government actors, as reported by GDELT. Elite support is based on UNGA voting
alignment scores. See Box 1b for additional information on the data sources that we use to construct these
measures. With respect to the relative change (double delta) scores for the public opinion and media
sentiment metrics (reported in Table A13), the higher the score, the larger the magnitude of the gain
achieved by China at the expense of the U.S. The relative change (double delta) scores for the elite support
(idealpoint) metric (reported in Table A13) reflect raw scores that have not been inverted—so the lower the
idealpoint score (reported in Table A13), the larger the size of China’s gain at the expense of the U.S.

Table A13: China’s soft power gains and losses vis-à-vis the U. S. during early and late BRI periods

Table A.13

China’s soft power gains and losses vis-à-vis the U. S. during early and late BRI
periods
Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

Afghanistan 3.0% 0.532 0.050 2.1% (0.170) 0.031

Albania 0.1% (0.031) 0.003 -4.5% 0.827 0.020

Algeria -3.3% 0.548 0.050 0.7% (0.650) 0.031

Angola 1.9% 0.041 0.050 (0.266) 0.031

Antigua and
4.219 0.050 (3.747) 0.031
Barbuda

Argentina 3.0% 0.921 (0.018) -12.4% (0.626) (0.023)

Armenia 3.8% 0.743 0.020 -0.2% 0.843 (0.032)

Azerbaijan 5.2% 0.849 0.050 -3.4% (0.607) 0.031

Bahamas 0.424 (0.059) 0.233 0.031

Bangladesh 1.6% 0.663 0.050 -2.8% (0.583) 0.031

Barbados 0.637 0.028 (1.020) 0.031

Belarus 4.1% 0.395 0.023 -7.7% 1.232 (0.039)

365
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

Benin -1.1% (1.410) 0.050 -0.2% (0.429) 0.031

Bolivia 5.7% 0.675 0.050 -12.4% (0.193) 0.031

Bosnia and
3.2% (0.047) -4.5% (0.086)
Herzegovina

Botswana 4.9% (0.401) (0.005) 0.7% 1.350 0.031

Brazil 3.0% 0.194 0.092 -8.8% (0.159) 0.159

Brunei
0.036 0.050 0.218 0.031
Darussalam

Bulgaria 4.9% 0.442 0.003 -6.8% 0.284 0.024

Burkina Faso 0.0% 0.332 0.050 -0.6% 0.676 0.031

Burundi -15.1% 0.536 0.050 (1.229) 0.031

Cabo Verde (1.005) 0.050 0.031

Cambodia 0.4% 1.321 0.050 0.0% (0.799) 0.031

Cameroon 2.1% 0.983 (0.070) -3.6% (1.027) (0.092)

Central African
4.4% (0.460) 0.141 (0.550) (0.112)
Republic

Chad 6.8% (1.028) 0.050 -0.9% 1.290 0.031

Chile 8.1% 0.288 (0.113) -11.1% 0.197 0.053

Colombia 6.5% (0.181) (0.090) -9.0% 1.164 0.061

Comoros (2.648) 0.050 -0.5% (5.837) 0.031

Congo 0.4% 1.656 0.050 -2.4% (0.214) 0.031

Cook Islands (1.629) (1.198)

Costa Rica 10.0% 0.644 (0.040) -11.5% (0.336) (0.004)

Cote d'Ivoire -1.3% 0.363 (0.075) -1.2% 1.009 0.080

Cuba 0.541 0.050 0.591 0.031

North Korea 0.160 0.131 0.180 (0.032)

366
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

DRC 0.0% 2.784 0.050 (1.029) 0.151

Djibouti 1.137 0.050 (1.032) 0.031

Dominica (1.015) (0.057) (2.684) 0.031

Dominican
5.7% 0.309 (0.015) -5.7% 1.319 0.053
Republic

Ecuador 7.2% 0.810 0.050 -8.6% 0.059 0.031

Egypt 7.8% 0.537 0.050 -36.7% 0.338 0.031

El Salvador 10.9% (0.649) (0.030) -6.8% (0.926) 0.031

Equatorial
1.972 (0.037) 1.241 0.031
Guinea

Eritrea 2.531 0.050 0.710 0.031

Ethiopia -0.1% 1.054 0.050 9.4% 0.074 0.031

Fiji 0.665 0.027 (0.202) 0.031

Gabon 0.1% 1.519 0.050 -0.5% (2.304) 0.031

Gambia (0.236) 0.050 (0.248) 0.031

Georgia 2.3% 0.015 -3.5% (14.520) (0.080)

Ghana 2.8% (0.152) 0.050 -1.9% 0.382 0.031

Grenada 0.498 (0.088) 0.062 0.031

Guinea 1.5% 0.728 0.050 -3.6% (0.611) 0.031

Guinea-Bissau 2.370 0.050 (3.296) 0.031

Guyana (0.265) 0.050 0.401 0.031

Haiti 4.3% 0.945 0.000 -5.1% (2.191) 0.056

Honduras 5.8% 0.040 -11.2% (3.421) (0.088)

India -2.3% (0.184) 0.145 -5.4% (0.230) 0.015

Indonesia 2.9% 0.191 0.050 -3.5% (0.426) 0.031

367
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

Iran 7.2% 0.723 0.050 2.3% (0.109) 0.031

Iraq -1.0% 0.630 0.050 -5.4% (0.551) 0.031

Israel -1.2% 0.454 (0.062) 4.7% (0.309) (0.064)

Jamaica 16.1% 0.377 0.027 -12.1% (0.290) 0.031

Jordan 0.360 0.050 -26.8% 0.108 0.031

Kazakhstan 5.0% 0.425 0.037 -9.4% (0.774) 0.031

Kenya 2.3% 0.350 0.039 -5.0% (0.593) 0.031

Kiribati (0.397) (1.334) 0.031

Kyrgyz Republic 2.5% (0.041) 0.050 -7.2% 0.217 0.031

Laos 0.839 0.050 (0.253) 0.031

Lebanon -1.1% 1.400 0.050 -4.6% 0.375 0.031

Lesotho 13.5% (0.155) 0.050 0.6% 1.063 0.031

Liberia 0.0% 0.472 0.001 1.8% (0.072) 0.235

Libya 59.7% 0.520 0.050 -3.7% 0.168 0.031

Madagascar 0.1% 1.377 0.046 1.4% (0.174) 0.052

Malawi 1.1% 0.086 (0.162) -0.4% 0.936 0.037

Malaysia -18.2% 0.233 0.050 8.2% (0.517) 0.031

Maldives 0.943 0.023 (0.791) 0.031

Mali 0.2% 1.235 0.159 -0.3% (1.396) (0.078)

Marshall Islands (0.064) (8.304) (0.068)

Mauritania 5.2% 1.199 0.050 -0.6% (0.630) 0.031

Mauritius 5.1% 1.527 0.040 -4.9% (0.270) 0.031

Mexico 8.4% 0.323 0.031 -8.7% 0.100 (0.080)

368
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

Micronesia 1.101 (0.341) (1.693) 0.062

Moldova 6.8% (0.430) (0.118) -2.2% (1.467) (0.088)

Mongolia 4.1% (0.180) 0.006 -1.1% 0.228 0.031

Montenegro -0.6% 0.078 -0.3% (0.006)

Morocco 9.5% (0.588) 0.050 -7.5% (0.655) 0.031

Mozambique 9.4% 0.407 0.050 -3.3% (0.604) 0.031

Myanmar 3.3% 0.429 0.050 -8.5% (0.345) 0.031

Namibia 9.5% (0.215) 0.050 -6.5% (0.605) 0.031

Nauru 0.029 (8.454) (0.124)

Nepal 0.5% 0.015 0.050 2.0% (0.059) 0.031

Nicaragua 5.5% 0.515 0.050 -5.2% 1.161 0.031

Niger -3.7% 0.523 0.050 -5.2% (0.024) 0.031

Nigeria 1.7% 0.414 0.050 0.5% 0.201 0.031

North
3.5% 1.412 -0.4% (1.283)
Macedonia

Oman 2.068 0.050 (1.293) 0.031

Pakistan 4.6% 0.409 0.125 -6.4% (0.182) 0.042

Palau 0.055

Panama 6.7% 0.135 (0.180) -11.3% 0.179 (0.046)

Papua New
(0.071) (0.156) 0.009 (0.001)
Guinea

Paraguay 4.6% 4.208 (0.122) -1.0% 1.445 (0.126)

Peru 7.3% 0.997 (0.067) -5.2% 0.224 (0.066)

Philippines 3.0% 0.496 0.040 -2.9% (0.674) 0.031

Romania 5.3% 0.051 -6.6% (0.018)

369
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)

Russia 5.7% 0.411 0.032 0.5% (0.085) (0.021)

Rwanda 2.5% 0.068 0.142 -1.6% 0.143 (0.013)

Saint Lucia (5.200)

Samoa (0.502) (0.170) (0.795) (0.007)

Sao Tome and

4.726 0.050 0.031
Principe

Senegal 4.4% 0.370 0.050 -0.9% 0.796 0.031

Serbia 6.3% 0.535 (0.080) -3.1% (0.674) (0.087)

Seychelles 1.317 0.008 (0.996) 0.031

Sierra Leone 2.4% 1.104 0.038 -2.6% (0.203) 0.031

Solomon Islands 0.495 (0.189)

Somalia -2.9% 1.837 0.050 0.202 0.031

South Africa 3.1% 0.858 0.022 -3.9% (0.698) 0.031

South Sudan -1.2% 0.124 0.185 1.337 (0.490)

Sri Lanka -6.5% 0.192 0.050 -3.1% (0.379) 0.031

Sudan 0.0% 0.533 0.050 (0.092) 0.031

Suriname (1.848) 0.050 2.828 0.031

Syrian Arab
8.8% 0.327 0.613
Republic

Tajikistan -2.9% 1.527 0.050 (2.183) 0.031

Tanzania 2.1% 0.422 0.050 -0.9% (0.036) 0.031

Thailand 4.2% 0.164 (0.003) -4.2% (0.959) 0.031

Timor-Leste 0.011 0.031

Togo -0.2% 1.958 0.019 -2.6% (2.328) (0.088)

Tonga 0.153 (0.123) 0.422 (0.002)

370
 Early BRI (2014-2017) Late BRI (2018-2021)
Public Media Public Media
Country Elite Support Elite Support
Opinion Sentiment Opinion Sentiment
(idealpoint) (idealpoint)
(%) (score) (%) (score)
Trinidad and
0.107 0.006 0.855 0.031
Tobago

Tunisia 3.6% (1.372) 0.050 -7.8% (0.233) 0.031

Turkey -0.6% 0.713 0.086 1.7% (0.590) (0.022)

Turkmenistan 0.2% 0.582 0.050 2.8% (1.278) 0.031

Uganda 2.8% (0.290) 0.050 -3.0% 1.343 0.031

Ukraine -0.2% 1.101 0.141 -0.8% (0.607) (0.025)

Uruguay 8.8% 0.660 (0.000) -11.1% (0.254) 0.094

Uzbekistan 5.2% 1.340 0.022 -7.6% (1.164) (0.038)

Vanuatu 1.750 (0.160) 1.448 (0.060)

Venezuela -0.8% 0.211 0.050 -3.4% 0.849 0.031

Viet Nam -9.8% 0.355 0.050 4.5% (0.070) 0.031

West Bank and

1.7% 0.453 -0.1% (0.309)
Gaza Strip

Yemen -0.3% (0.063) 0.050 8.1% (1.016) 0.031

Zambia 4.0% 0.745 0.050 -7.7% (0.126) 0.031

Zimbabwe 2.7% 0.158 0.050 -2.1% (0.157) 0.031

Table A14 presents the soft power cohorts (safe bet, toss-up leaning China, toss-up, and moonshot) to
which each LIC/MIC belongs based on its relative change (double delta) scores during the early BRI
(2014-2017) period. Table A14 also reports observed and expected development finance commitments
from Beijing during the late BRI period (2018-2021) to each LIC/MIC. Expected development finance
commitments are based on a hypothetical scenario in which China allocated development finance to each
country on a non-strategic basis (i.e., based only on the population size of each country). The differences
between expected and observed allocations (in constant 2021 USD millions between 2018 and 2021 and
constant 2021 USD millions as a percentage of total ODA and OOF from China between 2018 and 2021)
therefore provide an indication of how much China may have prioritized or deprioritized each cohort based
on strategic considerations.

371
 Table A14: Country-by-country breakdown of expected versus observed development finance allocations during late BRI period

Table A.14

Country-by-country breakdown of expected versus observed development

finance allocations during late BRI period
Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Afghanistan Toss-up Toss-up (China) Toss-up 2,741.7 82.4 2,659.3 0.7%

Albania Toss-up Moonshot Toss-up (China) 203.0 1.6 201.5 0.1%

Algeria Moonshot Toss-up (China) Toss-up 3,076.3 68.9 3,007.3 0.8%

Angola Toss-up Moonshot Toss-up 2,349.4 4,658.9 (2,309.5) -0.6%

Antigua and
Safe Bets Moonshot 6.6 108.9 (102.3) 0.0%
Barbuda

Argentina Toss-up (China) Toss-up (China) Safe Bets 3,225.5 (85,906.1 (82,680.6) -22.4%

Armenia Toss-up (China) Toss-up (China) Toss-up (China) 201.0 2.0 199.0 0.1%

Azerbaijan Toss-up (China) Toss-up (China) Toss-up 717.8 492.6 225.2 0.1%

Bahamas Toss-up Safe Bets 28.9 0.9 28.0 0.0%

Bangladesh Toss-up Toss-up (China) Toss-up (11,893.1 (13,488.6 (1,595.4) -0.4%

Barbados Toss-up (China) Toss-up (China) 20.0 1.3 18.8 0.0%

Belarus Toss-up (China) Toss-up Toss-up (China) 671.1 1,837.4 (1,166.3) -0.3%

Benin Moonshot Moonshot Moonshot 890.6 1,240.0 (349.4) -0.1%

Bolivia Safe Bets Toss-up (China) Moonshot 846.5 131.1 715.4 0.2%

Bosnia and
Toss-up (China) Safe Bets 238.4 469.5 (231.1) -0.1%
Herzegovina

Botswana Toss-up (China) Moonshot Safe Bets 180.1 5.0 175.1 0.0%

Brazil Toss-up (China) Toss-up Moonshot (15,170.0 7,384.9 7,785.1 2.1%

Brunei
Moonshot Toss-up 31.4 1,926.4 (1,895.0) -0.5%
Darussalam

Bulgaria Toss-up (China) Toss-up Safe Bets 496.7 583.4 (86.7) 0.0%

372
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Burkina Faso Moonshot Toss-up Toss-up 1,517.4 446.9 1,070.5 0.3%

Burundi Moonshot Toss-up (China) Toss-up 859.7 34.4 825.3 0.2%

Cabo Verde Moonshot Moonshot 41.4 2.5 38.9 0.0%

Cambodia Toss-up Safe Bets Toss-up (China) 1,164.7 6,476.1 (5,311.4) -1.4%

Cameroon Toss-up Safe Bets Safe Bets 1,867.0 504.5 1,362.5 0.4%

Central
African Toss-up (China) Moonshot Moonshot 376.9 286.7 90.2 0.0%
Republic

Chad Safe Bets Moonshot Moonshot 1,170.7 381.8 788.9 0.2%

Chile Safe Bets Toss-up Safe Bets 1,366.8 5,050.4 (3,683.6) -1.0%

Colombia Safe Bets Moonshot Safe Bets 3,605.8 954.4 2,651.4 0.7%

Comoros Moonshot Moonshot 57.1 125.0 (67.9) 0.0%

Congo Toss-up Safe Bets Toss-up 402.7 106.8 295.9 0.1%

Costa Rica Safe Bets Toss-up (China) Safe Bets 364.3 79.8 284.5 0.1%

Cote d'Ivoire Moonshot Toss-up Safe Bets 1,891.7 3,938.2 (2,046.4) -0.6%

Cuba Toss-up (China) Toss-up 807.2 151.5 655.8 0.2%

North Korea Toss-up Moonshot 1,843.5 1,163.9 679.6 0.2%

DRC Moonshot Safe Bets Toss-up 6,531.5 1,005.7 5,525.8 1.5%

Djibouti Safe Bets Moonshot 77.3 301.5 (224.2) -0.1%

Dominica Moonshot Safe Bets 5.1 36.3 (31.2) 0.0%

Dominican
Safe Bets Toss-up Safe Bets 781.6 55.0 726.5 0.2%
Republic

Ecuador Safe Bets Toss-up (China) Toss-up 1,245.5 1,388.4 (142.9) 0.0%

Egypt Safe Bets Toss-up (China) Moonshot 7,609.1 (19,114.8 (11,505.7) -3.1%

El Salvador Safe Bets Moonshot Safe Bets 449.4 287.3 162.1 0.0%

373
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)
Equatorial
Safe Bets Safe Bets 112.2 1,257.7 (1,145.4) -0.3%
Guinea

Eritrea Safe Bets Moonshot 252.2 1,762.9 (1,510.8) -0.4%

Ethiopia Moonshot Safe Bets Toss-up 8,263.4 3,367.4 4,896.0 1.3%

Fiji Toss-up (China) Toss-up (China) 65.8 35.3 30.4 0.0%

Gabon Toss-up Safe Bets Moonshot 161.9 187.2 (25.2) 0.0%

Gambia Moonshot Moonshot 181.6 16.4 165.2 0.0%

Georgia Toss-up Toss-up (China) 265.7 94.4 171.2 0.0%

Ghana Toss-up Moonshot Moonshot 2,275.2 3,298.0 (1,022.8) -0.3%

Grenada Toss-up (China) Safe Bets 8.8 57.8 (49.0) 0.0%

Guinea Toss-up Toss-up (China) Toss-up 931.6 3,196.5 (2,264.8) -0.6%

Guinea-Bissa
Safe Bets Toss-up (China) 142.4 59.3 83.0 0.0%
u

Guyana Moonshot Toss-up 56.9 340.1 (283.2) -0.1%

Honduras Safe Bets Toss-up (China) 717.0 36.0 681.1 0.2%

India Moonshot Moonshot Moonshot (99,221.0 214.6 99,006.4 26.8%

Indonesia Toss-up Toss-up Toss-up (19,327.2 7,265.8 12,061.4 3.3%

Iran Safe Bets Toss-up (China) Toss-up (China) 6,203.8 6,689.4 (485.5) -0.1%

Iraq Moonshot Toss-up (China) Toss-up 3,004.5 5,700.4 (2,695.9) -0.7%

Israel Moonshot Toss-up (China) Safe Bets 652.1 140.0 512.1 0.1%

Jamaica Safe Bets Toss-up Toss-up (China) 201.3 76.6 124.8 0.0%

Jordan Toss-up Toss-up 772.1 32.2 739.9 0.2%

Kazakhstan Toss-up (China) Toss-up Toss-up (China) 1,331.3 5,501.6 (4,170.4) -1.1%

Kenya Toss-up Toss-up Toss-up (China) 3,676.9 1,310.8 2,366.2 0.6%

374
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Kiribati Safe Bets 9.0 161.4 (152.5) 0.0%

Kyrgyz
Toss-up Moonshot Moonshot 465.2 230.1 235.1 0.1%
Republic

Laos Toss-up (China) Moonshot 519.0 5,734.9 (5,216.0) -1.4%

Lebanon Moonshot Safe Bets Toss-up 410.5 334.1 76.4 0.0%

Lesotho Safe Bets Moonshot Toss-up 160.0 279.7 (119.7) 0.0%

Liberia Moonshot Toss-up (China) Safe Bets 359.9 139.8 220.1 0.1%

Libya Safe Bets Toss-up (China) Toss-up 472.1 10.7 461.4 0.1%

Madagascar Toss-up Safe Bets Toss-up (China) 1,991.5 352.2 1,639.4 0.4%

Malawi Toss-up Moonshot Safe Bets 1,366.2 422.3 943.9 0.3%

Malaysia Moonshot Toss-up Toss-up (China) 2,356.9 (10,235.3 (7,878.4) -2.1%

Maldives Safe Bets Toss-up (China) 36.3 195.5 (159.2) 0.0%

Mali Toss-up Safe Bets Moonshot 1,493.5 203.5 1,290.0 0.3%

Marshall
Safe Bets 3.1 1,379.5 (1,376.3) -0.4%
Islands

Mauritania Toss-up (China) Safe Bets Toss-up 317.3 292.9 24.4 0.0%

Mauritius Toss-up (China) Safe Bets Toss-up (China) 90.4 205.3 (114.9) 0.0%

Mexico Safe Bets Toss-up Toss-up (China) 8,961.3 1,255.2 7,706.1 2.1%

Micronesia Safe Bets Safe Bets 8.0 81.0 (73.0) 0.0%

Moldova Safe Bets Moonshot Safe Bets 189.7 2.8 186.9 0.1%

Mongolia Toss-up (China) Moonshot Toss-up (China) 232.8 8,698.4 (8,465.6) -2.3%

Montenegro Moonshot Moonshot 44.4 70.3 (26.0) 0.0%

Morocco Safe Bets Moonshot Toss-up (China) 2,607.2 334.6 2,272.6 0.6%

Mozambique Safe Bets Toss-up Toss-up 2,195.9 416.5 1,779.5 0.5%

375
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Myanmar Toss-up (China) Toss-up Moonshot 3,802.5 679.9 3,122.6 0.8%

Namibia Safe Bets Moonshot Toss-up (China) 176.3 146.6 29.7 0.0%

Nauru Toss-up (China) 0.9 0.0 0.9 0.0%

Nepal Toss-up Moonshot Toss-up 2,084.5 550.9 1,533.6 0.4%

Nicaragua Safe Bets Toss-up (China) Toss-up 479.4 32.6 446.8 0.1%

Niger Moonshot Toss-up (China) Moonshot 1,707.4 1,201.0 506.4 0.1%

Nigeria Toss-up Toss-up Toss-up (14,704.8 6,488.8 8,216.0 2.2%

North
Toss-up (China) Safe Bets 148.1 239.6 (91.6) 0.0%
Macedonia

Oman Safe Bets Toss-up 326.2 732.5 (406.3) -0.1%

Pakistan Toss-up (China) Toss-up Moonshot (16,101.3 (43,114.6 (27,013.3) -7.3%

Panama Safe Bets Toss-up Safe Bets 304.4 385.6 (81.2) 0.0%

Papua New
Moonshot Safe Bets 688.8 276.6 412.2 0.1%
Guinea

Paraguay Toss-up (China) Safe Bets Safe Bets 469.6 3.3 466.3 0.1%

Peru Safe Bets Safe Bets Safe Bets 2,358.2 4,566.7 (2,208.5) -0.6%

Philippines Toss-up (China) Toss-up (China) Toss-up (China) 7,947.2 2,223.9 5,723.3 1.6%

Romania Toss-up (China) Moonshot 1,379.2 181.9 1,197.3 0.3%

Russia Safe Bets Toss-up Toss-up (China) (10,293.5 (18,081.5 (7,788.0) -2.1%

Rwanda Toss-up Moonshot Moonshot 928.2 480.0 448.2 0.1%

Samoa Moonshot Safe Bets 15.3 2.0 13.3 0.0%

Sao Tome
Safe Bets Toss-up (China) 15.5 95.8 (80.3) 0.0%
and Principe

Senegal Toss-up (China) Toss-up Moonshot 1,158.8 1,410.4 (251.6) -0.1%

Serbia Safe Bets Toss-up (China) Safe Bets 494.0 2,403.6 (1,909.6) -0.5%

376
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Seychelles Safe Bets Toss-up (China) 7.0 7.3 (0.3) 0.0%

Sierra Leone Toss-up Safe Bets Toss-up (China) 581.5 2,213.5 (1,632.0) -0.4%

Solomon
Toss-up (China) 48.8 162.0 (113.2) 0.0%
Islands

Somalia Moonshot Safe Bets Moonshot 1,160.7 26.9 1,133.8 0.3%

South Africa Toss-up (China) Toss-up (China) Toss-up (China) 4,172.0 7,312.3 (3,140.3) -0.9%

South Sudan Moonshot Toss-up Moonshot 753.6 1,312.3 (558.8) -0.2%

Sri Lanka Moonshot Toss-up Toss-up (China) 1,563.4 6,793.4 (5,230.0) -1.4%

Sudan Moonshot Toss-up (China) Toss-up 3,131.0 101.4 3,029.6 0.8%

Suriname Moonshot Toss-up 43.1 947.6 (904.5) -0.2%

Syrian Arab
Safe Bets Toss-up 1,456.0 34.5 1,421.5 0.4%
Republic

Tajikistan Moonshot Safe Bets Moonshot 674.3 1,479.9 (805.6) -0.2%

Tanzania Toss-up Toss-up Toss-up (China) 4,344.1 578.3 3,765.8 1.0%

Thailand Toss-up (China) Toss-up Safe Bets 5,098.7 319.2 4,779.6 1.3%

Timor-Leste Toss-up (China) 92.2 5.5 86.7 0.0%

Togo Moonshot Safe Bets Toss-up (China) 596.1 35.1 560.9 0.2%

Tonga Toss-up Safe Bets 7.5 43.4 (35.9) 0.0%

Trinidad and
Moonshot Toss-up (China) 108.4 142.4 (34.0) 0.0%
Tobago

Tunisia Toss-up (China) Moonshot Toss-up 864.5 112.5 751.9 0.2%

Turkey Moonshot Toss-up (China) Moonshot 5,986.1 (17,859.1 (11,873.0) -3.2%

Turkmenistan Toss-up Toss-up (China) Toss-up 443.2 1.6 441.5 0.1%

Uganda Toss-up Moonshot Toss-up 3,120.2 948.9 2,171.3 0.6%

Ukraine Moonshot Safe Bets Moonshot 3,159.7 598.5 2,561.2 0.7%

377
 Early BRI (2014-2017) Late BRI (2018-2021)
(Observed Soft Power Changes) (Expected and Observed Allocations)
Difference
Expected Observed
(percent of
Media ODA/OOF ODA/OOF Difference
Country Public Opinion Elite Support total
Sentiment Allocation Allocation (USD millions)
Chinese
(USD millions) (USD millions)
ODA/OOF)

Uruguay Safe Bets Toss-up (China) Safe Bets 244.8 36.4 208.4 0.1%

Uzbekistan Toss-up (China) Safe Bets Toss-up (China) 2,423.1 5,152.0 (2,729.0) -0.7%

Vanuatu Safe Bets Safe Bets 22.0 176.1 (154.1) 0.0%

Venezuela Moonshot Toss-up Moonshot 2,062.4 319.1 1,743.3 0.5%

Viet Nam Moonshot Toss-up Toss-up 6,872.0 3,313.0 3,559.0 1.0%

West Bank
and Gaza Toss-up Toss-up 339.0 12.4 326.6 0.1%
Strip

Yemen Moonshot Moonshot Toss-up 2,278.8 27.3 2,251.4 0.6%

Zambia Toss-up (China) Toss-up (China) Moonshot 1,332.5 3,060.8 (1,728.3) -0.5%

Zimbabwe Toss-up Toss-up Moonshot 1,108.5 429.5 679.0 0.2%

378
Section A-13: China’s official sector lending portfolio in LICs
and MICs: a comparison of AidData, IDS, and CODF
Table A15 presents the aggregate monetary value (in constant 2021 USD) of loan commitments from official
sector creditors in China to borrowers in LICs and MICs, as measured by three different sources: the 3.0
version of AidData’s GCDF dataset, the World Bank’s International Debt Statistics (IDS), and Boston
University’s China’s Overseas Development Finance (CODF) dataset.393 The country-level summary statistics
are organized by level of public liability.394 IDS and CODF provide data on public and publicly-guaranteed
debt (PPG) for several official sector creditors in China. IDS provides coverage for 89 LICs and MICs
(excluding China) from 2000 to 2021 and CODF provides coverage for 96 LICs and MICs from 2008 to
2021. The 3.0 version of AidData’s GCDF dataset covers PPG and non-PPG debt from 180 official sector
creditors in China to 165 LICs and MICs (of which 126 contracted loans from official sector creditors in
China) from 2000 to 2021.395

AidData measures PPG debt by aggregating lending commitments to government and majority
state-owned institutions as well as other institutions that secured central government repayment guarantees
or repayment guarantees from state-owned entities other than the central government in the host country
(i.e., by using the “Level of Public Liability” variable in the 3.0 version to the GCDF dataset to identify all
loan commitments assigned to the “central government debt,” “central government-guaranteed debt,”
and “other public sector debt" categories). In addition to PPG debt, the “Level of Public Liability” variable
in the 3.0 version of AidData’s GCDF dataset captures “potential public sector debt” (loans to minority
state-owned institutions without public sector repayment guarantees), “private debt,” and debt that cannot
be easily categorized based on the level of public liability (referred to as “unallocable” debt in Table A15).
The “Total Debt to China” column in Table A15) represents the sum of all loan commitments from official
sector creditors in China to borrowers in LICs and MICs between 2000 and 2021 that qualify as PPG or
non-PPG debt.

393
As of August 2023, the Chinese Loans to Africa (CLA) database captured $203 billion in Chinese lending
commitments to government and majority state-owned institutions in Africa from 2000 to 2020 (deflated to
constant USD 2021). Over the same time period, the 3.0 version of AidData’s GCDF dataset captures $266
billion in Chinese lending commitments to government and majority state-owned institutions as well as
other institutions that secured central government repayment guarantees or repayment guarantees from
state-owned entities other than the central government in the host country (i.e., “central government
debt,” “central government-guaranteed debt,” and “other public sector debt") in Africa. Additionally, in
2021, AidData records $5.3 billion in Chinese lending commitments to Africa that qualify as “central
government debt,” “central government-guaranteed debt,” and “other public sector debt.” AidData also
captures Chinese lending commitments to Africa that are excluded from the CLA database by definition:
$31 billion to private sector borrowing institutions (“private debt”) in Africa and $8.6 billion to special
purpose vehicle and joint venture borrowers that are minority-owned by public sector institutions in Africa
without repayment guarantees from public sector institutions in host countries (“potential public sector
debt”) between 2000 and 2021.
394
Section A-5 in the Appendix provides more details on how AidData uses the “level of public liability”
measure to categorize lending to different types of borrowers.
395
In Table A15, the loan commitment totals from the 3.0 version of AidData’s GCDF dataset exclude the
short-term "rollover" facilities described in Box 2c and Section A-3.
379
 The AidData, CODF and IDS estimates of PPG and non-PPG debt to China in Table A15 should be
interpreted with caution since they do not represent amounts outstanding and as such do not account for
disbursements or repayments.396

Table A15: Official sector lending commitments from China to LICs and MICs: Country-by-country comparison of AidData, IDS, and CODF in constant 2021 USD millions

Table A.15

Official sector lending commitments from China to LICs and MICs:

Country-by-country comparison of AidData, IDS, and CODF in constant 2021
USD millions
Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Albania 100 NA NA 0 0 0 100

Algeria 114 2 NA 0 0 0 114

Angola 63,022 38,590 38,715 1,422 9 362 64,816

Antigua and 310 NA 221 0 0 9 320

Barbuda
Argentina 32,870 8,581 21,723 38 3,331 1,484 37,724

Armenia 28 28 NA 0 0 0 28

Azerbaijan 288 NA NA 776 118 0 1,182

Bahamas 132 NA 119 0 3,412 0 3,544

Bangladesh 17,168 11,845 17,301 0 2,584 318 20,070

Barbados 216 NA 191 0 0 0 216

Belarus 10,822 12,633 8,491 124 118 0 11,064

Benin 1,397 1,037 390 24 840 0 2,260

Bolivia 2,382 2,923 3,744 0 6 0 2,388

396
AidData recently launched a new data collection initiative to track disbursements, repayments, and
amounts outstanding on a loan-by-loan basis. In the 3.0 version of the GCDF dataset, we have documented
disbursements, repayments, and amounts outstanding in the “description” field for a subset of countries.
However, in the future, we intend to publish loan-level data on disbursements, repayments, and amounts
outstanding in a user-friendly format for a more complete set of countries.
380
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Bosnia and 1,359 NA 1,016 0 773 0 2,132

Herzegovina
Botswana 1,660 194 NA 0 0 0 1,660

Brazil 39,240 23,824 36,180 0 13,597 1,505 54,341

Brunei 0 NA NA 2,102 24 0 2,126

Bulgaria 911 50 268 0 102 357 1,371

Burkina Faso 242 178 89 0 0 0 242

Burundi 48 198 NA 38 0 24 109

Cabo Verde 198 63 NA 0 0 0 198

Cambodia 6,173 6,776 4,306 1,278 8,792 51 16,293

Cameroon 7,444 6,899 6,454 44 943 3 8,433

Central African 266 93 NA 0 200 0 466

Republic
Chad 1,073 603 771 86 0 56 1,214

Chile 312 NA NA 719 5,173 245 6,449

Colombia 296 NA NA 76 1,169 56 1,597

Comoros 138 138 134 0 0 0 138

Congo (Republic) 7,149 2,892 4,502 0 59 50 7,258

Cook Islands 38 NA NA 0 0 0 38

Costa Rica 753 265 504 0 0 0 753

Cote d'Ivoire 7,014 5,404 3,888 0 0 0 7,014

Cuba 4,086 NA 427 0 84 0 4,171

North Korea 76 NA NA 0 0 0 76

381
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Congo (DRC) 4,053 8,130 2,356 1,217 7,227 641 13,137

Djibouti 2,171 1,598 1,423 0 0 0 2,171

Dominica 76 58 60 0 0 0 76

Dominican 0 NA 654 0 0 139 139

Republic
Ecuador 25,672 15,451 22,471 0 123 458 26,254

Egypt 14,360 6,503 8,593 0 489 137 14,986

El Salvador 0 NA NA 0 53 79 132

Equatorial 9,009 NA 2,581 0 0 0 9,009

Guinea
Eritrea 1,053 270 626 121 0 1,712 2,886

Ethiopia 17,440 12,154 9,728 0 2,887 91 20,419

Fiji 444 390 664 0 0 0 444

Gabon 2,868 2,123 1,587 12 0 0 2,880

Gambia 30 30 27 0 0 0 30

Georgia 9 9 NA 0 127 330 466

Ghana 7,854 7,911 3,084 543 1,412 2 9,812

Grenada 84 79 76 0 0 0 84

Guinea 3,016 2,181 2,263 0 927 824 4,768

Guinea-Bissau 27 NA NA 0 0 0 27

Guyana 408 493 342 0 210 82 700

Honduras 336 NA NA 0 114 0 450

Hungary NA NA 2,122 NA NA NA NA

382
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

India 968 NA NA 0 9,672 14 10,654

Indonesia 22,620 4,503 17,267 3,250 27,213 1,917 54,999

Iran 24,825 2,310 33,399 0 430 2,696 27,951

Iraq 6,782 NA NA 0 637 393 7,812

Israel 0 NA NA 0 2,787 2,866 5,653

Jamaica 1,704 1,709 2,438 0 517 201 2,422

Jordan 36 72 NA 0 1,948 0 1,984

Kazakhstan 27,113 4,856 22,126 27,508 3,741 5,818 64,180

Kenya 11,793 11,472 10,568 0 860 0 12,653

Kiribati 0 NA NA 115 0 0 115

Kyrgyzstan 2,698 2,546 2,717 0 689 0 3,386

Laos 10,440 9,253 10,870 6,526 2,586 1,024 20,577

Lebanon 317 NA NA 0 0 0 317

Lesotho 317 295 278 0 0 0 317

Liberia 62 72 58 0 0 564 626

Libya 451 NA NA 0 0 0 451

Madagascar 557 564 461 0 15 0 572

Malawi 641 464 318 0 49 0 689

Malaysia 12,584 NA 5,277 0 2,563 789 15,936

Maldives 1,916 1,253 1,106 0 0 0 1,916

Mali 1,257 1,207 687 0 0 0 1,257

383
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Marshall Islands 144 NA NA 0 4,989 0 5,133

Mauritania 963 1,223 817 0 165 0 1,129

Mauritius 887 833 585 0 283 71 1,241

Mexico 832 NA 1,128 0 2,759 474 4,064

Micronesia 4 NA NA 0 0 0 4

Moldova 16 NA NA 0 0 0 16

Mongolia 5,478 1,806 2,241 0 329 751 6,558

Montenegro 1,163 1,166 1,163 0 45 24 1,233

Morocco 1,586 1,502 1,422 0 518 0 2,104

Mozambique 3,177 3,758 2,581 1,962 381 93 5,613

Myanmar 10,640 3,469 5,760 243 2,180 638 13,701

Namibia 2,020 NA 139 0 0 155 2,175

Nepal 498 418 438 0 346 0 844

Nicaragua 0 NA NA 0 20 0 20

Niger 1,719 1,746 387 1,397 250 0 3,366

Nigeria 12,472 8,767 7,715 836 1,144 75 14,528

North 1,319 896 884 0 0 27 1,346

Macedonia
Oman 6,944 NA 3,687 186 0 0 7,130

Pakistan 67,221 45,999 26,001 0 910 787 68,918

Panama 105 NA NA 0 534 0 640

Papua New 1,650 1,553 1,581 2,989 1,986 300 6,924

Guinea

384
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Paraguay 56 NA NA 0 20 0 76

Peru 421 NA 75 0 14,177 2,293 16,890

Philippines 2,504 3,180 1,247 1,181 4,090 0 7,774

Romania 292 NA NA 0 285 146 722

Russia 129,191 NA 75,822 0 37,184 2,883 169,258

Rwanda 719 652 647 0 26 0 745

Samoa 288 272 233 0 0 0 288

Senegal 3,610 2,938 2,071 0 0 49 3,660

Serbia 4,776 4,406 2,704 0 68 869 5,712

Seychelles 2 NA NA 0 0 0 2

Sierra Leone 878 113 34 195 2,051 1,409 4,533

South Africa 11,678 4,812 5,133 0 9,088 561 21,326

South Sudan 5,283 NA 446 0 0 0 5,283

Sri Lanka 17,356 13,097 14,420 430 1,118 567 19,472

Sudan 18,001 5,926 2,150 0 0 0 18,001

Suriname 1,514 NA 967 0 0 0 1,514

Syria 150 77 55 0 49 2,228 2,427

Tajikistan 3,595 2,413 1,165 683 599 231 5,108

Tanzania 2,859 3,017 2,422 0 0 0 2,859

Thailand 3,267 NA NA 0 1,557 241 5,065

Timor-Leste 0 NA 56 0 0 0 0

385
 Country Total PPG Total PPG Total PPG Potential Public Private Unallocable Total
Debt to Debt to Debt to Sector Debt to Sector Debt Debt to Debt to
China China China China to China China China
Source AidData IDS CODF AidData AidData AidData AidData

Year Range 2000-2021 2000-2021 2008-2021 2000-2021 2000-2021 2000-2021 2000-202

1

Togo 1,128 890 655 0 57 0 1,185

Tonga 193 176 126 0 0 0 193

Trinidad and 646 NA 761 0 0 0 646

Tobago
Tunisia 314 189 120 0 0 0 314

Türkiye 17,664 3,093 1,389 789 9,834 45 28,331

Turkmenistan 12,217 11,727 9,257 0 0 0 12,217

Uganda 4,169 4,137 3,933 0 0 269 4,438

Ukraine 3,007 1,761 2,571 0 230 101 3,339

Uruguay 130 NA NA 0 109 23 262

Uzbekistan 10,232 3,920 9,136 7,209 367 205 18,013

Vanuatu 323 270 260 0 0 0 323

Venezuela 112,782 NA 71,341 0 0 0 112,782

Vietnam 18,595 5,365 11,613 2,362 7,738 102 28,797

Yemen 357 448 NA 0 0 0 357

Zambia 12,317 7,141 7,377 696 310 147 13,469

Zimbabwe 3,899 3,873 2,869 10 637 444 4,990

Regional 57 NA 2,048 0 1,322 8,449 9,828

Total 1,109,793 378,195 605,294 67,184 216,362 49,954 1,280,125

Section A-14: Comparison of AidData and IDS estimates of

PPG and non-PPG debt exposure to China
Table A16 compares total loan commitments (in constant 2021 USD millions and as a percentage of GDP)
from official sector creditors in China to borrowers in LICs and MICs from 2000-2021, as captured by the 3.0

386
version of AidData’s GCDF dataset and the World Bank’s IDS.397 The country-level summary statistics are
organized by level of public liability.398

89 LICs and MICs (excluding China) voluntarily report on public and publicly-guaranteed debt (PPG) to the
World Bank through its Debtor Reporting System (DRS). The World Bank’s definition of PPG debt includes
(a) long-term external obligations of public debtors, including the national government, a political
subdivision (or an agency of either), and autonomous public bodies; and (b) long-term external obligations
of private debtors that are guaranteed for repayment by a public entity (World Bank 2000). “Public
debtors” include entities in which the host government holds at least fifty percent ownership.

To maximize comparability with IDS, the 3.0 version of AidData’s GCDF dataset classifies all loans according
to their levels of public liability. AidData measures PPG debt by aggregating lending commitments to
government and majority state-owned institutions as well as other institutions that secured central
government repayment guarantees or repayment guarantees from state-owned entities other than the
central government in the host country (i.e., by using the “Level of Public Liability” variable in the 3.0
version of the GCDF dataset to identify all loan commitments assigned to the “central government debt,”
“central government-guaranteed debt,” and “other public sector debt" categories).

While similar to Table A27 in Malik et al. (2021), Table A16 in this report represents an updated
methodological approach. Malik et al. (2021) defines “sovereign debt” as central government and
central-government guaranteed debt, and it defines “hidden debt” as debt incurred by state-owned
entities (including SPVs) with any level of host government ownership. In Table A16, we align more closely
with the DRS definition of PPG debt by separately recording loans to majority state-owned entities
(including SPVs)—with and without repayment guarantees from public sector institutions other than the
central government—as well as private sector lending that is guaranteed by majority state-owned entities in
a new (“other public sector debt”) category. We classify loans to minority state-owned institutions without
public sector repayment guarantees as “potential public sector debt” since such loans may benefit from
implicit forms of host government liability protection (Malik et al. 2021; Malik and Parks 2021).

The AidData and IDS estimates of PPG and non-PPG debt exposure to China that are reported in Table
A16 should be interpreted with caution as they are based on cumulative loan commitments over a 22-year
period. They do not represent amounts outstanding and as such do not account for disbursements or
repayments.399

397
In Table A16, cumulative loan commitments from the 3.0 version of AidData’s GCDF dataset exclude the
short-term "rollover" facilities described in Box 2c and Section A-3.
398
Section A-5 in the Appendix provides more details on how AidData uses the “level of public liability”
measure to categorize lending to different types of borrowers.
399
AidData recently launched a new data collection initiative to track disbursements, repayments, and
amounts outstanding on a loan-by-loan basis. In the 3.0 version of the GCDF dataset, we have documented
disbursements, repayments, and amounts outstanding in the “description” field for a subset of countries.
However, in the future, we intend to publish loan-level data on disbursements, repayments, and amounts
outstanding in a user-friendly format for a more complete set of countries.
387
 Table A16: Country-by-country comparison of AidData and IDS estimates of PPG and non-PPG debt exposure to China in constant 2021 USD millions

Table A.16

Country-by-country comparison of AidData and IDS estimates of PPG and non-PPG debt exposure to China in
constant 2021 USD millions

(1) (2) (3)

PPG debt PPG debt PPG and potential Difference (1b) (2b) (3b) Difference
exposure to exposure to public debt between PPG debt to PPG debt to PPG and potential between
Reports China China exposure to China AidData (3) China as a % of China as a % public debt to China as AidData (3b)
Country to DRS (AidData) (DRS) (AidData) and DRS (2) GDP (AidData) of GDP (DRS) % of GDP (AidData) and DRS (2b)
Afghanistan N 0 NA 0 0 0.0% NA 0.0% 0.0%
Albania N 100 NA 100 100 0.5% NA 0.5% 0.5%
Algeria Y 114 2 114 112 0.1% 0.0% 0.1% 0.1%
American Samoa N 0 NA 0 0 0.0% NA 0.0% 0.0%
Angola Y 63,022 38,590 64,444 25,854 89.4% 54.7% 91.4% 36.7%
Antigua and Barbuda N 310 NA 310 310 21.8% NA 21.8% 21.8%
Argentina Y 32,870 8,581 32,909 24,328 6.7% 1.8% 6.8% 5.0%
Armenia Y 28 28 28 (0) 0.2% 0.2% 0.2% 0.0%
Aruba N 0 NA 0 0 0.0% NA 0.0% 0.0%
Azerbaijan N 288 NA 1,064 1,064 0.5% NA 1.9% 1.9%
Bahamas N 132 NA 132 132 1.2% NA 1.2% 1.2%
Bangladesh Y 17,168 11,845 17,168 5,323 4.1% 2.9% 4.1% 1.3%
Barbados N 216 NA 216 216 4.5% NA 4.5% 4.5%
Belarus Y 10,822 12,633 10,946 (1,687) 15.9% 18.5% 16.0% -2.5%
Belize N 0 NA 0 0 0.0% NA 0.0% 0.0%
Benin Y 1,397 1,037 1,420 383 7.9% 5.9% 8.0% 2.2%
Bhutan N 0 NA 0 0 0.0% NA 0.0% 0.0%
Bolivia Y 2,382 2,923 2,382 (541) 5.9% 7.2% 5.9% -1.3%
Bosnia and Herzegovina N 1,359 NA 1,359 1,359 5.8% NA 5.8% 5.8%
Botswana Y 1,660 194 1,660 1,466 9.4% 1.1% 9.4% 8.3%
Brazil Y 39,240 23,824 39,240 15,416 2.4% 1.5% 2.4% 1.0%
British Virgin Islands N 0 NA 0 0 0.0% NA 0.0% 0.0%

388
Brunei N 0 NA 2,102 2,102 0.0% NA 15.0% 15.0%
Bulgaria Y 911 50 911 861 1.1% 0.1% 1.1% 1.0%
Burkina Faso Y 242 178 242 65 1.2% 0.9% 1.2% 0.3%
Burundi Y 48 198 85 (113) 1.2% 5.1% 2.2% -2.9%
Cabo Verde Y 198 63 198 135 10.2% 3.2% 10.2% 7.0%
Cambodia Y 6,173 6,776 7,451 675 23.1% 25.4% 27.9% 2.5%
Cameroon Y 7,444 6,899 7,488 589 16.4% 15.2% 16.5% 1.3%
Cayman Islands N 0 NA 0 0 0.0% NA 0.0% 0.0%
Central African Republic Y 266 93 266 173 10.6% 3.7% 10.6% 6.9%
Chad Y 1,073 603 1,159 555 6.5% 3.7% 7.1% 3.4%
Chile N 312 NA 1,031 1,031 0.1% NA 0.3% 0.3%
Colombia N 296 NA 372 372 0.1% NA 0.1% 0.1%
Comoros Y 138 138 138 (0) 10.3% 10.3% 10.3% 0.0%
Congo (DRC) Y 4,053 8,130 5,270 (2,860) 7.7% 15.4% 10.0% -5.4%
Congo (Republic) Y 7,149 2,892 7,149 4,258 55.7% 22.5% 55.7% 33.2%
Cook Islands N 38 NA 38 38 11.7% NA 11.7% 11.7%
Costa Rica Y 753 265 753 489 1.2% 0.4% 1.2% 0.8%
Cote d'Ivoire Y 7,014 5,404 7,014 1,610 10.1% 7.7% 10.1% 2.3%
Cuba N 4,086 NA 4,086 4,086 3.2% NA 3.2% 3.2%
Curaçao N 0 NA 0 0 0.0% NA 0.0% 0.0%
Djibouti Y 2,171 1,598 2,171 573 58.6% 43.2% 58.6% 15.5%
Dominica Y 76 58 76 18 13.8% 10.5% 13.8% 3.3%
Dominican Republic N 0 NA 0 0 0.0% NA 0.0% 0.0%
Ecuador Y 25,672 15,451 25,672 10,221 24.2% 14.6% 24.2% 9.6%
Egypt Y 14,360 6,503 14,360 7,857 3.4% 1.5% 3.4% 1.8%
El Salvador N 0 NA 0 0 0.0% NA 0.0% 0.0%
Equatorial Guinea N 9,009 NA 9,009 9,009 72.5% NA 72.5% 72.5%
Eritrea Y 1,053 270 1,174 904 46.7% 12.0% 52.1% 40.1%
Eswatini N 0 NA 0 0 0.0% NA 0.0% 0.0%
Ethiopia Y 17,440 12,154 17,440 5,286 17.6% 12.2% 17.6% 5.3%
Fiji Y 444 390 444 54 10.3% 9.1% 10.3% 1.3%
French Polynesia N 0 NA 0 0 0.0% NA 0.0% 0.0%
Gabon Y 2,868 2,123 2,880 757 15.5% 11.5% 15.6% 4.1%

389
Gambia Y 30 30 30 (0) 1.5% 1.5% 1.5% 0.0%
Georgia Y 9 9 9 (0) 0.0% 0.0% 0.0% 0.0%
Ghana Y 7,854 7,911 8,397 487 9.9% 10.0% 10.6% 0.6%
Grenada Y 84 79 84 6 7.6% 7.0% 7.6% 0.5%
Guam N 0 NA 0 0 0.0% NA 0.0% 0.0%
Guatemala N 0 NA 0 0 0.0% NA 0.0% 0.0%
Guinea Y 3,016 2,181 3,016 836 18.8% 13.6% 18.8% 5.2%
Guinea-Bissau N 27 NA 27 27 1.7% NA 1.7% 1.7%
Guyana Y 408 493 408 (85) 5.1% 6.1% 5.1% -1.1%
Haiti N 0 NA 0 0 0.0% NA 0.0% 0.0%
Honduras N 336 NA 336 336 1.2% NA 1.2% 1.2%
India N 968 NA 968 968 0.0% NA 0.0% 0.0%
Indonesia Y 22,620 4,503 25,869 21,366 1.9% 0.4% 2.2% 1.8%
Iran Y 24,825 2,310 24,825 22,515 4.2% 0.4% 4.2% 3.8%
Iraq N 6,782 NA 6,782 6,782 3.3% NA 3.3% 3.3%
Israel N 0 NA 0 0 0.0% NA 0.0% 0.0%
Jamaica Y 1,704 1,709 1,704 (5) 11.6% 11.7% 11.6% 0.0%
Jordan Y 36 72 36 (36) 0.1% 0.2% 0.1% -0.1%
Kazakhstan Y 27,113 4,856 54,621 49,765 14.0% 2.5% 28.3% 25.8%
Kenya Y 11,793 11,472 11,793 321 10.7% 10.4% 10.7% 0.3%
Kiribati N 0 NA 115 115 0.0% NA 50.4% 50.4%
Kosovo N 0 NA 0 0 0.0% NA 0.0% 0.0%
Kyrgyzstan Y 2,698 2,546 2,698 152 30.9% 29.1% 30.9% 1.7%
Laos Y 10,440 9,253 16,966 7,713 54.7% 48.5% 88.9% 40.4%
Lebanon N 317 NA 317 317 0.8% NA 0.8% 0.8%
Lesotho Y 317 295 317 22 13.4% 12.4% 13.4% 0.9%
Liberia Y 62 72 62 (10) 2.5% 3.0% 2.5% -0.4%
Libya N 451 NA 451 451 1.2% NA 1.2% 1.2%
Madagascar Y 557 564 557 (7) 3.9% 3.9% 3.9% 0.0%
Malawi Y 641 464 641 177 5.3% 3.8% 5.3% 1.4%
Malaysia N 12,584 NA 12,584 12,584 3.4% NA 3.4% 3.4%
Maldives Y 1,916 1,253 1,916 663 35.5% 23.2% 35.5% 12.3%
Mali Y 1,257 1,207 1,257 49 6.6% 6.3% 6.6% 0.3%

390
Marshall Islands N 144 NA 144 144 56.0% NA 56.0% 56.0%
Mauritania Y 963 1,223 963 (260) 9.6% 12.2% 9.6% -2.6%
Mauritius Y 887 833 887 54 7.7% 7.2% 7.7% 0.5%
Mexico N 832 NA 832 832 0.1% NA 0.1% 0.1%
Micronesia N 4 NA 4 4 0.3% NA 0.3% 0.3%
Moldova N 16 NA 16 16 0.1% NA 0.1% 0.1%
Mongolia Y 5,478 1,806 5,478 3,672 36.3% 12.0% 36.3% 24.3%
Montenegro Y 1,163 1,166 1,163 (2) 20.0% 20.1% 20.0% 0.0%
Morocco Y 1,586 1,502 1,586 84 1.1% 1.1% 1.1% 0.1%
Mozambique Y 3,177 3,758 5,139 1,382 20.1% 23.8% 32.6% 8.8%
Myanmar Y 10,640 3,469 10,883 7,414 18.2% 5.9% 18.6% 12.7%
N. Mariana Islands N 0 NA 0 0 0.0% NA 0.0% 0.0%
Namibia N 2,020 NA 2,020 2,020 16.5% NA 16.5% 16.5%
Nauru N 0 NA 0 0 0.0% NA 0.0% 0.0%
Nepal Y 498 418 498 80 1.4% 1.2% 1.4% 0.2%
New Caledonia N 0 NA 0 0 0.0% NA 0.0% 0.0%
Nicaragua N 0 NA 0 0 0.0% NA 0.0% 0.0%
Niger Y 1,719 1,746 3,116 1,370 11.5% 11.7% 20.9% 9.2%
Nigeria Y 12,472 8,767 13,308 4,541 2.9% 2.0% 3.1% 1.1%
Niue N 0 NA 0 0 0.0% NA 0.0% 0.0%
North Korea N 76 NA 76 76 0.5% NA 0.5% 0.5%
North Macedonia Y 1,319 896 1,319 423 9.5% 6.5% 9.5% 3.0%
Oman N 6,944 NA 7,130 7,130 7.9% NA 8.1% 8.1%
Pakistan Y 67,221 45,999 67,221 21,222 19.6% 13.4% 19.6% 6.2%
Palau N 0 NA 0 0 0.0% NA 0.0% 0.0%
Panama N 105 NA 105 105 0.2% NA 0.2% 0.2%
Papua New Guinea Y 1,650 1,553 4,639 3,086 6.2% 5.8% 17.4% 11.6%
Paraguay N 56 NA 56 56 0.1% NA 0.1% 0.1%
Peru N 421 NA 421 421 0.2% NA 0.2% 0.2%
Philippines Y 2,504 3,180 3,685 505 0.6% 0.8% 0.9% 0.1%
Puerto Rico N 0 NA 0 0 0.0% NA 0.0% 0.0%
Romania N 292 NA 292 292 0.1% NA 0.1% 0.1%
Russia N 129,191 NA 129,191 129,191 7.3% NA 7.3% 7.3%

391
Rwanda Y 719 652 719 67 6.5% 5.9% 6.5% 0.6%
Saint-Martin N 0 NA 0 0 0.0% NA 0.0% 0.0%
Samoa Y 288 272 288 16 33.6% 31.7% 33.6% 1.9%
Sao Tome and Principe N 0 NA 0 0 0.0% NA 0.0% 0.0%
Senegal Y 3,610 2,938 3,610 672 13.1% 10.6% 13.1% 2.4%
Serbia Y 4,776 4,406 4,776 370 7.6% 7.0% 7.6% 0.6%
Seychelles N 2 NA 2 2 0.2% NA 0.2% 0.2%
Sierra Leone Y 878 113 1,073 960 20.7% 2.7% 25.3% 22.6%
Sint Maarten N 0 NA 0 0 0.0% NA 0.0% 0.0%
Solomon Islands N 0 NA 0 0 0.0% NA 0.0% 0.0%
Somalia N 0 NA 0 0 0.0% NA 0.0% 0.0%
South Africa Y 11,678 4,812 11,678 6,866 2.8% 1.1% 2.8% 1.6%
South Sudan N 5,283 NA 5,283 5,283 122.8% NA 122.8% 122.8%
Sri Lanka Y 17,356 13,097 17,787 4,690 20.3% 15.4% 20.8% 5.5%
St. Kitts and Nevis N 0 NA 0 0 0.0% NA 0.0% 0.0%
St. Lucia N 0 NA 0 0 0.0% NA 0.0% 0.0%
St. Vicent N 0 NA 0 0 0.0% NA 0.0% 0.0%
Sudan Y 18,001 5,926 18,001 12,074 50.2% 16.5% 50.2% 33.7%
Suriname N 1,514 NA 1,514 1,514 47.0% NA 47.0% 47.0%
Syria Y 150 77 150 74 0.8% 0.4% 0.8% 0.4%
Tajikistan Y 3,595 2,413 4,278 1,865 41.1% 27.6% 48.9% 21.3%
Tanzania Y 2,859 3,017 2,859 (157) 4.1% 4.3% 4.1% -0.2%
Thailand N 3,267 NA 3,267 3,267 0.6% NA 0.6% 0.6%
Timor-Leste N 0 NA 0 0 0.0% NA 0.0% 0.0%
Togo Y 1,128 890 1,128 238 13.8% 10.9% 13.8% 2.9%
Tonga Y 193 176 193 18 41.0% 37.2% 41.0% 3.8%
Trinidad and Tobago N 646 NA 646 646 2.6% NA 2.6% 2.6%
Tunisia Y 314 189 314 125 0.7% 0.4% 0.7% 0.3%
Türkiye Y 17,664 3,093 18,453 15,360 2.2% 0.4% 2.3% 1.9%
Turkmenistan Y 12,217 11,727 12,217 490 22.6% 21.7% 22.6% 0.9%
Turks and Caicos N 0 NA 0 0 0.0% NA 0.0% 0.0%
Tuvalu N 0 NA 0 0 0.0% NA 0.0% 0.0%
U.S. Virgin Islands N 0 NA 0 0 0.0% NA 0.0% 0.0%

392
Uganda Y 4,169 4,137 4,169 32 9.8% 9.7% 9.8% 0.1%
Ukraine Y 3,007 1,761 3,007 1,246 1.5% 0.9% 1.5% 0.6%
Uruguay N 130 NA 130 130 0.2% NA 0.2% 0.2%
Uzbekistan Y 10,232 3,920 17,441 13,521 14.8% 5.7% 25.2% 19.5%
Vanuatu Y 323 270 323 53 32.9% 27.5% 32.9% 5.5%
Venezuela N 112,782 NA 112,782 112,782 100.9% NA 100.9% 100.9%
Vietnam Y 18,595 5,365 20,956 15,591 5.1% 1.5% 5.7% 4.3%
West Bank and Gaza N 0 NA 0 0 0.0% NA 0.0% 0.0%
Strip
Yemen Y 357 448 357 (92) 3.6% 4.5% 3.6% -0.9%
Zambia Y 12,317 7,141 13,013 5,872 57.8% 33.5% 61.1% 27.6%
Zimbabwe Y 3,899 3,873 3,909 36 16.2% 16.1% 16.2% 0.1%
Notes: For details on AidData’s “level of public liability” measure, see Section A-5. Columns 1b, 2b, and 3b present cumulative 2000-2021 loan
commitments from China (over the 2000-2021 period) as a percentage of recipient country GDP. GDP data are drawn from UNstats.un.org and
measured in nominal 2021 USD.

393