BANK OF GREECE

EUROSYSTEM

Working Paper
Financial soundness indicators and
financial crisis episodes

Maria Th. Kasselaki

Athanasios O. Tagkalakis

BANK OF GREECE
EUROSYSTEM

Economic Research Department

158
Special Studies Division
21, E. Venizelos Avenue
GR - 102 50, Athens

Tel.:+30 210 320 3610

Fax:+30 210 320 2432
www.bankofgreece.gr

KPAPERWORKINKPAPERWORKINKPAPERWORKINKPAPER
ISSN: 1109-6691 WORKINKPAPERWORKINKPAPERWORKINKPAPERWORKINKPAPERWO
MAY 2013
BANK OF GREECE
Economic Research Department – Special Studies Division
21, Ε. Venizelos Avenue
GR-102 50 Athens
Τel: +30210-320 3610
Fax: +30210-320 2432

www.bankofgreece.gr

Printed in Athens, Greece

at the Bank of Greece Printing Works.
All rights reserved. Reproduction for educational and non-commercial purposes is permitted
provided that the source is acknowledged.

ISSN 1109-6691
FINANCIAL SOUNDNESS INDICATORS AND FINANCIAL CRISIS EPISODES

Maria Th. Kasselaki

Bank of Greece

Athanasios O. Tagkalakis
Bank of Greece

ABSTRACT
This paper studies the links between financial soundness indicators and financial crisis
episodes controlling for several macroeconomic and fiscal variables in 20 OECD
countries. We focus our attention on aggregate capital adequacy, asset quality and bank
profitability indicators compiled by the IMF. Our key findings suggest that, in times of
severe financial crisis, regulatory capital to risk weighted assets increases (by about 0.5-
0.6 percentage points –p.p.) to abide by regulatory and supervisory demands, non
performing loans (NPL) to total loans increase dramatically (by about 0.5-0.6 p.p.), but
loan loss provisions lag behind NPLs (they fall by about 12.3-18.8 p.p.) and profitability
deteriorates dramatically (returns on assets (equity) fall by about 0.3-0.4 (5.0-7.0) p.p.).

Keywords: Bank profitability, capital adequacy, asset quality, financial crisis.

JEL classification: E44, E58, G21, G28, E61, E62, H61, H62, E32.

Acknowledgments: We would like to thank Anne Villamil for useful comments and
suggestions. The views of the paper are our own and do not necessarily reflect those of
the Bank of Greece. All remaining errors are ours.

Correspondence:
Athanasios Tagkalakis
Bank of Greece,
21, El. Venizelos Ave.
10250 Athens, Greece
Tel.:0030-210-3202442
Fax: 0030-210-3232025
Email: [email protected]
1. Introduction
During the recent financial and economic crisis preserving banking sector stability
was one of the main problems that governments around the world had to deal with. To
this end, government assistance to the banking sector involved equity injections,
subsidies, asset purchases, loan guarantees and other forms of assistance which all had
very high cost for public finances and contributed to rising debt levels. According to
European Commission (2011) government assistance to the banking sector has been
“sizeable and in more than half of EU Member States has exceeded 5% of GDP” and
“currently, sizable rescue measures to the banking sector weigh heavily on the public
finances, most particularly in Ireland, the UK, Denmark, Belgium, the Netherlands,
Austria and Germany.” Building on this, the aim of the current study is to asses the risks
to banking sector stability that policy makers will have to address in the event of severe
financial crisis episodes.

The recent crisis has revealed that there are complex links between fiscal policy and
the financial sector and feedback loops between government activity and banking sector
stability. For example, unsound fiscal policies, by impacting negatively on market
confidence and sovereign bonds, could represent a risk to financial and consequently
economic stability1. The government borrowing operations in financial markets and its
tax decisions could also have repercussion for interest rates and asset price behaviour,
which could become a risk to financial market stability (IMF 2009a).

The increased interest of policy makers in understanding what leads to systemic

banking crises; the need to design early warning mechanisms to prevent them from
occurring has led to the creation of the so-called macro-prudential analysis.2 The IMF
initiated work on identifying a small set of macro-prudential indicators comprising
information on the aggregate banking system of each country (see Sundararajan et al.

1
As stated by Peter Praet “Public debt is commonly held as a low-risk asset by financial institutions and it
is also used as collateral in refinancing operations… When the financial markets doubt the sustainability of
public debt, the liquidity and even the solvency of financial institutions can deteriorate, in turn potentially
destabilising the financial sector.” See Bank for International Settlements (BIS, 2011) and discussion
therein on the implication that fiscal policy has on monetary and financial stability.
2
As discussed in Galati and Moessner (2010) the origin of the term macro-prudential traces back to the late
1970s, and became much more commonly used in the post 2007 period. For more information see also
Borio (2010). Tsomocos (2003) and Goodhart et al. (2004, 2005, 2006) have examined theoretically the
importance of financial fragility/stability and its consequence for economic policy.

5
2002). Several other IMF studies followed using these financial soundness or stability
indicators (FSIs) in cross country studies, e.g., Babihuga (2007), Cihak and Schaeck
(2007, 2010). Babihuga (2007) analyzed the relationship between selected FSIs and
macroeconomic variables and Cihak and Schaeck (2007, 2010) examined whether these
financial soundness indicators can predict systemic banking crisis.

Financial stability indicators are affected by a series of factors, macroeconomic,

fiscal, institutional, etc. Several country level studies have investigated the determinant of
FSIs, e.g., Berger and De Young (1997) link cost efficiency and non performing loans
(NPLs) in the US banking industry; Podpiera and Weill (2008) link cost efficiency and
NPLs in the Czech Rebublic; Salas and Saurina (2002) examine the determinants of
NPLs in the Spanish banking sector. Sorge and Virolainen (2006) examine the
relationship between loan loss provisions and macroeconomic determinants in Finland.
Quagliarello (2007) discusses the links between business cycle developments and NPLs
in Italy, and more recently Louzis at al. (2012) investigate the determinants of NPLs in
the Greek banking industry. Nkusu (2011) and De Bock and Demyanets (2012)
investigate the determinants of NPLs in developed and emerging market economies,
respectively. Gropp and Heider (2009) examine the determinants of bank capital
structure of large EU and the US banks. While, Saurina (2009a,b) examines the role of
provisioning as a tool that avoids pro-cyclicality of capital adequacy requirements and
strengthens the solvency of banks.

This paper building on the abovementioned country specific studies as well as

earlier cross-country IMF studies like Babihuga (2007) and Cihak and Schaeck (2007,
2010) extend previous work by investigating the effect of financial crisis episodes on
FSIs, while controlling for several macroeconomic and fiscal variables in a set of 20
advanced OECD economies over the periods 1997-2009. Our findings will reveal the
degree of the fragility of the banking system (in terms of capital adequacy, asset quality
and profitability) and the risks to financial stability that policy makers will have to
address in the event of financial crisis episodes.

Moreover, the effects of the financial crisis on banking sector stability will signal to
policy makers the likely future costs for public finances, in the event that these costs

6
might have to be borne by the public.3 This reinforces the argument for a more proactive
stance on the side of the regulatory and supervisory authorities of the financial sector in
order to preserve financial stability, as well as enhanced cooperation between fiscal,
monetary and macro-prudential authorities in order to contain the effects of financial
crisis. 4

It is worth stressing that there have been a number of theoretical papers linking the
financial sector with macroeconomic developments, highlighting the important role of
banking sector stability, as well as the need to avoid pro-cyclical capital requirements.5
Whereas, other theoretical studies have discussed how monetary policy decisions impact
on the risk taking behavior of financial intermediaries that are relevant for financial
stability.6

Our key findings suggest that in times of severe financial crisis the ratio of
regulatory capital to risk weighted assets is increased by about 0.5-0.6 percentage points
(to abide by regulatory and supervisory demands as also pointed out by Cihak and
Schaeck, 2010)7, non performing loans (NPL) to total loans increase dramatically (by
about 0.5-0.6 percentage points in the short run and 3.0-4.6 p.p. in the long run), but loan
loss provisions lag behind NPLs (fall by about 12.3-18.8 p. p.), and profitability
deteriorates dramatically (returns on assets (equity) fall by about 0.3-0.4 (6.0-7.0)

3
The Greek and the Irish debt problems are very useful examples of inter-linkages between the government
and the financial sectors. Despite the fact that the two problems are interrelated, the origin of each one of
them is different. The Irish public debt problem stems from the fact that the Irish government assumed the
debt and vulnerabilities of the private and banking sector created by the 2008-2009 financial crisis; while,
in the case of Greece the sovereign debt problem was passed on to the local banking sector, which had been
unaffected by the 2008-2009 financial crisis (see IMF, 2010a, 2010b). As stated in IMF (2010b) “sovereign
downgrades, increasing loan impairment, and the deteriorating economic outlook have undermined
confidence in the Greek banking sector.”
4
In this study we investigate the effects of financial crisis on banking sector stability. Several previous
studies have examined the impact of financial and banking crisis instability on public finance. For example,
some previous studies have investigated the direct fiscal implications of past banking system support
schemes (Honohan and Klingebiel 2003), the determinants of fiscal recovery rates (European Commission,
2009), as well as whether costly fiscal interventions reduce output loss (Claessens et al. 2005; Detragiache
and Ho 2010). Other studies have investigated the effect of financial crisis on the debt to GDP ratio and
GDP growth (European Commission 2009b; Furceri and Zdzienicka 2010, 2011; Reinhart and Rogoff
2008, 2009, 2010, Tagkalakis, 2013).
5
See for example Benanke et al. 1999, Goodhart et al. (2005, 2006), Goodfried and McCallum (2007),
Covas and Fujita (2009), Repullo and Suarez (2008), Meh and Moran (2010).,
6
See for example Adrian and Shin (2009), Borio and Zhu (2008) and Brunnermeier and Pedersen (2009).
7
Other studies such as Gropp and Heider (2009) find that find that unobserved time invariant bank fixed
effects are ultimately the most important determinant of banks’ capital structures.

7
percentage points in the short run and by 0.8-0.9 p.p. (10.0-15.0 p.p.) in the long run ).
Increased real short term interest rates are associated with higher regulatory-capital to
risk weighted assets, increased NPLs and lower profitability. Increases in the real long
term government bonds rates deteriorate asset quality and reduce profitability.

Our findings should be looked at in close association with the findings reported by
Adrian and Shin (2009, 2010) suggesting that broader balance sheet aggregates of
financial intermediates such as total assets and leverage could be incorporated into
macroeconomic analysis in order to improve the conduct of monetary policy making.8

The remaining of the paper is as follows: Section 2 provides a brief overview of

related studies and data information on FSIs and financial crisis indicators. In Sections 3
to 5 we discuss methodology and econometric estimation of capital adequacy, asset
quality and profitability equations, respectively. Section 6 summarizes the main findings
and concludes. A Data Appendix provides additional information on the data used in the
analysis.

2. Financial soundness indicators: data issues and related studies

The FSI data used in this study are drawn from successive IMF Global Financial
Stability Reports (from March 2002 to October 2010)9. FSI data have been previously
employed by several IMF studies, like Babihuga (2007) and Cihak and Schaeck (2007,
2010). Due to their wide coverage, FSIs are able to capture a range of factors that may
pose risks to the financial system as a whole (Sundararajan et al. 2002).

The paper employs the following core FSIs: capital adequacy (measured by the
ratios of capital to assets and regulatory capital to risk weighted assets), asset quality
(measured by the ratio of non-performing loans (NPLs) to total loans and by loan loss
provisions to non performing loan) and profitability (measured by return on assets and

8
As pointed out by Goodhart et al (2004) there may be a trade-off between efficiency and financial
stability, not only for regulatory policies, but also for monetary policy. However, in subsequent research
Goodhart et al. (2011) show that interest rate setting is an appropriate instrument in order to maintain
financial stability, because in times of a panic or financial crisis the Central Bank automatically satisfies the
increased demand for money, i.e. preventing sharp losses in asset values and enhanced asset volatility.
9
See Data Appendix. The IMF has created a website (http://fsi.imf.org/) disseminating data and metadata
on selected FSI provided by several countries.

8
return on equity). Capital adequacy, asset quality and profitability are all important
indicators of bank performance and fragility.

The FSI data start in 1997, reflecting the fact that many countries began collecting
FSI data in the context of the IMF’s Financial Sector Assessment Programme (FSAP),
which began in 1999 (Babihuga 2007). Despite the short time dimension of the dataset
(1997-2009), the sample size (20 countries) is sufficient to allow for consistent estimators
by taking into account the asymptotic properties (of the relatively larger sample of
countries). 10

A few IMF studies have used these FSI indicators. Babihuga (2007) analyzes the
relationship between selected macroeconomic and financial soundness indicators (FSIs)
for 96 countries covering the period 1998-2005. The analysis covers key macroeconomic
indicators and capital adequacy, asset quality and profitability. The paper finds that the
business cycle—measured as the cycle component of real GDP, obtained using the
Hodrick-Prescott filter (1980)—has a robust, negative relationship with capital adequacy,
and non-performing loans (NPL), and a robust, positive relationship with profitability.
Furthermore, inflation, the real effective exchange rate, and real interest rates also emerge
to different degrees as important determinants of FSIs. Cross country differences in
income, size of the financial sector, quality of banking supervision, and market
concentration robustly explain cross country differences in the cyclicality of FSIs.

Cihak and Schaeck (2007, 2010) working with financial soundness indicators
investigate how well these aggregate banking system ratios identify systemic banking
crises. The authors also estimate a duration model to investigate whether these ratios help
determine the timing of a banking crisis. As is shown by the authors, bank regulatory
capital to risk weighted assets does not show any variability prior to the crisis, it only
increases as a consequence of the crisis, i.e., authorities impose or markets require a
higher capital requirement after a financial crisis. The capital to asset ratio increases prior

10
The second drawback which is also stated in Babihuga (2007), Cihak and Schaeck (2007) and IMF
(2009b) is that FSI metadata is sourced from national sources, implying that due to differences in national
accounting, taxation, and supervisory regimes, FSI data might not be strictly comparable across countries.
However, contrary to Babihuga (2007) and Cihak and Schaeck (2007) we decide to focus on a smaller
sample, i.e., 20 industrialized countries (excluding other emerging market and developing economies for
which the IMF reports analogous data). This way we try to avoid major problems in terms of data quality,
as well as in terms of non comparability or great diversity and heterogeneity of national definitions.

9
to the crisis possibly because banks build up capital buffers in anticipation of regulatory
pressure. The non-performing loans to total loans increase prior to the crisis deteriorating
the asset quality of institutions and gradually fall following the crisis. Consequently, bank
provisions to non performing loans increase following the recognition of nonperforming
loans. Finally, bank profitability (returns on earnings) is not much affected prior to the
banking crisis, but deteriorates rapidly at the time of the crisis.

After controlling for macroeconomic conditions Cihak and Schaeck (2007)

investigate, by estimating a logit regression, which financial soundness indicators can
provide an accurate signal for the probability of observing systemic banking
vulnerabilities. They use a sample of 100 countries of which 51 experience serious
banking problems during the period 1994-2004. A high capital to risk weighted assets
and a high return on equity lowers the probability of a systemic banking crisis occurring.
On the contrary, declining asset quality, i.e., an increase in nonperforming loans to total
loans is indicative of an impending banking turmoil. The duration analysis performed by
the authors reiterates that increasing profitability of financial institutions increases in a
significant manner the survival time of banking systems. A low capital adequacy ratio
and a high ratio of nonperforming loans to total loans decreases the survival time of the
banking system, but the effect is not statistically significant.11

There are several other studies that use macro-economic variables to explain
banking crisis episodes, e.g., Demirguc-Kunt and Detragiache (1998) and Kaminsky
(1998). Prodriera (2004) controlling for macroeconomic variables and other relevant
factors investigates the extent to which quality of supervision and regulation impact on
the asset quality and profitability of the banking sector.

Several studies have focused on country specific cases, linking FSIs with
macroeconomic and other institutional and bank specific factors. For example, Wong et

11
Poghosyan and Cihak (2009) present a new database on individual bank distress across the European
Union from mid-1990s to 2008. Building on this dataset, they analyze the causes of banking distress and
identify a set of indicators and thresholds that can help distinguish sound from vulnerable banks and can
help as an early warning system. The authors estimate a logistic random effects model robust to
heteroskedasticity to identify the determinant of the probability of distress. According to the findings the
probability of distress is negatively associated with the level of bank capitalization and earnings. Moreover,
the probability of distress is inversely related to asset quality, i.e., the higher loan loss provision profile
implies a riskier loan portfolio.

10
al. (2005) have investigated the determinants of asset quality, profitability and capital
adequacy in Hong-Kong. Brewer et al. (2008) demonstrated that if the banking sector is
relatively small, the banks maintain a higher capital adequacy ratio. According to
Marques and Santos (2003) capital regulation is the first external determinant of the
banks’ capital structure. Barth et al. (2005), Berger et al. (2008) and Brewer et al. (2008)
observe that the levels of bank capital are much higher than the regulatory minimum,
which means that banks could hold capital buffers in excess of the regulatory minimum
(see Ayuso et al. 2004). Banks may hold capital buffers in order to reduce the probability
that they incur high costs by raising equity on short notice so as to abide by regulatory
requirements.

Gropp and Heider (2009) examining the determinants of bank capital structure of
large EU and the US banks find that unobserved time invariant bank fixed effects are
ultimately the most important determinant of banks’ capital structures. Contrary to
previous beliefs (Mishkin 2000) they suggest that capital regulation and buffers may only
be of second order importance in determining the capital structure of most banks.12

Bikker and Metzemakers (2002) have investigated the determinants of loan loss
provisioning in OECD countries, while Sorge and Virolainen (2006) examine the
relationship between loan loss provisions to total loans and various macroeconomic and
other determinants in the Finish banking system.

Other country specific studies have linked cost efficiency with non-performing
loans. Berger and De Young (1997) using US banking industry data investigate the
relationship between loan quality, cost efficiency, and bank capital. They find a negative
correlation between cost efficiency and NPLs, which runs both ways, as well as that that
reductions in capital at thinly capitalized banks, precede increases in problem loans.
Hence, cost efficiency may be an important indicator of future problem loans and

12
Several authors have developed theories of optimal bank capital structure, in which capital requirements
are not necessarily binding (e.g. Flannery 1994; Myers and Rajan 1998; Diamond and Rajan 2000; Allen et
al. 2009). Based on the market discipline view, banks’ capital structures are the outcome of pressures
emanating from shareholders, debt holders and depositors which implies that regulatory intervention is non-
binding and of secondary importance (see e.g. Flannery and Sorescu 1996; Morgan and Stiroh 2001;
Martinez Peria and Schmuckler 2001; Calomiris and Wilson 2004; Ashcraft 2008; Flannery and Rangan
2008).

11
problem banks. Salas and Saurina (2002) investigating the Spanish banking system use
both macroeconomic and efficiency variables to explain NPLs and find a significant
negative effect of output and bank capitalization on NPLs. Podpiera and Weill (2008)
looking into the Czech banking sector examine the links between cost efficiency and
NPLs and finding significant evidence that low efficiency lead to rising future NPLs.
More recently, Louzis et al. (2011) focusing on the Greek banking system and
investigating NPLs (on consumer, business and mortgage loans) find that rising NPLs can
be explained mainly by macro fundamentals (GDP, unemployment, interest rates) and
management quality (cost efficiency). Quagliarello (2007) has investigated and found
significant business cycle effects on NPLs in the Italian banking industry. While Nkusu
(2011) and De Bock and Demyanets (2012) investigated the macroeconomic
determinants of NPLs in developed and emerging market economies, respectively.

2.1 Financial crisis indicators and financial soundness indicators

In our analysis we consider 20 advanced OECD economies over the period 1997-
2009. 13 The decisions on the dataset is based on two grounds: 1) the time dimension is
dictated by the fact that aggregate FSI are collected by the IMF since 1997 and 2) the
group of 20 OECD countries was chosen because of the deficiencies in the collection of
the FSI (described above), i.e., advanced OECD economies are more homogeneous group
reducing to the extent possible data coverage and definition problems linked to FSI
indicators.

Building on the work of Babihuga (2007) and Cihak and Schaeck (2007, 2010) we
extend their work by studying the relationship between financial soundness indicators and
financial crisis. Before doing that we need first to discuss the financial crisis indicators
that will be considered in the analysis.14 In our investigation we do not employ the

13
We consider the following OECD countries: Australia, Austria, Belgium, Canada, Switzerland,
Germany, Denmark, Spain, Finland, France, UK, Greece, Ireland, Italy, Japan, Netherlands, Norway,
Portugal, Sweden, and the US, see Data Appendix.
14
Allen and Gale in a series of contributions have investigated the root causes of financial crisis. For
example, Allen and Gale (1998) describe a model where financial crisis are caused by exogenous asset
returns shocks, whereas Allen and Gale (2004a, 2004b) describe endogenous crises, where small or
negligible shocks set off self-reinforcing and self-amplifying price changes, with a key role played by
liquidity (which affects asset prices).

12
Laevan and Valencia (2008) banking crisis data base. Instead we define two types of
financial crisis episodes, a severe one and a weak one. The first definition relates to the
2008-2009 financial market crash. As can be seen in Table 1, on average, in the 20
OECD countries considered in the sample, real share prices fell by 26.2% per year. At the
same time the real GDP fell, on average, by 1.7% per year and the unemployment rate
increased, on average, by 1 percentage point per year. Automatic and discretionary fiscal
policy responses resulted in debt increasing, on average, by 7.4% of GDP per year.
Hence, in the 2008-2009 financial crises, we observed a substantial fall in real GDP, an
increase in unemployment, a collapse in asset prices and a significant worsening of fiscal
positions. The first financial crisis indicator, capturing the effects of the 2008-2009 Great
Recession is called “severe financial crisis” indicator.15

[Table 1 about here]

As an alternative definition we want to consider financial market crashes that are

significant but have far less important effects compared to the 2008-2009 Great
Recession. We call this second indicator “weaker financial crisis” indicator. It captures
the year-country observations where real share prices fell by more than 5%. The second
indicator encompasses the country-year observations reflected in the severe financial
crisis indicator.16 As can be seen in Table 2 the weak financial crisis indicator results in
financial crisis episodes that are not associated with dramatic falls in real GDP or sharp
increases in unemployment and debt ratios. During the periods captured by this weaker
definition of financial crisis, real share prices fell on average by 21.0% per year,17 the
debt ratio increased by 3.4% per year, real GDP increased by 0.4% per year and

15
Borio and Drehmann (2009) building on earlier work on Borio and Lowe (2002a,b) show that indicators
based on asset price and credit developments provide a fairly successful signal for subsequent banking
system stability issues.
16
This definition captures the real share prices falls in late 1990s, early 2000s and in 2008-2009. The
exclusion of the 2008-2009 financial crisis would have resulted in having less pronounced effects on FSIs
in weak crisis episodes. At the same time it would have implied that we would have to reduce our sample to
1997-2007 and lose valuable information. In any case the main message comes through i.e., that in severe
or extreme event financial crisis the effects are bigger than in more normal or reoccurring weaker financial
crisis events.
17
Note that in the severe financial crisis definition (Table 1) the average per year fall in real share prices
was more pronounced than in the weak financial crisis definition (Table 2) in all OECD countries
considered (except in Canada where the average per year fall of real share prices is about the same;
however, even in Canada’s case during the 2008-2009 financial crisis the debt ratio increased by much
more, real GDP fell to a greater extent and the unemployment rate increase was bigger).

13
unemployment increased by 0.4 percentage points per year in the 20 OECD countries
examined.

While in the case of severe financial market episode we have in total 40 country
year observations (those corresponding to 2008 and 2009) , in the weak financial crisis
indicator we have in total 99 country-year observations, i.e., 4.95 observations per
country.

[Table 2 about here]

Financial stability indicators are affected significantly in financial crisis.18 As

shown in Table 3 bank profitability (reflected in returns to asset and returns to equity) is
reduced dramatically in financial crisis episodes (be it severe ones like the 2008-2009
financial crisis –column 2- or less severe ones-column 5) compared to non crisis periods.
Bank capital to assets ratios are reduced in times of crisis, however, bank regulatory
capital to risk weighted assets is increased possibly reflecting regulatory authorities’
requirements and interventions. Furthermore, asset quality deteriorates during financial
crisis as reflected by the higher value of non-performing loans to total loans in crisis
episodes. At the same time loan loss provisioning to non performing loans remains well
below its non crisis levels, implying increased riskiness and exposure on the side of
banks.

Therefore, the behavior of financial stability indicators can provide useful signals to
policy makers on risks stemming from the banking sector and its likely consequences for
public finances. If the deterioration of banking sector stability coincides with falling
economic activity then fiscal risks could be much higher.19

[Tables 3 and 4 about here]

18
As discussed by Adrian and Shin (2008, 2009) price changes contributed to financial contagion in the
post 2007 financial crisis era. As the authors point out “when balance sheets are marked to market, asset
price changes show up immediately on balance sheets and elicit response from financial market
participants. Even if exposures are dispersed widely throughout the financial system, the potential impact
of a shock can be amplified many-fold through market price changes”.
19
During the 2008-2009 financial crisis bank profitability ratios (returns on assets and on equity) were
substantially lower that in the other financial crisis episodes. At the same time capital to asset ratio was
lower, but regulatory capital to risk-weighted assets were higher. Non performing loans to total loans were
slightly lower that in past financial crisis, but this time loan loss provisioning to non-performing loans was
significantly lower (see Table 4).

14
3. Capital adequacy
Building on earlier work by Shrieve and Dahl (1992), Jacques and Nigro (1997),
Aggarwal and Jacques (1998), Wong et al. (2005), and Babihuga (2007) we estimate the
following regression: 20 21

Capital/assets it = α1Capital/assetsit-1 + α2ygapit + α3Xit-1+ α4FCit+ α5Creditit +λi+εit (1)

λi stand for unobserved country effects, Capital/assets is the capital adequacy ratio
(regulatory capital to risk weighted assets, capital to assets), Capital/assetst-1 stands for
the lagged change in the capital adequacy ratio. ygapt is the cyclical indicator (output
gap), FCit stands for the financial crisis indicators, Credit stands for credit growth, and Xt-
1 are variables that are likely to affect the capital adequacy ratio. These are the GDP
deflator based inflation rate, the percentage change in the real effective exchange rate, the
change in the real short term interest rate, the change in the real long term interest
(government bond) rate, the debt ratio, the change in the debt ratio, the level of financial
intermediation, and the percentage change in real share prices (see Data Appendix).
These additional explanatory variables, and the country effects, were also included to
reduce the omitted variables problem related to institutional and other bank related
variables identified by earlier country specific studies (see sections 1 and 2).22

20
The underlying specification assumes that changes to bank capital reflect (partial) adjustment towards a
target capital rate and exogenous factors (see Shrieve and Dahl 1992). A detailed look on theories of capital
structure is provided in Puerra and Keppo (2006), Allen et al. (2011), Mehran and Thakor (2011), Diamond
and Rajan (2000). Diamond and Rajan (2000) and Allen et al. (2009) suggest that banks are different and
that we should be looking for bank specific factors to explain bank leverage. While Bertrand and Schoar
(2003) and Frank and Goyal (2007), indicate that managers’ preferences have a direct impact on capital
structure, with less risk averse managers choosing a more aggressive strategy and higher leverage. See
Guegan and Tarrant (2012) for a discussion of risk management measures that are relevant for the
determination of banks’ capital requirements.
21
Goodhart et al. (2006) develop a theoretical model that allows one to assess the role and impact of capital
requirements for the soundness of the financial system and the macroeconomy. As the authors highlight “it
is important for crisis management and prevention to study situations where banks’ capital depletes and
capital requirements are ‘biting’.”
22
Jokipii and Milne (2011) and Allen and Gale (2003) discuss how liquidity affects capital and risk. Jokipii
and Milne (2011) claim that banks with higher liquidity can decrease their capital and increase their levels
of risk. However, banks may hold liquidity as self-insurance against liquidity shocks. In turn, high levels of
liquidity expose banks, mainly small ones, to risk-taking (Allen and Gale, 2003) leading to increasing
levels of capital in order to control risk-taking. In some cases liquidity requirements can be as effective as

15
 In order to address endogeneity issues due to the presence of a lagged dependent
variable and possible feedback effects from the output gap and credit growth we estimate
equation (1) with a dynamic panel data two-step system GMM estimator (see Arellano
and Bover 1995; Blundell and Bond 1998; Roodman 2009b). We use a subset of the
available instrument matrix, i.e., we use the t-1 to t-3 lags of the lagged dependent
variable and the lagged values of the output gap and credit growth.23

The specific decision on the subset of instruments to be used in each case that will
be presented below takes into account the performance of the Hansen test of
overidentifying restrictions and the absence of second order autocorrelation in first
difference errors (i.e., that moment conditions are valid).24 To transform the equation of
interest and remove fixed effects we consider both first differencing and forward
orthogonal deviations proposed by Arellano and Bover (1995). The forward orthogonal
deviations transformation, rather than subtracting the previous observation, subtracts the
average of all available future observations. The use of forward orthogonal deviations
transformation is more appropriate in unbalanced panels.25

The likely impact of the explanatory variables on the dependent variable is the
following: the lagged dependent variable reflects the cost of raising capital or reflects
adjustment costs; hence, the higher the adjustment costs the more capital the banks will
hold. In a cyclical upswing (increase in ygap) banks hold less capital.26 The impact of the

capital requirements - in this case, the effect of liquidity on capital will be positive the effect on risk will be
ambiguous.
23
The system GMM estimator is less affected by the weak instrument problem compared to the differenced
GMM (Arellano and Bond, 1991). Omitting the more distant lags might not lead to significant loss of
information, see Bond (2002) and Roodman (2009a) on the implication of using too many instruments.
Moreover, the two-system GMM estimator is more efficient than the one-step system GMM estimator. The
finite-sample correction to the two-step covariance matrix derived by Windmeijer (2005) is implemented.
24
In all specifications, the test on overidentifying restrictions indicates that the hypothesis that instruments
are valid cannot be rejected and that there is no higher-order autocorrelation.
25
On balanced panels, GMM estimators based on the two transformations return numerically identical
coefficient estimates, holding the instrument set fixed (Arellano and Bover 1995). As a robustness test we
consider also the two-step difference GMM estimator where we apply the forward orthogonal deviations
transformation. We shall refer to these findings in the next sections but we do not present them here due to
space limitation. However, they are available in the supplementary material appendix.
26
Jokipii and Milne (2011) argue that higher risk-taking can increase the probability of default and
encourage banks to increase regulatory capital. Jacques and Nigro (1997), Rime (2001) and Roy (2008)
find that weakly capitalized banks increase their capital faster than well-capitalized banks.

16
inflation rate and the real effective exchange rate on bank capital is ambiguous, as
Babihuga (2007) points out the effects depends on the impact of high inflation to bank
income and on the share of banking system assets held abroad. Changes in the real short
term interest rate reflect the impact of monetary policy; the impact is ambiguous
depending on the pass through to deposit vis-à-vis lending rates. Changes in real share
prices will impact the capital adequacy ratios to the extent that part of the capital is
depicted in current market and not on historic prices. Moreover, the movements of share
prices incorporate expectation about future economic prospects and outlook, which could
impact on the capital raising decision of the banking sector.

Long term interest rates, debt ratio and change in debt ratio, all reflect the impact of
the government sector on the banking system stability. As the government sector expands
and its debt increases this could lead to higher borrowing costs for the official sector.
This would imply that government assets held by banks might lose value, e.g., in case of
sovereign downgrade and fall in the price of government bonds. On the other hand, if the
government securities are still creditworthy the purchase of new government securities at
lower prices and higher yields will increase banks’ profits. The overall effect depends on
the creditworthiness of the sovereign, e.g. in the current sovereign debt crisis,
government assets held by banks have lost significant part of their value, deteriorating
banks’ positions and leading to successive recapitalizations by the state. The financial
crisis indicator reflects the impact of severe and/or weak financial crisis on the capital
adequacy indicators. Credit growth and financial intermediation control for the leverage
and level of development of the banking system.

3.1 Findings

In Tables 5-6 and in Tables 7-8 we report the findings for the specifications using
as dependent variable the regulatory capital to risk weighted assets and the capital to

17
assets ratio, respectively. In Tables 5 and 7 we control for the severe financial crisis
episodes, while in Tables 6 and 8 we control for the weak financial crisis episodes.27

Starting from the case of regulatory capital to risk weighted assets (Tables 5 and 6)
we see that the lagged dependent variable is positive and statistically significant
confirming the existence of adjustment costs (as in Babihuga 2007). The output gap
coefficient is negative and statistically significant in most cases, implying that as
economic conditions improve (the output gap variable increases) the capital adequacy
ratio declines. Alternatively, the capital ratio increases when economic conditions worsen
(a 1% fall in the output gap increases the regulatory capital adequacy ratio by 0.10 to 0.16
p.p.).28 This finding is in line with Wong et al (2005) and Babihuga (2007). It implies that
as economic conditions worsen and the quality and value of assets held by banks
deteriorate, banks increase capital to address vulnerabilities or alternatively reduce their
assets to attain regulatory capital ratios. This could also reflect the procyclicallity of
capital requirements which propagates recessions, i.e., when economic conditions
deteriorate, supervisory authorities impose tougher rules on banks, which in turn reduces
credit to the private sector propagating the fall in output.29

The financial crisis indicator, in particular the one controlling for severe financial
crisis episodes, confirms that in times of crisis the regulatory capital to risk weighted
assets is increased to abide by regulatory and supervisory authorities’ demands.
According to our findings the increase in the regulatory capital adequacy ratio ranges
from about 0.5 to 0.6 p.p. (see Table 5, columns 1-4)30. This highlights the procyclicality

27
Apart from the other macroeconomic and financial crisis variables, in columns 1 and 5 we control for the
real short term interest rate, in columns 2 and 6 we control for the real long term interest rate, in columns 3
and 7 we control for the real short term interest rate and the debt ratio, while in columns 4 and 8 we control
for the real short term interest rate and the change in debt ratio. This applies to Tables 5-16.
28
A similar result is found when considering the two step difference GMM estimator (with the findings
being more robust in the weak financial crisis estimations); see supplementary material appendix.
29
Saurina (2009a) finds that the effect of provisioning in Spain had only a small effect on credit growth
while strengthening the solvency of banks through countercyclical buffers. While Drehmann et al (2010)
propose several options of system-wide countercyclical capital buffers. More recently, Drehmann et al.
(2011) find that the gap between the ratio of credit-to-GDP and its long-term backward-looking trend
performs best as an indicator for the accumulation of capital as this variable captures the build-up of
system-wide vulnerabilities that typically lead to banking crises. As these authors point out “credit spreads
are better in indicating the release phase as they are contemporaneous signals of banking sector distress
that can precede a credit crunch”.
30
However, the long run effects are not statistically significant.

18
of capital requirements, which can in themselves propagate the crisis, if credit conditions
are tightened and economic activity deteriorates further. As pointed out by Cihak and
Schaeck (2010), “in the aftermath of a crisis, regulatory capital increases, which is due
to the frequently higher capital requirements in the period after an episode of financial
turmoil and the shoring up of reserves in financial institutions.” 31 However, the findings
reported in Table 6 (on weak financial crisis) and in Table 5 (columns 5-8) point to a
positive, but not statistically significant coefficient estimate.

[Tables 5 and 6 about here]

The coefficient estimates of inflation rate and real effective exchange rate are
insignificant as reported in Tables 5 and 6. There is statistically significant evidence that
increased short term real interest rates are associated with higher regulatory capital to risk
weighted assets ratios (see Table 5, columns 1, 3-4). Long term interest rates, debt ratio
and the change in debt ratio are all insignificantly estimated. The same applies for the
coefficient estimates of credit growth and the level of financial intermediation. Finally,
there is evidence that higher real share prices lead to lower regulatory capital to risk
weighted assets ratio, possibly because increases in stock market reflect improved current
and future outlook, which leads to lower regulatory capital asset ratios (see Table 5,
columns 1, 3-4 and 6).

Turning to the capital-to-assets ratio (see Tables 7 and 8), we see that most
independent variables are not statistically significant. Moreover, there are some
differences compared to the results presented in Tables 5-6 that stand out, we will focus
our attention to these ones. The coefficients of the lagged dependent variable, the output
gap, and the real short term interest rate appear to have no significant effect on the
capital-to-assets ratio. Moreover, both the severe and the weak financial crisis dummy
variables appear to lower the capital-to-assets ratio, but not in a statistically significant

31
Earlier work by Gropp and Heider (2009) find that unobserved time invariant bank fixed effects are
ultimately the most important determinant of banks’ capital structures. While in others studies banks’
capital structures are the outcome of pressures emanating from shareholders, debt holders and depositors
which makes regulatory intervention non-binding and of secondary importance (see e.g. Calomiris and
Wilson 2004; Ashcraft 2008; Flannery and Rangan 2008).

19
manner.32 Note that Cihak and Schaeck (2010) have found that the capital-to-assets ratio
decreases in the event of a financial crisis.

[Tables 7 and 8 about here]

4. Asset quality
Building on the work of Salas and Saurina (2002), Qualgliarello (2007), Bubihuga
(2007), Louzis et al. (2012) and Nkusu (2011) we estimate the following equation33:

Asset quality it = α1Asset quality it-1 + α2ygapit + α3Xit-1+ α4FCit+ α5Creditit-1 +λi+εit (2)

λi stand for unobserved country effects, Asset quality is the asset quality index, i.e., non-
performing loans to total loans and provisions to non performing loans. Asset qualityt-1
stands for the lagged value of the depended variable. ygapt is the cyclical indicator
(output gap), FCit stands for the financial crisis indicators, Creditt-1 stands for credit
growth, and Xt-1 are variables that are likely to impact on the asset quality ratios. These
are the GDP deflator based inflation rate, the unemployment rate, the percentage change
in the real effective exchange rate, the change in the real short term interest rate, the
change in the real long term interest rate the debt ratio, the change in the debt ratio, the
level of financial intermediation (see Data Appendix).34

In order to address endogeneity issues due to the presence of a lagged dependent

variable and possible feedback effects from the output gap and credit growth we estimate

32
When considering the two step difference GMM estimator (see supplementary material appendix) we
find statistically significant evidence that the real effective exchange rate has a negative impact effect on
the capital-to-assets ratio (i.e., a real appreciation lowers the capital to assets ratio) in line with Babihuga
(2007). Moreover, we find that an increase in the level of financial intermediation (domestic credit to
private sector over GDP) reduces in a significant manner the capital to assets ratio. Finally, we find that an
increase in real share prices leads to lower capital-to-assets ratio.
33
The specification reflects the inter-linkages between the financial sector and the real economy, focusing
mostly on the macroeconomic determinants of NPLs. The financial accelerator theory lies behind the
modeling of both NPL and its interaction with macroeconomic performance (see e.g Bernanke and Gertler
1989; Bernanke and Gilchrist 1999; Kiyotaki and Moore 1997.
34
These additional explanatory variables, and the fixed effects, were also included to reduce the omitted
variables problem related to institutional and other bank related variables identified by country specific
studies (see sections 1 and 2).

20
equation (1) with a dynamic panel data two-step system GMM estimator (see Arellano
and Bover 1995; Blundell and Bond 1998; Roodman 2009b). We use a subset of the
available instrument matrix, i.e., in our benchmark model we use the t-1 to t-3 lags of the
lagged dependent variable and the lagged values of the output gap and credit growth. The
specific decision on the subset of instruments to be used in each case that will be
presented below takes into account the performance of the Hansen test of overidentifying
restrictions and the absence of second order autocorrelation in first difference errors (i.e.,
that moment conditions are valid). To transform the equation on interest and remove
fixed effects we consider both first differencing and forward orthogonal deviations
proposed by Arellano and Bover (1995).35

In line with explanations put forward by earlier studies the likely impact of the
explanatory variables on the dependent variable is the following: improved economic
conditions (an increase in the output gap and lower unemployment) are associated with
deterioration in asset quality. Higher short term interest rates worsen repayment
opportunities and deteriorate asset quality.36 The inflation rate and the real exchange rate
have ambiguous effect. Increases in long term interest rates and the other fiscal variables
could signal a worse budgetary position and could be associated with increase taxes in the
near future, impacts on the ability to repay loans, reducing asset quality. A high credit
growth and high degree of financial intermediation are likely to be both associated with
deteriorating asset quality. Finally, in times of financial crisis asset quality is expected to
deteriorate dramatically.

4.1 Findings

In Tables 9 and 10 we report the findings for non performing loans to total loans
and in Tables 11 and 12 the findings for provisions to non-performing loans. In Tables 9
and 11 we control for the severe financial crisis definition, while in Tables 10 and 12 we
control for the weak financial crisis definition.

35
As a robustness test we consider also the two-step difference GMM estimator where we apply the
forward orthogonal deviations transformation.
36
Lawrence (1995) and Rinaldi and Sanchis-Arellano (2006) consider GDP, unemployment rate and the
lending rate as the key determinants of NPLs.

21
 Starting from non-performing loans to total loans, we find significant persistence
effects, i.e., the lagged dependent variable is positive and highly significant. The
coefficient of the output gap variable is negative and statistically significant in most
specifications (see Tables 9 and 10) in line with the findings reported by Nkusu (2011)
and De Bock and Demyanets (2012). When economic conditions worsen, borrowers’
creditworthiness and the value of collaterals deteriorate increasing NPLs. A 1% fall in the
output gap increases the ratio of NPLs to total loans by about 0.10 to 0.19 percentage
points. This effect is magnified in times of financial crisis episodes.

NPLs increase by about 0.46-0.59 p.p. in severe financial crisis (see Table 9
columns 1-8). The long run effect of the severe financial crisis ranges from 3.0-4.6 p.p. In
weak financial crisis, the coefficient estimate of the financial crisis dummy is positive,
but insignificantly estimated in most cases (see Table 10, columns 1-8). The coefficient
estimates of the unemployment rate, inflation, credit growth and financial intermediation
are not statistically significant.37 The coefficient estimate of the real effective exchange
rate is negative as in Babihuga (2007), but not statistically significant.

[Tables 9 and 10 about here]

Increases in both short and long term interest rates worsen asset quality, increasing
non-performing loans to total loans (see Tables 9 and 10). A 1 p.p. increase in short and
long term interest rates increases the ratio of NPLs to total loans by about 0.07-0.1 and by
0.07 - 0.10 p.p., respectively.38 Fiscal variables have no particular impact effect on non
performing loans.39

The results for the other dependent variable (provisions to non-performing loans)
are shown in Tables 11-12. The lagged dependent variable is positive and highly
significant, indicating that there is high degree of persistence. The output gap has a
negative (and in some cases statistically significant) correlation with the provisions to
37
It is worth noting that in the two-step difference GMM estimation the coefficient estimate of financial
intermediation is positive and significant across all specifications, i.e., the higher the ratio of domestic
credit to private sector to GDP, the bigger the ratio of NPLs to total loans (see supplementary material
appendix).
38
Nkusu (2011) finds that an increase in the policy rate increases NPLs.
39
We do not account for differences in supervisory practices and quality of banking supervision, but the
wider explanatory variable set used and the presence of country effects control for unaccounted country
characteristics.

22
NPL ratio (see Table 11 and 12, columns 1-3), i.e., when economic conditions worsen
and NPLs increase, the provisions that banks set aside to cover NPLs increases as well. A
1% fall in the output gap increases provisions to NPLs by about 1.4 to 1.6 p.p.; exactly
the opposite holds in booms. However, when examining financial crisis episodes we see
that NPLs increase dramatically (as shown also by Cihak and Schaeck, 2010) but at the
same time provisions to NPLs fall dramatically (i.e., provisioning lags behind the
recognition of NPLs). This effect is quite pronounced in the case of severe financial crisis
episodes, where the fall reaches 12.3-18.8 p.p. (see Table 11, columns 1-8). The fall
ranges from 4.6-5.6 percentage points in times of weak financial crisis (see Table 12,
columns 1-4).40

The unemployment rate is insignificantly estimated in most cases.41 Inflation rate

and the real effective exchange rate have a negative (and in most cases statistically
significant) correlation with the ratio of provisions to NPLs (see Tables 11 and 12,
columns 1-4). The coefficient estimates of the both interest rates and the fiscal variables
are insignificantly estimated. 42

The level of financial intermediation is associated positively with the NPL ratio (as
seen before) and negatively with the ratio of provisions to NPLs (see Table 11 and 12).
Credit growth is associated negatively with the ratio of provisions to NPLs (see Tables 11
and 12).43

[Tables 11 and 12 about here]

5. Profitability
Building on the work of Demirguc-Kunt and Huizinga (1999) and Abreu ad
Mendes (2002) and Bubihuga (2007) we estimate the following equation:

40
The long run effects of both severe and weak financial crisis are not statistically significant.
41
Interestingly, contrary to the two step system GMM findings in the two step difference GMM estimation
the unemployment rate has a positive and at times significant effect on provisions to NPLs ratio.
42
In the two step difference GMM estimation we find a positive response following an increase in real
short term interest rate (see supplementary material appendix).
43
We find a negative and quite statistically significant coefficient estimate in case of credit growth and
financial intermediation in the two step difference GMM estimation (see supplementary material appendix).

23
Profitability it = α1Profitabilityit-1 + α2ygapit + α3C/Ait +α4NPL/TLit +α5Creditit +α6FCit+
α7Xit-1 +λi+εit (3)

λi stand for unobserved country effects, profitability is returns on assets (equity).

Profitabilityt-1 stands for the lagged value of the depended variable. ygapt is the cyclical
indicator (output gap), C/A stands for capital-to-asset ratio, NPL/TL stands for non-
performing loans to total loans, FCit stands for the financial crisis indicators, Credit
stands for credit growth, and Xt-1 are variables that are likely to impact on the asset
quality ratios. These are the GDP deflator based inflation rate, the percentage change in
the real effective exchange rate, the change in the real short term interest rate, the change
in the real long term interest rate the debt ratio, the change in the debt ratio, the level of
financial intermediation.44

In order to address endogeneity issues due to the presence of a lagged dependent

variable and possible feedback effects from the output gap, the capital to assets ratio,
non-performing loans to total loans and credit growth we estimate equation (3) with a
dynamic panel data two-step system GMM estimator (see Blundell and Bond 1998;
Roodman 2009b). We use a subset of the available instrument matrix, i.e., in our
benchmark model we use the t-1 to t-2 lags of the lagged dependent variable and the
lagged value of the output gap, the capital to assets ratio and the non performing loans to
total loans and credit growth. To transform the equation on interest and remove fixed
effects we consider both first differencing and forward orthogonal deviations proposed by
Arellano and Bover (1995).45

In line with explanations put forward in ealier studies like Demirguc-Kunt and
Huizinga (1999) and Babihuga (2007) the likely impact of the explanatory variables on
the dependent variable is the following: improved economic conditions (an increase in
the output gap) are associated with increased profitability. Banks with higher capital to

44
These additional explanatory variables, and the fixed effects, were also included to reduce the omitted
variables problem related to institutional and other bank related variables identified by country specific
studies (see sections 1 and 2).
45
As a robustness test we consider also the two-step difference GMM estimator where we apply the
forward orthogonal deviations transformation.

24
assets ratio face lower funding costs and thus have higher profits46, a higher credit risk as
reflected by a high ratio of nonperforming loans to total loans is associated with lower
profitability. Changes in short term interest rates impact both on deposit and lending
rates; the pass-through to each respective component will determine whether profits will
increase or fall. Inflation impacts the interest margin – the effect is ambiguous.47
Ambiguous is also the impact of the real effective exchange rate. As Babihuga (2007)
points out it depends on the share of assets held abroad. Higher credit growth and a
greater degree of financial intermediation could be associated with increased profits.

Higher long term interest rates and increases in the debt ratio signal a worse
budgetary position and could be associated with increased taxes in the near future,
impacting negatively on profits. At the same time, they might also reflect increased profit
opportunities if a higher debt ratio requires increased funds for its refinancing paying
higher interest rates on government bonds and so on. However, as the recent crisis has
shown this could lead to negative effects for the banking sector as a whole if the
government sector defaults (or is near default) and the value of its bonds deteriorates
dramatically (following sovereign rating downgrades by credit rating agencies). Finally,
bank profitability is expected to fall in financial crisis episodes.

5.1 Findings

In Tables 13-14 (15-16) we report the findings for the return on assets (equity). In
Tables 13 and 15 we control for the severe financial crisis definition, while in Tables 14
and 16 we control for the weak financial crisis definition.

In all cases the lagged dependent variable is positive and highly significant
indicating the presence of significant persistence effects. Interestingly, the output gap
coefficient is insignificantly estimated across all specifications (see Tables 13-16).
However, there is robust evidence that profitability deteriorates dramatically at times of
financial crisis, with the effect being much more pronounced in severe financial crisis

46
Berger (1995) finds a positive association between capital to assets and return on equity ratios.
47
However, as Demirguc-Kunt and Huizinga (1999) point out inflation is associated with higher realized
interest margins and greater profitability. Inflation brings higher costs--more transactions and generally
more extensive branch networks--and also more income from bank float. Bank income increases more with
inflation than bank costs do.

25
episodes (see Tables 13-16).48 Return on assets fall by about 0.24-0.28 p.p. and 0.34-0.45
p.p., respectively, in weak and severe financial crisis. With the long run effect in severe
financial crisis reaching about 0.9 p.p. Returns on equity fall by about 4.8-7.0 p.p. and
4.7-7.4 percentage points, respectively, in weak and severe financial crisis. With the long
run effect in severe financial crisis reaching 15.1 p.p.

[Tables 13 and14 about here]

The coefficient estimate of the inflation rate is insignificantly estimated, though in

most cases it has a positive sign as in Demirguc-Kunt and Huizinga (1999) and Babihuga
(2007). The real effective exchange rate has no particular effect on returns on assets,
however, in most cases it has a positive and significant effect on returns on equity, i.e., a
real appreciation is associated with higher returns (see Tables 15-16). Increases in both
interest rate variables exerts a negative effect on both profitability ratios, in particular in
the estimations that control for severe financial crisis (see Tables 13 and 15). Increases in
long term government bond yields reduce the price of government bonds that are held by
banks affecting negatively their profits. Moreover, higher long term yields on government
paper lead to even higher yields on long term bonds issued by banks reducing their
profits.49 The variables that control for the fiscal situation are insignificantly estimated at
all times.

Both credit growth and financial intermediation are associated negatively with the
two aggregate profitability ratios. The findings for credit growth are statistically
significant when we control for the severe financial crisis dummy (Tables 13-14), while
in the case of financial intermediation the effects are significant when we control for the
weak financial crisis dummy (but in some of the returns-on-equity specifications; see
Table 16).

The evidence as regards capital-to-assets ratio is mixed. However, when its

coefficient estimate is statistically significant, its sign is positive (see Table 13, columns
12; Table 14, columns 5-8; Table 16, column 8) in line with Babihuga (2007) and

48
Cihak and Schaeck (2010) report that return on equity fall dramatically after the crisis, but at the time of
the crisis profits do not deteriorate dramatically.
49
In the two-step difference GMM estimation returns on equity respond negatively (and statistically
significant) to increases in real short and long term interest rates.

26
Demirguc-Hunt and Huizinga (1999). As pointed out in Babihuga (2007) “banks with
higher capital face lower funding costs and higher profits”.50 Non-performing loans to
total loans, proxying the risk exposure of the banking system, are associated negatively
with profitability in line with the findings reported in Babihuga (2007), but the coefficient
estimates are not statistically significant.

[Tables 15 and 16 about here]

6. Conclusions
This paper building on earlier IMF studies, like Babihuga (2007), Cihak and
Schaeck (2007, 2010) investigates the relationship between financial soundness
indicators (compiled by the IMF) and financial crisis episodes while controlling for
several macroeconomic and fiscal variables. Our findings reveal the degree of the
fragility of the banking system and the risks to financial stability that policy makers will
have to address in severe (and weak) financial crisis episodes. Actions to address banking
systems stability problems could carry heavy burden for public finances as was shown
during the recent crisis. Hence, proactive action and enhanced cooperation is required by
all relevant authorities.51

We report evidence the regulatory-capital-to-risk-weighted-assets increases as

economic conditions worsen (the output gap variable falls). The financial crisis indicator
confirms that in times of severe crisis regulatory capital to risk weighted assets ratios
increase (by about 0.5-0.6 p.p.) to abide by regulatory and supervisory authorities’
demands. This highlights the procyclicality of capital requirements, which can in
themselves propagate the crisis, if credit conditions are tightened and economic activity

50
As noted by Demirguc-Hunt and Huizinga (1999) “well-capitalized banks have higher net interest
margins and are more profitable. This is consistent with the fact that banks with higher capital ratios have
a lower cost of funding because of lower prospective bankruptcy costs. “
51
As shown in Aspachs et al. (2007) when banks do not have to comply with capital adequacy
requirements, shocks that induce a decline in banks profits and an increase in banks’ default rates also
produce a fall in GDP. When capital adequacy requirements are in place, most shocks do not result in a fall
in bank profits. The reason for this is that banks need to maintain or top up their capital, and they do this by
choosing (riskier) investments that raise their profits. The authors also show that a shock to banks’
probability of default and equity induces welfare and output losses.

27
deteriorates further. On the contrary the capital-to-assets ratio has a negative but not
statistically significant association with the financial crisis indicators. Moreover, we find
evidence that increases in short term real interest rates are associated with increases in the
regulatory capital to risk weighted assets ratio. Increases in real share prices, to the extent
they reflect improved future outlook, are associated negatively with both capital
adequacy ratios. We also find evidence that increases in the level of financial
intermediation, i.e., in the ratio of domestic credit to private sector to GDP leads to lower
capital-to-assets ratios.

Turning to asset quality, when economic conditions worsen (the output gap falls)
borrowers’ creditworthiness and the value of collaterals deteriorate, increasing NPLs (by
about 0.10-0.19 p.p.) and increasing banks’ provisions to NPLs (by about 1.4-1.6 p.p.).
However, when examining financial crisis episodes the NPL ratio increases dramatically
(by about 0.5-0.6 p.p.), but at the same time provisions to NPL fall (by about 4.6 to 18.8
percentage points). This effect is quite pronounced in the case of severe financial crisis
episodes, indicating that provisioning is inadequate. The long run effect of severe
financial crisis on the NPL ratio reaches 3.0-4.6. p.p.

Higher short and long term real interest rates increase the NPL ratio. A 1 p.p.
increase in the real short and long term interest rates increases the NPL ratio by about
0.08-0.10 p.p. A high level of financial intermediation is associated with a higher NPL
ratio and a lower ratio of provisions to NPLs; whereas rapid credit growth is associated
with lower provisions to NPL ratio. We find evidence that the inflation rate and the real
effective exchange rate have a negative and significant effect on the ratio of provisions to
NPLs.

There is robust evidence that profitability deteriorates dramatically at times of

financial crisis, with the effect being much more pronounced in severe financial crisis
episodes (returns on assets fall by about 0.3-0.4 p.p. and returns on equity fall by about
6.0-7.0 p.p.). Severe financial crisis reduce return-on-assets (equity) by about 0.8-0.9
(10.0-15.0) p.p. in the long run. There is some evidence that banks with higher capital-to-
assets ratio face lower funding costs and higher profits (in the case of returns-on-assets).

28
Increases in the short and the long term real interest rate exert a negative effect on both
profitability ratios, with the effect being statistically significant when controlling for
severe financial crisis. An appreciation of the real effective exchange rate has a positive
and significant effect on returns on equity. Rapid credit growth is associated negatively
with the two profitability ratios, when controlling for severe financial crisis. A high level
of financial intermediation is associated negatively with returns-on-equity when
controlling for weak financial crisis episodes.

Despite the data limitations summarized above, our findings indicate that the
soundness of the aggregate banking system, controlling for a series of macroeconomic
and fiscal factors, is impacted heavily in times of severe financial crisis. Policy makers
should take this into account in order to develop early warning systems of whether
banking system stability is put at risk and to avoid passing the burden of rescue operation
to society, as was the case in the recent years. This reinforces the argument for a more
proactive stance on the side of the regulatory and supervisory authorities of the financial
sector in order to preserve financial stability. At the same time it is important to improve
both the supervisory and regulatory framework of financial markets in order to contain
risks stemming from the financial sector.

As discussed in Adrian and Shin (2009) binding regulatory constraints can come
through leverage bounds, forward-looking provisioning, explicit countercyclical capital
rules, or systemic capital rules. All these actions will reduce excessive risk taking and
unsustainable patterns and behavior like the hunt for rents, the propensity to herd and
create bubbles, the misalignment of incentives, and the proliferation of complex
innovative financial instruments that might carry hidden risk etc (See Bini-Smaghi,
2010). 52

52
In this context Andrian and Shin (2009) point out that shadow banking and, in particular the role of
securitization, should be examined more thoroughly considering also more stringent financial regulation
and the recognition of the importance of preventing excessive leverage and maturity mismatching (which
undermines financial stability).

29
 Moreover out findings improve our understanding on the likely consequences and
spill-overs from the government to the banking sector (as higher long term government
bond rates increase NPLs and reduce profitability) and hint to the need for enhanced
cooperation and coordination between the fiscal, monetary and macro-prudential
authorities to contain such risks.

Last but not least, the present paper shows that additional work should be carried
out to better understand the complex links between banking sector stability, fiscal policy
and financial crisis. These complex inter-relations have led to the recent worldwide
financial crisis and the subsequent sovereign debt crisis is Europe.

30
A. Data Appendix
We used a yearly unbalanced panel data set (1997-2009) of 20 OECD economies:
Australia, Austria, Belgium, Canada, Switzerland, Germany, Denmark, Spain, Finland,
France, UK, Greece, Ireland, Italy, Japan, Netherlands, Norway, Portugal, Sweden, and
the US.

Financial soundness indicators

The financial soundness indicators are taken from successive issues of the IMF’s Global
Financial Stability Report (GFSR) from March 2002 to October 2010 (see e.g. the
statistical appendix of the October 2010 GFSR, tables 22-27; IMF, 2010c).

Capital adequacy is measured by the following variables: Bank capital to assets, bank
regulatory capital to risk-weighted assets. We measure asset quality with the ratio on
bank non performing loans to total loans and bank provisions to non performing loans.
Return on assets and Return on equity measure bank profitability.

The FSI data start in 1997, reflecting the fact that many countries began collecting FSI
data in the context of the IMF’s Financial Sector Assessment Programme (Babihuga
2007), which began in 1999. Despite the short time dimension of the dataset (1997-2009),
the sample size (20 countries) is sufficient to allow for consistent estimators by taking
into account the asymptotic properties (of the relatively larger sample of countries).

The second drawback which is also stated in Babihuga (2007), Cihak and Schaeck (2007)
and IMF (2009b) is that FSI metadata is sourced from national sources, implying that due
to differences in national accounting, taxation, and supervisory regimes, FSI data might
not be strictly comparable across countries. However, contrary to Babihuga (2007) and
Cihak and Schaeck (2007) we decided to focus on a smaller sample, i.e., 20 industrialized
countries (excluding other emerging market and developing economies for which the
IMF reports analogous data). This way we try to avoid major problems in terms of data
quality, as well as in terms of non comparability or great diversity and heterogeneity of
national definitions.

31
Macroeconomic variables
The fiscal and macroeconomic variables used extent from 1997 to 2009 and are taken
from the Economic Outlook of the OECD (OECD, 2010).

The change in the debt ratio is calculated as the change in the debt ratio between t and t-1.
The inflation rate is the percentage change in the GDP deflator (GDP deflator based
inflation rate). The real short term interest rate is calculated as nominal short term interest
rate minus the GDP deflator based inflation rate. The real long term interest rate is
calculated as nominal long term interest rate minus the GDP deflator based inflation rate.
The percentage change in real share prices is calculated as the percentage change in share
prices minus the percentage change in GDP deflator. The real effective exchange rate
variable used is the percentage change in the real effective exchange rate. As financial
intermediation we use domestic credit to private sector as a percent of nominal GDP.
Credit growth is the percentage change in the domestic credit to private sector.

The descriptive statistics of the variables used in the analysis are shown in Table A.1.

[Table A.1 about here]

32
REFERENCES
Abreu, M., Mendes, V.: Commercial Bank Interest Margins and Profitability:Evidence
For Some E.U. Countries. University of Porto, Working Paper Series, No.122 (2002).

Aggarwal, R., Jacques, K.: Assessing the Impact of Prompt Corrective Action on Bank
Capital and Risk in Financial Services at the Crossroads: Capital Regulation in the 21st
Century. Proceedings of a Conference Held at the Federal Reserve Bank of New York on
February 26-27 (1998).

Allen, F., Carletti, E., Marquez, R.: Credit Market Competition and Capital Regulation.
Review of Financial Studies 24, 983-1018 (2011).

Allen, F., Gale, D.: Optimal Financial Crises, Journal of Finance 53, 1245–1284 (1998).

Allen, F., Gale, D.: Financial fragility, liquidity and asset prices.Journal of the European
Economic Association 2, 1015–84 (2004a).

Allen, F., Gale, D.: Financial intermediaries and markets. Econometrica 72 , 1023-1061
(2004b).

Allen, F., Carletti, E., Marquez, R.: Credit Market Competition and Capital Regulation.
European University Institute, Economics Working Papers ECO, No.08, (2009).

Adrian, T., Shin, H.S.: Liquidity and Financial Contagion. Banque de France, Financial
Stability Review – Special Issue on Liquidity, No.11 (2008).

Adrian, T., Shin, H.S.: The shadow banking system: Implications for financial regulation,
Banque de France, Financial Stability Review –The Future of Financial Regulation, No.
13 (2009).

Adrian, T., Shin, H.S.: Financial intermediaries and monetary economics. In: Friedman,
B., Woodford, M. (eds.) Handbook of Monetary Economics Vol. 3, pp.601-650., North
Holland, Amsterdam (2010).

Arellano, M., Bond, S. : Some tests of specification for panel data : Monte Carlo
evidence and an application to employment equations. Review of Economic Studies 58,
277-97 (1991).

Arellano, M., Bover, O.: Another look at the instrumental variables estimation of error-
components models. Journal of Econometrics 68, 29-51 (1995)

Aspachs, O., Goodhart, C., Tsomocos, D, Zicchino,L.: Towards a measure of financial

fragility. Annals of Finance 3 37-74 (2007).

33
Ashcraft, A.: Does the market discipline bank? New evidence from regulatory capital
mix. Journal of Financial Intermediation 17, 543-561 (2008).

Ayuso, J., Perez, D., Saurina, J.: Are capital buffers pro-cyclical? Evidence from Spanish
panel data. Journal of Financial Intermediation 13, 249-264 (2004).

Babihuga, R. : Macroeconomic and financial soundness indicators: An empirical

investigation, IMF Working paper, No. 115 (2007).

Bank for International Settlements: Fiscal policy and its implications for monetary and
financial stability. 10th BIS Annual Conference, 23-24 June 2011, BIS papers, No. 59
(2011).

Barth, J., Caprio, G., Levine, R.: Rethinking Bank Regulation: Till Angels Govern,
Cambridge University Press, Cambridge and New York (2005).

Berger, N.A.: The Relationship between Capital and Earnings in Banking. Journal of
Money, Credit and Banking 27, 432-456 (1995).

Berger, A. N., DeYoung R.: Problem Loans and Cost Efficiency in Commercial Banks.
Journal of Banking and Finance 21, 849-870 (1997).

Berger, A., DeYoung, R., Flannery, M., Lee, D., Öztekin, Ö.: How do large banking
organizations manage their capital ratios? Journal of Financial Services Research 34,
123-149 (2008).

Bernanke, B., Gertler, M.: Agency Costs, Net Worth and Business Fluctuations.
American Economic Review 79, 14–31 (1989).

Bernanke, B., Gertler, M.: Monetary policy and asset price volatility. In: New challenges
for monetary policy, proceedings of a symposium sponsored by the Federal Reserve Bank
of Kansas City, Jackson Hole, Wyoming, 26-29 August (1999).

Bernanke, B., Gertler, M., Gilchrist, S.: The Financial Accelerator in a Quantitative
Business Cycle Framework. In: Taylor, J., Woodford, M. (eds.) Handbook of
Macroeconomics, Vol. 1C, North Holland, Amsterdam (1999).

Bertrand, M. Schoar, A., Managing with style: The effect of managers on firm policies.
Quarterly Journal of Economics 118, 1169-1208 (2003).

Bikker, J.A., Metzemakers, P.A.J.: Bank provisioning behaviour and procyclicality. Bank
of Netherlands, Research Series Supervision, No.50 (2002).

Bini-Smaghi, L.: Has the financial sector grown too big? Speech by Lorenzo Bini
Smaghi, Member of the Executive Board of the ECB, Nomura Seminar. The paradigm
shift after the financial crisis, Kyoto, (15 April 2010).

34
http://www.ecb.int/press/key/date/2010/html/sp100415.en.html.

Blundell, R.W., Bond, S. : Initial conditions and moment restrictions in dynamic panel
data models. Journal of Econometrics 87, 115-43 (1998).

Borio, C.: Implementing a macroprudential framework: blending boldness and realism.

Keynote address for the BIS-HKMA research conference on: Financial stability: Towards
a Macroprudential approach. Honk Kong SAR, 5-6 July 2010.
http://www.bis.org/repofficepubl/hkimr201007.12c.htm.

Borio, C., Drehmann, M.: Assessing the risk of banking crises – revisitedBIS Quarterly
Review, March 29-46 (2009).

Borio, C., Lowe, P.: Asset prices, financial and monetary stability: exploring the nexus”,
BIS Working Papers, no 114, (2002a).

Borio, C., Lowe, P.: Assessing the risk of banking crises, BIS Quarterly Review,
December, pp 43–54 (2002b).

Borio, C.,Zhu, H.: Capital regulation, risk-taking and monetary policy: a missing link
in the transmission mechanism?, BIS Working Papers, no 268, (2008).

Bond, S.: Dynamic panel data models : A guide to micro data methods and practice.
CEMMAP, Working paper, No. CWP 09 (2002).

Brewer, E., Kaufman, G., Wall, L.: Bank capital ratios across countries: Why do they
vary? Journal of Financial Services Research 34, 177-201 (2008).

Brunnermeier, M, Pedersen,L.H.: Market Liquidity and Funding Liquidity. Review of

Financial Studies, Society for Financial Studies 22, 2201-2238 (2009).

Calomiris, C., Wilson, B.: Bank capital and portfolio management: the 1930’s “capital
crunch” and the scramble to shed risk. Journal of Business 77, 421-455 (2004).

Cihak, M., Schaeck K.: How well do aggregate bank ratios identify banking problems?
IMF Working paper, No. 275 (2007).

Cihak, M., Schaeck K.: How well do aggregate prudential ratios identify banking system
problems? Journal of Financial Stability 6, 130-144 (2010).

Claessens, S., Klingebiel, D., Laeven, L.: Crisis resolution, policies, and institutions, in:
Honohan, P. and Laeven, L. (eds.), Systemic financial crises. Containment and resolution,
Cambridge University Press, pp.169-194 (2005).

35
Covas, F., Fujita, S.: Time-varying capital requirements in a general equilibrium model of
liquidity dependence. Federal Reserve Bank of Philadelphia, Working paper No. 23
(2009).

De Bock, R., Demyanets, A.: Bank asset quality in emerging markets: Determinants and
spillovers. IMF Working Paper, No.71 (2012).

Demirguc-Kunt, A., Detragiache, E.: Financial liberalization and financial fragility. IMF,
Working Paper, No 83 (1998).

Demirguc-Kunt, A., Huizinga, H.: Determinants of Commercial Bank Interest Margins

and Profitability: Some International Evidence. World Bank Economic Review 13, 379-
408 (1999).

Demirguc-Kunt, A., Huizinga, H.: Financial structure and bank profitability. The World
Bank, Policy Research Working Paper Series, No. 2430 (2000).

Detragiache, E., Ho, G. : Responding to banking crises: Lessons from cross-country

evidence. IMF, Working Paper, No 18 (2010)

Drehmann, M., Borio, C., Gambacorta,L., Jimenez, G., Trucharte, C.: Countercyclical
capital buffers: Exploring options”, BIS Working Papers, no 317 (2010).

Drehmann,M., Borio, C., Tsatsaronis K.: Anchoring countercyclical capital buffers: the
role of credit aggregates. BIS Working Paper No. 355 (2011)

Diamond, D., Rajan, R.:A theory of bank capital. Journal of Finance 55, 2431-2465
(2000).

European Commission: Public Finances in EMU. European Economy No 3. European

Commission: Brussels (2010).

European Commission: Public Finances in EMU. European Economy No 3. European

Commission: Brussels (2009).

Flannery, M.: Debt maturity and the deadweight cost of leverage: Optimally financing
banking firms, American Economic Review 84, 320-331 (2008).

Flannery, M., Rangan, K.: What caused the bank capital build-up of the 1990s? Review
of Finance 12, 391-429 (2008).

Flannery, M., Sorescu, S.: Evidence of bank market discipline on subordinated debenture
yields: 1983-1991, Journal of Finance 51, 1347-1377 (1996).

36
Frank, M.,Goyal, V.: Corporate Leverage: How Much Do Managers Really Matter?
(2007). Available at SSRN: http://ssrn.com/abstract=971082 or
http://dx.doi.org/10.2139/ssrn.971082.

Furceri, D., Zdzienicka, A.: The consequences of banking crisis for public debt. OECD,
Working Paper, No. 81 (2010).

Furceri, D., Zdzienicka, A.: How costly are debt crises? MPRE, Available online at
http://mpra.ub.uni-muenchen.de/30953/ (2011).

Galati, G., Moessner, R.: Macroprudential policy – a literature review. BIS, Working
paper, No.337 (2011).

Goodfriend, M., McCallum, B.: Banking and interest rates in monetary policy analysis: A
quantitative exploration, Journal of Monetary Economics 54, 1480-1507 (2007).

Goodhart, C., Sunirand P., Tsomocos, D.: A Model to Analyse Financial Fragility:
Applications, Journal of Financial Stability 1, 1-30 (2004).

Goodhart, C., Sunirand, P., Tsomocos,D.: A Risk Assessment Model for Banks. Annals
of Finance 1, 197-224 (2005).

Goodhart, C., Sunirand P., Tsomocos, D.: A Model to Analyse Financial Fragility.
Economic Theory 27, 107-142 (2006).

Goodhart, C., Sunirand P., Tsomocos, D.: The Optimal Monetary Instrument for
Prudential Purposes. Journal of Financial Stability 7, 70-77 (2011).

Gropp, R., Heider, F.: The Determinants of Bank Capital Structure. European Central
Bank, Working Paper, No. 1096 (2009).

Guegan, D., Tarrant, W.: On the necessity of five risk measures. Ann. Finance 8, 533-552
(2012).

Honohan, P., Klingebiel. D. : The fiscal cost implications of an accommodating approach

to banking crises. Journal of Banking and Finance 27, 1539-1560 (2003).

Jacques, K., Nigro, P.: Risk-based Capital Portfolio Risk and Bank Capital:A
Simultaneous Equations Approach. Journal of Economics and Business 49, 533-547
(1997).

Jokipii,T., Milne, A.: Bank capital buffer and risk adjustment decisions. Journal of
Financial Stability 7, 165-178 (2011).

IMF: The IMF-FSB early warning exercise. Design and methodological toolkit,
(available online at:http://www.imf.org/external/np/pp/eng/2010/090110.pdf) (2011)

37
IMF: Ireland: Staff report on request for a Stand-By Arrangement, IMF, Country report
No. 10/366 (2010a)

IMF: Greece: Staff report on request for a Stand-By Arrangement, IMF, Country report
No. 10/110 (2010b).

IMF: Global Financial Stability Report, October, Washington: IMF (2010c).

IMF: Global Financial Stability Report, April, Washington: IMF (2010d).

IMF: Debt Bias and Other Distortions: Crisis-Related Issues in Tax Policy. IMF’s Fiscal
Affairs Department, June, Washington: IMF (2009a).

IMF: Global Financial Stability Report, October, Washington: IMF (2009b).

IMF: Global Financial Stability Report, April, Washington: IMF (2009c).

IMF: Global Financial Stability Report, October, Washington: IMF (2008a).

IMF: Global Financial Stability Report, April, Washington: IMF (2008b).

IMF: Global Financial Stability Report, September, Washington: IMF (2007a).

IMF: Global Financial Stability Report, April, Washington: IMF (2007b).

IMF: Global Financial Stability Report, September, Washington: IMF (2006a).

IMF: Global Financial Stability Report, April, Washington: IMF (2006b).

IMF: Global Financial Stability Report, September, Washington: IMF (2005a).

IMF: Global Financial Stability Report, April, Washington: IMF (2005b).

IMF: Global Financial Stability Report, September, Washington: IMF (2004a)

IMF: Global Financial Stability Report, April, Washington: IMF (2004b).

IMF: Global Financial Stability Report, September, Washington: IMF (2003a)

IMF: Global Financial Stability Report, March, Washington: IMF (2003b).

IMF: Global Financial Stability Report, December, Washington: IMF (2002a).

IMF: Global Financial Stability Report, September, Washington: IMF (2002b).

IMF: Global Financial Stability Report, June, Washington: IMF (2002c).

38
IMF: Global Financial Stability Report, March, Washington: IMF (2002d)

Kaminsky, G. : Currency and banking crises: The early warnings of distress. Federal
Reserve System, International Finance Discussion Paper, No. 629 (1998).

Kiyotaki, N., Moore, J.: Credit Cycles, Journal of Political Economy 105, 211–47 (1997).

Laeven L., Valencia, P.: Systemic banking crises: A new database. IMF, Working paper,
No. 224 (2008)

Lawrence, E. C. : Consumer Default and the Life Cycle Model. Journal of Money Credit
and Banking 27, 939-954 (1998).

Louzis, D., Vouldis, A.T., Metaxas, V.L.: Macroeconomic and bank-specific

determinates of non-performing loans in Greece: A comparative study of mortgage,
business and consumer loan portfolios, Journal of Banking and Finance 36, 1012-1027
(2012).

Marques O. M., Santos C. M.: Capital Structure Policy and Determinants: Theory and
Managerial Evidence. FFMA 2004 Basel Meetings paper (2003).

Martinez Peria, M. Schmuckler, S.: Do depositors punish banks for bad behaviour?
Market discipline, deposit insurance and banking crises. Journal of Finance 56, 1029-
1051 (2001).

Meh, C., Moran, K. : The role of bank capital in the propagation of shocks. Journal of
Economic Dynamics and Control 34, 555-576 (2010).

Mehran, H., Thakor, A.: Bank Capital and Value in the Cross-Section. Review of
Financial Studies 241019-67 (2011).

Mishkin, F.:The economics of money, banking and financial markets, Addison Wesley,
New York, 6th edition (2000).

Morgan, D., Stiroh, K.: Market discipline of banks: The asset test, Journal of Financial
Services Research 20, 195-208 (2001).

Myers, S., Rajan, R.: The paradox of liquidity, Quarterly Journal of Economics 113, 733-
771 (1998).

Nkusu, M.: Nonperforming Loans and macro-financial vulnerabilities in advanced

economies. IMF Working Paper, 161 (2011).

Organization of Economic Cooperation and Development: Economic Outlook. OECD:

Paris (2010)

39
Podpiera, R.: Does compliance with Basel core principles bring any measurable benefits?
IMF Working Paper ,No. 204 (2004).

Podpiera, J., Weill L.: Bad Luck or Bad Management? Emerging Banking Market
Experience. Journal of Financial Stability 4, 135-148 (2008).

Poghosyan, T., Cihak., M.: Distress in European Banks: an analysis based on a new data
set, IMF Working paper, No 9 (2009).

Puera, S., Keppo, J.: Optimal Bank Capital with Costly Recapitalization. Journal of
Business 79, 2163-2201 (2006).

Quagliarello M.: Banks’ Riskiness Over the Business Cycle: a Panel Analysis on Italian
Intermediaries. Applied Financial Economics 17, 119-138 (2007).

Reinhart, C. M., Rogoff, K.S.: Banking crises: An equal opportunity menace. NBER,
Working Paper No. 14587 (2008).

Reinhart, C.M., Rogoff., K.S.: The aftermath of financial crises. American Economic
Review 99, 466-72 (2009)

Reinhart C.M., Rogoff, K.S.: From financial crash to debt crisis, NBER, Working Paper,
No. 15795 (2010).

Repullo, R., Suarez, J.: The Procyclical Effects of Basel II. CEPR Discussion Papers, No.
6862, (2008).

Rime, B.: Capital requirements and bank behaviour, Empirical Evidence of Switzerland.
Journal of Banking and Finance 25, 789-805 (2001).

Rinaldi L., Sanchis-Arellano A.: Household Debt Sustainability: What explains

Household Non-performing Loans? An Empirical Analysis. ECB Working Paper, 570
(2006).

Roodman, D.: A note on the theme of too may instruments. Oxford Bulletin of
Economics and Statistics 71, 135-58 (2009a).

Roodman, D.: How to do xtabond2: An introduction to difference and system GMM in

Stata. The Stata Journal 9, 86-136 (2009b).

van Roy, P. :Capital Requirements and Bank Behaviour in the Early 1990: Cross-Country
Evidence. International Journal of Central Banking 4, 29-60 (2008).

40
Salas, V., Saurina, J.: Credit Risk in Two Institutional Regimes: Spanish Commercial and
Savings Banks. Journal of Financial Services Research 22, 203-224 (2002).

Saurina, J.: Loan loss provisions in Spain. A working macroprudential tool. Estabilidad
Financiera, No. 17, Banco de Espana, pp. 11-26 (2009a).

Saurina, J.: Dynamic provisioning. The experience of Spain., Crisis Response, Public
Policy for the Private Sector, Note Number 7, July, The World Bank (2009b).

Shrieves, R. E., Dahl, D.: The relationship between risk and capital in commercial banks.
Journal of Banking and Finance 16, 439-457 (1992).

Sorge, M., Virolainen, K. : A comparative analysis of macro stress-testing methodologies

with application to Finland. Journal of Financial Stability 2, 113-151 (2006).

Sundararajan, V., Enoch, C., San Jose, A., Hilbers, P., Krueger, R., Moretti, M., Slack,
G.: Financial soundness indicators: Analytical aspects and country practices. IMF
Occasional paper, No. 212. (2002).

Tagkalakis, A.: The effects of financial crisis on fiscal positions. European Journal of
Political Economy 29, 197-213 (2013).

Tsomocos, D.P.: Equilibrium Analysis, Banking and Financial Instability', Journal of

Mathematical Economics 39, 619-655 (2003).

Windmeijer, F.: A finite sample correction for the variance of linear efficient two-step
GMM estimators. Journal of Econometrics 126, 25-51 (2005).

Wong, J., Choi., K., Fong, T.: Determinants of the capital level of banks in Hong-Kong.
Hong-Kong Monetary Authority, Quarterly Bulletin, No.44 (2005).

41
Table A.1: Descriptive statistics
Variable Mean Std. Dev. Min Max
Severe financial crisis 0.1538 0.3615 0 1
Weak financial crisis 0.3808 0.4865 0 1
Change in debt ratio 1.1149 5.4293 -11.1118 21.9433
Output gap -0.3880 2.2228 -9.1863 6.2350
Debt ratio 69.6086 30.671 13.6087 192.856
Inflation rate 2.4697 2.4661 -3.8385 19.7877
Change in real share prices 4.707 24.621 -45.406 95.432
Change in real long term interest rate -0.0882 2.0252 -6.7814 9.3808
Change in real short term interest rate -0.2763 2.3249 -9.9056 9.5158
Credit growth 9.2662 20.772 -41.925 313.855
Financial intermediation 104.6605 45.3401 30.77 231.629
Change in real effective exchange 0.4277 5.8117 -26.644 20.3667
rate
Regulatory capital/risk weighted 12.061 1.5895 9.1 19.1
assets
Capital/assets 5.8189 1.6420 2.4 11
Nonperforming loans/total loans 2.6386 2.4197 0.2 15.5
Provisions to non performing loans 79.038 45.763 24.1 322.1
Return on assets 0.6452 0.4393 -1.3 2.4
Return on equity 11.654 8.1573 -36.5 29

42
Table 1: Severe financial crisis definition – how key macroeconomic and other variables behaved in the 2008-2009 financial
crisis

Change in real share Change in debt ratio Change in real GDP Change in
prices unemployment rate
cumulative Average cumulative Average cumulative Average cumulative Average Number of severe
per year per year per year per year financial crisis
episodes
Australia -43.9 -22.0 4.9 2.5 3.6 1.8 1.2 0.6 2
Austria -66.2 -33.1 9.6 4.8 -1.9 -0.9 0.9 0.5 2
Belgium -63.6 -31.8 12.9 6.5 -1.9 -0.9 0.4 0.2 2
Canada -28.9 -14.5 17.5 8.7 -2.2 -1.1 2.2 1.1 2
Switzerland -43.7 -21.9 -4.3 -2.1 0.0 0.0 0.9 0.5 2
Germany -49.1 -24.6 10.9 5.4 -4.0 -2.0 -0.9 -0.4 2
Denmark -53.4 -26.7 17.7 8.8 -5.8 -2.9 0.5 0.2 2
Spain -42.1 -21.1 20.3 10.1 -2.9 -1.4 9.8 4.9 2
Finland -59.1 -29.5 11.1 5.5 -7.1 -3.6 1.4 0.7 2
France -50.8 -25.4 16.4 8.2 -2.5 -1.2 1.1 0.6 2
UK -36.2 -18.1 24.9 12.5 -4.4 -2.2 2.2 1.1 2
Greece -71.4 -35.7 21.5 10.8 -1.0 -0.5 1.2 0.6 2
Ireland -83.6 -41.8 42.0 21.0 -11.1 -5.6 7.3 3.6 2
Italy -64.0 -32.0 16.4 8.2 -6.4 -3.2 1.7 0.9 2
Japan -53.8 -26.9 25.8 12.9 -6.5 -3.2 1.2 0.6 2
Netherlands -57.9 -28.9 16.5 8.2 -2.0 -1.0 0.3 0.2 2
Norway -50.2 -25.1 -9.4 -4.7 0.3 0.1 0.6 0.3 2
Portugal -48.8 -24.4 17.5 8.8 -2.5 -1.3 1.5 0.7 2
Sweden -43.1 -21.6 4.4 2.2 -5.8 -2.9 2.2 1.1 2
US -38.8 -19.4 21.1 10.6 -2.6 -1.3 4.7 2.3 2
Un-weighted 2
average -26.2 7.4 -1.7 1.0 (total: 40)

43
Table 2: Weaker financial crises definition – how key macroeconomic and other variables behaved in
these 99 country year-observations
Change in real share Change in debt ratio Change in real GDP Change in
prices unemployment rate
cumulative Average per cumulative Average cumulative Average cumulative Average Number of weak
year per year per year per year financial crisis
episodes
Australia -56.4 -14.1 1.5 0.4 10.9 2.7 0.4 0.1 4
Austria -83.9 -21.0 12.3 3.1 5.1 1.3 -0.4 -0.1 4
Belgium -117.1 -19.5 -3.3 -0.6 4.8 0.8 -1.8 -0.3 6
Canada -59.6 -14.9 15.9 4.0 2.5 0.6 3.1 0.8 4
Switzerland -90.4 -18.1 0.2 0.0 1.4 0.3 2.8 0.6 5
Germany -122.1 -24.4 15.8 3.2 -2.8 -0.6 2.8 0.6 5
Denmark -80.9 -20.2 15.5 3.9 -4.6 -1.1 -0.2 0.0 4
Spain -86.6 -17.3 9.2 1.8 6.6 1.3 4.5 0.9 5
Finland -142.2 -28.4 10.2 2.0 -1.1 -0.2 -2.2 -0.4 5
France -119.2 -23.8 22.2 4.4 1.5 0.3 0.5 0.1 5
UK -84.5 -16.9 21.3 4.3 3.0 0.6 1.8 0.4 5
Greece -158.4 -26.4 32.4 5.4 17.0 2.8 -1.3 -0.2 6
Ireland -116.6 -29.1 38.7 9.7 -0.2 0.0 7.9 2.0 4
Italy -122.7 -24.5 11.6 2.3 -4.1 -0.8 0.0 0.0 5
Japan -120.9 -20.2 62.1 10.3 -6.5 -1.1 2.6 0.4 6
Netherlands -137.5 -27.5 14.5 2.9 0.3 0.1 3.1 0.6 5
Norway -93.1 -15.5 7.2 1.2 6.8 1.1 1.4 0.2 6
Portugal -132.5 -22.1 20.6 3.4 3.3 0.6 3.3 0.5 6
Sweden -109.1 -21.8 -0.6 -0.1 0.6 0.1 2.4 0.5 5
US -57.2 -14.3 23.4 5.9 0.3 0.1 6.4 1.6 4
Un-weighted 4.95
average -21.0 3.4 0.4 0.4 (Total 99)

44
Table 3: FSIs in crisis and non crisis periods – average effects for the 20 OECD
countries of the sample
1 2 3 4 5 6
Average in Average in Difference Average in Average Difference
the 2008- all periods all excluding
2009 excluding financial periods
financial the 2008- crisis where
crisis 2009 crisis (weak financial
(severe financial crisis
financial crisis occurred
crisis definition)
definitions)
Regulatory
capital/risk
weighted
assets 12.884 12.011 0.873 12.007 11.879 0.127
Capital/assets 5.584 5.858 -0.274 5.672 5.7198 -0.047
Nonperformin
g loans/total
loans 2.766 2.713 0.052 2.754 2.5772 0.176
Provisions to
non
performing
loans 62.826 77.690 -14.864 60.454 74.147 -13.693
Return on
assets 0.2205 0.701 -0.481 0.438 0.739 -0.300
Return on
equity 3.523 12.967 -9.443 7.352 14.024 -6.672

Table 4: FSIs in times of financial crisis – average effects for the 20 OECD countries
of the sample
1 2 3
2008-2009 financial All financial crisis Late 1990s and early
crisis (severe financial (weak financial crisis 2000s financial crisis
crisis definition) definition)
Regulatory capital/risk
weighted assets 12.884 12.007 11.975
Capital/assets 5.584 5.672 5.936
Nonperforming
loans/total loans 2.766 2.754 2.831
Provisions to non
performing loans 62.826 60.454 78.990
Return on assets 0.220 0.438 0.632
Return on equity 3.523 7.352 10.675

45
Table 5: Regulatory capital/risk weighted assets –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Regulatory
capital/risk
weighted
assets
Regulatory 0.864 0.792 0.868 0.873 0.849 0.869 0.846 0.853
capital/risk (9.34)** (8.89)** (9.51)** (8.82)** (6.26)** (7.63)** (6.05)** (6.15)*
weighted * * * * * * * **
assets (t-1)
Output gap -0.124 -0.176 -0.123 -0.097 -0.110 -0.113 -0.104 -0.082
(t) (- (- (- (-1.67) (-1.27) (-1.38) (-1.18) (-0.61)
2.32)** 3.44)** 2.34)**
*
Inflation rate -0.011 0.003 -0.012 -0.002 -0.025 -0.027 -0.020 -0.015
(t-1) (-0.33) (0.04) (-0.37) (-0.06) (-0.58) (-0.54) (-0.53) (-0.31)
Change in -0.011 -0.0009 -0.010 -0.011 -0.012 -0.012 -0.012 -0.011
real effective (-0.91) (-0.06) (-0.88) (-0.95) (-1.12) (-1.04) (-1.05) (-0.87)
exchange rate
(t-1)
Change in 0.049 0.049 0.054 0.018 0.017 0.031
real short (2.17)** (2.15)** (2.65)** (0.45) (0.46) (0.96)
term interest
rate (t-1)
Change in -0.029 -0.036
real long (-0.79) (-1.03)
term interest
rate (t-1)
Debt ratio (t- -0.001 -
1) (-0.43) 0.00004
(-0.01)
Change in 0.019 0.020
debt ratio (t- (0.88) (0.60)
1)
Credit growth -0.0002 0.008 0.0003 -0.0004 0.009 0.014 0.009 0.009
(t) (-0.02) (0.91) (0.03) (-0.04) (0.64) (0.94) (0.61) (0.68)
Financial -0.0003 0.0004 -0.0003 -0.0009 -0.0004 -0.0004 -0.0003 -
intermediatio (-0.13) (0.23) (-0.12) (-0.39) (-0.18) (-0.18) (-0.12) 0.0007
n (t-1) (-0.24)
Change in -0.005 -0.003 -0.005 -0.004 -0.008 -0.011 -0.008 -0.008
real share (- (-1.00) (-1.82)* (-1.78)* (-1.48) (- (-1.45) (-1.45)
prices (t-1) 1.98)** 2.26)**
Severe 0.587 0.497 0.596 0.609 0.489 0.547 0.506 0.512
financial (2.07)** (1.53) (2.15)** (1.97)** (1.02) (1.20) (1.06) (0.88)
crisis (short
run)
Severe 4.314 2.393 4.497 4.815 3.249 4.186 3.299 3.477
financial (1.45) (1.52) (1.46) (1.24) (0.86) (1.02) (0.87) (0.75)
crisis (long
run)
Constant 1.795 2.459 1.818 1.729 1.939 1.687 1.939 1.928
(1.59) (2.30)** (1.59) (1.47) (1.29) (1.32) (1.25) (1.24)

46
Table 5 (continued)
Obs. 226 226 226 225 226 226 226 225
Wald –Chi2 Wald Wald Wald Wald Wald Wald Wald Wald
(df) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9) chi2(10) chi2(1
(p-value) : : : : : : : 0) :
425.85 434.65 500.63 464.34 122.86 192.75 265.35 211.99
(0.000) (0.0000) (0.000) (0.000) (0.000) (0.0000) (0.0000) (0.000
0)
Residual’s 0.988 0.925 0.989 0.978 0.960 0.964 0.965 0.946
2nd order AR
(p-values)
Hansen test 0.396 0.307 0.375 0.382 0.246 0.285 0.262 0.215
of
overidentifyi
ng
restrictions
(p-values)
No of 19 19 20 20 19 19 20 20
instruments
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

Table 6: Regulatory capital/risk weighted assets –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Regulator
y
capital/risk
weighted
assets
Regulatory 0.903 0.863 0.908 0.902 0.851 0.883 0.849 0.869
capital/risk (9.30)** (7.21)** (9.81)** (8.16)** (4.38)** (5.37)** (4.31)** (4.54)**
weighted * * * * * * * *
assets (t-1)
Output -0.157 -0.201 -0.159 -0.158 -0.161 -0.170 -0.162 -0.168
gap (t) (- (- (- (- (- (- (- (-
3.09)*** 4.61)*** 3.04)*** 2.75)*** 2.67)*** 3.45)*** 2.64)*** 2.28)**
Inflation 0.008 0.028 0.006 0.018 0.001 0.0002 0.002 0.002
rate (t-1) (0.17) (0.41) (0.12) (0.35) (0.04) (0.00) (0.05) (0.05)
Change in -0.011 -0.009 -0.011 -0.013 -0.013 -0.014 -0.013 -0.014
real (-0.95) (-0.67) (-0.90) (-0.97) (-1.15) (-1.24) (-1.16) (-1.07)
effective
exchange
rate (t-1)

47
Table 6(continued)
Change in real short 0.049 0.053 0.052 0.025 0.026 0.035
term interest rate (t- (1.33) (1.39) (1.45) (0.46) (0.47) (0.73)
1)
Change in real long -0.058 -0.050
term interest rate (t- (-1.31) (-0.90)
1)
Debt ratio (t-1) -0.0006 -0.001
(-0.25) (-0.34)
Change in debt ratio - 0.0005
(t-1) 0.00006 (0.02)
(-0.00)
Credit growth (t) - 0.005 0.0009 0.0004 0.006 0.009 0.006 0.006
0.00007 (0.56) (0.07) (0.03) (0.42) (0.60) (0.40) (0.45)
(-0.01)
Financial 0.0004 0.0004 0.0005 0.0003 0.0004 0.0004 0.0004 0.0004
intermediation (t-1) (0.19) (0.23) (0.29) (0.16) (0.15) (0.16) (0.16) (0.16)
Change in real share -0.003 -0.004 -0.003 -0.002 -0.005 -0.007 -0.005 -0.005
prices (t-1) (-0.91) (-1.35) (-0.82) (-0.69) (-1.24) (-1.64) (-1.22) (-1.28)
Weak financial 0.218 0.206 0.201 0.172 0.136 0.221 0.125 0.116
crisis (short run) (0.80) (0.78) (0.73) (0.56) (0.48) (0.91) (0.45) (0.40)
Weak financial 2.235 1.502 2.192 1.762 0.912 1.883 0.830 0.887
crisis (long run) (0.49) (0.50) (0.47) (0.37) (0.34) (0.46) (0.33) (0.29)
Constant 1.209 1.566 1.161 1.182 1.755 1.344 1.842 1.551
(0.94) (1.12) (0.88) (0.84) (0.80) (0.74) (0.79) (0.71)
Obs. 226 226 226 225 226 226 226 225
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9) chi2(10) chi2(10)
: : : : : : : :
372.99 217.27 568.30 398.62 103.73 128.79 161.66 153.76
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.0000) (0.000)
Residual’s 2nd order 0.988 0.990 0.988 0.984 0.988 0.964 0.990 0.978
AR (p-values)
Hansen test of 0.274 0.341 0.266 0.254 0.216 0.294 0.214 0.206
overidentifying
restrictions (p-
values)
No of instruments 19 19 20 20 19 19 20 20
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

48
Table 7: Capital/Assets –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable : deviations
Capital/Asse
ts
Capital/Asse 0.128 0.152 0.139 0.147 0.116 0.118 0.139 0.105
ts (t-1) (0.73) (0.84) (0.75) (0.91) (0.80) (0.84) (0.98) (0.83)
Output gap 0.019 0.020 0.008 0.089 -0.021 -0.025 -0.029 0.049
(t) (0.34) (0.35) (0.15) (1.00) (-0.31) (-0.42) (-0.42) (0.59)
Inflation rate 0.037 0.028 0.021 0.034 0.027 0.022 0.011 0.034
(t-1) (0.70) (0.59) (0.48) (0.57) (0.46) (0.38) (0.22) (0.52)
Change in -0.020 -0.019 -0.023 -0.019 -0.029 -0.029 -0.033 -0.025
real effective (-1.36) (-1.35) (-1.40) (-1.36) (-1.06) (-1.12) (-1.19) (-1.06)
exchange
rate (t-1)
Change in 0.015 0.011 0.011 -0.007 0.011 -0.015 -0.004
real short (1.31) (0.84) (0.89) (-0.37) (0.42) (-0.57) (-0.18)
term interest
rate (t-1)
Change in 0.029
real long (1.32)
term interest
rate (t-1)
Debt ratio (t- -0.006 -0.007
1) (-0.69) (-0.76)
Change in 0.045 0.036
debt ratio (t- (1.14) (0.81)
1)
Credit 0.0004 0.0005 -2.97e- 0.0009 0.004 0.004 0.004 0.005
growth (t) (0.07) (0.08) 06 (0.16) (0.44) (0.44) (0.43) (0.51)
(-0.00)
Financial -0.004 -0.004 -0.004 -0.004 -0.005 -0.005 -0.005 -0.005
intermediatio (-0.49) (-0.54) (-0.54) (-0.50) (-0.79) (-0.89) (-0.71) (-0.73)
n (t-1)
Change in -0.002 -0.001 -0.002 -0.003 -0.002 -0.002 -0.002 -0.003
real share (-0.60) (-0.29) (-0.45) (-0.78) (-0.60) (-0.44) (-0.47) (-0.91)
prices (t-1)
Severe -0.209 -0.195 -0.224 -0.056 -0.204 -0.236 -0.191 -0.121
financial (-1.36) (-1.20) (-1.30) (-0.27) (-0.91) (-1.09) (-0.76) (-0.58)
crisis (short
run)
Severe -0.239 -0.229 -0.261 -0.065 -0.230 -0.267 -0.222 -0.136
financial (-1.32) (-1.22) (-1.30) (-0.27) (-0.90) (-1.10) (-0.77) (-0.57)
crisis (long
run)
Constant 5.233 5.158 5.605 5.116 5.569 5.622 5.932 5.643
(2.93)** (2.75)** (3.15)** (3.10)** (3.89)** (4.07)** (3.78)** (4.05)**
* * * * * * * *
Obs. 222 222 222 221 222 222 222 221
Wald –Chi2 Wald Wald Wald Wald Wald Wald Wald Wald
(df) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) : chi2(9) : chi2(10) chi2(10)
(p-value) :52.62 : : : 61.71 74.63 : :
(0.000) 102.76 58.37 39.55 (0.000) (0.0000) 71.68 62.21
(0.000) (0.000) (0.000) (0.0000) (0.0000)

49
Table 7 (continued)
Residual’s 2nd order 0.406 0.381 0.408 0.362 0.463 0.451 0.458 0.452
AR (p-values)
Hansen test of 0.554 0.616 0.514 0.605 0.322 0.334 0.276 0.342
overidentifying
restrictions (p-
values)
No of instruments 19 19 20 20 19 19 20 20
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

Table 8: Capital/Assets –two step system GMM estimations

1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable : deviations
Capital/Assets
Capital/Assets 0.169 0.188 0.191 0.143 0.175 0.188 0.181 0.152
(t-1) (0.85) (1.03) (0.95) (0.81) (1.11) (1.19) (1.28) (1.03)
Output gap (t) 0.023 0.025 0.013 0.083 -0.031 -0.034 -0.038 0.029
(0.40) (0.46) (0.26) (1.02) (-0.43) (-0.55) (-0.59) (0.34)
Inflation rate 0.040 0.033 0.013 0.031 0.021 0.020 -0.007 0.020
(t-1) (0.77) (0.68) (0.28) (0.56) (0.44) (0.45) (-0.15) (0.37)
Change in real -0.020 -0.019 -0.023 -0.020 -0.026 -0.026 -0.031 -0.024
effective (-1.60) (-1.58) (-1.65) (-1.50) (-1.15) (-1.19) (-1.39) (-1.16)
exchange rate
(t-1)
Change in real 0.011 0.009 0.009 -0.019 -0.024 -0.018
short term (0.58) (0.39) (0.69) (-0.74) (-0.83) (-0.78)
interest rate
(t-1)
Change in real 0.024 -0.005
long term (1.45) (-0.16)
interest rate
(t-1)
Debt ratio (t- -0.007 -0.007
1) (-0.91) (-0.88)
Change in 0.042 0.029
debt ratio (t-1) (0.99) (0.70)
Credit growth 0.001 0.001 0.001 0.002 0.005 0.006 0.005 0.006
(t) (0.20) (0.18) (0.14) (0.27) (0.47) (0.49) (0.44) (0.53)
Financial -0.004 -0.005 -0.005 -0.004 -0.005 -0.005 -0.006 -0.005
intermediation (-0.53) (-0.56) (-0.62) (-0.55) (-0.81) (-0.84) (-0.85) (-0.71)
(t-1)

50
Table 8 (continued)
Change in -0.002 -0.002 -0.002 -0.002 -0.001 -0.0007 -0.0007 -0.002
real share (-0.73) (-0.51) (-0.59) (-0.80) (-0.28) (-0.19) (-0.20) (-0.51)
prices (t-1)
Weak -0.121 -0.139 -0.131 -0.043 -0.006 -0.031 -0.023 0.041
financial (-0.72) (-1.07) (-0.79) (-0.27) (-0.04) (-0.21) (-0.16) (0.24)
crisis (short
run)
Weak -0.145 -0.171 -0.162 -0.050 -0.008 -0.039 -0.028 0.048
financial (-0.62) (-0.88) (-0.67) (-0.26) (-0.04) (-0.20) (-0.16) (0.25)
crisis (long
run)
Constant 5.0619 5.012 5.458 5.152 5.180 5.127 5.786 5.313
(2.68)** (2.59)** (2.98)** (3.23)** (3.69)** (3.64)** (3.70)** (3.80)**
* * * * * * * *
Obs. 222 222 222 221 222 222 222 221
Wald –Chi2 Wald Wald Wald Wald Wald Wald Wald Wald
(df) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9): chi2(10) chi2(10)
(p-value) : : : : 30.39 : 85.00 : :
74.58 94.30 112.81 (0.000) 69.19 (0.0000) 86.91 57.47
(0.000) (0.000) (0.000) (0.0000) (0.0000) (0.0000)
Residual’s 0.404 0.377 0.392 0.389 0.447 0.429 0.447 0.444
2nd order AR
(p-values)
Hansen test 0.549 0.636 0.529 0.546 0.296 0.307 0.263 0.305
of
overidentifyi
ng
restrictions
(p-values)
No of 19 19 20 20 19 19 20 20
instruments
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample orrection have been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

51
Table 9: Non performing loans to total loans –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Non
performing
loans to
total loans
Nonperform 0.838 0.819 0.883 0.848 0.866 0.878 0.887 0.875
ing loans to (15.11)* (14.10)* (12.37)* (14.59)* (14.85)* (13.89)* (13.33)* (11.98)*
total loans ** ** ** ** ** ** ** **
(t-1)
Output gap -0.147 -0.121 -0.151 -0.129 -0.118 -0.103 -0.121 -0.119
(t) (- (- (- (-1.96)* (-1.94)* (-1.94)* (- (-1.81)*
2.18)** 2.12)** 2.00)** 2.07)**
Inflation 0.002 -0.005 0.009 0.003 0.0004 -0.007 0.003 0.001
rate (t-1) (0.10) (-0.30) (0.61) (0.21) (0.02) (-0.24) (0.08) (0.04)
Unemploym 0.003 0.011 -0.012 -0.002 0.0006 0.002 -0.005 -0.001
ent rate (t-1) (0.09) (0.37) (-0.66) (-0.09) (0.03) (0.09) (-0.24) (-0.06)
Change in -0.008 -0.008 -0.006 -0.009 -0.009 -0.007 -0.009 -0.009
real (-1.07) (-0.94) (-0.76) (-1.25) (-0.86) (-0.63) (-0.84) (-0.84)
effective
exchange
rate (t-1)
Change in 0.083 0.092 0.090 0.096 0.099 0.099
real short (3.60)** (3.85)** (3.58)** (3.45)** (3.47)** (3.74)**
term interest * * * * * *
rate (t-1)
Change in 0.084 0.097
real long (3.17)** (3.41)**
term interest * *
rate (t-1)
Debt ratio 0.002 0.0009
(t-1) (0.34) (0.18)
Change in 0.002 -0.001
debt ratio (t- (0.08) (-0.05)
1)
Credit -0.0007 0.002 -0.002 -0.003 0.002 0.006 0.002 0.002
growth (t) (-0.08) (0.26) (-0.22) (-0.31) (0.20) (0.67) (0.23) (0.23)
Financial 0.001 0.001 0.0009 0.0009 0.003 0.003 0.003 0.003
intermediati (0.53) (0.71) (0.47) (0.40) (1.17) (1.49) (1.23) (1.17)
on (t-1)
Severe 0.486 0.591 0.540 0.502 0.465 0.548 0.460 0.463
financial (3.14)** (3.73)** (3.50)** (3.31)** (2.30)** (2.77)** (2.28)** (2.24)**
crisis (short * * * * *
run)
Severe 2.989 3.263 4.604 3.310 3.475 4.499 4.063 3.695
financial (1.95)* (2.08)** (1.36) (1.92)* (1.73)* (1.64) (1.48) (1.66)*
crisis (long *
run)
Constant 0.147 0.090 0.025 0.190 -0.153 -0.264 -0.261 -0.179
(0.41) (0.24) (0.07) (0.58) (-0.44) (-0.78) (-0.75) (-0.46)
Obs. 223 223 223 222 223 223 223 222

52
Table 9 (Continued)
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9) chi2(10) chi2(10)
: : : : : : : :
704.73 522.65 1296.16 904.15 616.43 560.18 873.12 1191.50
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Residual’s 2nd order 0.178 0.155 0.168 0.255 0.189 0.124 0.180 0.233
AR (p-values)
Hansen test of 0.293 0.346 0.220 0.281 0.270 0.278 0.246 0.259
overidentifying
restrictions (p-
values)
No of instruments 19 19 20 20 19 19 20 20
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

Table 10: Non performing loans to total loans –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Non
performing
loans to
total loans
Nonperform 0.777 0.757 0.818 0.759 0.845 0.839 0.869 0.877
ing loans to (16.68)* (14.48)* (13.03)* (13.13)* (12.91)* (12.37)* (12.79)* (11.38)*
total loans ** ** ** ** ** ** ** **
(t-1)
Output gap -0.125 -0.122 -0.158 -0.135 -0.156 -0.157 -0.152 -0.187
(t) (-1.53) (-1.75)* (-1.73)* (-1.58) (-1.86)* (-1.87)* (-1.92)* (-1.73)*
Inflation 0.014 0.011 0.025 0.012 0.006 0.004 0.0009 0.007
rate (t-1) (0.73) (0.51) (0.99) (0.57) (0.20) (0.18) (0.03) (0.20)
Unemploym 0.024 0.034 0.003 0.026 0.010 0.017 0.012 0.002
ent rate (t-1) (0.78) (0.84) (0.12) (0.74) (0.36) (0.56) (0.46) (0.08)
Change in -0.009 -0.011 -0.007 -0.008 -0.014 -0.012 -0.013 -0.016
real (-1.00) (-1.00) (-0.62) (-0.86) (-1.08) (-0.89) (-1.05) (-1.05)
effective
exchange
rate (t-1)
Change in 0.074 0.091 0.084 0.096 0.097 0.103
real short (2.19)** (2.80)** (2.48)** (2.45)** (2.58)** (2.45)**
term interest *
rate (t-1)

53
Table 10 (Continued)
Change in real 0.069 0.081
long term (2.09)** (2.06)**
interest rate (t-1)
Debt ratio (t-1) 0.005 -0.002
(0.62) (-0.33)
Change in debt 0.014 -0.017
ratio (t-1) (0.51) (-0.51)
Credit growth -0.003 -0.001 -0.003 -0.006 -0.001 0.003 -0.002 -0.0001
(t) (-0.36) (-0.14) (-0.38) (-0.57) (-0.10) (0.30) (-0.16) (-0.02)
Financial 0.002 0.003 0.001 0.001 0.004 0.005 0.004 0.005
intermediation (0.92) (1.08) (0.64) (0.44) (1.35) (1.65)* (1.41) (1.61)
(t-1)
Weak financial 0.188 0.194 0.144 0.145 0.129 0.151 0.137 0.083
crisis (short (1.55) (2.14)** (1.05) (1.00) (0.85) (0.98) (0.96) (0.43)
run)
Weak financial 0.844 0.798 0.794 0.604 0.836 0.936 1.049 0.673
crisis (long run) (1.55) (1.85)* (0.99) (1.09) (0.83) (0.94) (0.82) (0.45)
Constant -0.016 -0.120 -0.172 0.212 -0.294 -0.423 -0.235 -0.352
(-0.03) (-0.22) (-0.38) (0.38) (-0.52) (-0.79) (-0.44) (-0.67)
Obs. 223 223 223 222 223 223 223 222
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9) chi2(10) chi2(10)
: : : : : : : :
519.99 388.89 1018.10 655.53 449.48 448.56 924.24 800.85
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.000)
Residual’s 2nd 0.404 0.418 0.316 0.392 0.248 0.227 0.234 0.370
order AR (p-
values)
Hansen test of 0.388 0.295 0.195 0.259 0.153 0.152 0.171 0.137
overidentifying
restrictions (p-
values)
No of 19 19 20 20 19 19 20 20
instruments
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

54
Table 11: Provisions to non performing loans –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable : deviations
Provisions to
non
performing
loans
Provisions to 0.992 1.002 0.999 0.984 0.877 0.889 0.886 0.894
non (8.69)* (8.49)* (9.48)* (8.31)* (11.05)* (14.99)* (10.04)* (10.98)*
performing ** ** ** ** ** ** ** **
loans (t-1)
Output gap -1.657 -1.628 -1.626 -1.377 -0.598 0.095 -0.649 -1.518
(t) (-1.83)* (-1.83)* (-1.74)* (-1.20) (-0.58) (0.07) (-0.61) (-1.62)
Inflation rate -0.911 -0.906 -0.894 -1.022 -0.695 -1.232 -0.947 -1.001
(t-1) (- (- (- (- (-0.60) (-0.98) (-0.74) (-1.23)
2.25)** 2.39)** 2.00)** 2.79)**
*
Unemploym -0.268 -0.266 -0.226 -0.256 0.163 2.026 0.316 0.063
ent rate (t-1) (-0.62) (-0.55) (-0.58) (-0.57) (0.21) (1.19) (0.40) (0.08)
Change in -0.277 -0.256 -0.272 -0.222 -0.087 0.039 -0.140 -0.141
real effective (- (- (- (- (-0.29) (0.17) (-0.62) (-0.58)
exchange 2.19)** 2.51)** 2.13)** 2.04)**
rate (t-1)
Change in -0.219 -0.203 -0.204 0.354 0.545 0.216 -0.153
real short (-0.50) (-0.45) (-0.37) (0.24) (0.68) (0.15) (-0.12)
term interest
rate (t-1)
Change in -0.047
real long (-0.05)
term interest
rate (t-1)
Debt ratio (t- 0.008 -0.052
1) (0.15) (-0.48)
Change in 0.187 -0.371
debt ratio (t- (0.43) (-0.81)
1)
Credit -0.307 -0.321 -0.312 -0.287 -0.312 -0.343 -0.319 -0.298
growth (t) (-1.29) (-1.78)* (-1.30) (-1.09) (-1.32) (-1.45) (-1.35) (-1.29)
Financial -0.095 -0.094 -0.087 -0.097 -0.069 -0.081 -0.071 -0.051
intermediatio (- (- (-1.79)* (-1.90)* (-1.72)* (-1.93)* (-1.82)* (-1.35)
n (t-1) 2.18)** 2.26)**
Severe -12.337 -12.294 -12.260 -12.033 -15.196 -18.829 -14.106 -14.796
financial (-1.70)* (-1.72)* (-1.66) (-1.59) (-2.12)** (- (-2.03)** (-1.94)*
crisis (short 3.10)***
run)
Severe - 6465.94 - - - -169.621 -123.622 -139.703
financial 1559.27 7 33141.5 756.819 123.3674 (-1.54) (-1.07) (-1.13)
crisis (long (-0.07) (0.02) 6 (-0.13) (-1.31)
run) (-0.00)
Constant 21.091 20.344 19.003 21.785 23.785 24.049 26.595 21.013
(1.97)* (1.78)* (2.02)* (1.97)* (2.55)** (2.41)** (2.05)** (2.56)**
* *
Obs. 184 184 184 183 184 184 184 183

55
Table 11 (Continued)
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(8) chi2(8) chi2(9) chi2(9): chi2(9) chi2(9) chi2(10) chi2(10)
: : : 209.10 : : : :
196.52 221.79 305.87 (0.000) 600.10 759.74 573.46 494.65
(0.000) (0.0000) (0.000) (0.000) (0.0000) (0.0000) (0.0000)
Residual’s 2nd 0.165 0.159 0.161 0.161 0.143 0.111 0.146 0.190
order AR (p-
values)
Hansen test of 0.501 0.501 0.491 0.423 0.300 0.430 0.313 0.320
overidentifying
restrictions (p-
values)
No of instruments 19 19 20 20 19 19 20 20
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

Table 12: Provisions to non performing loans –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable : deviations
Provisions to
non
performing
loans
Provisions to 1.017 1.034 1.018 1.009 0.878 0.872 0.881 0.899
non (7.37)** (7.23)** (7.99)** (7.13)** (9.76)** (9.97)** (10.11)** (9.65)**
performing * * * * * * * *
loans (t-1)
Output gap -1.525 -1.472 -1.516 -0.682 -0.439 0.164 -0.429 -0.124
(t) (- (- (- (-0.65) (-0.49) (0.11) (-0.49) (-0.10)
2.02)** 2.18)** 2.00)**
Inflation rate -1.099 -1.129 -1.035 -1.186 -0.685 -0.589 -0.819 -1.012
(t-1) (-1.63) (-1.86)* (-1.53) (-1.85)* (-0.65) (-0.57) (-0.65) (-1.03)
Unemployme -0.459 -0.423 -0.490 -0.471 -0.139 -0.068 -0.034 -0.236
nt rate (t-1) (-1.08) (-0.84) (-1.81)* (-1.22) (-0.19) (-0.08) (-0.05) (-0.30)
Change in -0.246 -0.225 -0.235 -0.213 -0.164 -0.145 -0.175 -0.089
real effective (-1.30) (-1.35) (-1.24) (-1.17) (-0.71) (-0.62) (-0.87) (-0.37)
exchange rate
(t-1)
Change in -0.521 -0.523 -0.509 -0.794 -0.809 -0.967
real short (-1.20) (-1.21) (-0.94) (-0.52) (-0.55) (-0.73)
term interest
rate (t-1)

56
Table 12 (Continued)
Change in real -0.228 0.705
long term (-0.26) (0.31)
interest rate (t-1)
Debt ratio (t-1) 0.015 -0.032
(0.29) (-0.34)
Change in debt 0.463 0.248
ratio (t-1) (1.20) (0.55)
Credit growth (t) -0.176 -0.192 -0.178 -0.208 -0.173 -0.328 -0.169 -0.157
(-0.64) (-0.87) (-0.67) (-0.65) (-0.95) (- (-0.90) (-0.76)
2.05)**
Financial -0.111 -0.109 -0.110 -0.122 -0.074 -0.090 -0.072 -0.079
intermediation (t- (- (- (- (- (- (- (-1.88)* (-1.75)*
1) 2.45)** 2.35)** 2.35)** 2.43)** 1.77)* 2.17)**
Weak financial -5.591 -5.616 -5.585 -4.559 -3.555 -6.396 -3.344 -2.749
crisis (short run) (- (- (- (-1.69)* (-0.68) (-1.38) (-0.67) (-0.51)
1.97)** 2.00)** 1.99)**
Weak financial 321.478 166.434 310.623 487.609 - -49.824 -28.043 -27.120
crisis (long run) (0.12) (0.23) (0.14) (0.07) 29.193 (-1.28) (-0.81) (-0.65)
(-0.84)
Constant 22.320 20.753 21.222 24.264 23.478 27.706 24.840 23.524
(1.67)* (1.48) (1.52) (2.08)** (1.87)* (2.53)** (1.76)* (1.82)*
Obs. 184 184 184 183 184 184 184 183
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(8) chi2(8) chi2(9) chi2(9) chi2(9) chi2(9) chi2(10) chi2(10)
: : : : : : : : 656.49
451.53 454.64 522.19 389.62 562.36 481.27 510.03 (0.0000)
(0.000) (0.0000) (0.0000) (0.0000) (0.000) (0.0000) (0.000)
Residual’s 2nd 0.180 0.173 0.178 0.157 0.172 0.142 0.173 0.174
order AR (p-
values)
Hansen test of 0.525 0.556 0.533 0.439 0.385 0.381 0.384 0.372
overidentifying
restrictions (p-
values)
No of instruments 19 19 20 20 19 19 20 20
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-3 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable.

57
Table 13: Returns on assets –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Returns on
assets
Returns on 0.597 0.592 0.703 0.566 0.419 0.417 0.449 0.469
assets (t-1) (3.10)*** (2.84)*** (2.71)*** (2.55)** (2.83)*** (3.03)*** (2.16)** (2.15)**
Output gap (t) -0.007 -0.008 -0.024 0.0004 -0.019 -0.026 -0.021 -0.019
(-0.27) (-0.30) (-0.90) (0.01) (-0.55) (-0.98) (-0.47) (-0.70)
Inflation rate (t- -0.007 -0.008 0.005 -0.008 0.030 0.029 0.022 0.019
1) (-0.29) (-0.33) (0.16) (-0.39) (1.13) (1.21) (0.73) (0.76)
Change in real 0.006 0.004 0.004 0.005 0.0002 -0.0004 -0.0002 0.001
effective (1.26) (0.74) (0.87) (0.91) (0.03) (-0.07) (-0.02) (0.22)
exchange rate (t-
1)
Change in real -0.006 -0.003 -0.005 -0.027 -0.026 -0.024
short term (-0.39) (-0.17) (-0.26) (-1.91)* (-1.38) (-1.69)*
interest rate (t-1)
Change in real -0.014 -0.029
long term (-0.86) (-2.67)**
interest rate (t-1)
Debt ratio (t-1) 0.001 -0.001
(0.33) (-0.54)
Change in debt 0.004 0.002
ratio (t-1) (0.23) (0.16)
Credit growth -0.004 -0.004 -0.003 -0.004 -0.005 -0.005 -0.004 -0.005
(t) (-1.80)* (-1.58) (-1.00) (-1.27) (-2.11)** (-2.42)** (-1.84)* (-
2.26)**
Financial -0.001 -0.001 0.00007 -0.001 -0.002 -0.002 -0.002 -0.002
intermediation (-0.64) (-0.55) (0.02) (-0.50) (-1.30) (-1.43) (-1.08) (-1.45)
(t-1)
Severe financial -0.360 -0.339 -0.428 -0.395 -0.419 -0.432 -0.452 -0.405
crisis (short (- (- (- (- (- (- (- (-
run) 4.84)*** 4.01)*** 4.59)*** 3.20)*** 3.94)*** 4.97)*** 3.32)*** 4.06)***
Severe financial -0.895 -0.833 -1.442 -0.906 -0.722 -0.742 -0.822 -0.762
crisis (long run) (-1.87)* (-1.87)* (-1.03) (- (- (- (- (-
2.11)** 3.49)*** 4.09)*** 2.62)*** 2.43)**
Capital/assets (t) 0.128 0.147 0.049 0.084 -0.007 -0.008 -0.035 -0.003
(2.22)** (1.96)* (0.69) (1.20) (-0.07) (-0.08) (-0.27) (-0.04)
Non performing -0.029 -0.035 -0.019 -0.011 -0.056 -0.063 -0.044 -0.059
loans to total (-0.39) (-0.42) (-0.24) (-0.10) (-1.16) (-1.42) (-1.04) (-1.12)
loans (t)
Constant -0.198 -0.279 -0.139 0.026 0.881 0.890 1.112 0.855
(-0.30) (-0.33) (-0.21) (0.04) (1.06) (1.14) (0.99) (1.12)
Obs. 224 224 224 223 224 224 224 223
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(9) : chi2(9) : chi2(10) : chi2(10) chi2(10) : chi2(10) : chi2(11) chi2(11)
440.49 256.74 397.79 : 257.14 495.43 : :
(0.000) (0.000) (0.000) 189.80 (0.000) (0.000) 370.23 226.72
(0.000) (0.000) (0.000)

58
Table 13 (Continued)
Residual’s 2nd order 0.221 0.323 0.187 0.182 0.227 0.282 0.294 0.214
AR (p-values)
Hansen test of 0.577 0.355 0.505 0.520 0.555 0.787 0.654 0.518
overidentifying
restrictions (p-values)
No of instruments 21 21 22 22 21 21 22 22
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-2 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable and the lagged values of capital-to-assets and non-performing loans to total loans ratios.

Table 14: Returns on assets –two step system GMM estimations

1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Returns on
assets
Returns on 0.444 0.501 0.460 0.445 0.485 0.501 0.524 0.474
assets (t-1) (4.66)*** (2.77)*** (3.09)*** (3.17)*** (3.99)*** (5.33)*** (3.21)*** (2.78)***
Output gap (t) 0.003 0.008 0.009 0.009 -0.0003 0.0006 -0.011 -0.011
(0.14) (0.27) (0.40) (0.44) (-0.01) (0.02) (-0.36) (-0.38)
Inflation rate 0.034 0.017 0.029 0.033 0.023 0.022 0.018 0.027
(t-1) (1.15) (0.67) (0.89) (1.07) (0.57) (0.66) (0.49) (0.70)
Change in real 0.002 0.0005 0.005 0.002 0.007 0.007 0.007 0.004
effective (0.33) (0.08) (0.79) (0.48) (1.15) (1.46) (1.22) (0.83)
exchange rate
(t-1)
Change in real -0.008 -0.006 -0.008 -0.010 -0.011 -0.017
short term (-0.40) (-0.33) (-0.40) (-0.43) (-0.52) (-0.78)
interest rate
(t-1)
Change in real 0.006 -0.005
long term (0.26) (-0.41)
interest rate
(t-1)
Debt ratio (t- -0.0008 0.0009
1) (-0.29) (0.23)
Change in 0.003 -0.004
debt ratio (t-1) (0.22) (-0.21)
Credit growth -0.002 -0.002 -0.001 -0.002 -0.0005 -0.001 -0.0009 -0.0009
(t) (-0.71) (-0.64) (-0.80) (-0.71) (-0.18) (-0.42) (-0.43) (-0.31)
Financial -0.001 -0.001 -0.0007 -0.001 0.0005 0.0002 0.0003 -0.0002
intermediation (-0.56) (-0.45) (-0.75) (-0.62) (0.25) (0.11) (0.11) (-0.08)
(t-1)

59
Table 14 (Continued)
Weak financial -0.266 -0.275 -0.260 -0.258 -0.248 -0.247 -0.238 -0.242
crisis (short run) (- (- (- (- (- (- (- (-
6.44)*** 4.91)*** 4.91)*** 4.94)*** 3.69)*** 5.00)*** 3.30)*** 2.65)***
Weak financial -0.478 -0.551 -0.482 -0.466 -0.481 -0.494 -0.501 -0.459
crisis (long run) (-4.28) (- (- (- (- (- (-1.83)* (-1.57)
*** 2.01)** 2.81)*** 2.77)*** 2.24)** 3.11)***
Capital/assets (t) 0.080 0.077 0.139 0.090 0.286 0.266 0.289 0.224
(0.57) (0.67) (1.12) (0.82) (1.76)* (1.75)* (2.23)** (1.53)
Non performing -0.001 0.016 0.002 0.003 -0.002 -0.003 -0.023 -0.019
loans to total loans (-0.01) (0.17) (0.04) (0.04) (-0.05) (-0.08) (-0.46) (-0.27)
(t)
Constant 0.061 0.020 -0.289 -0.0009 -1.301 -1.158 -1.304 -0.814
(0.05) (0.02) (-0.32) (-0.00) (-1.09) (-1.03) (-1.16) (-0.67)
Obs. 224 224 224 223 224 224 224 223
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(9) chi2(9) chi2(10) chi2(10): chi2(10) chi2(10) chi2(11) chi2(11)
: : : 295.26 : : : :
212.15 187.81 435.91 (0.000) 298.72 173.99 622.16 185.06
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Residual’s 2nd order 0.307 0.304 0.378 0.305 0.896 0.803 0.900 0.743
AR (p-values)
Hansen test of 0.725 0.753 0.622 0.724 0.743 0.728 0.791 0.713
overidentifying
restrictions (p-
values)
No of instruments 21 21 22 22 21 21 22 22
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-2 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable and the lagged values of capital-to-assets and non-performing loans to total loans ratios.

60
Table 15: Returns on equity –two step system GMM estimations
1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Returns on
equity
Returns on 0.504 0.535 0.497 0.638 0.418 0.453 0.507 0.436
equity (t-1) (2.76)*** (3.29)*** (2.13)** (3.30)*** (2.83)*** (3.38)*** (3.31)*** (2.64)**
Output gap (t) 0.027 -0.332 0.021 0.326 -0.128 -0.296 -0.199 -0.479
(0.05) (-0.65) (0.02) (0.75) (-0.25) (-0.80) (-0.40) (-1.17)
Inflation rate 0.153 0.151 0.312 0.369 0.569 0.234 0.777 0.281
(t-1) (0.50) (0.50) (0.44) (0.88) (0.86) (0.37) (1.08) (0.56)
Change in real 0.122 0.115 0.127 0.149 0.161 0.193 0.245 0.096
effective (1.44) (1.50) (0.95) (1.90)* (2.13)** (3.29)*** (2.56)** (0.79)
exchange rate
(t-1)
Change in real -0.319 -0.356 -0.468 -0.672 -0.757 -0.825
short term (-0.91) (-0.95) (-2.00)** (- (- (-
interest rate (t- 2.73)*** 2.85)*** 3.32)***
1)
Change in real -0.377 -0.747
long term (-1.49) (-
interest rate (t- 3.60)***
1)
Debt ratio (t-1) -0.008 0.011
(-0.09) (0.24)
Change in debt 0.231 -0.109
ratio (t-1) (0.96) (-0.37)
Credit growth -0.078 -0.086 -0.080 -0.055 -0.074 -0.089 -0.066 -0.087
(t) (-1.60) (-1.60) (-1.60) (-0.98) (-1.30) (-1.87)* (-1.27) (-2.14)**
Financial -0.021 -0.021 -0.017 -0.025 -0.026 -0.029 -0.028 -0.047
intermediation (-0.44) (-0.67) (-0.23) (-0.78) (-0.83) (-1.09) (-0.84) (-1.55)
(t-1)
Severe -6.892 -7.025 -6.947 -2.657 -6.871 -5.915 -4.849 -6.178
financial crisis (- (- (- (-1.06) (- (- (-2.49)** (-
(short run) 3.77)*** 5.30)*** 3.18)*** 3.58)*** 3.93)*** 2.88)***
Severe -13.908 -15.107 -13.809 -7.336 -11.809 -10.803 -9.834 -
financial crisis (-2.59)** (- (- (-1.39) (- (- (- 10.95611
(long run) 2.94)*** 2.52)** 3.35)*** 3.39)*** 2.70)*** (-
3.15)***
Capital/assets 1.303 1.753 1.592 -0.213 0.924 2.359 1.767 0.054
(t) (0.54) (0.98) (0.55) (-0.15) (0.36) (1.14) (0.71) (0.02)
Non -0.206 -0.464 -0.432 0.152 -0.816 -1.269 -0.894 -1.257
performing (-0.15) (-0.39) (-0.25) (0.10) (-1.02) (-1.63) (-1.11) (-1.19)
loans to total
loans (t)
Constant 2.163 -0.117 0.816 7.648 6.933 0.245 -0.387 15.797
(0.11) (-0.01) (0.03) (0.92) (0.42) (0.02) (-0.02) (0.80)
Obs. 224 224 224 223 224 224 224 223

61
Table 15: (Continued)
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(9) chi2(9) chi2(10) chi2(10) chi2(10) chi2(10) chi2(11): chi2(11)
: : : : : : 194.55 :
213.28 212.99 227.51 542.73 267.35 247.14 (0.000) 282.07
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Residual’s 2nd order 0.131 0.187 0.156 0.097 0.153 0.269 0.157 0.187
AR (p-values)
Hansen test of 0.339 0.448 0.357 0.726 0.312 0.469 0.623 0.395
overidentifying
restrictions (p-values)
No of instruments 21 21 22 22 21 21 22 22
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-2 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable and the lagged values of capital-to-assets and non-performing loans to total loans ratios.

Table 16: Returns on equity –two step system GMM estimations

1 2 3 4 5 6 7 8
Dependent Transformation: first differencing Transformation: forward orthogonal
variable deviations
:Returns on
equity
Returns on 0.500 0.535 0.525 0.519 0.470 0.471 0.513 0.380
equity (t-1) (4.75)*** (4.99)*** (4.84)*** (4.49)*** (5.01)*** (4.73)*** (4.27)*** (3.13)***
Output gap (t) -0.026 0.021 -0.113 0.145 0.081 0.149 -0.114 0.129
(-0.07) (0.06) (-0.33) (0.38) (0.15) (0.31) (-0.18) (0.38)
Inflation rate 0.618 0.529 0.617 0.576 0.655 0.368 0.706 0.413
(t-1) (1.47) (1.35) (1.15) (1.22) (0.78) (0.73) (0.66) (0.67)
Change in real 0.078 0.068 0.093 0.081 0.191 0.249 0.236 0.235
effective (0.89) (0.90) (1.11) (1.00) (0.91) (2.72)*** (1.81)* (3.26)***
exchange rate
(t-1)
Change in real -0.339 -0.380 -0.321 -0.344 -0.457 0.056
short term (-1.08) (-1.87)* (-1.22) (-0.83) (-1.28) (0.11)
interest rate
(t-1)
Change in real -0.247 -0.169
long term (-0.91) (-0.59)
interest rate
(t-1)
Debt ratio (t- 0.002 0.028
1) (0.05) (0.55)
Change in 0.205 -0.032
debt ratio (t-1) (1.11) (-0.20)
Credit growth -0.049 -0.056 -0.053 -0.057 -0.011 -0.026 -0.006 0.009
(t) (-0.98) (-0.83) (-1.24) (-0.96) (-0.26) (-0.68) (-0.15) (0.16)

62
Table 16: (Continued)
Financial -0.040 -0.038 -0.037 -0.042 -0.019 -0.017 -0.004 -0.017
intermediation (t-1) (-1.65) (-1.37) (- (- (-0.53) (-0.63) (-0.08) (-0.74)
2.08)** 2.24)**
Weak financial -4.909 -4.948 -4.777 -4.672 -5.024 -4.806 -4.695 -7.441
crisis (short run) (- (- (- (- (- (- (- (-
5.85)*** 5.68)*** 4.21)*** 4.05)*** 5.43)*** 9.14)*** 4.13)*** 2.72)***
Weak financial -9.822 -10.652 -10.048 -9.711 -9.485 -9.078 -9.636 -12.009
crisis (long run) (- (- (- (- (- (- (- (-
4.14)*** 4.74)*** 3.48)*** 3.30)*** 3.55)*** 4.48)*** 2.55)** 3.64)***
Capital/assets (t) -0.157 0.346 0.448 -0.106 2.267 3.227 3.638 3.059
(-0.10) (0.30) (0.27) (-0.08) (0.53) (1.21) (0.84) (2.48)**
Non performing -0.229 -0.036 -0.645 -0.242 -0.080 -0.139 -0.502 -0.060
loans to total loans (-0.19) (-0.03) (-0.60) (-0.20) (-0.13) (-0.22) (-0.76) (-0.11)
(t)
Constant 12.849 8.871 9.424 12.699 -4.096 -9.535 -15.268 -6.917
(0.91) (0.85) (0.64) (1.14) (-0.14) (-0.52) (-0.54) (-0.72)
Obs. 224 224 224 223 224 224 224 223
Wald –Chi2 (df) Wald Wald Wald Wald Wald Wald Wald Wald
(p-value) chi2(9) chi2(9) chi2(10) chi2(10) chi2(10) chi2(10) chi2(11) chi2(11)
: : : : : : : :
542.84 1117.82 752.04 391.29 392.45 386.06 319.59 683.76
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Residual’s 2nd order 0.078 0.111 0.100 0.072 0.382 0.463 0.593 0.441
AR (p-values)
Hansen test of 0.645 0.615 0.657 0.708 0.643 0.763 0.683 0.889
overidentifying
restrictions (p-
values)
No of instruments 21 21 22 22 21 21 22 22
Notes: z-statistics in parenthesis, standard errors are robust in the presence of any pattern of heteroskedasticity and autocorrelation
within panels; the Windmeijer finite-sample correction has been used. ***,**,* denote significance at 1%, 5%, and 10%, respectively.
Estimators used: Two step system GMM; see Arelano and Bond 1991, Arellano and Bover, 1995; Blundell and Bond (1998) and
Roodman (2009a, 2009b). To remove fixed effects we use both first differencing transformation (columns 1-4) and forward
orthogonal deviations transformation (columns 5-8); forward orthogonal deviations perform better in unbalanced panels. In system
GMM with orthogonal deviations, the levels, or untransformed, equation is still instrumented with differences. A collapsed subset of
the available instrument matrix was used: namely the t-1 to t-2 lags of the lagged output gap, the lagged credit growth, and the lagged
dependent variable and the lagged values of capital-to-assets and non-performing loans to total loans ratios.

63
64
 BANK OF GREECE WORKING PAPERS
127. Gazopoulou, E. “Assessing the Impact of Terrorism on Travel Activity in
Greece”, April 2011.
128. Athanasoglou, P. “The Role of Product Variety and Quality and of Domestic
Supply in Foreign Trade”, April 2011.
129. Galuščắk, K., M. Keeney, D. Nicolitsas, F. Smets, P. Strzelecki, and Matija
Vodopivec, “The Determination of Wages of Newly Hired Employees: Survey
Evidence on Internal Versus External Factors”, April 2011.
130. Kazanas, T., and E. Tzavalis, “Unveiling the Monetary Policy Rule In Euro-
Area”, May 2011.
131. Milionis, A. E., and D. Patsouri, “A Conditional CAPM; Implications for the
Estimation of Systematic Risk”, May 2011
132. Christodoulakis, N., and V. Sarantides, “External Asymmetries in the Euro Area
and The Role Of Foreign Direct Investment”, June 2011.
133. Tagkalakis, A., “Asset Price Volatility and Government Revenue”, June 2011.
134. Milionis, E. A., and E. Papanagiotou, “Decomposing the Predictive Performance
of The Moving Average Trading Rule of Technical Analysis: The Contribution of
Linear and Non Linear Dependencies in Stock Returns”, July 2011.
135. Lothian, J. R., and J. Devereux, “Exchange Rates and Prices in the Netherlands
and Britain over the Past Four Centuries, July 2011.
136. Kelejian, J., G. S. Tavlas, and P. Petroulas, “In the Neighbourhood: the Trade
Effects of the Euro in a Spatial Framework”, August 2011.
137. Athanasoglou, P.P., “Bank Capital and Risk in the South Eastern European
Region”, August 2011.
138. Balfoussia, H., S. N. Brissimis, and M. D. Delis, “The Theoretical Framework of
Monetary Policy Revisited”, September 2011.
139. Athanasoglou, P. P., and I. Daniilidis, “Procyclicality in the Banking Industry:
Causes, Consequences and Response”, October 2011.
140. Lazaretou, S., “Financial Crises and Financial Market Regulation: The Long
Record of an ‘Emerger’, October 2011.
141. Papapetrou, E, and S. E. G. Lolos, “Housing credit and female labour supply:
assessing the evidence from Greece”, November 2011.
142. Angelopoulos, K., J. Malley, and A. Philippopoulos, “Time-consistent fiscal
policy under heterogeneity: conflicting or common interests?”, December 2011.
143. Georgoutsos, D. A., and P. M. Migiakis, “Heterogeneity of the determinants of
euro-area sovereign bond spreads; what does it tell us about financial stability?”,
May 2012.

65
144. Gazopoulou, E. “A note on the effectiveness of price policy on tourist arrivals to
Greece”, May 2012.
145. Tagkalakis, A. “The Effects of Financial Crisis on Fiscal Positions”, June 2012.
146. Bakas, D., and E. Papapetrou, “Unemployment in Greece: Evidence from Greek
Regions”, June 2012.
147. Angelopoulou, E, H. Balfoussia and H. Gibson, “Building a Financial Conditions
Index for the Euro Area and Selected Euro Area Countries: What Does it Tell Us
About The Crisis?”, July 2012.
148. Brissimis, S, E. Garganas and S. Hall, “Consumer Credit in an Era of Financial
Liberalisation: an Overreaction to Repressed Demand?”, October 2012
149. Dellas, H., and G. Tavlas, “The Road to Ithaca: the Gold Standard, the Euro and
the Origins of the Greek Sovereign Debt Crisis”, November 2012.
150. Philippopoulos, A., P. Varthalitis, and V. Vassilatos, “On The Optimal Mix of
Fiscal and Monetary Policy Actions”, December 2012.
151. Brissimis, N. S. and P. M. Migiakis, “Inflation Persistence and the Rationality of
Inflation Expectations”, January 2013.
152. Tagkalakis, O. A., “Audits and Tax Offenders: Recent Evidence from Greece”,
February 2013.
153. Bageri, V., Y. Katsoulacos, and G.Spagnolo, “The Distortive Effects of Antitrust
Fines Based on Revenue”, February 2013.
154. Louzis, P. D., “Measuring Return and Volatility Spillovers in Euro Area Financial
Markets”, March 2013
155. Louzis, P. D., and A.T. Vouldis,“A Financial Systemic Stress Index for Greece”,
March 2013.
156. Nikolitsas, D., “Price Setting Practices in Greece: Evidence From a Small-Scale
Firm-Level Survey”, April 2013
157. Bragoudakis, G. Z., S.T. Panagiotou and H. A. Thanopoulou, “Investment
Strategy and Greek Shipping Earnings: Exploring The Pre & Post "Ordering-
Frenzy" Period”, April 2013.

66