The Internet and International Trade in Services

Author(s): Caroline Freund and Diana Weinhold

Source: The American Economic Review , May, 2002, Vol. 92, No. 2, Papers and
Proceedings of the One Hundred Fourteenth Annual Meeting of the American Economic
Association (May, 2002), pp. 236-240
Published by: American Economic Association

Stable URL: https://www.jstor.org/stable/3083409

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms
 The Internet and International Trade in Services

By CAROLINE FREUND AND DIANA WEINHOLD*

The scope for growth of trade in services is communication, and non-standardization con-
vast. Although services currently make up over tribute to quality, the Internet would not be
60 percent of world production, they account expected to have a large impact on international
for only about 20 percent of world trade. A trade.
primary reason why international trade in ser- To determine whether the Internet has signif-
vices has been limited is that the performance of icantly affected international service provision
many services necessitates physical contact be- in practice, we estimate a general model of
tween producers and consumers, a condition services trade across countries and examine
that renders service provision to distant loca- whether the inclusion of data on Internet pene-
tions infeasible. New technology, in particular, tration, as measured by the number of Internet
the Internet, provides a medium of exchange hosts in a country, is statistically significant.
that overcomes such historical trading hurdles Overall, our results offer evidence that the In-
for many services, effectively reducing trans- ternet is related to growth in services trade.
port costs from infinity to virtually nothing. After controlling for GDP and exchange-rate
There is ample anecdotal evidence that the movements, we find that a 10-percent increase
Internet is having just this sort of an effect on in Internet penetration in a foreign country is
services trade. The accounting firm Netlink associated with about a 1.7-percentage-point in-
maintains the books for 6,000 employees in crease in export growth and a 1.1-percentage-
Reyanosa, Mexico, from their offices in Man- point increase in import growth. The results are
hattan. Infosys of India provides software- robust to a number of alternative specifications.
consulting services to international clients,
including Apple Computers, Lucent Technolo- I. Data on Services Trade and the Internet
gies, and Microsoft. A medical-transcription
company in South Africa, ITS, receives digital We focus our discussion on U.S. trade data,
recordings from abroad electronically and re- as there are no comparable bilateral services
turns a transcribed text file the next day. trade data across countries. International trade
Still, the question remains as to whether elec- in services is strongly dominated by the United
tronic sharing of information is an important States, which traded over $500 billion worth of
enough development to alter significantly the services in 2000. We use detailed data on U.S.
geography of service provision. Indeed, many "Other Private Services" from 31 countries ' and
services need to be tailored to the consumer's 14 industries2 that appear in the U.S. Interna-
needs and monitored for quality, and these are tional Transactions Accounts compiled by the
likely to be more effective if the provider is
close by and speaks the same language. In ad-
dition, in the event of a dispute, resolution will
1 Belgium-Luxembourg, South Africa, France. Israel.
be less complicated if both parties are subject to Germany, Australia, Italy, China, Netherlands, Hong Kong,
the same legal system. Finally, there may be Norway, India, Spain, Indonesia, Sweden, Japan, Switzer-
security concerns with allowing foreign access land, Korea, United Kingdom, Malaysia, Argentina, New
Zealand, Brazil, Philippines, Chile, Singapore, Mexico, Tai-
to some documents or systems. Thus, for some
wan, Venezuela, Thailand, and Bermuda.
services, especially those where familiarity,
2 Education; financial services; telecommunicatioins; ad-
vertising; computer and data processing services; database
and other information services; research, development, and
* Federal Reserve Board, Washington, DC 20551, and testing services; management, consulting, and public rela-
Development Studies Institute, London School of Econom- tions; legal services; construction, engineering, architecture,
ics, Houghton Street, London WC2A 2AE, respectively. and mining; industrial engineering; installation, mainte-
We thank Beth Allen, Simeon Djankov, Hali Edison, Jane nance, and repair of equipment; other business, profes-
Ihrig, and Andrew Levin for comments. sional, and technical services; and other services.

236

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms
VOL. 92 NO. 2 THE ECONOMICS OF TECHNOLOGY AND INNOVATION 237

16 16 61
142 14 -
C i 12

01990-1994 El1 990-1994

8
2 B 1995-1999
6 c 6
2~~~~~ ~~~~~~ 199|- Sl519990 o 4
E 2 w2

0*
uj
0
2 -|
Internet-Related Total Imports of Internet-Related Total Exports of
Services Goods and Services Services Goods and Servces

FIGURE 1. ANNUAL U.S. IMPORT GROWTH FIGURE 2. ANNuAL U.S. EXPORT GROWTH

Bureau of Economic Analysis (U.S. Depart- Internet-related services has been very rapid as
ment of Commerce, 2000). "Other Private Ser- compared with the previous four-year period and
vices" are primarily business services, and in has far outpaced growth of overall goods and
1999 they made up 27 percent of total service services. Moreover, one of the fastest-growing
imports and 38 percent of total service exports categories of services over this period was
in the United States. The remaining share of "computer and data processing." U.S. exports of
total services trade is composed of payments for computing services grew 71 percent between 1995
transportation, royalties, and license fees. and 1999, while U.S. imports grew 243 percent.
The disaggregated data on business services The United States is also by far the world leader
are available only for trade among unaffiliated in Internet usage, so if there is an effect of the
parties (i.e., the data exclude intra-firm trade, Internet on trade it is likely to show up in U.S.
such as trade between IBM and one of its sub- data. To measure Internet penetration we use data
sidiaries abroad). Although intra-firm trade is on the number of top-level domain names attrib-
important (accounting for about 35 percent of uted to each country from the Internet Software
U.S. trade in business services from 1995 to Consortium (ISC). A top-level domain name is
1999), it is determined by different forces and either an ISO country code or one of the generic
should therefore be studied separately. In par- domains (com/org/net/etc).4 In our sample, the
ticular, affiliated-party trade is by definition de- number of host sites in a country increased rapidly
pendent on direct investment and is therefore from, on average, 49,000 in 1995 to 378,000 in
greatest in industries and countries where direct 1999. There is also a lot of variation in the data:
investment is large. We hypothesize that this the standard deviation of the host variable was
omission will bias our analysis against finding a 68,000 in 1995 and 517,000 in i999.5
strong role for the Internet, as many of the
difficulties of foreign service provision over
the web, such as monitoring, communication,
and security, can be overcome in a shared- legal services; research, development, and testing; computer

ownership environment. and data processing; financial services; and management,

consulting, and public relations.
Examining the growth of services trade by
4 A site's domain name does not necessarily correspond to
industry provides some evidence that the Internet where it is located. However, in Freund and Weinhold (2000)
has stimulated trade in recent years. Figures 1 we argue that, even if a Venezuelan domain name, for exam-
and 2 show U.S. nominal trade growth in total ple, is located in another country, it is likely that the content of
the site is aimed at Venezuelans. In all of the specification
goods and services (excluding oil and agricul-
reported in the paper, we do not attribute hosts under the
ture) and Internet-related services, which consistdomains ".org," ".edu," ".net," ".com," or ".int" to any partic-
of services that can be transferred electroni- ular country.
cally.3 From 1995 to 1999, expansion of trade in In the regressions reported, the Internet variable is not
scaled. While this may overstate Internet intensity in large
countries, scaling the variable by population is likely to under-
state the importance of the Internet in large countries like India
3Internet-related services include accounting, auditing, and China. In robustness checks, the results were qualitatively
and bookkeeping; database and other information services; similar if we used the variable scaled by population.

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms
 238 AEA PAPERS AND PROCEEDINGS MAY 2002

II. Empirical Evidence TABLE 1-GROWTH OF U.S. IMPORTS AND U.S. EXPORTS,
VARIOUS SERVICES, 1995-1999

Previous empirical work finds that move-

A. U.S. Import Growth, 1995-1999
ments in economic activity and the real ex-
change rate are the most important determinants Services

of services trade flows (William Helkie and All BPTa All

Lois Stekler, 1987; Juann Hung and Sandra Independent variable (i) (ii) (iii)
Viana, 1995; Alan Deardorff et al., 2000).
Log Internet variable 0.17 0.23 0.17
Using these insights, we construct two models (4.06) (3.76) (224)
of services trade. In the first model, the depen- GDP growth 1.71 1.97 1.58
dent variable is the growth of U.S. trade in 14 (3.89) (4.00) (3.22)
Real-exchange appreciation 0.76 1.04 0.59
service industries from 1995 to 1999. Services
(1.85) (2.52) (1.30)
trade growth is regressed on industry fixed ef- Log trade in 1995 -0.30 -0.39 -0.32
fects, the Internet variable, GDP growth, real- (-6.70) (-5.81) (-6.63)
exchange-rate appreciation, and the initial level Log GDP in 1995 0.03
of trade.6 We use the log-level of the Internet (0.28)
Log population in 1995 0.02
variable at the end of the period because it is a
(0.31)
new technology and all of the growth in the Financial depth 0.02
Internet should contribute to trade over this (0.26)
period. We distinguish between exports and im-
No. observations: 361 239 361
ports in the regression equations because for- R 2 : 0.24 0.22 0.24
eign growth and exchange-rate movements are
likely to affect them in different ways. We B. U.S. Export Growth, 1995-1999

expect that GDP growth abroad will positively Services

impact U.S. export growth because of increased
All BPTa All
demand for services in these countries. U.S.
Independent variable (iv) (v) (vi)
import growth is also likely to be positively
impacted if GDP growth abroad is associated Log Internet variable 0.11 0.14 0.06
(3.79) (3.03) (1.15)
with a greater supply of these service industries.
GDP growth 0.84 1.04 0.62
A real appreciation of the dollar would be ex- (2.93) (2.65) (1.94)
pected to increase U.S. imports in the long run, Real-exchange appreciation 0.42 0.72 0.18
but it could lower the nominal value of imports (1.50) (1.85) (0.64)
Log trade in 1995 -0.25 -0.30 -0.30
in the short run, as the price of services de-
(-6.11) (-5.47) (-6.67)
creases for a given quantity demanded (the J- Log GDP in 1995 0.16
curve effect). We include the log of the initial (1.75)
value of trade to allow for more rapid adjust- Log population in 1995 -0.00
ment when levels are low. (-0.09)
Financial depth -0.09
As shown in Table 1, the effect of the
(-1.35)
Internet on services trade growth is positive
No. observations: 394 253 394
and significant. Countries that have greater In-
R 2: 0.25 0.26 0.28
ternet penetration also have had higher growth
in services trade in recent years. Specifically, Notes: Industry fixed effects were included in all regres-
for all service industries we find that a 10- sions (not shown). Table entries are regression coefficients,
percent increase in the level of the Internet with heteroskedasticity-consistent t statistics in parentheses.
a BPT = business, professional, and technical services.
variable is associated with a 1.7-percentage-
point increase in the growth of U.S. imports and
a 1. 1-percentage-point increase in the growth of
U.S. exports from 1995 to 1999. If this dynamic
relationship were to continue to hold over the
6 Trade and GDP are from the World Bank's World long run, it would imply that a 10-percent in-
Development Indicators. The real exchange rate is from the crease in the Internet variable abroad is associ-
IMF's International Financial Statistics. ated with about a 6-percent increase in the level

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms
VOL. 92 NO. 2 THE ECONOMICS OF TECHNOLOGY AND INNOVATION 239

of U.S. imports and about a 4-percent increase TABLE 2-GRAVITY EQUATIONS OF THE LEVEL OF TRADE,
VARIOUS SERVICES, 1995-1999
in the level of U.S. exports. We also estimate
the regression equation on a subset of industries
Log U.S. imports Log U.S. exports
called "Business, Professional, and Technical
Services," which exclude telecommunications, All All
services BPT' services BPT'
education, and financial services. The results are
(i) (ii) (iii) (iv)
even stronger for this subcategory.
It is important to note that, while our results Log Internet 0.05 0.12 -0.02 -0.01
vanable (1.60) (3.04) (-0.80) (-0.31)
imply that there is a correlation between Internet
Financial depth 0.34 0.45 0.30 0.41
adoption and international services trade, they say (5.12) (5.70) (5.10) (5.06)
little about causality. Using the twice-lagged In- Log GDP 0.78 0.68 0.73 0.75
temet variable in the regression equation yields (12.38) (8.80) (13.23) (9.88)
Log population -0.09 -0.04 -0.11 -0.11
similar results (not reported), suggesting that
(-2.23) (-0.83) (-3.01) (-2.13)
Internet development promotes services trade. Log distance -0.39 -0.42 -0.08 -0.08
However, causality probably runs both ways: (-5.14) (-4.81) (- 1.10) (-0.85)
increasing trade in services leads firms to adopt Common language 0.49 0.46 0.07 0.02
the Internet to facilitate that trade, and greater (6.93) (5.50) (1.20) (0.24)
Adjacency 0.61 0.40 1.29 1.21
Internet penetration causes firms to use the In-
(3.63) (2.12) (8.88) (6.85)
ternet for trade in services.
Apart from endogeneity, it is also possible No. observations: 1,977 1,289 2,056 1,328
R 2: 0.71 0.63 0.67 0.59
that the Internet is not actually correlated with
services trade, but that the Internet variable is Notes: Industry and year fixed ef
correlated with important variables for trade sions (not shown). Table entries are regression coefficients,
growth that are omitted from the regression with heteroskedasticity-consistent t statistics in parentheses.
a BPT = business, professional, and technical services.
equation. In particular, countries with higher
per capita incomes might have experienced
greater trade growth over this period, and the
Internet variable might be a good proxy for temet variable in the base regression is due to a
relative income; or the Internet variable could correlation between the Internet variable and
be correlated with comparative advantage in income, and that higher-income countries have
service production. To control for country size experienced faster import growth.
or income effects, we include the log of the The second regression that we estimate is a
initial value of GDP and population. We also modified gravity equation of services trade. A
include a measure of financial depth, M2/GDP basic gravity equation depicts trade between
(calculated from the IMF' s International Finan- countries as proportional to the product of their
cial Statistics), to proxy for overall comparative economic masses (GDP's) and inversely related
advantage in services across countries. The in- to the distance between them. Using panel data
tuition is that a country with a deeper financial from 1996 to 1999, we regress log levels of
system is likely to have a labor force that is exports and imports on the logs of income and
skilled in electronic services and has a demand population of the partner country, the log of
for such services. distance to the United States, indicator variables
Columns (iii) and (vi) of Table 1 report the for adjacency and sharing a common language,
results when these variables are included. We and industry and year fixed effects.7 We also
find that GDP, population, and financial depth include the index of financial depth and the log
are not significant in the U.S. import-growth of the Internet variable.
equation, and the coefficient on the Internet The results, reported in Table 2, provide fur-
variable remains unchanged and highly signifi- ther evidence of an effect of the Internet on U.S.
cant. The coefficient on the Internet variable in imports of services. In particular, a 10-percent
the U.S. export equation, however, declines to
about half of its previous value. The log level of
GDP is significant at the 10-percent level, sug- 7 See Freund and Weinhold (2000) for a detailed descrip-
gesting that part of the significance of the In- tion of the additional data.

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms
 240 AEA PAPERS AND PROCEEDINGS MAY 2002

increase in Internet penetration abroad leads to REFERENCES

about a 1.2-percent increase of U.S. imports in
business, professional, and technical services. Deardorff, Alan; Hymans, Saul; Stern, Robert M.
We do not find evidence that Internet penetra- and Xiang, Chong. "Forecasting U.S. Trade in
tion abroad has increased U.S. exports of ser- Services." Mimeo, University of Michigan,
vices. However, one problem with the export 2000.
equation is that we do not control for trade Freund, Caroline and Weinhold, Diana. "On the
policy, and it is likely that there are vast trade Effect of the Internet on International Trade."
policy differences abroad that affect the cross- Board of Governors of the Federal Reserve
sectional distribution of U.S. services exports.8 System International Finance Discussion Pa-
U.S. import data, on the other hand, do not per No. 693, 2000.
suffer from this problem, since all countries Helkie, William and Stekler, Lois. "Modeling In-
face the same U.S. regulations on their exports. vestment Income and Other Services in the
U.S. International Transactions Accounts."
III. Conclusion Board of Governors of the Federal Reserve
System International Finance Discussion Pa-
Our results suggest that Internet development per No. 319, 1987.
abroad has already begun facilitating increased Hoekman, Bernard and Primo Braga, Carlos.
exports of services to the United States. Specif- "Protection and Trade in Services." World
ically, we find that the short-run impact is about Bank Policy Research Working Paper No.
a 1.7-percentage-point boost to growth for a 1747, 1997.
10-percent increase in the Internet variable. Hufbauer, Gary C. and Warren, Tony. "The
The expansion of trade in services has impor- Globalization of Services." Institute for In-
tant implications for growth around the world. ternational Economics (Washington, DC)
The standard gains from trade can be expected: Working Paper, October 1999.
Resources will be reallocated to their most pro- Hung, Juann and Viana, Sandra. "Modeling U.S.
ductive uses and welfare will increase. More- Services Trade Flows: A Cointegration-
over, there are likely to be large multipliers on ECM Approach." Federal Reserve Bank of
these effects, as other studies have found that New York Research Paper No. 9518, 1995.
services are important inputs into nearly all Park, Se Hark and Chan, Kenneth S. "A Cross-
sectors (Se Hark Park and Kenneth S. Chan, Country Input-Output Analysis of Intersec-
1989). This implies that, while the Internet will toral Relationships between Manufacturing
affect growth directly through its impact on and Services." World Development, February
productivity, it should also increase growth in- 1989, 17(2), pp. 199-212.
directly via its effect on openness. U.S. Department of Commerce. "U.S. Interna-
tional Services: Cross Border Trade in 1999
and Sales Through Affiliates in 1998." Sur-
x See Bernard Hoekman and Carlos Primo Braga (1997)
and Gary C. Hufbauer and Tony Warren (1999) for evi- vey of Current Business, October 2000,
dence on protection in services. 80(10), pp. 119-61.

This content downloaded from

178.128.222.215 on Tue, 03 Jun 2025 04:13:55 UTC
All use subject to https://about.jstor.org/terms