Review of International Economics,11(2),296–316,2003
Financial Dependence and International Trade Thorsten Beck*
Abstract
Does ?nancial development translate into a comparative advantage in industries that use more external ?nance? The author uses industry-level data on ?rms’ dependence on external ?nance for 36 industries and 56 countries to examine this question.It is shown that countries with better-developed ?nancial systems have higher export shares and trade balances in industries that use more external ?nance.These results are robust to the use of alternative measures of external dependence and ?nancial development and are not due to reverse causality or simultaneity bias.
1.Introduction
The international trade literature focuses on factor endowments,technology,and scale economies as sources of comparative advantage and therefore determinants of trade ?ows between countries.Theory,however,also suggests that the level of ?nancial devel-opment may importantly in?uence the pattern of international trade ?ows.Kletzer and Bardhan (1987) augment the Heckscher–Ohlin trade model by incorporating a ?nan-cial sector and show that ?nancial sector development gives countries a comparative advantage in industries that rely more on external ?nancing.This paper explores empirically whether the level of ?nancial sector development constitutes a source of comparative advantage and explains the variance of the trade structure across coun-tries.Speci?cally,I assess whether a high level of ?nancial development translates into a comparative advantage in industries that rely more heavily on external ?nance. This is an interesting question for several reasons.First,if we ?nd that the level of ?nancial development does have an effect on the industrial structure of the trade balance,this emphasizes the importance of ?nancial sector development for economic development beyond its positive impact on economic growth and therefore increases the priority that ?nancial sector reforms should have on policymakers’ agendas. Second,if the level of ?nancial development is a determinant of the structure of the trade balance,this has implications for policy reforms in both the ?nancial and the trade sectors.On the one hand,a reform of the ?nancial sector that raises the level of external ?nance available to ?rms in an economy might have an impact on the indus-trial structure of this country’s exports.On the other hand,the effect of trade reforms on the industrial structure of the trade balance might depend on the level of external ?nance available in the economy.
While there are certainly a variety of channels through which the level of ?nancial sector development can translate into a comparative advantage,I concentrate on just *Beck:The World Bank,1818 H Street NW,Washington,DC 20433,USA.Tel:(202) 473-3215;Fax:(202) 522-1155;E-mail:tbeck@https://www.doczj.com/doc/b73636665.html,.Helpful comments from Marianne Baxter,Robert Cull,Bob King, Ross Levine,Colin Xu,two anonymous referees,and seminar participants at the University of Virginia, the World Bank,the University of California at Davis,and the Annual Meeting of the Southern Economic Association are gratefully acknowledged.The usual disclaimer applies.The ?ndings,interpretations and conclusions in this paper are entirely those of the author and do not necessarily represent the view of the World Bank,its Executive Directors,or the countries they represent.
? Blackwell Publishing Ltd 2003,9600 Garsington Road,Oxford OX4 2DQ,UK and 350 Main Street,Malden,MA 02148,USA
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE297 one.Speci?cally I focus on the ?nancial sector’s function to channel funds from savers to ?rms.By economizing on the costs of acquiring and processing information about ?rms and monitoring managers,?nancial institutions and markets can help overcome the problems of moral hazard and adverse selection,thus reducing the cost of exter-nal ?nance for ?rms.Countries with better developed ?nancial institutions and markets should therefore have a comparative advantage in industries that rely relatively more on external ?nance.1
To illustrate the test undertaken in this paper,consider Mexico and Korea,two middle-income developing countries that differ considerably in their level of ?nancial development.Credit to the private sector by ?nancial intermediaries as share of GDP was 12% for Mexico and 66% for Korea in the period 1980–89.Pottery is an industry that uses no external ?nance,whereas plastic products rely heavily on external ?nance. Whereas Mexico’s trade balance in plastic goods was 0.05 percentage points lower than its trade balance in pottery in 1980–89,Korea’s trade balance in plastic goods was 0.29 percentage points higher than its trade balance in pottery.
To test the hypothesis empirically,I follow a technique proposed by Rajan and Zingales (1998).They show that industries that use more external ?nance grow faster in countries with a higher level of ?nancial sector https://www.doczj.com/doc/b73636665.html,ing the methodo-logy developed by Rajan and Zingales,this paper explores whether the level of ?nan-cial sector development has an impact on trade patterns across countries.Thus I do not re-assess whether different levels of ?nancial development differently in?uence the growth of industries with greater or smaller ?nancing needs.Instead,I test the hypothesis that economies with higher levels of ?nancial development have higher export shares and trade balances in industries that use more external ?nance.
Using Rajan and Zingales’ data on external dependence,I ?nd robust evidence for the hypothesis that countries with better-developed ?nancial systems have higher export shares and trade balances in industries that rely more heavily on external ?https://www.doczj.com/doc/b73636665.html,ing different measures of ?nancial development across countries and the reliance on external ?nance across industries,the results are consistent in indicating a large positive relation of the interaction of external dependence and ?nancial devel-opment with export shares and trade balances across industries and countries.These results are robust to the use of instrumental variables,thus controlling for simultane-ity bias and possible reverse causality.
As an alternative test of the hypothesis,I also consider a simpli?ed variation of the factor-content studies of trade ?ows by Bowen et al.(1987) and Tre?er (1993,1995). Speci?cally,I calculate the weighted average external dependence of a country’s exports and trade balance and explore correlations between these measures and indi-cators of ?nancial sector development.
Section 2 discusses the theoretical underpinnings of the link between ?nancial devel-opment and the industrial structure of the trade balance.Section 3 presents the econo-metric methodology.Section 4 describes the data.Section 5 presents the main results. Section 6 explores the robustness of the ?ndings and alternative explanations.Section 7 concludes.
2.External Dependence,Comparative Advantage,and International Trade Classical models of international trade explain the comparative advantage of countries with differences in technology or endowments.2 Both the Ricardian and the Heckscher–Ohlin model,however,can be easily augmented to show the effect of ?nancial sector development on international trade ?ows.
? Blackwell Publishing Ltd 2003
298Thorsten Beck
Baldwin (1989) developed one of the ?rst models,in which ?nancial markets are a source of comparative advantage.In his two-country,two-sector,and one-factor model, the demand for one of the goods is subject to demand shocks,while the other is not. He shows that in economies with better-developed ?nancial markets and therefore better possibilities to diversify risk stemming from the demand shocks,?rms produc-ing the risky good face lower risk premiums and therefore lower marginal costs.Coun-tries with better developed ?nancial markets and therefore better diversi?cation possibilities thus specialize in the risky good.
While Baldwin stresses the risk-diversi?cation function of ?nancial markets,Kletzer and Bardhan (1987) focus on the role of ?nancial institutions and markets in chan-neling external ?nance to industries that are in need of it.They present two interna-tional trade models in the Heckscher–Ohlin tradition with two countries,two sectors, and two factors.While both sectors depend on land and labor,one sector also depends on external ?nance for working capital.They show that the country with a lower level of credit market restrictions specializes in the sector that uses external ?nance.The country with the higher level of credit market restrictions faces either a higher price of external ?nance or credit rationing and will therefore specialize in the sector that does not require working capital or external ?nance.
We can derive the link between ?nancial development and international trade also in the context of a simple back-of-the-envelope extension of the Ricardian trade model.Assume an economy with two sectors,food and manufacturing.While ?rms in the food sector produce with an inherited technology,manufacturing ?rms need working capital to purchase the technology every period before the production process.Firms have available different technologies whose quality increases in their price.The available external funds for the working capital thus determine the quality of technology and therefore the labor productivity in manufacturing.If we assume that country 1 and country 2 differ only in the amount of external ?nancing available to the manufacturing sector,this difference will drive comparative advantage and there-fore trade ?ows.If country 1 has more external ?nance available than country 2,it will have a comparative advantage in manufacturing and therefore export manufactured goods.Country 2,on the other hand,will have a comparative advantage in food and export food.3
Both the Kletzer and Bardhan and the augmented Ricardian model predict that countries with better-developed ?nancial sectors have a comparative advantage in industries with higher external ?nancing needs.4 We should therefore observe that, holding other things constant,countries with better-developed ?nancial sectors have higher exports and trade balances in industries that rely more on external ?nancing. To test this hypothesis,we therefore have to use trade ?ows at the industry level.
The relation between the structure of the trade balance and ?nancial sector development can also be derived from the recent literature on the link between ?nancial development and economic growth.5 Financial intermediaries and markets arise to overcome the problems of moral hazard and adverse selection that drive a wedge between the price of external and internal ?nance.By decreasing the cost of external ?nance,?nancial intermediaries allow a higher return on capital and thus more investment opportunities realized,which in turn enhances economic growth. Industries that rely more heavily on external ?nance should pro?t more than propor-tionally from a higher level of ?nancial development and therefore a lower cost of external ?nance.Rajan and Zingales (1998) ?nd evidence that industries relying more heavily on external ?nance grow faster in countries with a better-developed ?nancial system.
? Blackwell Publishing Ltd 2003
If industries that rely more on external ?nance pro?t relatively more from a higher
level of external ?nance,this should also affect the structure of the trade balance.As
Rajan and Zingales note,countries with a higher level of ?nancial development should
have a comparative advantage in industries relying more on external ?nance and there-
fore higher export shares and higher trade balances in these industries.
Many economists have argued that the development of the ?nancial sector follows
rather than leads the development of the real sector.In terms of our work,this would
mean that the specialization of a country in speci?c industries creates the demand for
a well-developed ?nancial sector.An empirical test of the link between the interaction
of ?nancial development and external dependence and the structure of the trade
balance therefore has to control for this possibility of reverse causality.
3.The Econometric Model
The hypothesis derived from the Kletzer and Bardhan (1987) model and the back-of-
the-envelope extension of the Ricardian model,described in the previous section,is
that economies with a higher level of ?nancial development have higher export shares
and trade balances in industries that rely more on external ?nance.I run the follow-
ing regression to empirically assess the relation of the interaction of ?nancial devel-
opment and reliance on external ?nance with exports and trade balances:
(1)
where y i,k will be the ratio of exports or the trade balance to GDP in industry k and
country i over the period 1980–89,Country j ,j =1,...,56 are country dummy variables,
Industry l ,l =1,...,36 are industry dummy variables,Ext k is the external depend-
ence for industry k as measured for a sample of US ?rms over the period 1980–89,
Finance i is the level of ?nancial development for country i and e i,k is the error term.
All data are averaged over the period 1980–89.6
Industry characteristics are interacted with country characteristics,in this case the
reliance on external ?nance across industries and indicators of ?nancial development
across countries.The dummy variables for countries and industries control for country-
and industry-speci?c effects that might determine the structure of the trade balance.7
We thus isolate the effect that the interaction of external dependence and ?nancial
development has on the exports and trade balances relative to country and industry
averages.The coef?cient of interest in equation (1) is d .If we ?nd a positive coef?cient
on this interaction term,this can be interpreted as evidence in favor of our hypothe-
sis.A negative or insigni?cant coef?cient might signal the irrelevance of ?nancial
development for the structure of the trade balance.A signi?cantly positive d pro-
vides evidence in favor of the Kletzer and Bardhan model that countries with better-
developed ?nancial systems have a comparative advantage in industries that use more
external ?nance.
I use the legal origin of countries as instrumental variables for the level of ?nancial
sector development to control for simultaneity bias and reverse causality.8 Previous
research has shown that the legal origin of a country materially in?uences its legal
treatment of creditors and shareholders,its accounting standards,and the ef?ciency of
contract enforcement,and thus the ef?ciency of ?nancial intermediaries and markets
(La Porta et al.,1997,1998;Levine,1998).
As an alternative test of the Kletzer and Bardhan model,I also consider a sim-
pli?ed variation of the empirical factor-content studies of international trade.The
y Country Industry Ext Finance i k j j j l l l k i i k ,,*,
=++()+??a b d e FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE 299
? Blackwell Publishing Ltd 2003
300Thorsten Beck
Heckscher–Ohlin–Vanek theorem states that countries export the commodity,which uses intensively its relatively abundant resource.Similarly,the results of the Kletzer and Bardhan model can be stated as follows:countries with relatively high levels of ?nancial development export the good whose production depends intensively on external ?nance.Bowen et al.(1987) and Tre?er (1993,1995) test the Heckscher–Ohlin–Vanek theorem using data on factor endowments,trade ?ows on the industry level,and the US input–output table.In this simpli?ed version,I use the level of ?nancial development,trade ?ows on the industry level,and dependence on external ?nance to test the Kletzer and Bardhan model.Speci?cally,I calculate the weighted-average external dependence of a country’s exports and trade balance and explore correlations between these measures and indicators of ?nancial sector devel-opment.A positive correlation can be considered evidence in favor of the Kletzer and Bardhan model,since it would indicate that countries with higher levels of ?nancial development are net exporters of goods that are produced by industries that rely more on external ?nance.Thus,I do not reassess the Heckscher–Ohlin–Vanek theorem. Rather,I use the logic behind this theorem for this alternative test of the Kletzer and Bardhan model.Unlike the regression analysis described above,this test does not control for other industry-speci?c and country-speci?c determinants of exports and the trade balance.
4.The Data
The present sample contains 56 countries and 36 industries.This section describes the measure of external dependence,the indicators of ?nancial development and the trade data.Table 1 presents descriptive statistics and correlations.
External Dependence
The industry-level data on external dependence are from a study by Rajan and Zingales (1998).The underlying assumption for their and for our work is that for technological reasons some industries rely more on external ?nance than others. Scale economies,gestation period,or intermediate product intensity might constitute some of these technological reasons.However,in general equilibrium the amount of external ?nance in an economy is the result of both production and funding deci-sions of producers and consumption–savings decisions of consumers,so both demand and supply.In countries with perfect and atomistic capital markets,however,Rajan and Zingales note that individual large ?rms that constitute a fraction of the overall market and therefore have no market power might face a perfectly elastic supply curve, so that the actual use of external ?nance by these ?rms would primarily re?ect the demand for it.9 Assuming that the variance of the reliance on external ?nance across industries persists across countries,we can thus use the actual external dependence of industries as observed for large ?rms in a country with a relatively well-developed ?nancial system as proxying for the “natural”dependence of industries on external ?nance in other countries.Rajan and Zingales use a sample of publicly listed ?rms in the US to compute the natural external dependence of industries,and then con?rm the results using a sample of Canadian ?rms to compute the external dependence of industries.
Rajan and Zingales use data from Standard and Poor’s Compustat for US ?rms in 36 industries.A ?rm’s dependence on external ?nance is de?ned as the share of invest-ment that cannot be ?nanced through internal cash ?ows;or as capital expenditures ? Blackwell Publishing Ltd 2003
Table 1.Summary Statistics and Correlation between Measures of Financial Development
Mean Standard deviation Median Minimum Maximum Number of observations Export share0.39 1.440.060.0038.891,981
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE Trade balance-0.12 1.50-0.07-11.1637.811,981
Private Credit45.9834.4635.15 2.36160.2057
Liquid Liabilities48.2327.6946.1912.58156.8457
Commercial–Central Bank76.6918.4980.7023.4799.5457
Market Capitalization17.8918.8710.090.4567.2342
Value Traded8.8621.49 2.710.02131.2641
Turnover30.2843.8324.750.17273.9541
Total Capitalization71.5751.2859.7114.93218.8342
Accounting60.0913.9362248335
External dependence (US,80s)0.320.410.24-0.45 1.4936
External dependence (US,young ?rms)0.680.640.68-1.53 2.0634
External dependence (US,70s)0.080.190.07-0.450.5435
External dependence (Canadian ?rms)0.430.770.38-0.80 3.5127
Private Liquid Commercial–Market Value Total
Credit Liabilities Central Bank Capitalization Traded Turnover Capitalization Accounting
Private Credit1
Liquid Liabilities0.821
? Blackwell Publishing Ltd 2003
(0.001)
Commercial–Central Bank0.680.511
(0.001)(0.001)
Market Capitalization0.710.620.541
(0.001)(0.001)(0.001)
301
302Thorsten Beck ? Blackwell Publishing Ltd 2003Table 1.Continued
Private Liquid Commercial–Market Value Total
Credit Liabilities Central Bank Capitalization Traded Turnover Capitalization Accounting
Value Traded 0.690.640.360.641
(0.001)(0.001)(0.022)(0.001)
Turnover 0.580.480.300.440.921
(0.001)(0.002)(0.059)(0.004)(0.001)
Total Capitalization 0.970.810.670.870.720.571
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Accounting 0.450.190.590.620.270.240.551
(0.007)(0.273)(0.001)(0.001)(0.120)(0.161)(0.001)
Notes :
Export share =share of an industry’s exports in GDP .
Trade balance =ratio of an industry’s trade balance to GDP .
External dependence (US,80s) =fraction of capital expenditures not ?nanced with internal funds for US ?rms in the same industry between 1980 and 1990.Source:Rajan and Zingales (1998).
External dependence (US,young ?rms) =fraction of capital expenditures not ?nanced with internal funds for US ?rms that went public during the previous 10 years in the same industry between 1980 and 1990.Source:Rajan and Zingales (1998).
External dependence (US,70s) =fraction of capital expenditures not ?nanced with internal funds for US ?rms in the same industry between 1970 and 1980.Source:Rajan and Zingales (1998).
External dependence (Canadian ?rms) =fraction of capital expenditures not ?nanced with internal funds for Canadian ?rms in the same industry between 1980 and 1990.Source:Rajan and Zingales (1998).
p -values are shown in parentheses.
Private Credit =credit by deposit money banks and other ?nancial institutions to the private sector divided by GDP ,times 100.
Liquid Liabilities =liquid liabilities of the ?nancial system (currency plus demand and interest-bearing liabilities of banks and nonbank ?nancial intermediaries) divided by GDP ,times 100.
Commercial–Central Bank =assets of deposit money banks divided by assets of deposit money banks plus central bank assets,times 100.
Market Capitalization =value of listed domestic shares on domestic exchanges divided by GDP ,times 100.
Value Traded =value of the trades of domestic shares on domestic exchanges divided by GDP ,times 100.
Turnover =value of the trades of domestic shares on domestic exchanges divided by the value of listed domestic shares on domestic exchanges,times 100.
Total Capitalization =Private Credit +Market Capitalization.
Accounting =Index of the comprehensiveness and quality of company reports.
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE303 minus cash ?ow from operations divided by capital expenditures.Both numerator and denominator are averaged over the 1980s to smooth temporal ?uctuations.The indus-try values are calculated as medians rather than means to thus prevent outliers from dominating the results.We have data for 36 industries varying from tobacco,an indus-try with no demand,to drugs,the industry with the highest use of external ?nance.
Indicators of Financial Development
The primary measure of ?nancial intermediary development is Private Credit,which equals the value of credits by ?nancial intermediaries to the private sector divided by GDP,expressed as a percentage.It captures the amount of credit channeled through ?nancial intermediaries to private ?rms.Recent research has shown a robust link between Private Credit and economic growth,as for example Levine et al.(2000).In the sensitivity analysis,I include two further measures of ?nancial intermediary devel-opment.Liquid Liabilities equals the liquid liabilities of the ?nancial system (currency plus demand and interest-bearing liabilities of banks and nonbank ?nancial interme-diaries) divided by G DP,and expressed as a https://www.doczj.com/doc/b73636665.html,mercial–Central Bank equals the percentage ratio of commercial banks’ domestic assets divided by com-mercial banks’ and central bank’s domestic assets.10
The primary measure of stock market development is Market Capitalization,which equals the value of listed domestic shares on domestic exchanges divided by GDP,and expressed as a percentage.This is an indicator of the size of the secondary stock market.Unlike Private Credit,however,Market Capitalization does not measure the amount of funding available to ?rms,but rather the discounted value of future earn-ings.It might therefore overestimate the importance of the stock market in obtaining external funds,as pointed out by Rajan and Zingales (1998).Furthermore,Market Cap-italization indicates the size rather than the activity of stock markets.In the sensitiv-ity analysis,I therefore use two percentage measures of stock market liquidity.Value Traded equals the value of the trades of domestic shares on domestic exchanges divided by GDP.Turnover equals the value of the trades of domestic shares on domes-tic exchanges divided by the value of listed domestic shares.11
Finally,I use two percentage measures of the overall importance of the ?nancial sector.Total Capitalization is the sum of Market Capitalization and Private Credit.It captures the overall size of the ?nancial sector by including all ?nancial institutions and the equity market and by combining our primary measures of ?nancial interme-diary and stock market development.In the sensitivity analysis,I use Accounting,a measure of the comprehensiveness of companies’ balance sheets and income state-ments.The higher is Accounting,the easier it should be for ?rms to obtain external ?nance,either from ?nancial intermediaries or ?nancial markets.
There is a signi?cant variation in ?nancial development across the countries included in this sample.Whereas Private Credit is only 2% in Ghana,it is 160% in Switzerland. Similarly,Market Capitalization is only 0.5% in Uruguay,but 67% in Japan.
The Trade Data
Data on exports and imports for the 36 industries are from the United Nations Sta-tistical Of?ce COMTRADE database.Exports are reported at f.o.b.(free on board) values and imports at c.i.f.(cost,insurance,freight) values.The data are in US dollars and averaged over the ten-year period 1980–89.They are de?ated by export and import price indices obtained from the World Development Indicators (WDI) of the World
? Blackwell Publishing Ltd 2003
304Thorsten Beck
Bank.To obtain the share of industry exports and imports in GDP,we divide by real GDP,using WDI data.There is a wide variation in industry export shares and trade balances.While Nigeria’s exports in the petroleum re?neries industry constitute 39% of GDP,Bolivia’s exports of petroleum and coal products constitute less than 0.01% of GDP.Nigeria has also the largest industry trade surplus in the sample,again in the petroleum re?neries industry,whereas Ghana has the largest industry trade de?cit in the sample,in the motor vehicles industry.
5.Basic Results
The results in Table 2 are evidence in favor of the hypothesis that countries with better-developed ?nancial institutions and markets have a comparative advantage in indus-tries that use more external ?nance.There are signi?cant and positive coef?cients on the interaction terms between external dependence and ?nancial development in the export and trade balance equations.This supports the theoretical model by Kletzer Table 2.Industry Exports,Trade Balances,and Financial Development
Export Export Export Export Dependent variable:share share share share Interaction (external dependence ￥log[Total 1.274
Capitalization])(0.001)
Interaction (external dependence ￥log[Private 1.259 1.256 Credit])(0.001)(0.005) Interaction (external dependence ￥log[Market0.7660.079 Capitalization])(0.001)(0.768) F-test joint signi?cance17.28
(0.001) Number of observations1,4201,9451,4201,420
Trade Trade Trade Trade Dependent variable:balance balance balance balance Interaction (external dependence ￥log[Total 1.430
Capitalization])(0.001)
Interaction (external dependence ￥log[Private 1.421 1.736 Credit])(0.001)(0.001) Interaction (external dependence ￥log[Market0.810-0.080 Capitalization])(0.001)(0.789) F-test joint signi?cance18.00
(0.001) Number of observations1,4201,9451,4201,420 Notes:
All regressions include country and industry dummies.
p-values from heteroskedasticity robust standard errors are reported in parentheses.
All regressions are estimated using the English,French,and German legal origin dummies as instrumental variables for the ?nancial development indicators.In column 4 the shares of Catholic,Muslim,and Protes-tant population in total population are included as additional instrumental variables.
The F-test is for the joint signi?cance of both interaction terms of external dependence with Private Credit and Market Capitalization.
? Blackwell Publishing Ltd 2003
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE305 and Bardhan (1987) that predicts higher exports and a larger trade balance in exter-nally dependent industries for countries with fewer capital market restrictions.While we cannot reject the possibility that there is a link from the structure of the trade balance to the development of the ?nancial sector and the use of external ?nance,using the legal origin to extract the exogenous component of ?nancial development and dropping the US from our sample allows us to conclude that the positive relation is not due only to reverse causality or simultaneity bias.
I start with Total Capitalization,the most comprehensive measure of ?nancial sector development.The results indicate that in countries with a higher level of Total Capitalization,industries using external ?nance have higher exports and larger trade balances.To illustrate this consider the furniture industry,which has the median value for external dependence (0.24) across the 36 industries.The regression results predict that a 10% higher level of ?nancial development results in a 3.1% higher export share of furniture in GDP and a 3.4% higher trade balance.
The results are also evidence for the importance of ?nancial development for the trade patterns across countries.Consider the industries at the 25th percentile (low external dependence) and 75th percentile (high external dependence),beverages and machinery,respectively.A 10% higher level of Total Capitalization implies 1% higher exports and a 1.1% larger trade balance in beverages (external dependence =0.08), and 5.7% higher exports and a 6.4% larger trade balance in machinery (external dependence =0.45).This indicates that the positive relation between ?nancial devel-opment and export share and trade balance increases in the external dependence of the industry.This result is consistent with the Kletzer and Bardhan model in that coun-tries with fewer capital market restrictions and therefore higher levels of ?nancial development have a comparative advantage in industries that use more external ?nancing.
The results in columns 2 and 3 of Table 2 suggest that both ?nancial-intermediary and stock-market development are important sources of comparative advantage. The coef?cients on the interaction terms of both Private Credit and Market Capital-ization are signi?cantly positive at the 1% level.When we include both Private Credit and Market Capitalization in a regression,however,only the interaction term with P rivate Credit exhibits a signi?cant coef?cient (column 4).This suggests that stock market development is not an independent source of comparative advantage or that the exogenous component of Market Capitalization does not contain any additional information about the development of the ?nancial sector that is not contained in Private Credit.12
As alternative to the regression analysis we can explore the correlation between the weighted-average external dependence of a country’s exports and trade balance and our indicators of ?nancial sector development.13 This amounts to a test whether coun-tries with better-developed ?nancial systems export goods produced by industries that use more external ?nance.
Table 3 ranks the countries in a sample according to the weighted-average external dependence of their exports and their trade balance.14 Note that two of the world’s ?nancially most developed countries—Japan and Switzerland—also have the highest weighted-average external dependence of exports and trade balance.There are certainly surprises:Chile has a relatively low weighted-average external dependence of exports and trade balance,while Paraguay has a relatively high weighted-average external dependence of the trade balance,although their respective levels of ?nancial development would predict otherwise.However,this ranking does not control for other country- and industry-speci?c effects,unlike the regression analysis performed
? Blackwell Publishing Ltd 2003
306Thorsten Beck
Table 3.Ranking of Countries According to Weighted-Average External Dependence of Exports and Trade Balance
Weighted-average Weighted-average
external dependence external dependence Country of exports Country of trade balance Japan0.52Japan0.21 Switzerland0.52Switzerland0.15 Ireland0.49Ireland0.07
Israel0.43Paraguay0.04 Malaysia0.40Denmark0.03
Great Britain0.39Israel0.01 Sweden0.37Great Britain0.00 Denmark0.37France0.00 France0.35Sweden-0.01 Korea0.35Mauritius-0.01 Austria0.35Italy-0.01
Italy0.32Belgium-0.02 Canada0.30Bolivia-0.03 Netherlands0.30Jamaica-0.03 Belgium0.30Austria-0.04 Finland0.28Netherlands-0.06
Spain0.28Korea-0.06 Costa Rica0.28Guatemala-0.07 Guatemala0.28Malaysia-0.07 Philippines0.28Nigeria-0.07 Thailand0.26Philippines-0.08 Portugal0.25Costa Rica-0.09 Mexico0.24Finland-0.10
India0.23India-0.10
Brazil0.22Togo-0.11 Norway0.21Portugal-0.11
El Salvador0.21Thailand-0.12 Jamaica0.21Ghana-0.13 Australia0.17Canada-0.13 Morocco0.16Spain-0.13 Argentina0.16Bangladesh-0.14 Mauritius0.16Senegal-0.15 Honduras0.16Morocco-0.15 Senegal0.16Norway-0.16 Paraguay0.16Mexico-0.18 Dominican Republic0.16Greece-0.19
New Zealand0.15El Salvador-0.19 Colombia0.15Brazil-0.19
Togo0.14Dominican Republic-0.19 Greece0.13Madagascar-0.20 Kenya0.13Colombia-0.21 Madagascar0.13Honduras-0.23 Bangladesh0.11New Zealand-0.24 Trinidad and Tobago0.09Pakistan-0.24 Pakistan0.09Syria-0.24 Bolivia0.09Kenya-0.25 Ecuador0.09Australia-0.25
? Blackwell Publishing Ltd 2003
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE307 Table 3.Continued
Weighted-average Weighted-average
external dependence external dependence Country of exports Country of trade balance Peru0.08Trinidad and Tobago-0.27 Uruguay0.08Argentina-0.28
Syria0.08Egypt-0.29
Chile0.07Peru-0.29 Ghana0.05Algeria-0.31 Venezuela0.05Uruguay-0.31 Algeria0.05Chile-0.32 Nigeria0.04Ecuador-0.32
Egypt0.04Venezuela-0.34 Notes:External dependence =share of capital expenditures not ?nanced with internal funds for US ?rms in the same industry between 1980 and 1990.
We calculate the weighted-average external dependence of a country’s exports by multiplying each indus-try’s external dependence with its share in total manufacturing exports of the country.To get the weighted-average external dependence of a country’s trade balance,we take the difference between the country’s weighted-average external dependence of exports and imports.
Table 4.Correlations between Weighted-Average External Dependence of Exports and Trade Balance and Financial Development
Weighted-average external dependence of:
Exports Trade balance Total Capitalization0.7150.612
(0.001)(0.001)
4141 Private Credit0.6690.473
(0.001)(0.001)
5656 Market Capitalization0.6560.541
(0.001)(0.001)
4141 Notes:See Tables 1 and 3.p-values are reported in parentheses and number of countries in the third line.
in the rest of the paper.Unlike the factor content studies of the Heckscher–Ohlin–Vanek theorem,this ranking does also not take into account other production factors and country endowments.It thus constitutes only a partial analysis.
The correlations reported in Table 4 support the present hypothesis and the Kletzer and Bardhan model.The weighted-average external dependence of the exports and the trade balance is signi?cantly and positively correlated with all three measures of ?nancial development.This con?rms the previous regression results that the exports and the trade balance of ?nancially more developed countries tend to be dominated by goods that are produced by industries using more external ?nancing.
? Blackwell Publishing Ltd 2003
308Thorsten Beck
6.Robustness Tests
Simultaneity Bias and Reverse Causality
Are the results reported so far due to simultaneity bias or reverse causality? Although the results obtained from the two-stage least squares (TSLS) regressions control for simultaneity bias and possible reverse causality,there are more direct ways to test for this.Following Rajan and Zingales (1998),I split the sample into two subsamples, export shares and trade balances above and below the median for a country.
The results in the top panel of Table 5 indicate that the previous results are not due to a spurious correlation.By restricting the sample to the industries that have export shares and trade balances above the median,we take account of the fact that an economy’s endowment,such as natural resources,might give it a comparative advan-tage in industries that also use a lot of external ?nance.Although the size of the coef-?cients is smaller than in Table 2,the coef?cients on all interaction terms are still signi?cant at the 5% level,except for the interaction term of external dependence with Market Capitalization in the trade balance regression.As in Table 2,the interaction term with Market Capitalization enters insigni?cantly once we control for the inter-action of external dependence with Private Credit.
The results in the bottom panel of Table 5 provide somewhat weaker evidence that the previous results are not due to reverse causality.By restricting the sample to indus-tries with export shares and trade balances below the median,we take account of the concern that ?nancial institutions and markets might have arisen due to a demand from industries that are dependent on external ?nance and that constitute a large part of a country’s exports.The results are less assuring,although most coef?cients on the interaction terms still enter signi?cantly positive at the 10% level.Exceptions are the regressions where we include the interaction of external dependence with both Private Credit and Market Capitalization.15
Alternative Measures of External Dependence
The Rajan and Zingales (1998) dataset provides three alternative measures of exter-nal dependence that allow us to test the sensitivity of the present results to the measure of external dependence used so far.The three alternative measures of external depen-dence are signi?cantly correlated with our principal measure of external dependence at the 1% level,with correlation coef?cients being at least 60%.16
The results in Table 6 show that the results are robust to using a measure of exter-nal dependence calculated for US ?rms that went public during the previous ten years. Rajan and Zingales show that the use of external ?nance is highest during the early years of a https://www.doczj.com/doc/b73636665.html,ing a sample of young ?rms to calculate the dependence on external ?nance might therefore give a more appropriate picture of the reliance on external ?nance.17 The coef?cients on most interaction terms are signi?cantly positive. The smaller size of the coef?cients re?ects the higher use of external ?nance by young ?rms.Whereas furniture—the median industry for both samples—has an external dependence ratio of 0.24 in the sample for all ?rms,it has an external dependence of 0.68 in the young ?rms’ https://www.doczj.com/doc/b73636665.html,ing the results from Table 7,a 10% increase in Total Capitalization therefore increases the export share of the median industry by 3.4%,a stronger effect than when using the results from the sample with all ?rms.18 Interest-ingly,when including interaction terms with both Private Credit and Market Capital-ization,only the latter enters signi?cantly positive in the export regressions,while only ? Blackwell Publishing Ltd 2003
Table 5.Robustness Tests
Export Export Export Export Trade Trade Trade Trade Dependent variable:share share share share balance balance balance balance Above median
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE Interaction (external dependence ￥log[Total Capitalization])0.7770.512
(0.001)(0.044)
Interaction (external dependence ￥log[Private Credit])0.784 1.4170.530 1.056
(0.001)(0.001)(0.014)(0.030)
Interaction (external dependence ￥log[Market Capitalization])0.359-0.4010.144-0.212
(0.018)(0.114)(0.454)(0.520)
F-test joint signi?cance8.26 4.24
(0.001)(0.015)
Number of observations709969709709709969709709 Below median
Interaction (external dependence ￥log[Total Capitalization])0.4790.506
(0.076)(0.022)
Interaction (external dependence ￥log[Private Credit])0.5340.2020.3490.177
(0.028)(0.616)(0.054)(0.570)
Interaction (external dependence ￥log[Market Capitalization])0.3460.1180.4190.054
(0.083)(0.636)(0.019)(0.800)
F-test joint signi?cance 1.290.79? Blackwell Publishing Ltd 2003
(0.277)(0.455)
Number of observations709970709709709969709709 Notes:The top panel reports the regression results using export shares and trade balances that are above the respective country median.The bottom panel reports the
regression results using export shares and trade balances that are below the respective country median.
See also the notes to Tables 1 and 2.
309
310? Blackwell Publishing Ltd 2003
Thorsten Beck
Table 6.External Dependence measured using Young Firms
Export Export Export Export Trade Trade Trade Trade Dependent variable:share share share share balance balance balance balance Interaction (external dependence ￥log[Total Capitalization])0.5020.646
(0.001)(0.001)
Interaction (external dependence ￥log[Private Credit])0.4310.2140.6390.623
(0.001)(0.386)(0.001)(0.024)
Interaction (external dependence ￥log[Market Capitalization])0.3370.2800.3690.168
(0.001)(0.050)(0.001)(0.269)
F-test joint signi?cance12.9713.61
(0.001)(0.001)
Number of observations1,3411,8401,3411,3411,3411,8401,3411,341 Notes:See Tables 1 and 2.
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE 311? Blackwell Publishing Ltd 2003Table 8.External Dependence measured using Canadian Firms
Export Export Export Export Trade Trade Trade Trade
Dependent variable:share share share share balance balance balance balance Interaction (external dependence ￥log[Total Capitalization ])0.3570.359
(0.018)(0.017)
Interaction (external dependence ￥log[Private Credit ])0.3590.3910.3940.529(0.011)(0.175)(0.004)(0.087)Interaction (external dependence ￥log[Market Capitalization ])0.216-0.0470.209-0.132
(0.035)(0.772)(0.037)(0.455)
F -test joint signi?cance 2.81 2.88
(0.061)(0.057)
Number of observations 1,0651,4561,0651,0651,0651,4561,0651,065Notes :See Tables 1 and 2.Table 7.External Dependence measured using Firms in the 1970s
Export Export Export Export Trade Trade Trade Trade
Dependent variable:share share share share balance balance balance balance Interaction (external dependence ￥log[Total Capitalization ]) 3.026 3.034
(0.001)(0.001)
Interaction (external dependence ￥log[Private Credit ]) 2.700 1.717 2.488 2.121
(0.001)(0.078)(0.001)(0.067)
Interaction (external dependence ￥log[Market Capitalization ]) 2.0740.889 1.9730.768
(0.001)(0.142)(0.001)(0.252)
F -test joint signi?cance 18.6315.98
(0.001)(0.001)
Number of observations 1,3801,8891,3801,3801,3801,8891,3801,380Notes:See Tables 1 and 2.
312Thorsten Beck
the former enters signi?cantly in the trade balance regression.This seems to shed doubt on the previous results,that it is ?nancial intermediaries rather than ?nancial markets that provide the necessary external ?nancing for export industries.
Table 7 shows that the results are robust to using a measure of external dependence that is calculated for a sample of listed US ?rms over the period 1970–79.Rajan and Zin-gales propose that if countries other than the US use older technologies,the external dependence as measured over the 1980s might not re?ect well the reliance on external ?nance in other countries,especially developing countries.19 I therefore reran the regres-sions using the external dependence measured over the 1970s.The coef?cients on all interaction terms are signi?cantly positive and the coef?cient size increases.When including interaction terms with both Private Credit and Market Capitalization,they are jointly signi?cant,while the interaction term with Market Capitalization is insigni?cant and the interaction term with Private Credit is signi?cant only at the 10% level. Finally,the results in Table 8 indicate that the previous results are not due to pecu-liar characteristics of industries in the US.The external dependence as calculated for a sample of Canadian ?rms is used.Although Canada has a different ?nancial struc-ture,its ?nancial system can be considered as well developed as the ?nancial system of the https://www.doczj.com/doc/b73636665.html,ing data from Canadian ?rms should therefore not alter our results.20 The results in Table 8 show that the coef?cients on most interaction terms are signi?cantly positive at the 5% level.When including interaction terms with both Private Credit and Market Capitalization,however,they are jointly signi?cant only at the 10% level. Further Robustness Tests
The present results are robust to the use of other indicators of ?nancial development, proposed in the literature.The interaction terms of external dependence with all ?ve additional indicators of ?nancial development described in section 4 enter signi?cantly positive.21
Financial development does not proxy for other possible determinants of compara-tive advantage.To control for other potential determinants of comparative advantage, we interact external dependence with a measure of schooling (log of one plus the average years of schooling in the population over the age of 25 in 1980) and with the log of real per capita GDP in 1980 in the regressions.22 Industries that rely highly on external ?nance might also depend on human capital as an input.Financial devel-opment might also proxy for the general level of development,as measured by per capita income.Although the coef?cient on the interaction term with GDP per capita is signi?cant in some regressions,it is not robustly positive in all regressions.The coef-?cient on the interaction term with schooling does not enter signi?cantly in any of the regressions.Including these additional interaction terms does not change signi?cance or size of the coef?cients on the interaction term of ?nancial development and exter-nal ?nance.23
The results are robust to the use of alternative instruments for ?nancial develop-ment.Replacing legal origin and religious composition as instruments by measures of the legal rights of secured creditors and minority shareholder,and indicators of insti-tutional quality,does not change our results.24
7.Concluding Remarks
This paper has examined whether countries with better-developed ?nancial systems enjoy a comparative advantage in industries that use more external ?nance.I have used ? Blackwell Publishing Ltd 2003
FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE313 a methodological approach developed by Rajan and Zingales (1998) by regressing export and trade shares on the interaction between external dependence across indus-tries and ?nancial development across countries.
The results provide robust evidence for the stated https://www.doczj.com/doc/b73636665.html,ing different meas-ures of ?nancial development and external dependence,the results indicate that,every-thing else equal,countries with a higher level of ?nancial development have higher export shares and trade balances in industries that rely more on external ?nance.The possibility that the results are driven by reverse causality or simultaneity bias has been controlled by using the legal origin of countries as instrumental variables for ?nancial development.By restricting the sample to industries with export shares or trade bal-ances above or below the median for a country,further evidence is provided against the hypotheses that the present results might be driven by the simultaneous determi-nation of export specialization and ?nancial development or that the development of the ?nancial sector simply follows the real sector.While the results suggest that both ?nancial institutions and markets are important in channeling external funds to ?rms and thus determining international trade ?ows,we cannot determine whether they function through independent channels.
It has been shown also that there are positive correlations between the weighted-average external dependence of a country’s exports and trade balance and its level of ?nancial development.This simple variation of the factor-content studies by Bowen et al.(1987) and Tre?er (1993,1995) indicates that countries with higher levels of ?nan-cial sector development are net exporters of goods that are produced by industries with higher reliance on external ?nancing.
The results of this paper provide the ?rst empirical examination of a small literature on the link between international trade and ?nancial development.Speci?cally,it pro-vides supporting evidence for the models developed by Kletzer and Bardhan (1987) that show that countries with fewer capital market restrictions have a comparative advantage in industries with higher external ?nancing needs.
These results have implications for policy reforms in both the ?nancial and the trade sector.Suppose a country with a low level of ?nancial development undertakes ?nancial sector reforms that raise the level of external ?nance available to private enterprises.These reforms might include strengthening creditor rights and contract enforcement through judiciary and judicial reforms.25 A subsequent increase in exter-nal ?nance available to private enterprises implies a shift in comparative advantage, away from industries that do not rely on external ?nance towards industries that rely relatively more on external ?nance.While exporters in industries with no use of exter-nal ?nance might see their export shares decrease,relative to other industries,and face more competition from other ?nancially less developed countries,exporters in indus-tries that rely more on external ?nance might gain export shares.
Suppose a country embarks on trade reforms,lowering import tariffs across the board and thus exposing domestic industries of exportable goods to international com-petition.26 If the country has a high level of ?nancial development,this new competi-tion might affect industries that rely less on external ?nance more than other industries. If,on the other hand,the country has a low level of ?nancial development,industries that rely relatively more on external ?nance might be hurt most.
References
Baldwin,Richard E.,“Exporting the Capital Markets:Comparative Advantage and Capital Market Imperfections,”in David B.Audretsch,Leo Sleuwaegen,and Hideki Yamawaki (eds.),
? Blackwell Publishing Ltd 2003
314Thorsten Beck
The Convergence of International and Domestic Markets,Amsterdam:North-Holland (1989):135–52.
Barro,Robert J.and Jong-Wha Lee,“International Measures of Schooling Years and Schooling Quality,”American Economic Review86 (1996):218–23.
Bowen,Harry P.,Edward E.Leamer,and Leo Sveikauskas,“Multicountry,Multifactor Tests of the Factor Abundance Theory,”American Economic Review77 (1987):791–809.
Jones,Ronald W.,“The Positive Theory of International Trade,”in Ronald W.Jones and Peter B.Kenen (eds.),Handbook of International Economics,Vol.1,Amsterdam:Elsevier Science (1984):1–62.
King,Robert G.and Ross Levine,“Finance and G rowth:Schumpeter Might Be Right,”Quarterly Journal of Economics108 (1993a):717–37.
———,“Finance,Entrepreneurship,and Growth.Theory and Evidence,”Journal of Monetary Economics32 (1993b):513–42.
Kletzer,Kenneth and Pranab Bardhan,“Credit Markets and Patterns of International Trade,”Journal of Development Economics27 (1987):57–70.
La Porta,Rafael,Florencio Lopez-de-Silanes,Andrei Shleifer,and Robert W.Vishny,“Legal Determinants of External Finance,”Journal of Finance52 (1997):1131–50.
———,“Law and Finance,”Journal of Political Economy106 (1998):1113–55.
———,“The Quality of G overnment,”Journal of Law,Economics,and Organization15 (1999):222–79.
Levine,Ross,“Financial Development and Economic Growth:Views and Agenda,”Journal of Economic Literature35 (1997):688–726.
———,“The Legal Environment,Banks,and Long-Run Economic Growth,”Journal of Money, Credit,and Banking30 (1998):596–620.
Levine,Ross and Sara Zervos,“Stock Markets,Banks,and Economic G rowth,”American Economic Review88 (1998):537–58.
Levine,Ross,Norman Loayza,and Thorsten Beck,“Financial Intermediation and Economic Growth:Causality and Causes,”Journal of Monetary Economics46 (2000):31–77.
Loayza,Norman,Humberto Lopez,Klaus Schmidt-Hebbel,and Luis Serven,“The World Saving Data Base”,https://www.doczj.com/doc/b73636665.html,/research/projects/savings/policies.htm (1998).
Rajan,Raghuram G.and Luigi Zingales,“Financial Dependence and G rowth,”American Economic Review88 (1998):559–86.
Sachs,Jeffrey and Andrew Warner,“Economic Reform and the Process of Global Integration,”Brookings Papers on Economic Activity1 (1995):1–118.
Tre?er,Daniel,“International Factor Price Differences:Leontief Was Right!”Journal of Politi-cal Economy101 (1993):961–87.
———,“The Case of the Missing Trade and Other Mysteries,”American Economic Review85 (1995):1029–46.
Notes
1.In the following,I will use the terms “external dependence,”“reliance,”and “use of external ?nance”interchangeably.
2.For an overview see Jones (1984).
3.This can be casted also in the context of a formal model.De?ne a ci(c =1,2,i=1,2) as country c’s labor requirement per unit of output in sector i,where food is sector 1 and manufacturing sector 2.If country 1 has more external ?nance available than country 2,then a12 4.To the best knowledge of the author,no empirical study has been undertaken yet to explore the link between ?nancial development and the structure of international trade ?ows. 5.See Levine (1997) for an overview. 6.Both the trade indicators and Finance are included in logs,so that the results can be inter-preted as elasticities.Since data from the US are used to calculate the dependence ratios,all US observations are dropped from the regressions. ? Blackwell Publishing Ltd 2003 FINANCIAL DEPENDENCE AND INTERNATIONAL TRADE315 7.By including country dummies in the regression we control for other determinants of industry export shares and trade balances,such as higher G DP growth or a smaller size of the economy.By including industry dummies we control for the fact that some industries might be more export-oriented than others,for reasons that are not related to ?nancial development. 8.The countries in the sample have British,French,German,or Scandinavian legal origin.Data are from La Porta et al.(1999). 9.Even if capital markets are not perfect,but the elasticity of supply does not change substan-tially across industries,the actual amount of external ?nance used is still a reasonable measure of relative demand.See Rajan and Zingales (1998). 10.Other researchers have used both measures;see King and Levine (1993a,b). 11.Both measures were used by Levine and Zervos (1998). 12.Since these results might be due to the fact that Private Credit and Market Capitalization “share”the legal origin dummies as instrumental variables,I also include interaction terms of external dependence with the share of Catholic,Muslim,and Protestant population as instrumental variables.Anecdotal and statistical evidence suggests that the development of institutions is partly driven by the dominant religion in a country;see La Porta et al.(1999). However,since the legal origin and religious composition are correlated,even this extended instrumental variable set might not contain enough information to extract the exogenous component of ?nancial intermediary and the distinct exogenous component of stock market development. 13.The weighted-average external dependence of a country’s exports is calculated by multi-plying each industry’s external dependence with its share in total manufacturing exports of the country.To get the weighted-average external dependence of a country’s trade balance,we take the difference between the country’s weighted-average external dependence of exports and imports. 14.As before,the US is not included in the sample. 15.I also split the sample according to the industry’s share in total manufacturing production, rather than exports or trade balance,a test proposed by Rajan and Zingales.I ?nd that the inter-action of external dependence and ?nancial development is insigni?cant for industries above the median and signi?cant for industries below the median;results are available on request. While this seems to shed doubt on the previous results,two quali?cations have to be made.First, the number of countries decreases to 36 for these regressions,since Rajan and Zingales’ sample is smaller.Second,the industrial structure of a country is not necessarily determined by the same factors as the structure of its exports. 16.Data are available on request. 17.The average external dependence is more than twice as high for the sample of young ?rms (0.68) than for the sample of all ?rms (0.32). 18.As reported above,a 10% increase in Total Capitalization increases the export share of the median industry (furniture) by 3.1%,when using the sample for all industries. 19.The average external dependence over the 1970s was 0.08 as compared to 0.32 in the 1980s. 20.According to Rajan and Zingales,Canada is the only other country for which ?rm-level data on ?ow of funds are available.The present data indicate that the ?rms in the Canadian sample did not use signi?cantly more or less external resources to ?nance their capital expenditures than US ?rms. 21.Results are available on request.I also ran regressions including interaction terms with one indicator of ?nancial intermediary development (Liquid Liabilities or Commercial–Central Bank) and one stock market indicator (Value Traded or Turnover).While the interac-tion with Liquid Liabilities does not enter signi?cantly,the interactions with either stock market indicator do.While the interaction with Commercial–Central Bank enters signi?cantly positive in the export share regression and neither of the interaction terms with the stock market indi-cator does,it is the other way around in the trade balance regressions.This sheds additional doubts on whether ?nancial intermediary or stock market development is the dominating channel. ? Blackwell Publishing Ltd 2003
相关主题
文本预览