The Evolving Importance of Banks and Securities Markets ¨c Asli Demirgu ¸ -Kunt, Erik Feyen, and Ross Levine The roles of banks and securities markets evolve during the process of economic development. As countries develop economically, (1) the size of both banks and securities markets increases relative to the size of the economy, (2) the association between an increase in economic output and an increase in bank development becomes smaller, and (3) the association between an increase in economic output and an increase in securities market development becomes larger. These findings are consistent with theories predicting that as economies develop, the services provided by securities markets become more important for economic activity, whereas those provided by banks become less important. JEL codes: O16, G1, G2, O43 Several economic theories stress that banks provide services to the economy that differ from those provided by securities markets, predicting that both the operation of banks and the functioning of securities markets have independent influences on economic development. For example, Acemoglu and Zilibotti (1997), Allen and Gale (1997, 1999), Boot and Thakor (1997, 2000), Dewatripont and Maskin (1995), Holmstrom and Tirole (1993), and Rajan (1992) argue that banks have a comparative advantage in reducing the market frictions associated with financing standardized, shorter-term, lower- risk, well-collateralized endeavors, whereas decentralized markets are relatively more effective in custom designing arrangements to finance more novel, longer- run, higher-risk projects that rely on intangible inputs. Consistent with these Asli Demirgu ¨c¸ -Kunt (ademirguckunt@worldbank.org) is the Director of Development Policy in the World Bank’s Development Economics Vice Presidency (DEC) and Chief Economist of the Financial and Private Sector Network (FPD). Erik Feyen (corresponding author; efeijen@worldbank.org) is a Senior Financial Economist in the Financial Systems Practice of the World Bank. Ross Levine (ross_levine@haas.berkeley.edu) is the Willis H. Booth Professor of Banking and Finance at the Haas School of Business at the University of California, Berkeley. We received very helpful comments from Thorsten Beck; Norman Loayza; Yona Rubinstein; seminar participants at the World Bank’s conference, “Financial Structure and Economic Development,” on June 16, 2011; and three anonymous referees. Mauricio Pinzon Latorre provided outstanding research assistance. A supplemental appendix to this article is available at http://wber.oxfordjournals.org/. THE WORLD BANK ECONOMIC REVIEW, VOL. 27, NO. 3, pp. 476– 490 doi:10.1093/wber/lhs022 Advance Access Publication August 24, 2012 # The Author 2012. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com 476 ¨c Demirgu ¸ -Kunt, Feyen, and Levine 477 theories, Demirgu ¨c¸ -Kunt and Maksimovic (1998), Levine and Zervos (1998), and Beck and Levine (2004) provide evidence that better-functioning banks and securities markets exert robust, independent, and positive effects on economic activity. A substantial body of economic theory emphasizes that the comparative importance of banks and markets for economic activity changes during the process of economic development, with markets becoming relatively more important for economic activity. For example, the concepts articulated in Goldsmith (1969), Allen and Gale (1995, 2000), Boot and Thakor (1997, 2000), Boyd and Smith (1998), Weinstein and Yafeh (1998), Morck and Nakamura (1999), and Song and Thakor (forthcoming) suggest that (1) banks and markets provide different, though sometimes complementary, financial ser- vices and that (2) the services provided by markets become comparatively more important for promoting economic activity as countries develop economically. In particular, these theories suggest that as economies develop, a greater number of projects require customized financial arrangements rather than stan- dardized contracts and rely on intangible assets rather than easily collateralized capital inputs. Because these models also suggest that banks have a compara- tive advantage in financing standardized, well-collateralized endeavors, whereas securities markets are better at custom designing arrangements to finance novel projects that rely on intangible inputs, these theories imply that the services provided by securities markets have a greater impact on economic activity as economies grow, whereas those provided by banks become less important. Empirical research has been largely unsuccessful at clarifying the evolving importance of banks and markets during the process of economic development, as exemplified by Beck and Levine (2002), Demirgu ¸ -Kunt and Maksimovic ¨c (2002), and Levine (2002). Demirgu ¸ -Kunt and Levine (2001) show that banks ¨c and securities markets tend to become more developed as economies grow and that securities markets tend to develop more rapidly than banks. Thus, finan- cial systems generally become more market-based during the process of economic development. However, this pattern could simply reflect reverse causality. Economic progress may boost the development of securities markets more than it boosts the development of banks. The observation that financial systems tend to become more market-based as economies develop does not nec- essarily imply that securities markets exert a larger impact on economic activity in more economically advanced economies. In this paper, the changing importance of banks and securities markets with the development of economies is evaluated in two steps. First, using a newly developed database, the evolution of banks and securities markets during the process of economic development is reassessed. That is, as countries develop economically, what happens to the size of banks and securities markets relative to the size of the overall economy? Second, the changes to the associations between economic activity and bank and stock market development as 478 THE WORLD BANK ECONOMIC REVIEW countries develop economically are examined. This investigation involves regressing economic activity on both bank and securities market development and assessing how the estimated coefficients change as countries develop economically. This analysis provides information on how the associations between economic activity and the different components of the financial system evolve during economic development. The primary methodological contribution of this paper is the use of quantile regressions to assess how the associations between economic activity and bank and securities market development evolve as countries grow (Koenker and Basset 1978). Ordinary least squares (OLS) regressions provide information on the association between, for example, economic development and bank devel- opment for the average country, defined as a country at the average level of economic development. However, quantile regressions provide information on the relationship between economic activity and bank development at each per- centile of the distribution of economic development. As emphasized throughout the paper, these quantile analyses do not yield a sharp causal interpretation. Rather, they show how the estimated coefficients of the financial development indicators vary at different levels of economic development. In this way, the analyses illustrate how the associations between economic development and both bank and securities market development change during the process of economic development. New data contribute to these analyses of finance and development with the construction of a database that covers 72 countries over the period from 1980 through 2008. For the analyses, the data are aggregated into five-year averages (data permitting), so there are a maximum of six observations per country. In addition to using standard indicators, such as bank credit to the private sector as a share of gross domestic product (GDP), stock market capitalization rela- tive to GDP, and the value of stock market transactions relative to GDP, the analyses employ data on the capitalization of private domestic bond markets relative to GDP (for country averages, please see table S1 in the supplemental appendix, available at http://wber.oxfordjournals.org/). Both banks and securities markets become larger relative to the size of the overall economy as countries develop economically, confirming the results in Demirgu ¸ -Kunt and Levine (2001). These findings hold across various mea- ¨c sures of bank and securities market development, including measures incorpo- rating private domestic bond markets. It is important to note that the measures of bank and securities market development are scaled by GDP. Thus, the find- ings show that the growth of marketable securities and bank loans outpaces the growth of economic activity as countries develop economically. The analyses also indicate that (1) the association between economic activity and bank development decreases with economic development, but (2) the asso- ciation between economic activity and securities market development increases as countries grow. Put differently, as economies develop, the marginal increase in economic activity associated with an increase in bank development falls, ¨c Demirgu ¸ -Kunt, Feyen, and Levine 479 whereas the marginal increase in economic activity associated with an increase in securities market development rises. Although instrumental variables are not employed to identify a causal effect, these results are consistent with the predic- tions emerging from the large body of theoretical research discussed above: as economies develop economically, the services provided by securities markets will become more important for future economic development, whereas those provided by banks will become less important. This research is policy relevant. First, if the optimal mixture of banks and markets changes as an economy develops, such a relationship is an indication of the costs of policy and institutional impediments to the evolution of the financial system. This is the first paper to show that the association between economic activity and stock market development increases as economies grow, whereas the association between economic activity and bank development decreases. Furthermore, this work suggests that the associations between economic activity and both bank and securities market development change with economic development. This change implies that the estimated elasticities from previous research regarding the impact of changes in bank or stock market development on economic development will yield misleading informa- tion about countries with incomes that are far from the sample average. Previous studies do not account for the evolving importance of banks and markets during the process of economic development. The new data, methods, and analyses contribute to a better understanding of the dynamic relationships among economic development, financial institu- tions, and securities markets, but these contributions come with qualifications and limitations. First, although the analyses are policy relevant, direct policy instruments and levers are not studied. Thus, the results suggest that impedi- ments to the evolution of financial systems will hinder economic activity, but they do not provide guidance on exactly which types of policies foster the healthy development of financial systems. Second, although the analyses reduce concerns about reverse causality, they do not specify a particular causal mechanism, nor do they rule out reverse causality or omitted variable bias. A substantial body of theory predicts that as economies develop, financial systems will become more market based and the marginal impact of securities markets on economic activity will increase, whereas that of banks will decrease. Our findings are consistent with these pre- dictions and inconsistent with simple reverse causality scenarios. Although a simple reverse causality scenario might predict that economic development increases the size of banks and securities markets relative to the size of the overall economy and that securities markets grow faster than banks, a simple reverse causality scenario does not yield predictions about the differential change in the association between economic activity and bank and securities market development as economies grow. That is, a simple reverse causality sce- nario does not predict that the association between economic activity and bank development diminishes in magnitude while the association between economic 480 THE WORLD BANK ECONOMIC REVIEW activity and securities market development increases in magnitude as countries develop economically. Although these differential effects might be accounted for by sophisticated reverse causality scenarios, potential omitted variable biases, or as yet unformalized theories of finance and development, this paper provides the first empirical evidence that is consistent with an influential theo- retical body of literature predicting that securities markets become more impor- tant for economic activity and that banks become less important as countries develop economically. DATA AND S U M M A RY S TAT I S T I C S Several measures of bank and stock market development are constructed and used to analyze the relationship between economic activity and the structure of the financial system. Economic theory highlights the advantages of developing indicators reflecting the degree to which banks and markets ameliorate market frictions and thereby (1) improve ex ante information about possible invest- ments; (2) enhance the monitoring of investments after financing occurs; (3) facilitate the trading, diversification, and management of risk; (4) ease the mobilization and pooling of savings; and (5) foster the exchange of goods, services, and financial claims. However, such empirical proxies do not exist for a broad cross-section of countries over the last few decades. Instead, measures of the size and activity of banks and securities markets are compiled and employed to examine the relationship between financial systems and economic development. These measures are constructed over the period from 1980 to 2008, and the primary sources of these indicators are provided in table 1. Private credit is used to measure bank development and equals deposit money bank credit to the domestic private sector as a share of GDP. Private credit isolates the credit issued to the private sector and excludes the credit issued to governments, government agencies, and public enterprises. Private credit also excludes credits issued by central banks. Not surprisingly, there is enormous cross-country variation in private credit. For example, averaging the 1980–2008 period, private credit was less than 10 percent of GDP in Angola, Cambodia, and Yemen, and it was greater than 85 percent of GDP in Austria, China, and the United Kingdom. Table 2 indicates that the annual average value of private credit across countries was 39 percent, with a standard deviation of 36 percent. Stock value traded is used to measure market development and equals the value of the stock market transactions as a share of GDP. This market develop- ment indicator incorporates information on the size and activity of the stock market, not simply the value of the listed shares. Previous work by Levine and Zervos (1998) indicates that the trading of ownership claims on firms in an economy is closely tied to the rate of economic development. There is substan- tial variation across counties. As shown in table 2, although the mean value of the stock value traded is approximately 29 percent of GDP, the standard ¨c Demirgu ¸ -Kunt, Feyen, and Levine 481 T A B L E 1 . Variable Definitions and Sources Name Source Definition Dependent variable and baseline financial sector controls Log real GDP per World Development Logarithm of real GDP per capita capita Indicators (constant 2000 USD). Private credit International Financial Deposit money bank credit to the private Statistics sector as a percentage of GDP. Stock value traded Standard & Poor’s Value of stock market transactions as a percentage of GDP. Stock market Standard & Poor’s The value of listed shares on a country’s capitalization stock exchange as a percentage of GDP. Securities market Standard & Poor’s; Bank Stock market capitalization plus domestic capitalization of International private bond market capitalization as a Settlements percentage of GDP. Standard controls Log initial GDP per World Development Log initial real GDP per capita capita Indicators (constant 2000 USD). Log average years of Barro and Lee (2010) Log (1 þ average years of schooling). schooling Log openness to trade World Development Log sum exports and imports of goods Indicators and services as a percentage of GDP. Log government size World Development Log general government consumption as Indicators a percentage of GDP. Log inflation rate International Financial Log (1 þ annual change of CPI). Statistics T A B L E 2 . Descriptive Statistics Variable Mean Standard deviation Maximum Minimum Dependent variable and baseline controls Log real GDP per capita 7.58 1.57 10.94 4.13 (constant 2000 USD) Private credit 39.28 35.90 319.71 0.00 Stock value traded 28.80 57.44 632.34 0.00 Stock market capitalization 47.70 58.39 561.44 0.00 Securities market capitalization 59.08 71.20 588.27 0.00 Standard controls Log average years of schooling 1.86 0.50 2.65 0.03 Log openness to trade 4.26 0.61 6.12 2 1.18 Log inflation rate 0.15 0.37 5.48 2 0.52 Log government size 2.72 0.43 4.42 0.32 Note: Descriptive statistics are calculated on all available annual data for the 1980– 2008 period. deviation is approximately double this value. In Armenia, Tanzania, and Uruguay, the stock value traded annually averaged less than 0.23 percent over the 1980–2008 sample (10th percentile). In contrast, the stock value traded 482 THE WORLD BANK ECONOMIC REVIEW averaged over 75 percent in Hong Kong, Saudi Arabia, Switzerland, and the United States (90th percentile). This paper’s results are robust to using other market development indicators, such as stock market capitalization, which simply measures the value of the listed shares on a country’s stock exchanges as a share of GDP, and securities market capitalization, which equals the capi- talization of the stock market plus the capitalization of the private domestic bond markets divided by GDP. Log real GDP per capita is used to measure economic activity and equals the logarithm of GDP per capita in constant 2000 US dollars. Consistent with theories guiding the empirical analyses (which are discussed in the introduc- tion), log real GDP per capita is examined rather than GDP per capita growth to obtain estimates of the association between economic activity and both bank and securities market development. With the current specification, the estimated coefficients provide information on how log real GDP per capita changes when, for example, securities market development changes. To assess the independent link between finance and economic development, the regressions control for many country characteristics. Some specifications include standard controls, such as years of schooling, openness to trade, infla- tion, government size, initial GDP per capita of the economy in 1980, and dummy variables for the five-year periods of analysis. The correlations in table 3 highlight the key features of the financial system and economic development. First, both bank development and securities market development are positively correlated with economic development. Second, bank development and securities market development are positively correlated with each other, suggesting that financial development involves both larger banks and larger markets. Although these are simple correlations, these basic patterns hold when controlling for many other national traits. THE RELATION SHI PS A MON G BANKS, MARK ET S, A N D E CO N O M I C DE V E LO PM E N T To assess how the relationships between economic activity and both bank development and stock market development evolve with economic develop- ment, quantile regressions are used with data averaged over non-overlapping five-year periods. OLS provides information on the relationship between log real GDP per capita and financial development for a country at an average level of economic development. However, OLS does not provide information on how the relationship between economic activity and financial development differs for countries at different levels of economic activity. The quantile regressions model the relationship between log real GDP per capita and financial development at the specific percentiles (or quantiles) of the log real GDP per capita. Thus, in a quantile regression of log real GDP per capita on private credit, the procedure is able to yield a different estimated coefficient of private credit for each percentile (or quantile) of log real GDP per T A B L E 3 . Correlations Log real GDP Private Stock value Log average Log openness Log inflation Log government Correlations per capita credit traded years of schooling to trade rate size Private credit 0.67*** 1 Stock value traded 0.41*** 0.51*** 1 Log average years of schooling 0.71*** 0.49*** 0.26*** 1 Log openness to trade 0.25*** 0.21*** 0.08*** 0.31*** 1 Log inflation rate 2 0.15*** 2 0.16*** 2 0.12*** 2 0.03** 2 0.13*** 1 Demirgu Log government size 0.28*** 0.21*** 0.04** 0.25*** 0.28*** 2 0.08*** 1 Source: Author’s analysis based on data described in the text. ¨c¸ -Kunt, Feyen, and Levine Note: Correlations are calculated on all available annual data for the 1980– 2008 period. *, **, and *** denote the significance level of correlation at the 10, 5, and 1 percent levels, respectively. 483 484 THE WORLD BANK ECONOMIC REVIEW capita. For example, the estimated coefficient at the 50th percentile is a median regression, yielding the estimated relationship between the log real GDP per capita and private credit at the median level of economic activity. By comput- ing the quantile regression for each of the 5th to the 95th quantiles, the analyses provide information on how the relationship between economic activity and financial development differs across distinct levels of log real GDP per capita. Neither the OLS nor the quantile regressions identify the causal impact of bank and securities market development on economic development. Rather, the goal is to explore whether and how the relationship between changes in economic activity and changes in both bank and market development varies with the level of economic development. Illustrating the Quantile Regression Results In panel A of figure 1, the graph on the upper left side plots the coefficients from 91 separate quantile regressions for each percentile from the 5th through the 95th percentiles of log real GDP per capita, where the dependent variable is log real GDP per capita, and the main regressor is private credit. The regres- sions control for stock value traded. A circle represents each coefficient esti- mate produced by the quantile regression associated with the corresponding percentile. The left axis provides information on the values of the coefficient es- timates. Thus, the estimated coefficient depicts the sensitivity of log real GDP per capita associated with a change in private credit at each percentile of eco- nomic development. These estimates are statistically significant. Additional in- formation on the sensitivity of these estimates is provided below. The graph also plots the actual values of private credit at each percentile, which are desig- nated with a triangle. The scale of the values of private credit is provided on the right axis. The graphs in the remainder of panel A of figure 1 provide similar information on the relationship between economic activity and stock market development. The upper-right graph provides information for stock value traded. The lower graphs confirm the increasingly relevant role of securi- ties markets by documenting similar upward trends for both securities market capitalization and stock market capitalization. Panel B of figure 1 provides the same types of quantile analyses while con- trolling for other characteristics of the national economies. The standard con- trols are log real GDP per capita in 1980, government size, openness to trade, inflation, average years of schooling, and period fixed effects. Each of the eight graphs in panels A and B of figure 1 provides two addition- al pieces of information. First, the horizontal dotted line is the OLS estimate of the coefficient of the financial development indicator. Thus, in the graph on the upper left side of panel A in figure 1, this line is simply the coefficient of private credit from an OLS regression of log real GDP per capita on private credit for the full sample of country-year observations, controlling for stock value traded. When moving away for the mean log real GDP per capita, the quantile estimates become statistically different from the OLS estimates. The ¨c Demirgu ¸ -Kunt, Feyen, and Levine 485 nature of these deviations is explored below. Second, the solid line shows the estimated linear relationship between each estimated coefficient of the financial development indicator (i.e., the circles) and the GDP per capita percentile asso- ciated with the coefficient. As a specific example, consider the graph in the upper-right quadrant of panel B in figure 1. The estimated coefficients of stock value traded after conditioning on the standard controls and period fixed Figure 1. Quantile Coefficients for Private Credit and Securities Market Activity Source: Author’s analysis based on the data described in the text. Note: The dependent variable is log real GDP per capita. The figure depicts the coefficients of the quantile regressions of private credit, stock value traded, securities market capitalization, and stock market capitalization as independent variables for each of the 5th to 95th percentiles of the GDP per capita distribution on the left axis. Private credit is defined as deposit money bank credit to the private sector as a percentage of GDP. Stock value traded is the value of stock value transactions as a percentage of GDP. Stock market capitalization is the value of listed shares on a country’s stock exchanges as a percentage of GDP. Securities market capitalization is defined as stock market capitalization plus sdmestic private bond market capitalization as a percentage of GDP. Percentile values are reported on the right axis. Data are five-year non-overlapping country averages. Panel A does not control for additional variables. Panel B controls for standard controls: initial GDP per capita, government size, openness to trade, inflation, average years of schooling, and time-fixed effects. The horizontal dotted line depicts the OLS estimate. The solid lines represent the linear fit. 486 THE WORLD BANK ECONOMIC REVIEW Figure 1. Continued effects are collected. Then, these estimated coefficients are regressed on the GDP per capita percentile associated with the estimates. Panel A in table 4, column (4) provides the results of this regression. The estimated coefficient for each GDP per capita percentile provides the trend line graphed in figure 1. Discussing the Quantile Regression Results In terms of bank development, figure 1 shows that as the log real GDP per capita rises, (1) private credit rises (triangles) and (2) the marginal increase in the log real GDP per capita associated with an increase in private credit dimin- ishes (circles). Put differently, the level of bank development increases, but its association with economic activity diminishes. In panel A of table 4, the signifi- cance of this association is tested formally. In this panel, the dependent vari- able is the estimated linear association between economic activity and either bank development or securities market development at each percentile of the distribution of GDP per capita underlying figure 1. Regressions (1) and (2) show that this relationship is statistically significant: as economic activity T A B L E 4 . Robust Regression Results of the Linear Regression Fits from Figure 1 Panel A: Linear model Dep. Var.: Percentile regression coefficient Private credit Dep. Var.: Percentile regression coefficient Stock value traded 1 (No controls) 2 (With all controls) 3 (No controls) 4 (With all controls) Percentile 2 1.24E 2 04*** 2 1.02E 2 05*** 4.18E 2 05*** 3.79E 2 05*** [0.00] [0.00] [0.00] [0.00] Constant 2.51E 2 02*** 4.45E 2 03*** 2.05E 2 03*** 2 1.34E 2 03 [0.00] [0.00] [0.00] [0.00] Controls No Yes No Yes Observations 91 91 91 91 Panel B: Quadratic model Percentile 2 1.09E 2 04*** 2.67E 2 05*** 2.13E 2 04*** 4.88E 2 05*** [0.00] [0.00] [0.00] [0.00] Square of percentile 2 1.37E 2 07 2 3.69E 2 07*** 2 1.63E 2 06*** 2 1.02E 2 07** [0.00] [0.00] [0.00] [0.00] Constant 2.48E 2 02*** 3.77E 2 03*** 2 1.44E 2 03*** 2 1.58E 2 03 Demirgu [0.00] [0.00] [0.00] [0.00] Controls No Yes No Yes Observations 91 91 91 91 ¨c¸ -Kunt, Feyen, and Levine Source: Author’s analysis based on data described in the text. Note: The table displays the robust regressions results of the linear fits depicted in figure 1. The dependent variables are coefficients of the quantile regres- sions of private credit and stock value traded for each of the 5th to 95th percentiles of the GDP per capita distribution, respectively, on five-year non- overlapping country averages. Panel A reports a linear model where the regressors are a constant and the income percentile associated with the coefficient. Panel B shows the results for the quadratic model using the same independent variables in the linear model plus the square of the percentile associated with the coefficient. Columns 1 and 3 use the coefficients of quantile regressions without additional controls ( panel A of figure 1). Columns 2 and 4 use the coeffi- cients of quantile regressions that include standard controls: Initial GDP per capita, Government size, Openness to trade, Inflation, Average years of school- ing, and time-fixed effects ( panel B of figure 1). The p-values in brackets are based on robust country-level clustered standard errors. *, **, and *** denote the significance on the 10, 5, and 1 percent levels, respectively. 487 488 THE WORLD BANK ECONOMIC REVIEW increases, there is a significant reduction in the estimated coefficient of private credit. The results are the opposite for securities market development. As log real GDP per capita rises, (1) stock value traded rises, and (2) the marginal increase in log real GDP per capita associated with an increase in stock value traded rises. That is, as countries develop economically, securities market development increases, and its association with economic activity increases. Regressions (3) and (4) in panel A of table 4 show that this effect is statistically significant: the association between economic activity and stock value traded increases as the log real GDP per capita rises. These results suggest that the relationship between bank development and economic activity differs from the relationship between securities market development and economic activity. Figure 1 suggests that there might be a nonlinear relationship (1) between economic activity and bank development and (2) between economic activity and securities market development. To assess the sensitivity of our findings and provide more information on the nature of the relationship, we examine these relationships more rigorously in panel B of table 4, which includes a quadratic term to allow for a potential nonlinear, parabolic relationship. This potential relationship makes it possible to estimate the level of economic activity at which the associations between financial and economic development start to decrease as the economy develops further. Consistent with figure 1, the regression results in panel B of table 4 suggest that there are a nonlinear associations between economic activity and both bank and securities market development. At very low levels of economic devel- opment, the association between economic activity and bank development is increasing in economic development, but the slope quickly becomes negative. In particular, regressions (1) and (2) of panel B indicate that the slope of the association between economic activity and bank development becomes negative after real GDP per capita reaches $1,032 in 2000 US dollars (36th percentile). For securities market development, panel B of table 4 indicates that the associa- tion between economic activity and securities is always increasing in the 5th to 95th percentile interval of economic development, but at a decreasing rate (regressions (3) and (4)). In other words, only the upward sloping part of the estimated parabola is relevant. For example, the regression (4) estimates suggest that as economies grow and move to the next percentile, the coefficient increases by more than 5 percent for countries below the 20th percentile. In contrast, the coefficient increases by just 0.7–1.0 percent for each additional percentile from the 78th percentile upward. Broader Implications of Quantile Analyses The results of the above analysis are consistent with several lines of theoretical research on the evolving importance of banks and financial markets during the process of economic development. As noted in the introduction of this paper, Allen and Gale (2000), Boot and Thakor (1997, 2000), Boyd and Smith ¨c Demirgu ¸ -Kunt, Feyen, and Levine 489 (1998), Song and Thakor (forthcoming), and others stress that at higher levels of economic development, economies require the types of custom-designed fi- nancial arrangements that ease the financing of novel, longer-term investments that often employ more intangible inputs than the types of projects that domi- nate economic activity at lower levels of economic development. These theories predict that securities markets are comparatively better than banks at financing these activities. Thus, influential lines of theoretical analysis predict that the services provided by securities markets will become more important for foster- ing economic activity as economies grow, whereas those provided by banks will tend to become less important. The quantile regression results are consis- tent with these predictions. However, the quantile regression results are incon- sistent with the view that economic development is simply associated with an increase in bank and stock market development with no differential effect on their association with economic activity. That is, as countries develop economi- cally, the association between economic activity and bank development tends to weaken, whereas the association with securities market development tends to strengthen. CONCLUSIONS Banks and markets evolve during the process of economic development. As economies grow, both the banking system and financial markets become more developed, but the association between economic activity and bank develop- ment tends to fall, and the association between economic activity and securities market development tends to increase. For the first time, the quantile analyses employed in this paper directly assess the predictions emerging from an influen- tial line of theoretical work on financial structure. The findings are consistent with the view that (a) financial institutions provide different financial services from those provided by financial markets, and (b) as economies grow, the services provided by securities markets become more important for promoting economic activity, whereas those provided by banks become less important. As such, this research suggests that policies and institutions that impede the evolu- tion of the structure of financial systems as economies grow can have detrimen- tal ramifications for economic development. REFERENCES Acemoglu, D., and F. Zilibotti. 1997. “Was Prometheus Unbound by Chance? Risk, Diversification, and Growth.” Journal of Political Economy 105: 709–775. Allen, F., and D. Gale. 1995. “A Welfare Comparison of the German and U.S. Financial Systems.” European Economic Review 39: 179–209. ———. 1997. “Financial Markets, Intermediaries, and Intertemporal Smoothing.” Journal of Political Economy 105: 523– 546. 490 THE WORLD BANK ECONOMIC REVIEW ———. 1999. “Diversity of Opinion and Financing of New Technologies.” European Economic Review 39: 179–209. ———. 2000. Comparing Financial Systems. Cambridge, MA: MIT Press. Beck, T., and R. Levine. 2002. “Industry Growth and Capital Accumulation: Does Having a Market- or Bank-based System Matter?” Journal of Financial Economics 64: 147–180. ———. 2004. “Stock Markets, Banks, and Growth: Panel Evidence.” Journal of Banking and Finance 28: 423–442. Boot, A. W. A., and A. Thakor. 1997. “Financial System Architecture.” Review of Financial Studies 10: 693 –733. ———. 2000. “Can Relationship Banking Survive Competition?” Journal of Finance 55: 679– 713. Boyd, J. H., R. Levine, and B. D. Smith. 2001. “The Impact of Inflation on Financial Sector Performance.” Journal of Monetary Economics 47: 221–248. Boyd, J. H., and B. D. Smith. 1998. “The Evolution of Debt and Equity Markets in Economic Development.” Economic Theory 12: 519– 560. ¨c Demirgu ¸ -Kunt, A., and R. Levine. 2001. Financial Structures and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. Cambridge, MA: MIT Press. ¨c Demirgu ¸ -Kunt, A., and V. Maksimovic. 1998. “Law, Finance, and Firm Growth.” Journal of Finance 53: 2107– 2137. ———. 2002. “Funding Growth in Bank-based and Market-based Financial Systems: Evidence from Firm Level Data.” Journal of Financial Economics 65: 337–363. Dewatripont, M., and E. Maskin. 1995. “Credit Efficiency in Centralized and Decentralized Economies.” Review of Economic Studies 62: 541 –555. Goldsmith, R. W. 1969. Financial Structure and Development. New Haven, CT: Yale University Press. Holmstrom, B., and J. Tirole. 1993. “Market Liquidity and Performance Monitoring,” Journal of Political Economy 101: 678– 709. Koenker, T., and G. Basset. 1978. “Quantile Regressions.” Econometrica 46(1): 33 –50. Levine, R. 2002. “Bank-based or Market-based Financial Systems: Which Is Better?” Journal of Financial Intermediation 11: 398– 428. Levine, R., and S. Zervos. 1998. “Stock Markets, Banks, and Economic Growth.” American Economic Review 88: 537– 558. Morck, R., and M. Nakamura. 1999. “Banks and Corporate Control in Japan.” Journal of Finance 54: 319 –340. Rajan, R. G. 1992. “Insiders and Outsiders: The Choice between Informed and Arm’s Length Debt.” Journal of Finance 47: 1367– 1400. Song, F., and A.V. Thakor. Forthcoming. “Financial System Architecture and the Co-evolution of Bank and Capital Markets.” Economic Journal. Weinstein, D. E., and Y. Yafeh. 1998. “On the Costs of a Bank-centered Financial System: Evidence from the Changing Main Bank Relations in Japan.” Journal of Finance 53: 635 –672.