'Finance And Growth' Re-Loaded

We assess the relationship between finance and growth over the period 1980-2014. We estimate a cross-country growth regression for 48 countries during 20 periods of 15 years starting in 1980 (to 1995) and ending in 1999 (to 2014). We use OLS and IV estimations and we find that: 1) overall financial development had a positive effect on economic growth during all periods of our sample, i.e., we confirm that from 1980 to 2014 financial services provided by the various financial systems were significant (to various degrees) for firm creation, industrial expansion and economic growth; but that, 2) the structure of financial markets was particularly relevant for economic growth until the financial crisis; while 3) the structure of the banking sector played a major role since; and finally that, 4) the legal system is the primary determinant of the effectiveness of the overall financial system in facilitating innovation and growth in (almost) all of our sample period. Hence, overall our results suggest that the relationship between finance and growth matters but also that it varies over time in strength and in sector origination.


INTRODUCTION
We empirically assess the relationship between financial structure and economic growth for 48 countries over the period from 1980 to 2014. There are two theoretical approaches to explain the impact of financial structure on economic growth: the bank-based and the marketbased ones. The first highlights that bank-based systems are better at mobilizing savings, identifying good investments, and exerting sound corporate control, particularly during early stages of economic development and in weak institutional environments. On the other hand, the market-based view emphasizes the advantages of markets in allocating capital, providing risk-management tools, and mitigating the problems associated with excessively powerful banks (Levine, 2002).
Those approaches stand in pointed contrast: the bank-based approach argues that bank activity is better than stock market development for economic growth, because: 1) welldeveloped markets quickly reveal information in public markets, which reduces the incentives for individual investors to acquire information and for identifying innovative projects; 2) in liquid markets investors can inexpensively sell their shares, so they have fewer incentives to monitor managers rigorously; and 3) greater market development may hinder corporate control and national productivity (Beck et al, 2001).
Meanwhile, the proponents of the market-based view stress the deficiencies of banking systems: 1) by acquiring expensive information about enterprises that can extract large rents from firms, and reducing the incentives to undertake high-risk and high-return projects; 2) banks have an inherent bias toward low risk, and low return project, and that may retard innovation and growth; 3) powerful banks may collude with firm managers against other investors, which stymies competition, effective corporate control, the emergence of new firms, and economic growth (Beck et al, 2001).
We wade into this debate by comprehensively revisiting the assessment of the relationship between finance and growth over the period 1980-2014. We estimate a cross-country growth regression for 48 countries during 20 periods of 15 years in duration starting in 1980 (to 1995) and ending in 1999 (to 2014). We find that overall financial development had a positive effect on economic growth during (almost) all periods of our sample (from 1980 to 2014). When we consider the activity of the financial system -i.e., the ratio of Total Value Traded and Bank Credit -the market-based structure was relevant for economic growth during eleven periods (from 1984 to 2009), while the development of the bank-based financial system was strongly relevant for economic activity only during the last four periods in our sample (from 1996 to 2014). Therefore, our results suggest that the relationship between finance and growth varies (rather dramatically) over time in strength and in sector origination.
The rest of the paper proceeds as follows. Section I reviews the literature, Section II introduces the estimated models and data. Section III presents the estimates. Section IV concludes.

I. Literature Review
We are obviously not the first to study the link between financial structure and economic growth. Goldsmith (1969) for example analyzes whether the mixture of markets and intermediaries operating in a country influence economic growth. 1 He shows that banks tend to become larger relative to national output as countries develop; he also presents evidence suggesting that stock markets and nonbank financial intermediaries frequently grow relative to banks in size and importance as countries expand economically; and he documents a positive correlation between overall financial development and the level of economic activity for 35 countries, using data from 1860 to 1963. However he is unable to provide much in terms of cross-country evidence due to then prevailing data limitations (Demirgüç-Kunt and Levine, 2001).
For over a century, the relationship between financial structure and economic growth primarily involved country studies of Germany and Japan as bank-based systems and the United States and the United Kingdom as market-based systems, but the four countries have very similar long-run growth rates, so it was difficult to correlate differences in financial structure with differences in long-run growth rates (Beck et al., 2000). 2 That changed significantly in 2000, with the New Database on the Structure and Development of Financial Sector constructed by Beck, Demirgüç-Kunt and Levine. With more specific information of financial systems over the world, it was possible to improve the cross-country analysis between finance and economic growth with a wide variety of indicators that measure the size, activity and efficiency of financial intermediaries and markets. For example, Demirgüç-Kunt and Levine (2001) analyzes this data and observes that: 1) banks, nonbanks, stock markets, and bond markets are larger, more active, and more efficient in richer countries; 2) financial systems, on average, are more developed in richer countries; 3) in higher-income countries, stock markets tend to become more active and efficient relative to banks; and 4) there is a general tendency for a national financial systems to become more market-oriented as they become richer.
With the motivation to contribute to the long debate of financial structure and growth, Beck et al. (2001) use different datasets and econometric methodologies to assess the relationship between financial structure and economic development. They extend three papers on financial structure (Levine, 2002;Beck and Levine, 2001;Demirgüç-Kunt and Maksimovic, 2002).
They examine the bank-based view, the market-based view, and also add two theoretical approaches: the law and finance view, which stresses that the legal system is the primary determinant of the effectiveness of the overall financial system in facilitating innovation and growth; and the financial services view, which argue that overall financial services provided by financial systems are crucial for firm creation, industrial expansion and economic growth.
In both cases, it is not necessary the distinction between bank-based or market-based financial structure and those approaches deny that financial structure exert influence on growth.
First, they estimate standard cross-country growth regressions to answer the question: "Do countries with bank-based financial systems grow faster than countries with market-based systems, or is financial structure unrelated to the pace of economic development?" They explore the impact of financial structure on long-run economic growth in a sample of fortyeight countries, with data averaged over the period 1980-1995. They run an Ordinary Less Square (OLS) and Instrumental Variables (IV) regressions. Second, they explore the impact of financial development and financial structure on industry growth and new firm creation based on an extension of the Rajan and Zingales (2001) model.
They use a country-industry panel of 34 countries and 36 industries to test whether industries that depend more heavily on external finance grow faster in market-or bank-based financial systems, or whether it is the overall level of financial development that is critical in accounting for cross-country differences in industrial growth patterns. They also run OLS and IV models.
Third, they use firm-level data from a panel of 33 countries and six years between 1990 and 1995 to explore whether firms' access to external finance varies across financial systems with different structures, or whether the overall level of initial development and legal system determine firms' access to external finance. They run an IV regression on this account.
After comparing the three methodologies, Beck et al. (2001) conclude that financial structure (the difference between bank-based and market-based systems) is not an analytically useful way to distinguish the relationship between finance and economic growth, and they argue that the law and finance and the financial services views are more useful to understand the impact of finance on economic growth. The first view stresses that the legal system is the primary determinant of the effectiveness of the financial system in facilitating innovation and growth. Work in this area uses indicators to examines legal rules covering the protection of corporate shareholders (anti-director) and creditors (creditor), the origin of these rules (legal origin), and the quality of their enforcement (rule of law).
La Porta et al. (1998) show that national legal origin strongly influences the legal and regulatory environment governing financial sector transactions. They explain that, in general, commercial laws come from two broad traditions: common law, which is English in origin, and civil law, which derives from Roman law. Within the civil tradition, there are three major families that modern commercial laws originate from: French, German and Scandinavian.
They also argue that the quality of law enforcement is the highest in Scandinavian and German civil law countries, next highest in common law countries and the lowest in French civil law countries. Finally, they suggest that the French and the German civil traditions as well as the common law tradition (British), have spread around the world through a combination of conquest, imperialism, outright borrowing and more subtle imitation. Since legal origin explains cross-country differences in financial intermediary development and since legal origin is (reasonably) exogenous, Levine et al. (2000) use this set as instrumental variables to control for the simultaneity bias.
Creditor is an index of the degree to which the legal codes of the country protect the claims of secured creditors in the case of reorganization or liquidation of a company 3 , and it ranges from 0 to 4. The authors explain that for higher values of creditor, outside investors have more rights relative to the management and other stakeholders, and should therefore be more willing to provide the external resources that firms need. Anti-director is an index of the degree to which the legal codes of the country protect minority shareholder rights, and it ranges from 0 to 6. In economies with higher values of anti-director, minority shareholders are better protected against expropriation by management and large shareholders and should therefore be more willing to provide external financing to firms. Finally, Rule of law is an assessment of the law and order tradition of a country that ranges from 10, strong law and order tradition, to 1, weak law and order tradition. In countries with a higher law and order tradition, outside investors can more easily enforce their claims and rights and should therefore be more willing to provide external finance.
Moreover, Beck et al. (2001) argue that the financial-services view consider key financial services that are crucial for firm creation, industrial expansion and economic growth. In this approach, the division between banks and markets in providing these services, is of secondary importance. Thus, the financial services view predicts that overall financial development is important for economic development, but also that financial structure per se will not add much to our understanding of the process of economic development. 3 Creditor is the sum of four dummy variables that indicate whether: 1) the reorganization procedure does not impose an automatic stay on assets, thereby not preventing secured creditors from taking possession of loan collateral, 2) secured creditors are ranked first in the case of liquidation, 3) management does not stay in charge of the firm during reorganization, thereby enhancing creditors' power, and 4) management needs creditors' consent when filing for reorganization. Anti-director is the sum of six dummy variables that indicate whether: 1) shareholders are allowed to mail their proxy vote to the firm, 2) shareholders are not required to deposit their shares prior to the General Shareholders' Meeting, 3) cumulative voting or proportional representation of minorities on the board of directors is allowed, 4) an oppressed minority mechanism is in place, 5) the minimum percentage of share capital that entitles a shareholder to call for an Extraordinary Shareholders' Meeting is less than or equal to 10 percent, and 6) shareholders have preemptive rights that can only be waived by a shareholders' vote. Rule of law use the measure constructed by International Country Risk Guide (ICRG) and is an average over the period (1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995) (Beck et al., 2001, 203). However other authors have obtained different results. Tadesse (2002) takes industry-level data from a panel of 36 countries and 10 industries over the period 1980-1995 and estimated an OLS grow equation to examines whether real economic performance varies across countries of differing financial architecture (bank-based or market-based). He finds that across countries with developed financial sectors, industries supported by market-based financial systems grow faster than industries with bank-based systems. Conversely, bankbased financial systems significantly outperform market-based systems across countries with underdeveloped financial sectors. Furthermore, he showed that market-oriented systems retard economic growth, and conversely, bank-oriented systems promote growth in countries dominated by smaller firms.
Other studies have been focusing only on the effect of banking structure on economic growth, in particular on banking concentration. For example, Cetorelli and Gambera (2001) provide evidence of a positive relationship between the level of development of the banking sector of an economy and its long-run output growth. In particular, they show that bank concentration promotes growth of those industrial sectors that are more in need of external finance by facilitating credit access to younger firms. However, they also find a general dead weight loss that depresses economic growth, which impacts all sectors and all firms indiscriminately. They base themselves on Rajan and Zingales (2001) to estimate a OLS and Instrumental Variables (IV) using a sample of 41 countries and, for each of them, 36 industries over the period 1980-1990. Other researchers focus on the nonlinear relationship between financial structure and growth.
Ergungor (2008) highlights that there is a nonlinear (contingent) relationship between financial structure and growth and posits that "countries that have an inflexible judicial system grow faster when have a more bank-oriented financial system". He uses a sample of 46 countries from 1980 to 1995 and runs a growth equation with IV regression.
Recently, different papers have emphasized the effect of financial structure on the firms' size. Beck and Singer (2013) combine two data sets and explore the relationship between financial structure and firm's access to financial services in a sample of eighty-nine countries during the period 1995-2008. They highlight two important aspects: 1) dominance of the financial system by banks is associated with lower use of financial services by firms of all sizes, while low-end financial institutions and specialized lenders seem particularly suited to ease access to finance in low-income countries; and 2) there is no evidence that smaller institutions are better in providing access to finance. Focusing also in the size of the firms, Kim, Lin and Chen (2016) adapt the methodology of Beck el al. (2008) and use data for fortytwo countries and for each 33 industries over the period 1980-1990. They find evidence that industries dominated by small firms grow faster in a country with a more bank-based financial system. They also find that the effect of financial structure on industry growth runs mostly through growth in the number of establishments rather than through growth in the average size of establishments. Table 1 summarizes our empirical review.
Our paper represents an effort to contribute to the intense debate on whether financial structure promotes economic growth. We have concentrated on the methodology used by Beck et al (2001) to test the impact of financial structure on economic growth but we consider a larger period, over 1980-2014, and we used 20 event windows to test the hypothesis that financial structure matters for economic growth with OLS and IV estimates.

A. Estimated Models
We assess the impact of financial structure on economic growth in a sample of 48 countries with data averaged over the period 1980-2014. Using OLS, we estimate a modified growth regression used in Beck et al (2001) to test the validity of the bank-based or market-based approach as follow: where ℎ is the average annual growth rate of real per capita GDP, is a set of potential growth determinants, is a set of financial development indicators, is a set of financial structure indicators, and is the error term. Each financial structure predicts different signs for and . The bank-based approach predicts that bank-based systems are better for growth, implying 0 and 0; in that case, the overall financial system and the banking activity have a positive effect on economic growth. The market-based view expects that market-based financial systems grow faster, implying 0 and 0; when both, the overall financial system and the stock market, have a positive impact on economic growth. Then, when 0 and is not significant, the financial structure is not relevant to explain economic growth and the overall financial system have a positive impact on growth.
And finally, when both and are not significant, neither the financial system nor the financial structure have an effect on economic growth.
We estimate equation (1) 1981-1996, 1982-1997, 1983-1998… 1999-2014. development as control variables as follows: ℎ 2 include the set of potential growth determinants and the indicators, is the set of financial structure indicators and is the error term. Now we focus on signs of and the analysis is very similar. The market-based view expects that 0 and the bank-based view predicts 0. We run equation (2) for the 20 periods using OLS.
We also use instrumental variable (IV) estimations with Two-Stage Least-Squares (2SLS).
We consider legal origin (dummies), creditor, anti-director and rule of law as instrumental variables for financial development to extract the component of finance that is defined by the legal system. We examine the appropriateness of the instruments with Wooldridge's (1995) robust score test of overidentifying restrictions. The null hypothesis is that our instruments are valid. Beck et al. (2001) argue that we can interpret this result as indicating that the instruments affect real per capita GDP growth only through the financial development or structure indicators.

B. Data
We assess the impact of financial structure on economic growth in a sample of 48 countries took from different sources. All data definitions and sources are given in Table 2 4 Private Credit is the most comprehensive indicator of the activity of financial intermediaries as it includes both bank and nonbank intermediaries. Value Traded measures the activity of the stock market trading volume as a share of national output and thus indicates the degree of liquidity that stock markets provide to economic agents. Market Capitalization is defined as the value of listed shares divided by GDP and is a measure of the size of stock markets relative to the economy. Large Overhead Cost may reflect inefficiencies in the banking system and therefore proxy as a negative indicator of banking-sector inefficiency (Beck et al, 2001(Beck et al, , p. 196-1996. 5 Bank Credit equals the claims of the banking sector on the private sector as a share of GDP. This is focus on the commercial banking sector (Beck et al., 2001(Beck et al., , p. 1997.

3) Structure Efficiency is defined as the log of the product of Overhead Cost and Value
Traded and indicates the efficiency of the stock market relative to the banking sector; 4) Structure Aggregate combines the previous three measures and is thus a conglomerate indicator of the size, activity and efficiency of stock markets relative to banks. It is the first principal component of structure activity, structure size and structure efficiency; 5) Structure Dummy, is a bivariate classification of stock market versus bank-based financial systems. Is equals 1 if structure aggregate is greater than the sample median and 0 otherwise.
Appendix A shows the average of financial development and financial structure indicators from 1980 to 2014. We observe that Switzerland is the country with the highest level of financial activity with 9.81 followed by United States with 9.75 and Japan with 9.15 points.
The lowest levels of financial activity are in Honduras, Ghana and Ecuador with 0, 1.13 and 1.81 points each. For the Structure Activity indicator we observe that the United States, Zimbabwe, Turkey and Pakistan have a more stock market activity than banking activity, and that therefore the indicators are positive. On the other hand, the lowest levels of Structure Activity indicate that in some countries banking activity is larger than stock market activity.
While the classification of some countries is intuitively attractive, such as the United States being classified as market-based, the indicator also classifies Turkey and Zimbabwe as market-based; but, this classification is due to a low value of Bank Credit, rather than to a high level of Total Value of Equity Shares Traded (Beck et al., 2001).
Our set of potential growth determinants for each country is presented in Appendix B. We consider the initial GDP per capita to control for convergence; years of total schooling to control for the effect of human capital accumulation; the average rate of inflation as macroeconomic stability; the government size as share of GDP to proxy for government intrusion; the average black market premium and the exports plus imports as share of GDP to capture the degree of openness of economies. We use creditor, anti-director, rule of law and three dummies for legal origin (British, French, and German relative to Scandinavian origin) as instruments. The Appendix C shows the mean statistics of our cross-country data: mean, standard deviation, minimum., maximum, and the number of observations.
The results suggest that the magnitude and significance of financial development coefficients (size, efficiency, aggregate and dummy) also decrease during the last periods ( Figure 1). The results only indicate that the overall financial sector development has a positive effect on long-run economic growth over the first ten periods (I-X).
The results in Table 3 are not only significant, but also economically important. To explain the economic view of our results we use the Development Activity coefficient of 0.5673 in period I (1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995) XI (1990-2005) until XX (1990-2014, because the sign of Structure Activity indicator is negative during these periods, however, the coefficient is not significant. Then we observe that Structure Size indicators are significantly related with economic growth over eight different periods (IX, X, XI, XII, XV, XVI, XVIII, XIX and XX). But now the coefficient is negative. That result suggest that considering the size of financial institutions commercial banks dominate financial systems and that banking activity is significantly related with economic growth over these eight periods. When we consider efficiency, we find that market-based systems are significant in two periods (II and III). And considering Structure Dummy (equals 1 if the first principal of structure activity, structure size and structure efficiency is greater than the sample median and 0 otherwise) we find three periods (X, XI and XII) where bank-based systems are significant for economic growth.
Those results are too weak to conclude that financial structure matters for economic growth. Figure 2 shows the change of the coefficients across the time.
Third, we estimate the equation (2) with OLS again calculating heteroskedasticity-consistent standard errors and we also consider our Financial Development indicators as control variables. Table 5 show the results which indicates that bank-based financial systems dominate in our 48 countries sample over all 20 periods, because the sign of almost all the coefficients is negative. Figure 3 shows that coefficients on the last periods are more negative and banking activity is then higher (than stock market activity). We also find that the relationship between financial structure and economic growth was not significant during the first periods of the sample (I: 1980(I: -1995(I: and VIII: 1989(I: -2014. However, that relationship increases and become stronger from period IX (1988( -2003( ) until period XIV (1994( -2009.
Then, the results over the periods XV (1995-2010)-XX (1999-2014 show that bank-based financial system is relevant for economic growth. Then, Table 6 and Figure 4 present the IV coefficients for equation (1) with the instruments mentioned before. We confirm that financial development is positively correlated with longrun economic growth over (almost) all periods of our sample (since 1980-1995 until 1999-2014). Only in two regressions (of 100), 6 our coefficients were not statistically significant (numbers without asterisk). We also confirm that instruments affect real per capita GDP growth only through the financial development indicators (coefficients in bold). Table 7 and Figure 5 show that financial structure was not highly correlated with economic growth, but we find that stock-market activity (positive coefficients in blue) prevail over banking activity (Bank Credit) in eleven periods (since 1984-1999 until 1994-2009). These results change when we consider financial development as a control variable (Table 8 and Figure 6). In that case, banking activity (negative coefficients in blue) prevail over stock-6 Each coefficient represents one regression using real GDP growth as dependent variable and considering the policy conditioning information set: logarithm of initial income, schooling, inflation, black market premium, government size, and trade openness. market activity, but only it becomes stronger only during the last four periods of our sample data (from 1996-2011 to 1999). Those results are coherent, because at the same time that stock market prevail (in Table 7) banking activity becomes less significant (in Table 8), in particular considering the years of the financial crisis (2007, 2008 and 2009); and, at the end, becomes stronger in the last four periods.
We also find that creditor, anti-director, rule of law and three dummies for legal origin (British, French, and German relative to Scandinavian origin) were good instruments for the financial structure indicators. Only in eight regressions (of in total 500 regressions) we reject the null hypothesis that our instruments are valid (in Table 8).

Controlling for Development Dummy
Notes: all regressions include the policy conditioning information set: logarithm of initial income, schooling, inflation, black market premium, government size, and trade openness.