)VfS 3 IDJ POLICY RESEARCH WORKING PAPER 3 041 Bank Concentration and Crises Thorsten Beck Aslh Demirguc-Kunt Ross Levine The World Bank Development Research Group Finance May 2003 POLICY RESEARCH WORKING PAPER 3041 Abstract Beck, Demsrguc-Kunt, and Levine study the impact of concentrated banking systems, (2) in countries with bank concentration, regulations, and national institutions fewer regulatory restrictions on bank competition and on the likelihood of suffering a systemic banking crisis. activities, and (3) in economies with better institutions, Using data on 79 countries over the period 1980-97, that is, institutions that encourage competition and they find that crises are less likely (1) in more support private property rights. This paper-a product of Finance, Development Research Group-is part of a larger effort in the group to understand the impact of bank concentration and competition. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Kari Labrie, room MC3-456, telephone 202-473-1001, fax 202-522-1155, email address klabrie@worldbank.org. Policy Research Working Papers are also posted on the Web at http:// econ.worldbank.org. The authors may be contacted at tbeck@worldbank.org or ademirguckunt@worldbank.org. May 2003. (41 pages) The Polcy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues An objective of the series is to get the findtings out quickly, even if the presentations are less than fully polished. The papers carry the nanies of the authors and should be cited accordingly. The findings, interpretations, and conclusionis expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Research Advisory Staff Bank Concentration and Crises Thorsten Beck, Asli Demirguc-Kunt and Ross Levine Keywords: Banking Crises, Bank Concentration and Competition JEL Classification: E44, G21 Beck and Demirgti,-Kunt: World Bank; Levine: Carlson School of Management, University of Minnesota and the NBER. We thank Jerry Caprio and Patrick Honohan for comments and Paramjit K. Gill for outstanding research assistance. I. Introduction The consolidation of financial intermediaries around the globe is fueling an active public policy debate on the impact of bank consolidation on financial stability.' Unfortunately, economic theory provides conflicting predictions about the relationship between banking structure and bank fragility. This paper investigates empirically the impact of banking structure on financial stability. Some theoretical arguments and country comparisons suggest that a less concentrated banking sector with many small banks is more prone to financial crises than a concentrated banking sector with a few large banks as discussed in Allen and Gale (2000, 2003). First, proponents of the "concentration-stability" view hold that large banks can diversify better so that banking systems characterized by a few large banks will be less fragile than banking systems with many small banks.2 Second, concentrated banking systems may enhance profits and therefore lower bank fragility. High profits provide a "buffer" against adverse shocks and increase the franchise value of the bank, reducing incentives for bank owners to take excessive risk (Hellmann, Murdoch, and Stiglitz, 2000). Third, some hold that a few large banks are easier to monitor than many small banks, so that corporate control of banks will be more effective and the risks of contagion less pronounced in a concentrated banking system (Allen and Gale, 2000). The U.S., with its large number of small banks, supports this "concentration-stability" view since it has had a history of much greater financial instability than the U.K or Canada, where the banking sector is dominated by a few large banks. ' See Group of Ten (2001), Bank for International Settlements (2001), International Monetary Fund (2001), and Boyd and Graham (1998, 1991). 2Models by Diamond (1984), Ramakrishnan and Thakor (1984), Boyd and Prescott (1986), Williamson (1986), Allen (1990), and others predict economies of scale in intermediation. 2 An opposing view is that a more concentrated banking structure enhances bank fragility. First, advocates of the "concentration-fragility" view note that large banks frequently receive greater net subsidies than small banks through implicit "too big to fail" policies. 3 This greater subsidy for large banks may in turn intensify risk-taking incentives, increasing the fragility of concentrated banking systems (Boyd and Runkle, 1993 and Mishkin, 1999).4 Second, proponents of the concentration-fragility view would disagree with the proposition that a few large banks are easier to monitor than many small banks. If size is positively correlated with complexity, then large banks may be more opaque than small banks, which would tend to produce a positive relationship between concentration and fragility. Finally, Boyd and De Nicolo (2003) stress that banks with greater market power tend to charge higher interest rates to firms, which induces firms to assume greater risk. To the extent that the concentration is positively associated with banks enjoying greater market power, the Boyd and De Nicolo (2003) model predicts a positive relationship between concentration and bank fragility. Despite the importance of the topic for policymakers and the different theoretical predictions regarding the impact of banking structure on stability, cross-country empirical analysis of the relationship between bank structure and fragility is surprisingly limited. For the United States, Boyd and Runkle (1993) examine 122 bank holding companies. They find that there is an inverse relationship between size and the volatility of asset returns, but no evidence that large banks are less likely to fail. In fact they observe that large banks failed somewhat 3 Even in the absence of deposit insurance, banks are prone to excessive risk-taking due to limited liability for their equity holders and to their high leverage (Stiglitz, 1972). 4 There is a substantial literature that deals with deposit insurance and its effect on bank decisions. According to this literature -that includes Merton (1977), Sharpe (1978), Flannery (1989), Kane (1989), and Chan, Greenbaum and Thakor (1992) - mispriced deposit insurance produces an incentive for insured banks to take risk. If the regulatory treatment were the same for insured banks of all sizes, these models would predict no relationship between bank size and riskiness. Since regulators fear potential macroeconomic consequences of large bank failures, most countries have implicit "too large to fail" policies which protect all liabilities of very large banks whether they are insured or 3 more often in the 1971-90 period. They explain this result by showing that larger banks are more leveraged and less profitable in terms of asset returns. Although there is a growing cross-country empirical literature on banking crises, this literature does not address the issue of banking structure. Earlier work has mostly focused on identifying (i) the macroeconomic determinants of banking crises (Caprio and Klingebiel, 1996; Demirguc-Kunt and Detragiache, 1998), (ii) the relationship between banking and currency crises (Kaminsky and Reinhart, 1999), (iii) the impact of financial liberalization on bank stability (Demirguc-Kunt and Detragiache, 1999), and (iv) the impact of deposit insurance design on bank fragility (Demirguc-Kunt and Detragiache, 2003). This paper studies the impact of bank concentration, bank regulations, and national institutions on the likelihood of suffering a systemic banking crisis using data on 79 countries over the period 1980-1997 while controlling for many national characteristics. We believe this is the first paper to examine the impact of concentration on crises across a broad cross-section of countries while controlling for differences in regulatory policies, national institutions governing property rights and economic freedom, the ownership structure of banks, and macroeconomic and financial conditions. To draw accurate inferences about the independent impact of banking structure on crises, it is imperative to control for international differences in the generosity of deposit insurance regimes, capital-regulations, restrictions on bank entry, and regulatory restrictions on bank activities. Furthermore, to assess the impact of concentration on crises, we need to control for cross-country differences in bank ownership, i.e., the degree to which the state and foreigners own the country's banks. Finally, we control for the overall institutional not. Thus, largest banks generally receive a greater net subsidy from the government unless they are less risky. This subsidy may in turn increase the risk-taking incentives of the larger banks. 4 environment governing economic activity as well as the level of economic development, economic growth, inflation, terms of trade and exchange rate changes, credit growth, etc. The paper finds that crises are less likely in more concentrated banking systems. Furthermore, the data indicate that fewer regulatory restrictions on banks - lower barriers to bank entry and fewer restrictions on bank activities - reduce bank fragility. We also find that economic freedoms and better institutions in general promote bank stability. Thus, concentration and competition reduce bank fragility. To the extent that our regulatory and institutional indicators fully capture the competitive environment, these results suggest that concentration proxies for something beyond market power. We test whether concentration proxies for better diversification or easier monitoring. We find no evidence supporting the claim that concentrated banking systems are easier to monitor and weak evidence that concentrated banking systems are better diversified. Our results do not seem to be driven by reverse causality and are robust to an array of sensitivity checks. The rest of the paper is organized as follows. Section II describes the data set and presents summary statistics. Section III explains the methodology used in empirical tests. Section IV contains the main results and Section V concludes. I1. Data anmd Summary Statistis This section describes the variables and the data sources we use in our empirical analysis. Two variables of particular interest in our study are the crisis variable and the concentration measure. Table 1 presents these for the countries in our sample. Table 2 contains summary statistics and correlations of all variables we use in our analysis. The Appendix reports detailed variable definitions and data sources. 5 Crisis is a dummy variable that takes the value one if the country is going through a crisis, and zero if it is'not. We define a crisis only as a systemic crisis episode in which significant segments of the banking sector become insolvent or illiquid, and cannot continue to operate without special assistance from the monetary or supervisory authorities. Following Demirguc-Kunt and Detragiache (2003), we identify and date episodes of banking sector distress using primarily information from Lindgren, Garcia and Saal (1996) and Caprio and Klingebiel (1999). Then, these episodes of distress are classified as systemic if emergency measures were taken to assist the banking system (such as bank holidays, deposit freezes, blanket guarantees to depositors or other bank creditors), or if large-scale nationalizations took place. Episodes were also classified as systemic if non-performing assets reached at least 10 percent of total assets at the peak, of the crisis, or if the cost of the rescue operations was at least 2 percent of GDP. Multiple crises are allowed, but the years in which banking crises were under way were excluded from the panel since during a crisis the behavior of some of the explanatory variables is likely to be affected by the crisis itself, leading to reverse causality. For the period 1980-1997, the sample includes all countries covered in the International Financial Statistics, excluding only countries in transition and those for which data series was mostly incomplete. This results in 79 countries and 50 crisis episodes. Concentration equals the share of assets of the three largest banks. We compute a measure of bank concentration using the Bankscope database compiled by Fitch-IBCA, which reports bank balance sheet data in a large cross-section of countries beginning in 1988. However, because the sample of banks covered increased significantly over the sample period, changes in the measure of concentration may just reflect changes in coverage. To reduce this potential problem, we average the measure over the period 1988-1997. As reported in Tables 1 6 and 2, most countries have concentrated banking systems with a sample mean of 72 percent. Still, there is wide variation in the sample, with concentration levels ranging from less than 20 percent for the U.S. to 100 percent for many African countries. Simple correlations do not show a significant relationship between the crisis dummy and bank concentration, although the sign is negative. Using this measure of concentration may blur the interpretation of estimation results since for many observations the crisis date would precede the time period for which we have the concentration values. However, later in our analysis we also use initial level of concentration and focus on crises that occurred after this date. Although this halves the number of observations and reduces crisis episodes to 20, we confirm our results using this smaller sample. As an additional robustness test, we test the sensitivity of the results to using a concentration measure obtained from Barth, Caprio, and Levine's (2001, 2003) survey and analysis of national regulatory policies. To estimate the regressions, we adopt the specification in Demirguc-Kunt and Detragiache (2003). Thus, the control variables are the rate of growth of real GDP, the change in the external terms of trade, and the rate of inflation, to capture macroeconomic developments that are likely to affect the quality of bank assets. Short-term real interest rate is included to capture the banks' cost of funds and since increases in interest rates may also affect profitability through increasing default rates. Bank vulnerability to sudden capital outflows triggered by a run on the currency and bank exposure to foreign exchange risk are measured by the rate of exchange rate depreciation and by the ratio of M2 to foreign exchange reserves. Lagged credit growth is also a control since high rates of credit expansion may finance an asset price bubble that may cause a crisis when it bursts. We also include Demirguc-Kunt and Detragiache's index 7 of moral hazard caused by deposit insurance since they find that it contributes significantly to financial fragility.5 Finally, GDP per capita is used to control for the level of development of the country, which can proxy for the quality of regulations and the general institutional environment. Thus, we leave GDP per capita out of the benchmark when we explore the impact of specific banking regulations or institutional variables. Simple correlations in Table 2 suggest that banking crises are more likely in countries with higher levels of inflation and exchange rate depreciation, and less likely in growing countries with higher GDP per capita. In addition to bank concentration, we augment the benchmark specification in Demirguc- Kunt and Detragiache by using measures of bank regulation and supervision, bank ownership, measures of the competitiveness of the banking system and the economy in general, and a summary institutional index. Measures of bank regulation and supervision come from Barth, Caprio and Levine (2001). The data set is collected through surveys of government officials and is only cross-sectional and refers to the late 1990s, but according to Barth, Caprio and Levine, these aspects of bank regulation have not seen much change in the last twenty years. We use four measures of bank regulation and supervision. Fraction of Entry Denied is the number of entry applications denied as a fraction of the number of applications received from domestic and foreign entities. This is a measure of entry restrictions in banking and thus the contestability of the market. To the extent restricted entry increases bank profits, this variable would be associated with a lower rate of fragility. If 5 To build an aggregate index of moral hazard Demirguc-Kunt and Detragiache (2003) use principal component analysis of deposit insurance design features. Specifically, they use coinsurance, coverage of foreign currency and interbank deposits, type of funding, source of funding, management, membership, and the level of explicit coverage to create this index, which increases in moral hazard. The index varies over time since different countries adopted deposit insurance or revised its design features at different points in time. 8 however, restricted entry breeds inefficiency in the banking market, it could also lead to greater fragility. Activity Restrictions is an indicator of restrictions on banks' ability to engage in securities, insurance and real estate business and takes on higher values for higher restrictions. If these restrictions manage to keep banks from entering lines of business that are too risky for them to adequately evaluate or manage, banking systems with greater restrictions may be more stable. If however, restrictions prevent firms from diversifying outside their traditional lines of business, they may increase the fragility of the system. Required Reserves is the ratio of bank assets required to be held as reserves. Banking systems with higher ratios of required reserves may be more stable since they would have a greater buffer to absorb liquidity shocks. However, high required reserves are also a tax on the banking system leading to lower rates of profits, therefore making them more fragile. Capital Regulatory Index is a summary measure of capital stringency, and it is given by the sum of initial capital stringency and overall capital requirements. To the extent book capital is an accurate measure of bank solvency we expect better capitalized banks to be less fragile. Table 2 indicates that fraction of entry denied, activity restrictions, and required reserves are positively and significantly correlated with each other. Capital regulatory index is also positively correlated with required reserves but negatively correlated with fraction of entry denied. The moral hazard index is negatively and significantly correlated with all regulatory variables except capital regulatory index where the correlation is positive. It is also interesting that deposit insurance schemes in concentrated banking systems tend to be designed such that moral hazard is significantly lower. Among the regulatory variables only activity restrictions is significantly correlated -albeit at ten percent - with the crisis dummy, and the sign is positive. 9 We also control for ownership. If public banks are considered to have government guarantees, banking systems with a larger share of public banks may be less prone to banking runs. However, inefficiencies in public banks may also make them more fragile, destabilizing the system. Indeed, Caprio and Martinez-Peria (2000) and Barth, Caprio and Levine (2001) find evidence supporting the former argument. The extent of foreign bank ownership is another important control. To the extent foreign banks improve domestic banks' efficiency (as found in Claessens, Demirguc-Kunt and Huizinga, 2001), they may also make the system more stable. However, reduction in profits due to higher competition from foreign banks may also hurt the domestic banks making the system more fragile. Demirguc-Kunt, Levine and Min (1998) find that a larger foreign bank share is associaled with a lower probability of systemic crisis. State Ownership and Foreign Ownership are from Barth, Caprio and Levine (2001), defined as the percentage of banking system's assets in banks that are 50 percent or more govermnent or foreign owned. As in the case of regulatory variables, the assumption is that ownership of banks does not vary significantly over the years.6 Simple correlations in Table 2 do not reveal significant relationships between bank ownership variables and crisis occurrence. We also use three additional variables to capture the extent of banking freedoms and general economic freedoms and institutional environment. Banking Freedom is an indicator of relative openness of the banking system. Specifically, it is a composite index of whether foreign banks and financial services firms are able to operate freely, how difficult it is to open domestic banks and other financial services 6 We also use state bank data from La Porta, Lopez de Silanes, and Shleifer (LLS, 2002) who report figures on the percentage of assets of the largest 10 banks owned by the government. For each country there are two data points, one for 1995, and one referring to public ownership "before the privatizations of the 1990s." In the regression, we use the latter figures for the 1980s and the forrner for the 1990s. 10 firms, how heavily regulated the financial system is, the presence of state-owned banks, whether the government influences allocation of credit, and whether banks are free to provide customers with insurance and invest in securities. Higher values indicate fewer restrictions on banking freedoms. Economic Freedom is an indicator of how a country's policies rank in terms of providing economic freedoms. It is a composite of ten indicators ranking policies in the areas of trade, government finances, government interventions, monetary policy, capital flows and foreign investment, banking and finance, wages and prices, property rights, regulation, and black market activity. Higher scores indicate polices more conducive to competition and economic freedom. Both variables are available from the Heritage Foundation and are average values for the 1995- 97 period. To the extent freedoms allow banks to improve efficiency and to engage in different activities and diversify their risks, we expect increased level of freedoms to reduce fragility. However, it is also true that greater freedoms allow banks to undertake greater risks, particularly if the underlying institutional environment and existing regulations and supervision distort risk- taking incentives. Thus, overall greater freedoms may also lead to greater bank fragility. KKZ_Composite is an index of the overall level of institutional development constructed by Kaufman, Kraay and Zoido-Lobaton (1999). The underlying indicators are voice and accountability, government effectiveness, political stability, regulatory quality, rule of law, and control of corruption. This index is available for 1998. We expect better institutions to lead to reduced bank fragility, controlling for all other factors. Simple correlations indicate that the crisis dummy is negatively and significantly correlated with the two freedom indicators and the institutions variable. Countries with better institutions also tend to have more competitive banking systems with fewer regulatory restrictions. 11 HII. Methodology In estimating the crisis model, we follow Demirguc-Kunt and Detragiache (1998, 2003) and use a logit probability model. Using this model of banking crisis, we can test the hypothesis that bank concentration and competition has an impact on fragility when other factors are controlled for. Thus, we estimate the probability that a systemic crisis will occur at a particular time in a particular country, assuming that this probability is a function of our explanatory variables -X(i,t)- discussed above. Let P(i, t) denote a dummy variable that takes the value of one when a banking crisis occurs in country i and time t and a value of zero otherwise. j is a vector of n unknown coefficients and F(P3rX(i, t)) is the cumulative probability distribution function evaluated at PIVX(i, t). Then, the log-likelihood function of the model is: Ln L = 4 tPl .T 4 j=l .{P(i,t)ln[F(P<&X(i,t))] + (l-P(i,t)) ln[l- F(PfOX(i,t))]}. In modeling the probability distribution we use the logistic functional form, which is commonly used in studying banking difficulties.7 We estimate the model with robust standard errors since there may be heteroskedasticity across different observations. Observations within each country group may also be correlated. We also deal with this problem below, by relaxing the assumption that errors are independent within country observations. When interpreting the regression results, it is important to remember that the estimated coefficients do not indicate an increase in the probability of a crisis given a one-unit increase in the corresponding explanatory variables. Instead, the coefficients reflect the effect of a change in an explanatory variable on ln(P(i,t)/(l-P(i,t)). Therefore, the increase in probability depends on the original probability and thus upon the initial values of all the independent variables and their coefficients. While the sign of the coefficient does indicate the direction of the change, the 12 magnitude depends on the slope of the cumulative distribution function at P'X(i,t). In other words, a change in the explanatory variable will have different effects on the probability of a crisis depending on the country's initial crisis probability. Under the logistic specification, if a country has an extremely high (or low) initial probability of crisis, a marginal change in the independent variables has little effect on its prospects, while the same marginal change has a greater effect if the country's probability of crisis is in an intermediate range. In the analysis presented below, we investigate the impact of bank concentration variable and different regulatory, competition, ownership and institutional variables on bank fragility one at a time. We also analyze if the impact of concentration is robust to controlling for regulatory variables and indicators of competition and institutional development and whether there are significant interactions with concentration and these variables. Finally, we explore the potential non-linearity of the crisis-concentration relationship, and discuss the robustness of our results to different definitions of concentration and reverse causality. IV. Resuits A. Main findings 1. Concentration and crises The concentration variable enters the regressions negatively and significantly, suggesting that concentrated banking systems are less vulnerable to banking crises (Table 3). Evaluating the marginal impact of concentration on the probability of crisis at the mean values for all variables, we see that a one standard deviation increase in concentration leads to a decrease in crisis probability of one percent. Since crisis probabilities at any point in time are quite low, with a ' In addition to Demirguc-Kunt and Detragiache (1998,1999, 2003) also see Cole and Gunther (1993), Gonzalez- Hermosillo et al. (1997), and Demirguc-Kunt (1989). 13 mean value of four percent, this is a substantial reduction. This result is supportive of the concenlration-stability view, i.e., that concentration fosters a more stable banking system. Among the control variables, GDP growth and per capita GDP enter negatively, while the real interest rates enter positively, as suggested by economic theory and earlier empirical studies. Credit growth is positive, but significant at only ten percent level, which lends weak support to the argument that credit booms signal future fragility. Confirming the results of Demirguc-Kunt and DetTagiache (2003), moral hazard enters positive and significantly, indicating that deposit insurance design can have an important impact on fragility, and the result is weaker controlling for bank concentration. The model also fits the data well, classifying 65 percent of all observations and over 70 percent of crisis observations accurately.8 In column (3) we also add a squared concentration variable to the specification to check for potential nonlinearities in the relationship between concentration and banking crises. When including the squared term, the concentration variable retains a negative and significant coefficient at ten percent, while the squared concentration term is positive and insignificant. Testing for the joint significance of the two variables, we see that together the coefficients are significantly different from zero at the five percent level. This indicates that, although weak, at very high levels of concentration, there is an offsetting effect at work where concentrated banking systems are no longer as stable. Below, we try to understand the nature of this nonlinearity better. Finally, the table shows that the concentration result is not sensitive to excluding GDP per capita from the regression. 14 2. Concentration, regulations, and crises In Table 4, we include indicators of bank regulation to the specification. These specifications exclude GDP per capita since it is also a proxy for the institutional environment. We have also estimated specifications where we have left out the concentration variable and included only the regulation supervision variables one at a time. The results on these variables are virtually unchanged, thus we do not report them for brevity. The results indicate that tighter entry restrictions and more severe regulatory restrictions on bank activities boost bank fragility (Table 4). A higher fraction of entry applications denied- a proxy for tighter entry regulations - leads to higher levels of fragility in the banking system. This is consistent with the argument that restricted entry reduces the efficiency of the banking system, also making it more vulnerable to external shocks. Similarly, we find that restrictions on bank activities increase crisis probabilities. This result indicates that overall these restrictions prevent banks from diversifying outside their traditional business, reducing their ability to limit the riskiness of their portfolios. The required reserves and capital regulatory index do not enter with significant coefficients. The results also indicate that the concentration result is robust to inclusion of regulatory variables. The overall effect of bank concentration on crisis likelihood is still negative and significant. In unreported regressions we have also explored specifications where we have interacted the concentration variable with these regulatory variables, but the interaction terms did not enter significantly. 8 In classifying observations, predicted probabilities significantly higher than 4 percent (no of crisis observations divided by total number of observations which equals the sample mean of the crisis dummy) are classified as crisis 15 3. Concentration, ownership, institutions, and crises In Table 5, we explore the impact of concentration, bank ownership, and the overall institutional environment variables on bank fragility. We examined each of the ownership and institutional indicators both with and without concentration included in the regression. Since the coefficients on the ownership and institutional variables are not significantly different in either specification, we only report the results of the regressions that include concentration. The first two columns explore the impact of bank ownership on fragility. While we see a positive impact of state ownership on bank fragility, this result is not very robust.9 The impact of foreign ownership on fragility is negative, but insignificant. The variables that capture the general openness and competitiveness of the banking system and the economy, and the composite institutional variable enter with negative and very significant coefficients.10 Thus countries with greater freedoms in banking and generally more competitive economic policies are less likely to experience banking crises. This is the case despite the fact that these policies also tend to reduce entry barriers and are correlated with reduced levels of bank concentration. Better institutional environment is also associated with a lower probability of systemic crisis, as expected. The evidence is consistent with theories that emphasize the stabilizing effects of competition (Boyd and De Nicolo, 2003), but inconsistent with the many models that stress the destabilizing effects from competition.' observations and those below 4 percent are classified as no crisis. 9 In the specification that excludes bank concentration, state ownership is not significant. Replicating these regressions using LLS (2002) measure of state bank ownership confirm these findings. 10 In the specification without bank concentration, the economic freedom variable is significant at only ten percent. " Boyd and De Nicolo (2003) stress that competition exerts a stabilizing impact on banks because more competitive banks charge lower interest rates to firms and these lower rates reduce the likelihood of default. This prediction is consistent with our results. However, Boyd and De Nicolo (2003) use bank concentration as an indicator of bank competition. Thus, they stress that concentration will exert a destabilizing impact on banks, which is inconsistent with our results. 16 The results on bank concentration are robust to including bank ownership and general competition and institutional variables. In unreported regressions we also explored whether the impact of concentration on fragility differs in countries with different levels of freedoms and institutional development, by including interaction terms in the regressions. None of these interaction terms were significant, suggesting that bank concentration reduces fragility regardless of the competition environment or the institutional development of the country. 4. Concentration, regulations, ownership, institutions, and crises In Table 6, we simultaneously include bank concentration, regulations, ownership, and institutions. In each specification we enter bank concentration, the index of overall institutional development, and a measure of regulation. Bank concentration remains significantly, negatively associated with bank fragility even when controlling for the regulatory variables and overall institutional development. Indeed, the size of the coefficient on concentration is not substantially affected by expanding the conditioning information set. In contrast, the regulatory restriction variables and the overall institutional development indicators exhibit substantial multicollinearity. Their independent significance is materially weakened in Table 6 when they are included jointly. These results suggest that regulatory approaches to banking are part of the overall institutional approach to openness, competition, and private property in the economy. Thus, while regulations and institutions matter for bank fragility, they do not independently explain banking crises. Bank concentration is different. The evidence in Table 6 suggests that bank concentration is not a simple proxy for regulatory restrictions or national institutions. Bank concentration enters negatively in the crisis regression when controlling for regulations and institutions. This constitutes strong evidence for arguments that more 17 concentrated banking systems are more stable and is inconsistent with theories that predict more fragility in more concentrated banking systems. The findings that (i) concentration lowers fragility and (ii) low competition raises fragility imply that we need to move beyond a simple "concentration-stability" versus "concentration-fragility" debate where concentration is viewed as a simple proxy for market power. If our measures of regulatory restrictions and market openness do a reasonably good job of measuring the competitiveness of the banking industry, then the finding that concentration remains negatively associated with the probability of suffering a systemic banking crises, implies that concentration is measuring "something else" besides market power. As discussed in Allen and Gale (2000, 2003), concentrated banking systems may promote greater diversification and it may be easier to monitor a few large banks than many smaller banks. At the same time, the results suggest that competition and contestability promote stability. This is inconsistent with franchise value arguments, but consistent with theories that emphasize the benefits of operating in a competitive industry and a sound institutional environment (Boyd and De Nicolo, 2003). 5. Why does concentration matter? Next, we investigate if the results on concentration are due to the ability of banks in concentrated banking systems to better diversify, or the ease with which regulators and market participants monitor the riskiness of fewer banks. The argument that banks in more concentrated systems diversify better assumes concentrated banking systems have larger banks. Thus, in Table 7, we add to our baseline specification a mean bank size variable defined as the total bank assets divided by the number of banks. 12 As an alternative indicator of bank diversification, we use a diversification index from Barth et al. (2001) database, which asks whether regulators have diversification guidelines and whether banks are prohibited from investing abroad. 18 Another potential determiinant of whether banks can diversify their portfolios effectively is the size of the economy. To investigate if banks are able to better diversify in larger economies, we include level of GDP in the specification.13 Finally, to see if the number of banks makes a difference in monitoring risks and therefore preventing crises, we also include this variable. As the results in Table 7 show, none of these additional variables develop significant coefficients in the regressions, while the concentration coefficient remains negative and significant. However, the significance level is lower at ten percent when we control for mean bank size and the size of the economy, suggesting that diversification explanation may have some merit. B. Sensitivity analyses In Table 8, we try to better understand how the relationship between concentration and fragility changes at high levels of bank concentration. In columns (1) to (6) we define a high concentration dummy for different cut-off levels of concentration using 45th to 70th percentiles, where the dummy takes the value one at this cut-off value of concentration or higher. Results indicate that the high concentration dummy is significant between the 50t' and 70h percentiles, for concentration levels of 77 percent or higher. However, once we hit the 70' percentile, at concentrations levels of 87 percent or higher, the effect is no longer significant. The loss of significance may be due to the fact we only have 11 countries with concentration levels of over 87 percent that experienced a crisis, and when the sample becomes very imbalanced with respect to crisis/non-crisis observations estimation becomes imprecise. In the last column, we estimate a polynomial, including squared and cubed concentration terms. This does not yield significant results either. To see if in addition to the intercept change there is also a slope change at high 12 Using the mean bank size of the largest three banks does not change our results. '3 Replacing GDP by M2 to control for the size of the financial system does not change our results significantly. 19 concentration levels, we also explored specifications where we included an interaction term of the concentration variable and high concentration dummy. The coefficient estimate was not significant. We conclude that while there is some evidence that the impact of concentration on stability is less strong at high levels of concentration, this result is somewhat sensitive to how we define high concentration. In sum, our results indicate that the overall effect of concentration on fragility is negative at all levels of bank concentration, even after we control for bank regulation and supervision, differences in bank ownership, the level of competitiveness in banking and the economy and general institutional development. In Table 9, we investigate the sensitivity of our results to the way we define the concentration variable. We first replicate the regression replacing the concentration variable by the one obtained from Barth, Caprio and Levine (2001). This measure of concentration, obtained through surveys of bank regulators, is calculated as the fraction of deposits held by the five largest commercial banks in each country as of end-1999. We expect this measure not to suffer from problems of differential coverage in each country since the source is the bank regulators themselves. Using this different measure of bank concentration, we get very similar results. There may exist concerns regarding reverse causality. This would be the case if systemic crises led to lower levels of concentration in the banking system through greater entry or changes in general competition policies. When we inspect individual crisis cases in our sample, however, we do not see a significant pattern of reduced concentration after the crisis episodes and the concentration levels do not vary significantly from year to year. 14 Nevertheless, we estimate a specification where we define concentration as the initial level of concentration (1988 or the first available year) instead of the 1988-97 average. As can be seen in column 3 of Table 9, this does 20 not change our results significantly, consistent with the observation that concentration does not vary much over time. This estimation is still subject to problems though, since some of the crisis episodes have taken place before the date for which we have data for concentration. Thus, we drop all those crisis episodes, which precede the initial concentration date. This leaves us with only 20 crisis episodes and less than half of the total number of observations, yet the concentration variable still remains negative and significant (column 4, Table 9). These results lead us to believe that the negative impact of concentration on banking system fragility is not due to reverse causality. So far in the analysis, we have allowed for heteroskedasticity of errors and corrected for it, but assumed that the errors are independent. However, given that we use a panel data set, it is likely that the error terms within individual country observations are correlated with each other. Table 10, column 1 reports our results relaxing the assumption that within country observations are independent. Concentration still enters with a negative and significant coefficient. In column 2, we estimate a logit model with random country effects. Again, the results are not significantly different. We also investigate the sensitivity of our results to using alternative samples. In column 3 we exclude from the sample all countries with populations of less than 1 million. Results are not sensitive to excluding small countries. In column 4, we exclude all African countries since they tend to have very high bank concentration ratios. We see that our results are not driven by African observations. In column 5, we exclude all developed countries from the analysis. Again, we see that concentration significantly reduces fragility also in the sample of developing 4 Also note that the actual crisis period immediately following the crisis is taken out of the estimations. 21 countries. Finally, in column 6 we exclude a few outlier observations in terms of inflation and interest rates, which leaves the results unchanged.'5 V. Conclusions This paper investigates the impact of bank concentration, bank regulations, bank ownership, and the overall competitive/institutional environment on banking system fragility. We use cross-country data on 79 countries and 50 crisis episodes. In concluding, we emphasize the following findings. First, bank concentration has a stabilizing effect. Concentrated banking systems are less likely to experience systemic banking crises, even after controlling for a wide array of macroeconomic, regulatory and institutional factors. There is also some evidence that the stabilizing effect of bank concentration is weaker at higher levels of concentration, although this does not change the fact that the overall impact of concentration on fragility is negative and that the relationship holds regardless of the quality of bank regulations or the overall competitive/institutional climate. Second, entry barriers and activity restrictions have a destabilizing effect on banking systems. Banking systems where a larger fraction of entry applications are denied, and those with tighter activity restrictions that limit expansion into different lines of business suffer a greater likelihood of systemic crisis. The data do not support the view that more competition induces greater fragility. Quite to the contrary, more competitive banking systems and those with fewer entry regulations and activity restrictions tend to be more stable. Finally, we find that countries with better-developed institutions and with policies that promote competition throughout the economy are less likely to suffer from systemic banking 15Excluded observations are Gabon (1993) and Cote d'lvoire (1993) because their M2/reserves values are outliers 22 crises. The composite indicator of institutional development always has a negative and significant sign in the fragility regressions. Moreover, we find that it is difficult to identify the independent effect of bank regulations and bank policies that promote competition from the overall institutional environment. Countries with better institutions (property rights, rule of law, political openness, low corruption, etc.) also tend to be countries with bank regulations and bank policies that support openness and competition. Thus, while bank regulations and policies that foster competition and contestability promote bank stability, these regulations and policies cannot be viewed in isolation from the overall institutional environment. In sum, we find that (1) concentration lowers fragility and (2) low competition raises fragility. These results suggest that we need to complicate the simple "concentration-stability" versus "concentration-fragility" debate. If our indicators of regulations, bank policies, ownership, and institutions do a reasonably good job of measuring the competitiveness of the banking industry then the finding that concentration lowers fragility implies that concentration is measuring something beyond market power. We try to see if this "something else" is related to diversification or benefits from monitoring only a few large banks. Although we find weak evidence that big banks are more diversified, future research needs to more completely explain exactly why bank concentrations is associated with a lower likelihood of suffering a systemic banking crisis. and Peru (1991) because its inflation and real interest rate values are outliers. 23 References Allen, Franklin, 1990. "The Market for Information and the Origin of Financial Intermediation." Joumal of Financial Intermediation 1, no. 1, May. Allen, Franklin and Douglas Gale, 2000. Comparing Financial Systems. Cambridge and London: MIT press. Allen, Franklin and Douglas Gale, 2003. "Competition and Financial Stability." World Bank mimeo. Bank for Intemational Settlement, 2001."The Banking Industry in the Emerging Market Economies: Competition, Consolidation, and Systemic Stability," BIS Papers 4, Basle. Barth, James R., Gerard Caprio Jr., and Ross Levine, 2001, "Banking Systems Around the Globe: Do Regulation and Ownership Affect Performance and Stability?" In: F. Mishkin (ed.), Financial Supervision and Regulation: What Works and What Doesn't, Chicago, IL: Chicago University Press, 31-88. Barth, James R., Gerard Caprio Jr., and Ross Levine, 2003, "Bank Supervision and Regulation: What Works Best?" Journal of Financial Intermediation forthcoming. Beck, Thorsten, Asli Demirguc-Kunt, and Ross Levine, 2000, "A New Database on the Structure and Development of the Financial Sector," The World Bank Economic Review 14, 597- 605 Boyd, J.H., and G. De Nicolo, 2003. "Competition and Bank Risk Revisited" University of Minnesota mimeo. Boyd, J. H. and S. Graham (1991). "Investigating the Banking Consolidation Trend," Federal Reserve Bank of Minneapolis Quarterly Review. Spring, 1 -15. Boyd, J. H. and S. Graham (1998). "Consolidation in US Banking," in Y. Amihud and G. Miller (eds.), Bank Mergers and Acquisitions. Norwell, MA. Kluwer, pp. 113-135. Boyd John H., and David E. Runkle, 1993. "Size and Performance of Banking Firms: Testing the Predictions of Theory." Journal of Monetary Economics 31. 47-67. Boyd John H., and Edward C. Prescott, 1986. "Financial Intermnediary-Coalittions." Journal of Economic Theory 38, 211-232. Caprio, Gerard, Jr., and Daniela Klingebiel, 1996, "Bank Insolvencies: Cross Country Experience," Policy Research Working Paper No. 1620 (Washington: World Bank). 24 Caprio, Gerard, Jr., and Daniela Klingebiel, 1999, "Episodes of Systematic and Borderline Financial Distress", mimeo, The World Bank. Caprio, Gerard, Jr., and Maria Soledad Martinez-Peria, 2000. "Avoiding Disaster: Policies to Reduce the Risk of Banking Crises." Discussion Paper. Egyptian Center for Economic Studies. Chan, Yuk-Shee, Stuart Greenbaum, and Anjan Thakor, 1992. "Is Fairly Priced Deposit Insurance Possible?" Journal of Finance 47. 227-245. Claessens, Stijn, Demirguc-Kunt, Asli and Harry Huizinga, 2001. "How Does Foreign Entry Affect Domestic Banking Markets?" Journal of Banking and Finance. 25(5), Cole, R. A. and Gunther, J.W., 1993. "Separating the Likelihood and Timing of Bank Failure." Finance and Economics Discussion Series, Federal Reserve Board, 93-20, Washington D.C. Demirguic-Kunt, Asli, 1989. "Deposit-Institution Failures: A review of Empirical Literature," Economic Review, Federal Reserve Bank of Cleveland, Vol. 25, No.4. Denirguc-Kunt, Asli, and Enrica Detragiache, 1998, "The Determinants of Banking Crises in Developing and Developed Countries", IMF Staff Papers, 45 (1), pp. 81-109. Demirguic-Kunt, Asli, and Enrica Detragiache, 1999, "Financial Liberalization and Financial Fragility", in B. Pleskovic and J.E. Stiglitz (Eds.) Proceedings of the 1998 World Bank Conference on Development Economics, Washington, DC, The World Bank. Demirguci-Kunt, Asli, and Enrica Detragiache, 2003, "Does Deposit Insurance Increase Banking System Stability? An Empirical Investigation." Journal of Monetary Economics, forthcoming. Demirguc-Kunt, Asli, Levine, Ross and Hong-Ghi Min, 1998. "Opening to Foreign Banks: Issues of Stability, Efficiency, and Growth," in Proceedings of the Bank of Korea Conference on The Implications of Globalization of World Financial Markets, Seoul: Bank of Korea. Diamond, Douglas W., 1984. "Financial Internediation and Delegated Monitoring." Review of Economic Studies 51, 393414. Flannery, Mark J., 1989. "Capital Regulation and Insured Banks' Choice of Individual Loan Default Risks." Journal of Monetary Economics 24 235-258. Gonzales-Hermosillo, B, Pazarbasioglu, C. And Billings, R., 1997. "Banking System Fragility: Likelihood Versus Timing of Failure - An Application to the Mexican Financial Crisis," IMF Staff Papers. 25 Group of Ten, 2001. Report on Consolidation in the Financial Sector, Bank for International Settlements: Basel, Switzerland. Hellman, Thomas, Kevin Murdock, and Joseph E. Stiglitz, 2000, Liberalization, Moral Hazard in Banking and Prudential Regulation: Are capital Controls enough? American Economic Review 90(1), pp 147-165. International Monetary Fund, 2001. Financial Sector Consolidation in Emerging Markets, Chapter V, International Capital Market Report. Kaminsky, Garciela, and Reinhart, Carmen, 1999. "The Twin Crises: The Causes of Banking and Balance of Payments Problems." American Economic Review 89 (3), 473-500. Kane, Edward J., 1989. The S&L Insurance Mess: How Did It Happen? Washington: Urban Institute Press. Kaufman, D., Kraay, A., Lobaton, P.Z. 1999. "Governance Matters." World Bank Policy Research Department Working Paper No. 2196. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, 2002. "Government Ownership of Commercial Banks." Journal of Finance 57. 265-301. Lindgren, Carl-Johan, Gillian Garcia, and Matthew I. Saal, 1996, Bank Soundness and Macroeconomic Policy, (Washington: International Monetary Fund). Merton, Robert C., 1977." An Analytic Derivation of the Cost of Deposit Insurance and Loan Guarantees: An Application of Modern Option Pricing Theory." Journal of Banking and Finance 1, 3-11. Mishkin, Frederic S., 1999. " Financial Consolidation: Dangers and Opportunities." Journal of Banking and Finance 23, 675-691. Ramakrishnan, Ram and Anjan V. Thakor, 1984. "Information Reliability and A Theory of Financial Intermediation." Review of Economic Studies 51, 415-432. Sharpe, William F., 1978. "Bank Capital Adequacy, Deposit Insurance and Security Values." Journal of Financial and Ouantitative Analysis 13, 701-718. Stiglitz, Joseph E., 1972. "Some Aspects of the Pure Theory of Corporate Finance: Bankruptcies and Takeovers." Bell Journal of Economics 3 (3), 458-82. Williamson, Stephen D., 1986. " Costly Monitoring, Financial Intermediation, and Equilibrium Credit Rationing." Journal of Monetary Economics 18, 159-179. 26 Table 1. Bank Concenintradon and Compmetdoitn amd Bankimg Crses GDP per capita is in constant dollars, averaged over the entire sample period. Crisis period denotes the years in which each country experienced a systemic banking crisis and the duration of said crisis. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Sources are in the data appendix. GDP per capita Crisis Period Concentration Australia 17,913 0.65 Austria 25,785 0.75 Bahrain 9,398 0.93 Belgium 24,442 0.64 Belize 2,996 Benin 362 (1998-90) 1.00 Botswana 2,781 0.94 Burkina Faso 230 (1988-94) Burundi 186 1.00 Cameroon 790 (1987-93, 1995-98) 0.95 Canada 18,252 0.58 Chile 3,048 (1981-87) 0.49 Colombia 1,802 (1982-85) 0.49 Congo 940 1.00 Cote d'lvoire 843 (1988-91) 0.96 Cyprus 9,267 0.88 Denmark 31,049 0.78 Dominican Republic 1,426 0.65 Ecuador 1,516 (1995-97) 0.40 Egypt 905 0.67 El Salvador 1,450 (1989) 0.84 Finland 23,304 (1991-94) 0.85 France 24,227 0.44 Gabon 4,625 Gambia 369 Germany 27,883 0.48 Ghana 356 (1982-89) 0.89 Greece 10,202 0.79 Guatemala 1,415 0.37 Guinea 523 (1993-94) Guyana 653 (1993-95) 1.00 Honduras 694 0.44 India 313 (1991-97) 0.47 Indonesia 761 (1992-97) 0.44 GDP per capita Crisis Period Concentration Ireland 13,419 0 74 Israel 13,355 (1983-84) 0.84 Italy 17,041 (1990-95) 0.35 Jamaica 1,539 (1996-97) 0.82 Japan 35,608 (1992-97) 0.24 Jordan 1,646 (1989-90) 0 92 Kenya 336 (1993) 0.74 Korea 6,857 (1997) 0 31 Lesotho 356 1.00 Malawi 154 Malaysia 3,197 (1985-88, 1997) 0.54 Mali 260 (1987-89) 0 91 Mauritania 456 (1984-93) Mauritius 2,724 0 94 Mexico 3,240 (1982, 1994-97) 0 63 Nepal 179 (1988-97) 0 90 Netherlands 22,976 0 76 New Zealand 15,539 0 77 Niger 245 (1983-97) Nigeria 251 (1991-95) 0.83 Norvay 28,843 (1987-93) 0.85 Panama 2,824 (1988-89) 0.42 Papua New Guinea 1,024 0 87 Peru 2,458 (1983-90) 0 69 Philippmes 1,070 (1981-87) 0.49 Portugal 8,904 (1986-89) 0 46 Senegal 562 (1988-91) 0.94 Seychelles 5,719 Siena Leone 260 (1990-97) 1 00 Singapore 20,079 0 71 South Africa 3,680 (1985) 0.77 Sri Lanka 588 (1989-93) 0.86 Swaziland 1,254 (1995) 0.95 Sweden 24,845 (1990-93) 0 89 Switzerland 42,658 0 77 Tanzania 170 (1988-97) 1 00 28 GDP per capita Crisis Period Concentration Thailand 1,886 (1983-87, 1997) 0.54 Togo 366 1.00 Tunisia 1,831 0.63 Turkey 2,451 (1982,1991,1994) 0.45 United Kingdom 16,883 0.57 United States 24,459 (1980-92) 0.19 Uruguay 5,037 (1981-85) 0.87 Venezuela 3,558 (1993-97) 0.52 Zambia 464 0 84 29 Table 2. Summary Statistics and Correlations Summary statistics are presented in Panel A and correlations in Panel B and C. Banking crisis is a crisis dunmmy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the ratc of changc of thc GDP dcflator M/.eserves is tLhe ratio of M2 to international reserves Cred:t growvth is the real growth of domestic credit, lagged tvo periods Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the'three largest banks in each country, averaged over the samnple period. Fraction of entry denied is the number of entry applications denied as a fraction of the number of applications received from domestic and foreign entities Activity restrictions measures the degree to which a bank is able to engage in business of securities underwriting, tnsurance underwriting and selling, and in real estate investment, management, and development. Requtred reserves is the percentage of reserves required to be held by banks. Capital regulatory index measures capital strmgency in the banking system. State ownership measures the percentage of banking system's assets in banks that are 50%/o or more government owned, while foreign ownership measures percentage of banking system's assets in banks that are 50% or more foreign owned. Banking freedom is indicator of the relative openness of banking and financial system Economic freedom is a composite measure of institutional factors detemining economic freedom KKZ composite is a composite measure of govemance indicators Sources are given in the data appendix. Panel A: Summary Statistics: Mean Median St Dev Maximum Minimum Observations Banking crisis 0.04 0 00 0.20 1.00 0 00 1238 Growth 3 41 3.45 4.25 23.60 -17 15 1216 Termsoftradechange 0.15 0.01 10.30 63.24 -51.45 1191 Real interest rate 1.58 2.68 19.34 151.21 ' -283.00 1160 Inflation 14.07 7 75 23 42 350 56 -29.17 1220 M2/reserves 1987 6.56 68.86 128931 0 19 1222 Depreciation 0.10 0.04 0 22 2.62 -0.35 1238 Credit Growth,2 6.01 5.09 15.84 115.42 -54 62 1203 Real GDP per capita 7,813.94 2,302 37 10,299.92 45,950.46 134 54 1222 Moral hazard -1 09 -2.49 2.24 3.98 -2.49 1238 Concentration 0.72 0.77 0.21 1.00 0 19 1106 Fraction of entry denied 0.21 0 08 0.29 1.00 0.00 688 Activity restrictions 9 44 9.00 2.64 14.00 4.00 903 Required reserves 12 48 10.00 11.86 43 00 0 00 692 Capital regulatory index 5 41 5.50 1.70 8 00 2 00 871 State ownership 17.84 11.56 20.95 80.00 000 796 Foreign ownership 23 85 11.70 26 59 99.00 0.00 710 Banking freedom 3.36 3.00 0 88 5.00 1.00 1184 Economic freedom 3.17 3.05 0.61 4.50 1.9 1184 KKZ composite 0.28 -0.03 0.79 1.72 -1.03 1220 Panel B: Correlationns: Banking Crisis, Concentration, Regulations, and Institutions Banking crisis Concentration Fraction of Activity Required Capital Moral State Foreign Banking Economic entry denied restrictions reserves Regulatory hazard ownership ownership freedom freedom index Banking crisis Concentration -0.037 Fraction of entry 0.058 0.001 denied Activity 0.058* -0.027 0.461°44 restrictions Required reserves 0.016 0.183"4' 0.334"'* 0.23344° Capital regulatory -0.016 0.053 -0.048 -0.084444 0.229**' index Moral hazard 0.016 -0.396"*' -0.238"4' -0.24844 -0.105444 0.10744 State ownership 0.034 0.048 0.43344° 0.284444 0.356444 0.039 -0.022 Foreign ownership -0.050 0.394444 0.059 0.025 0.262**' 0.192444 -0.321444 -0.234444 Banling freedom -0.06344 -0.0249** -0.382444 -0.4774** -0.101444 0.077444 0.174444 -0.385444 0.190444 Economic freedom -0.0534 -0.390"* -0.450**' -0.515444 -0.401*** 0.06944 0.327**' -0.539444 -0.003 0.745444 KKZ_composite -0.066"'* 0.35444 o0.507444 -0.566444 0.4450** 0.06744 0.3544°4 -0.460444 0.029 0.560444 0.861"'* 444,44, and 4 indicate statistical significance at 1, 5, and 10 percent, respectively. Pamnel C: CorreRadons: Banking Crisis, Concentration, and Macro Indicators Banking crisis Real GDP Terms of trade Real interest Inflation M2/reserves Depreciation Credit Growtht2 Real GDP per growth change rate capita Banking crisis Real GDP growth -0.134444 Terms of trade change -0.014 0.0464 Real interest rate 0.003 0.085**' -0.050* Intlation 0.067**4 -0.103444 0.038 -0,98044 M2/reserves 0.034 -0.098"'* 0 007 0 010 -0 015 Depreciation 0.074444 -0.168444 -0.020 -0.546"'* 0 616"4 -0 031 Credit growth,2 0.044 0.024 0.000 0.003 -0.007 -0.0454 -0.05444 Real GDP per capita -0.05744 -0.05544 0 017 0.026 -0.0474 -0.033 -0.201"'* -0.008 Concentration -0.037 -0.076"44 -0.007 0.004 0.000 0.100444 0.0444 -0.001 -0.246444 444,44, and 4 indicate statistical significance at 1, 5, and 10 percent, respectively. Table 3. Banking Crisis and Concentration The logit probability model estimated is Bankng Crisis [Com T,,T,.,,t]= a + pi3 Real GDP growth,t+ 02 Terms of trade change,, + P33 Real interest rate,,+ 034 Infation,., P5M2/reserves,t + 1)6Depreciation,t + 037 Credit growth,.1.2 + P3s Real GDP per capitat,+09 Moral Hazard Index,,+0310 Average concentration,1+ p l l Concentration 2,,+ 6,,. The dependent vanable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to international reserves. Credit growth is the real growth of domestic credit, lagged two penods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the samnple period. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed vanable definitions and sources are given in the data appendix. (1) (2) (3) (4) Real GDP growth -0 142*** -0.166*** -0.162*** -0.163*** (0.031) (0.035) (0.035) (0.036) Tenns of trade change -0.007 -0.011 -0.010 -0.012 (0.010) (0.012) (0.012) (0.013) Real interest rate 0.009*** 0.010*** 0.011 *** 0 010*o* (0.004) (0.004) (0.004) (0.004) Inflation 0.008 0 009 0.010 0.009 (0.008) (0.009) (0.009) (0 008) M2/reserves 0.001 0.002* 0.002* 0.002* (0.001) (0.001) (O 001) (0.001) Depreciation 0.692 0.448 0.503 0 821 (1.067) (I 202) (1.217) (1.176) Credit Growtht.2 0.013* 0.014* 0.015* 0.016* (0.008) (0.009) (0 009) (0.010) Real GDP per capita -0.004* -0.004** -0.005** (0.002) (0 002) (0.002) Moral Hazard Index 0.157*** 0.103 0.120* 0.042 (0 071) (0.075) (0 081) (0.076) Concentration -1.749** -7.447* -1.567** (0.871) (4.711) (0.862) (Concentration)2 4.346 (3.602) No. of Crisis 50 46 46 46 No of Observations 1111 997 997 997 % cnsis correct 70 70 72 70 % correct 65 65 64 62 Model x2 4730*** 47.29*** 53 46*** 36.77*** * *,**, and * indicate statistical significance at 1, 5, and 10 percent, respectively. Tab1le 4. Banking Crisis, Regulatfion and Concenitrafion The logit probability model estimated is Banking Crisis [cO,,. Timestt= a + PI Real GDP growtht+ 02 Terms of trade change,,, + ,B3 Real interest rater,t + 14 Inflation j,, + ,B5M2/reserves;p + 06Depreciation p + 01 Credit growth13t-2 + j8Moral Hazard Indexj,&+09 Concentration,,4+ ,13 Regulatory measures,;t+ c;,,. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to international reserves. Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. The Regulatory measures in specifications (1)-(4) - Fraction of entry denied, Activity restrictions, Required reserves and Capital regulatory index, - are included one at a time. Fraction of entry denied measures the number of entry applications denied as a fraction of the total received. Activity restrictions captures bank's ability to engage in business of securities underwriting, insurance underwriting and selling, and in real estate investment, management, and development. Required reserves are ratio of reserves required to be held by banks. Capital regulatory index is a summary measure of capital stringency. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) Concentration - -2.320* -1.928** -2.695*** -2.375*** (1.554) (1.016) (1.203) (1.115) Fraction of Entry Denied 1.993*** (0.750) Activity Restrictions 0.182*** (0.073) Required Reserves 0.017 (0.017) Capital Regulatory Index -0.078 (0.129) No. of Crisis 21 34 27 33 No. of Observations 583 767 572 755 % crisis correct 67 74 67 70 % correct 77 75 72 73 Model x2 31.97*** 37.38*** 30.38t** 37.38*** §ub§ and b indicate statistical significance at 1, 5, and 10 percent, respectively. 33 Table 5. Banking Crisis, Ownership, Institutions, and Concentration The logit probability model estimated is Banking Crisis IcOun,Y=J T.,= t= a + 1 Real GDP growtht+ 02 Terms of trade changet, + 03 Real interest ratep,, + 04 Inflation,, + 05M2/reserves1,, + 036Depreciation1r + 07 Credit growth,1,.2 +138 Moral hazard indexN,, + 19 Concentrationj,, + 1 1o Regulatory measures1,,+ t,t. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to international reserves Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA. The sample penod is 1980-1997 The Regulatory measures in specifications (l)-(5) - State ownership, Foreign ownership, Banking freedom, Economic freedom, and KKZ composite - are included one at a time. State ownership is the percentage of banking system's assets in banks that are 50% or more government owned. Foreign ownership is the percentage of banking system's assets in banks that are 50% or more foreign owned. Banking freedom is an indicator of relative openness of banking and financial system, while economic freedom is a composite of 10 institutional factors determining economic freedom. KKZ_composite is an aggregate measure of six governance indicators. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) (5) Concentration -2.571*** -2 199** -I 828** -1.840*** -1.738** (1.132) (1.089) (0.861) (0.857) (0.828) State ownership 0 015t (0.008) Foreign ownership -0.003 (0.008) Banking Freedom -0.451*** (0.160) Economic Freedom -0.532*** (0.232) KKZ_composite -0.460*** (0 206) No. of Crisis 32 31 46 46 46 No. of obs. 686 609 963 963 997 % crisis correct 75 71 72 72 74 % correct 69 66 62 61 65 Model x2 30.90*** 33.66*** 50.57*** 44.99*** 47.22*** ***,**, and * indicate statistical significance at 1, 5, and 10 percent, respectively. 34 Table 6. Bankdng Crisis, Governance, Ownership, lInstitutions, and Concentration The logit probability model estimated is Banking Crisis [country=, T,me t]= a + L ' Real GDP growth1,t+ 02 Terms of trade change1,, + 03 Real interest ratei,, + P4 Inflation j,, + 05M2/reserves1 t + 06Depreciation,,t + 07 Credit growth, .2 +1s Moral hazard index,,, +139 KKZ_composite1, + 0 10 Concentration1 t+,t 13 ,, Regulatory measures;,t+ &s,t. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to intemational reserves. Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. KKZ_composite is an aggregate measure of six govemance indicators. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. The Regulatory measures in specifications (2)-(4) - Fraction of entry denied, Activity restrictions, and State ownership - are included one at a time. Fraction of entry denied measures the number of entry applications denied as a fraction of the total received. Activity restrictions captures bank's ability to engage in business of securities underwriting, insurance underwriting and selling, and in real estate investment, management, and development. State ownership is the percentage of baning system's assets in banks that are 50% or more govemment owned. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) Concentration -1.738** -2.324* -1.96244 -2.515t** (0.828) (1.588) (0.992) (1.141) KKZ_composite -0.460*4t 0.018 -0.138 -0.319 (0.206) (0.517) (0.329) (0.313) Fraction of Entry 2.016* Denied (1.182) Activity Restrictions 0.162* (0.992) State ownership 0.011 (0.009) No. of Crisis 46 21 34 32 No. of obs. 997 583 767 686 % crisis correct 74 67 74 75 % correct 65 77 74 70 Model x2 47.22*** 43.78*** 46.01** 40.04*** *,*¶*, and * indicate statistical significance at 1, 5, and 10 percent, respectively. 35 Table 7. Banking Crisis and Concentration: Diversification vs. Ease of Supervision The logit probability model estimated is Banking Crisis pu , j= r a + 3 1 Real GDP growth,,,+ 12 Terms of trade change,,, + 03 Real interest rate1,, + 04 Inflation ,t+ 135M2/reservesi,,+ 136Depreciation3, + 17 Credit growthJ t.2+ Moral hazard index3.,+ OConcentration3,t+ 103 Mean Bank Size,,+Pl I Diversification,t+ 012 GDP,t + 013 No. of Banksj + cs ,. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to international reserves. Credit growth is the real growth of domestic credit; lagged two periods. Depreciatton is the rate of change of the exchange rate Moral hazard is an aggregate mdex of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA The sample period is 1980- 1997. Mean Bank Size is given by average bank asset size (in billions of US dollars). Diversification is an indicator of whether there are guidelines regarding asset diversification. GDP is real GDP in billions of US$. No. of banks is given in hundreds and is from Barth et al database. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) (5) Concentration -1.576* -2.349** -1.479* -2.234** -2.019* (0.899) (1.083) (0.879) (I 162) (1.162) Mean Bank Size 0.000 -0.001 (0.001) (0.002) Diversification -0.207 -0 184 (0.346) (0.363) GDP ($) 0.003 0.001 (0.002) (0 001) No of banks 0.008 -0.000 (0 014) (0.001) No. of Crisis 46 34 46 34 34 No. of obs. 979 767 997 767 767 % crisis correct 74 71 72 68 76 % correct 63 74 64 73 75 Model X2 48.36*** 47.04*** 47.67*** 43.90** 56.05*** ***,**, and * indicate statistical significance at 1, 5, and 10 percent, respectively. 36 Table 8. Banking Crisis and High Concemtration: Robustness The logit probability model estimated is Banking Crisis JC,, Tut, I= a + Real GDP growth,,+ 02 Terms of trade change,,, + 03 Real interest rate,,+ 04 lnflation ,,+ 05M2/reservesj,, + j16Depreciationi,, + 07 Credit growthN, 2 + 38 Real GDP per capita,,t+03 Moral Hazard Index1 t+p I0 Average concentrationi&+ 3 1 High concentration, ,, E+. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to intemational reserves. Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. High concentration is a dummy taking a value of one in cases where the banking concentration is greater than or equal to the cutoff listed in the footnote of the table. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. Specifications (1)- (6) use high concentration at the 459, 50, 55h, 60", 65"', and 70"' percentile, respectively. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) (5) (6) (7) Concentration -3.428*** -4.052*4 -4.1694*4 44.357 437 -2.7264 -26.700* (1.407) (1.368) (1.311) (1.344) (1.344) (1.233) (15.259) High Concentration 0.832* 1.24644 1.362** 1.590*** 1.590*44 0.773 (0.561) (0.638) (0.652) (0.678) (0.678) (0.648) Concentration2 37.245 (25.923) Concentration3 -17.360 (13.772) No. of Crisis 46 46 46 46 46 46 46 No. of Observations 997 997 997 997 997 997 997 % crisis correct 76 76 76 74 74 70 76 % correct 65 66 64 65 65 64 66 Model X2 47.244*4 50.03*** 54.54*44 54.79*** 54.7944e 52.17444 54.19**4 Value of cutoff 0.73927 0.76707 0.78977 0.83955 0.84154 0.87530 444,44 and 4 indicate statistical significance at 1, 5, and 10 percent, respectively. 37 Table 9. Banking Crisis and Concentration: Robustness I The logit probability model estimated is Banking Crisis [crn-j,,Tint,=t a + I3 Real GDP growth,,t+ ,B2 Terms of trade change,, + 03 Real interest rateC, + D4 Inflation1,,+ 05M2/reserves1,, + 06Depreciation2, + 07 Credit growthj,,.2+ P8 Real GDP per capita,,,+09Moral Hazard lndexjl+jo 10Average concentration,,+ E,t. The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to intemational reserves. Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. In specification (2) Average Concentration is replaced by the concentration data from Barth, Caprio and Levine (2001 for the entire sample period. In specification (3) Average Concentration is replaced by the Initial Concentration, for the entire sample period. In specification (4) the sample is restricted to the actual date of Initial Concentration and the years following that date (for many of the countnes, the restricted sample is either 1988-97 or 1990-97) White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) Concentration -I.749** -1.536* -1.678** -3.481*** (0.871) (1.010) (0.869) (1.502) No. of Crisis 46 32 46 20 No. of Observations 997 730 997 415 % crisis correct 70 72 70 70 % correct 65 75 . 65 71 Model X2 47 29*** 47.99*** 46.22*** 40.68*** **, and * indicate statistical significance at 1, 5, and 10 percent, respectively. 38 Table 10. 1Banmkimg Cirisis and Concentratioua: IRobustmess 1131 The logit probability model estimated is Banking Crisis (Country Time= t= a + 1l Real GDP growth,,,+ 02 Terms of trade change,,, + P3 Real interest rate,,, + ,B4 Inflation t+ 15M2/reserves,, + , 6Depreciation1, + ,B7 Credit growths t.2 + 18 Real GDP per capita1.t+19 Moral Hazard Indexj,,+I 10Average concentration,, + E,t.The dependent variable is a crisis dummy, which takes on the value of one if there is a systemic crisis and the value of zero otherwise. Growth is the rate of growth of real GDP. Real interest rate is the nominal interest rate minus the contemporaneous rate of inflation. Inflation is the rate of change of the GDP deflator. M2/reserves is the ratio of M2 to international reserves. Credit growth is the real growth of domestic credit, lagged two periods. Depreciation is the rate of change of the exchange rate. Moral hazard is an aggregate index of moral hazard associated with deposit insurance schemes. Concentration is a measure of concentration in the banking industry, calculated as the fraction of assets held by the three largest banks in each country, averaged over the sample period. Bank data are from the BankScope database of Fitch IBCA. The sample period is 1980-1997. In specification (I) the sample is clustered by country. In specification (2) the estimation includes random country effects. In specification (3) the sample excludes all countries with populations of less than I million. In specification (4) the sample excludes all African countries. In specification (5) the sample excludes all developed countries. In specification (6) we remove outliers found in three control variables - real interest rate, inflation, and m2/reserves. White's heteroskedasiticy consistent standard errors are given in parentheses. Detailed variable definitions and sources are given in the data appendix. (1) (2) (3) (4) (5) (6) Concentration -4.357444 -1 .749444 -1.82944 -2.368*4 -1.60444 -1.80644 (1.033) (0.089) (0.917) (1.133) (0.852) (0.876) No. of Crisis 46 46 44 34 38 46 No. of Observations 997 997 935 742 670 995 % crisis correct 74 73 74 74 72 % correct 65 65 69 54 65 Model x2 59.10444 40.28444 56.014O4 43.394°° 29.19°44 47.73*** 440,44, and 4 indicate statistical significance at 1, 5, and 10 percent, respectively. 39 Data Appendix Vanable Name Definition Source Banking crisis Dummy takes on value of one durng episodes identified as a Demirguc-Kunt and Detragaiche (2003) systematic banking crises Growth Rate of growth of real GDP WDI (World Bank) Terms of trade change Change in the terms of trade WDI (World Bank) Real interest rate Nominal interest rate minus the contemporaneous rate of IFS (IMF) inflation Inflation Rate of change of GDP deflator IFS (IMF) M2/reserves Ratio of M2 to international reserves IFS (IMF) Depreciation Rate of depreciation IFS (IMF) Credit growth Rate of growth of real domestic credit to the private sector IFS line 32d divided by GDP deflator GDP/CAP Real GDP per capita WDI (World Bank) GDP Real GDP in billions of US dollars WDI (World Bank) Moral hazard index Aggregate index of moral hazard associated with deposit Demirguc-Kunt and Detragaiche (2003) insurance schemes, calculated using principal component analysis Concentration Degree of concentration in the banking industry, calculated as Beck, Demirguc-Kunt, Levine (2000) - Financial Structures Database the fraction of assets held by the three largest banks. Averaged over the 1988-97 period. Mean Bank Size Total banking assets divided by number of banks. BankScope database. Diversification Diversification index uses responses to survey questions 7.1 and Barth, Caprio, and Levine (2001) - Survey of bank Regulation and Supervision 7.2. Question 7.1 asks if there are explicit, verifiable, and quantifiable guidelines regarding asset diversification (yes=l, no=0). Question 7.2 asks if banks are prohibited from making loans abroad (yes=l, no=O). The index is calculated as Q7.1- ((Q7.2 - No. of Banks No. of banks m hundreds. Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision Fraction of entry denied Number of entry applications denied as a fraction of the number Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision of applications received from domestic and foreign entities Activity restrictions Indicator of bank's ability to engage in business of securities Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision underwriting, insurance underwriting and selling, and in real estate investment, management, and development Required reserves Ratio of reserves required to be held by banks Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision 40 Variable Name Defnition Source Capital regulatory index Summary measure of capital stringency: sum of overall and Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision initial capital stringency. Higher values indicate greater stringency. State ownership Percentage of banking system's assets in banks that are 50%/O or Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision more government owned Foreign ownership Percentage of banking system's assets in banks that are 50% or Barth, Caprio, and Levine (2001) - Survey of Bank Regulation and Supervision more foreign owned Banking Freedom Indicator of relative openness of banking and financial system: Index of Economic Freedom (Heritage Foundation) specifically, whether the foreign banks and financial services firms are able to operate freely, how difficult it is to open domestic banks and other financial services firms, how heavily regulated the financial system is, the presence of state-owned banks, whether the government influences allocation of credit, and whether banks are free to provide customers with insurance and invest in securities (and vice-versa). The index ranges in value from I (very low - banks are primitive) to 5 (very high - few restrictions). Averaged over 1995-97 period Economic Freedom Composite of 10 institutional factors determining economic Index of Economic Freedom (Heritage Foundation) freedom: trade policy, fiscal burden of government, govemment intervention in the economy, monetary policy, capital flows and foreign investment, banking and finance, wages and prices, property rights, regulation, and black market activity. Individual factors are weighted equally to determine overall score of economic freedom. A high score signifies an institutional or consistent set of policies that are most conducive to economic freedom, while a score close to I signifies a set of policies that are least conducive. Averaged over 1995-97 period. KKZ_composite Composite of six govemance indicators (1998 data): voice and Kaufman, Kraay and Zoido-Lobaton (1999) accountability, political stability, government effectiveness, regulatory quality, rule of law, and corruption. Individual factors are weighted equally to determine overall score of economic freedom. Higher values correspond to better govemance outcomes. 41 I Policy Research Working Paper Series Contact Title Author Date for paper WPS3011 Renegotiation of Concession J Luis Guasch April 2003 J Troncoso Contracts in Latin America Jean-Jacques Laffont 37826 Stephane Straub WPS3012 Just-in-Case Inventories. A Cross- J Luis Guasch Aprl 2003 J Troncoso Country Analysis Joseph Kogan 37826 WPS3013 Land Sales and Rental Markets in Klaus Deininger April 2003 M Fernandez Transition Evidence from Rural Songqing Jin 33766 Vietnam WPS3014 Evaluation of Financial Liberalization Xavier Gin6 April 2003 K. Labrie A General Equilibrium Model with Robert M Townsend 31001 Constrained Occupation Choice WPS3015 Off and Running? Technology, Trade, Carolina Sanchez-Paramo April 2003 H Sladovich and the Rising Demand for Skilled Norbert Schady 37698 Workers in Latin America WPS3016 Partisan Politics and Stuti Khemani April 2003 H Sladovich Intergovernmental Transfers in India 37698 WPS3017 Why is Unemployment so High Jan Rutkowski April 2003 J Rutkowski in Bulgaria? 84569 WPS3018 Integrating the Least Developed Paul Brenton April 2003 L Tabada Countries into the World Trading 36896 System. The Current Impact of EU Preferences under Everything but Arms WPS3019 Foreign Aid, Conditionality, and Ghost Thilak Ranaweera April2003 P Rathan Raj of the Financing Gap A Forgotten 33705 Aspect of the Aid Debate WPS3020 Consumption, Health, Gender, Anne Case April 2003 D Ballantyne and Poverty Angus Deaton 87198 WPS3021 Institutional Arrangements for Public Elizabeth Currie April 2003 M Rosenquist Debt Management Jean-Jacques Dethier 82602 Eriko Togo WPS3022 Sending Farmers Back to School. Gershon Feder April 2003 P Kokila The Impact of Farmer Field Schools Rinku Murgai 33716 in Indonesia Jaime B. Quizon WPS3023 Productivity Growth and Product Douglas M Addison April 2003 D Addison Variety. Gains from Imitation and 31188 Education WPS3024 R&D and Development Daniel Lederman April 2003 P. Soto William F. Maloney 37892 WPS3025 Trade Structure and Growth Daniel Lederman Apnl 2003 P Soto William F. Maloney 37892 WPS3026 Strategic Approaches to Science Robert Watson April 2003 R Watson and Technology in Development Michael Crawford 36965 Sara Farley Policy Research Working Paper Series Contact Title Author Date for paper WPS3027 Financial Intermediation and Growth Genevieve Boyreau- April 2003 P. Sintim-Aboagye Chinese Style Debray 38526 WPS3028 Does a Country Need a Promotion Jacques Morisset April 2003 M Feghali Agency to AHtract Foreign Direct 36177 Investment9 A Small Analytical Model Applied to 58 Countries WPS3029 Who Benefits and How Much? How Alessandro Nicita April 2003 P. Flewitt Gender Affects Welfare Impacts of a Susan Razzaz 32724 Booming Textile Industry WPS3030 The Impact of Bank Regulations, Aslh DemirgGc-Kunt April 2003 A. Yaptenco Concentration, and Institutions on Luc Laeven 31823 Bank Margins Ross Levine WPS3031 Imports, Entry, and Competition Law Hiau Looi Kee April 2003 P. FlewiHt as Market Disciplines Bernard Hoekman 32724 WPS3032 Information Diffusion in International Alejandro lzquierdo Apnl 2003 M. Feghali Markets Jacques Morisset 36177 Marcelo Olarreaga WPS3033 The Role of Occupational Pension Dimitri Vittas Apnl 2003 P Infante Funds in Mauritius 37642 WPS3034 The Insurance Industry in Dimitri Vittas April 2003 P. Infante Mauritius 37642 WPS3035 Traffic Fatalities and Economic Elizabeth Kopits April 2003 V. Soukhanov Growth Maureen Cropper 35721 WPS3036 Telecommunications Reform in George R. G. Clarke April 2003 P. Sintim-Aboagye Malawi Frew A Gebreab 37644 Henry R Mgombelo WPS3037 Regulation and Private Sector Sheoli Pargal Apnl 2003 S. Pargal Investment in Infrastructure. 81951 Evidence from Latin America WPS3038 The Debate on Globalization, Martin Ravallion May 2003 P. Sader Poverty, and Inequality: Why 33902 Measurement Matters WPS3039 Scaling Up Community-Driven Hans P. Binswanger May 2003 T. Nguyen Development. Theoretical Swaminathan S. Aiyar 31389 Underpinnings and Program Design Implications WPS3040 Household Welfare Impacts of Shaohua Chen May 2003 P. Sader China's Accession to the World Martin Ravallion 33902 Trade Organization