Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Inter-American Development Bank (IDB) Banco Interamericano de Desarrollo (BID) Research department Departamento de Investigación Should the Government be in the Banking Business? The Role of State-Owned and Development Banks Eduardo Levy Yeyati Alejandro Micco Ugo Panizza* Paper prepared for the seminar Governments and Banks: Responsibilities and Limits IDB-IIC Annual Meetings Lima, Peru March 28, 2004 * Eduardo Levy-Yeyati, Universidad Torcuato Di Tella, email: [email protected], Alejandro Micco and Ugo Panizza, Research Department, Inter-American Development Bank, email, [email protected] and [email protected]. We would like to thank Monica Yañez and Danielken Molina for excellent research assistance. This paper is circulated to stimulate discussion. The views expressed in this paper are the authors’ and do not necessarily reflect those of the Inter-American Development Bank. The usual caveats apply. 1 Introduction Arthur Lewis, Alexander Gerschenkron, Gunnar Myrdal and several other prominent development economists writing in the 1950s and 1960s tended to agree that the state should play a key role in the banking sector. Governments actual behavior was in line with this view and, by the 1970s, the state owned 50 percent of assets of the largest banks in industrial countries and 70 percent of assets of the largest banks in developing countries (Figure 1). The 1980s and 1990s witnessed a sea change in the view on the state’s role in the economy and privatization was at the very center of the neo-liberal economic policies codified in the Washington Consensus. Consequently, from 1987 to 2003 more than 250 banks were privatized, raising US$ 143 billion.1 But even after this big privatization wave, the presence of the state in the banking sector is still widespread and pervasive. In the mid 1990s, about one quarter of the assets of the largest banks in industrial countries and 50 percent of the assets of the largest banks in developing countries were still under state control.2 It is, therefore, interesting to ask whether there is a justification for such a massive public presence in the banking sector. Some argue that state presence in the banking sector is justified by market failures and development goals. They point out that financial markets in general and the banking sector in particular are different from other markets and that government intervention can improve the working of the financial sector and the overall functioning of the economy. In particular, the social view emphasizes the role of the public sector to compensate for market imperfections that leave socially profitable investments underfinanced.3 Also supportive of public participation in the banking sector is the development view (often identified with Gerscherkron, 1962) that stresses the need for public intervention in economies where the scarcity of capital, general distrust of the public, and endemic fraudulent practices among debtors may fail to generate the sizable financial sector required to facilitate economic development.4 Others argue that it is not necessarily true that banks are different from other businesses and that the case for financial market imperfection is often overstated. Furthermore, they suggest that market failures can be better addressed with regulation and subsidies rather than with direct state ownership of banks. This is the political view that argues that politicians create and maintain 1 See Meggison (2004). The data reported here are from La Porta et al. (2002) and refer to the assets of the ten largest banks. Data for the whole banking system (from Barth et al. 2001) are highly correlated with the data for the top ten banks (the regression coefficient between the two variables is 0.8 and the R2 is 0.5) but the former dataset shows somewhat lower presence of the public sector (11 percentage points lower on average). 3 See Atkinson and Stiglitz (1980) and Stiglitz (1993), among others. 4 See, for instance, Stiglitz (1994) 2 2 state-owned banks not to channel funds to socially efficient uses but rather as a political tool aimed at maximizing the politicians’ personal objectives (La Porta et al, 2001). According to the proponents of this view, state ownership of banks is just dictated by redistributive politics and by the fact that politicians are interested in appropriating the rents that may derive from controlling the banking sector. Somewhere in between the benign view of state intervention in the banking sector (as represented by the social and development views) and the more cynical political view, there is also an agency view that highlights the trade-off between the allocative efficiency motive stressed by the social and development views, and internal efficiency, namely the ability of stateowned enterprises to carry out their mandate. This view emphasizes that while market imperfections may exist, agency costs within government bureaucracies may more than offset the social gains of public participation. In order to understand whether the state should be in the banking business it is useful to decompose the issue into the following two questions: Are there market failures that justify state intervention in the banking sector? Are these market failures better addressed with subsidies and regulations or do they require direct state ownership? The Rationale for State Intervention Standard arguments for state intervention in the banking sector can be broadly classified into four groups: (i) maintaining the safety and soundness of the banking system; (ii) mitigating market failures due to the presence of costly and asymmetric information; (iii) financing socially valuable (but financially unprofitable) projects; and (iv) promoting financial development and giving access to competitive banking services to residents of isolated areas. The first group of reasons has to do with the fact that banks are inherently fragile institutions because their liabilities consist of demand deposits and their assets consist of more illiquid loans. Such a situation can lead to self-fulfilling bank runs and widespread bank failures. However, banking fragility by itself does not justify government intervention aimed at guaranteeing the stability of the banking system, unless bank failures generate large negative externalities. It is exactly in this sense that banks are special because, besides intermediating credit, they also provide two services that have a public-good nature: they are the backup source of liquidity for all other institutions and the transmission belt for monetary policy (Corrigan, 1982). The need for state intervention also arises from the fact that, due to the large leverage ratios that characterize financial institutions in general, bank managers and owners may have strong incentives to pursue investment activities that are riskier than the ones that would be 3 preferred by depositors.5 This would not be a problem if depositors could effectively monitor banks’ managers. However, there is a free rider problem in bank monitoring because banks’ liabilities are mostly held by small depositors who have very limited incentives and ability to monitor banks’ activities.6 The second set of explanations has to do with the fact that financial markets in general and banking in particular are informational intensive activities. It is generally accepted that the stock of information gathered by banks plays a role in increasing the pool of domestic savings that is channeled to available investment opportunities. However, as information has some publicgoods characteristics (non-rivalries in consumption and costly excludability) and often entails a fixed acquisition cost, competitive markets will undersupply information and the fixed cost nature will lead to imperfect competition in the banking system. Moreover, information can be easily destroyed, increasing the cost of bank failures as customers of the failed bank may lose access to credit. It has also been shown that asymmetric information may lead to credit rationing, that is, a situation in which good projects are underfinanced (or not financed at all) due to the lack of verifiable information.7 A similar case can be made for the relationship between depositors and banks: lack of bank-specific information can dissuade savers from depositing in banks, particularly in incipient banking systems where long-standing customer relationships are still to be built. The third group of reasons has to do with the fact that private lenders may have limited incentive to finance projects that produce externalities. In this line, direct state participation would be warranted to compensate for market imperfections that leave socially profitable (but financially unattractive) investments underfinanced.8 Alternatively, state intervention may be justified by big-push theories like the one originally formulated by Rosenstein-Rodan. It is also possible to argue that banks may frustrate expansionary monetary policy because they have limited incentives to lend during periods of economic downturns and low interest rates and do not internalize the fact that, by increasing lending, they would push the economy out of recession 5 See Jensen and Meckling (1976) and, for a textbook treatment, Freixas and Rochet (1997). The same problem underlies the role of banks as delegated monitors of depositors´ investments, as pointed out by Diamond (1984). These arguments have been invoked to motivate the need for more stringent prudential regulation, as opposed to direct state participation in banking activities. 7 Indeed, rationing may occur as an adverse selection phenomenon in which, by pooling good and bad projects, the lender may increase the financing costs to the point of driving good projects out of the market. For a detailed discussion of market failures arising from costly and asymmetric information, see Stiglitz (1994) 8 See Atkinson and Stiglitz (1982) and Stiglitz (1993), among others. 6 4 (this is the macroeconomic view).9 If this is the case, state intervention could solve a coordination problem and make monetary policy more effective. A related theoretical argument in favor of state intervention, borrowed from the literature on financial markets mix, points to the fact that effective prudential regulation (and, in some cases, the banks’ own incentives) tend to make private banks too risk averse to finance all potentially profitable investments.10 Then, in the absence of developed capital markets that allow for alternative sources of financing, which is the case in most developing countries, state intervention may be warranted. A last argument that is often invoked by supporters of state intervention in the banking sector is that private banks may not find it profitable to open branches in rural and isolated areas and that state intervention is necessary in order to provide banking services to residents of these areas. Underlying the argument are the beliefs that granting access to banking services may increase financial development with positive externalities on growth or poverty reduction (see, for instance, Burgess and Pande, 2003), and that access to financial services is, at any rate, a right and that the state should make an effort to guarantee its universal provision. Relatedly, the presence of public banks has also been advocated as a means to guarantee competitive behavior in an otherwise collusive banking sector. This rationale, however, is likely to be relevant only when the regulatory and monitoring capacity of the public sector is limited and prone to capture. How Should the State Intervene? While most --but not all-- economists would agree that the banking system is characterized but enough market failures to grant some form of government intervention, they disagree on whether the state should intervene through the regulation of, and the design of specific contracts with private suppliers, or through direct ownership. The recent literature on contracting suggests that addressing market failures and achieving social goals do not necessarily warrant direct state ownership. As Shleifer (1998, P. 135) states: “When the opportunities for government contracting are exploited the benefit of outright state ownership becomes elusive…The Contractual theory…also allows us to think about an imperfect government, which maximizes political goals such as patronage or simply the income of politicians through bribes.” 9 Prudential regulation may create an additional disincentive, as both the quality of banks´ portfolios and prospective investments tend to deteriorate during a recession. 10 There are at least two reasons why this may be the case. First, due to the presence of externalities in the banking sector, the regulator may aim at a suboptimal risk level. Second, reputation costs and significant market power may induce large private banks to shy away from risky investments in order to protect their charter value. 5 But what are the conditions under which state ownership can be preferable to contracting and regulation? Following Shleifer (1998) it is useful to start by assuming a benevolent government and then look at what happens if this assumption is relaxed. Shleifer starts by making the point that if the government knows exactly what it wants to produce and if the characteristics of the goods or services to be produced can be written in a contract or specified by regulation, then it will not matter whether a given good or service is directly provided by the government or contracted to a private provider.11 While the availability of perfect contracting and regulation weakens the case for direct state ownership, things become more complicated when it is not possible to fully describe or monitor (and hence contract or regulate) the goods or services to be produced. Oliver Hart, Andrei Shleifer and Robert Vishny (1997) use the theory of incomplete contracts to discuss the costs and benefits of public ownership in cases in which contracting opportunities are limited because the good or service to be provided has some “non-contractible” quality.12 In their set-up, the key variables are the cost of providing the good or service and the incentive to provide it (as all the other qualities are contractible, they are irrelevant in the choice between public and private ownership). A key assumption of their analysis is that cost reductions lead to a deterioration of the non-contractible quality of the good or service. Under this assumption, they find that while costs are always lower under private provision, quality may be higher or lower according to the effect of cost reduction activities on the quality of the good or service provided. Thus, public ownership would be preferable when there is limited potential for quality improvement or when the adverse effect of cost reduction on quality is likely to be substantial. Hart et al. (1997) also make the point that the presence of competition in the provision of the good or service would reduce the incentives of the private providers to reduce quality by overinvesting in cost-reducing activities.13 Finally, they deviate from the assumption of a benevolent government and allow for the presence of corruption and political patronage. They show that the presence of corruption weakens the case for privatization because privatization maximizes the bribes that can be collected by politicians. At the same time, in the presence of political patronage, privatization is preferable to public ownership because the latter may lead to excessive 11 This is because, from the government’s point of view, there is no difference between providing the right set of incentives to the private or public managers, and this holds even in the presence of moral hazard and adverse selection (Hart et al, 1997). 12 An example of such non-contractible quality could be the “development impact” of banking operations (i.e., by how much the lending activity of a given bank promotes economic development). 13 This is true only if those who choose the provider care about the non-contractible component. Therefore, this clearly applies to schools (one of the examples that Shleifer, 1998, and Hart et al., 1997, have in mind) but may not apply to banking if the non-contractible quality is, for example, the development impact of banking. 6 costs and misallocation of resources. The case is more ambiguous if corruption and political patronage go hand in hand. Following Shleifer (1998), it is possible to summarize the conditions under which state ownership is preferable to contracting and regulation as: (i) there are cost reducing actions that have a large negative effect on the non-contractible quality of the good or service; (ii) there is limited scope for innovation in the production and delivery of the good or service; (iii) there is limited scope for competition or competition does not concern the non-contractible quality; (iv) reputation considerations are not at play. The relationship between the costs and benefits of state ownership and the presence of corruption, patronage, and, more in general, a “weak” state is not very clear. While state ownership may increase the opportunities for corruption and patronage, a “weak” state makes contracting and regulation more difficult and hence may increase the benefits of state ownership. How Does This Apply to Banking? Most economists and policymakers agree that the inherent fragility of the banking system is indeed an issue but that this should not be addressed with direct state ownership. Banking regulation and supervision together with deposit insurance should be able to reduce banking fragility without eliminating the positive incentive to reduce costs and innovate that arise from private ownership.14 This is indeed the avenue followed by most industrial countries. It is, however, true that deposit insurance and regulation do not work satisfactorily in poor developing countries that are plagued by high levels of corruption and poor institutional quality (DemirgüçKunt and Detragiache, 2002). In such a situation, direct state ownership could increase the trust of the public in the banking system and lead to deeper financial markets. This was the original view of Alexander Gerschenkron that was recently formalized by Svetlana Adrianova, Panicos Demetriades, and Anja Shortland (2002). They justify their work using the example of Russia where public mistrust of banks induces most small savers to keep their funds outside the banking system and where 70 percent of retail deposits are with the largest state saving bank.15 Note that the argument can be made more generally in terms of a comparison of agency costs. Credible deposit insurance and effective regulation and supervision can offset the mistrust of depositors while limiting the contingent liability of the insurance agency (which in 14 In the case of banking, a possible source of cost reduction is better screening of potential debtors. This would reduce non-performing loans and hence reduce fragility of the banking system. 15 At the cross-country level there is a positive, but not statistically significant, correlation between the saving ratio and state ownership of banks. 7 this case becomes the principal of the problem). If regulation and supervision are ineffective, the cost in terms of insurance outlays of providing insured private banking may outweigh the agency costs of direct state ownership. Thus, the case for direct intervention hinges on the government’s ability to provide incentives to private bank owners and managers relative to its ability to do so for its own agents. The other arguments for state intervention in the banking sector relate to the presence of asymmetric information and externalities that lead to underprovision of banking services, during given periods of time (too little lending during recessions), or to certain economic sectors or geographical areas. The case for too little lending of private banks during recession, while still unexplored in the literature, would make a strong case for direct state ownership as opposed to the use of contracts. This is because, while private banks could be induced to increase lending by offering government guarantees or subsidies, the process would in practice require some sort of legislative action and would probably be longer than just instructing a manager of a public bank to increase lending.16 It should be pointed out, however, that there is some empirical evidence in support of the idea that the effectiveness of monetary policy is reduced (and not enhanced) by the presence of state-owned banks (Cecchetti and Krause, 2001). Moreover, a case can be made in favor of contingent contracts that activate in the event of a crisis.17 Direct intervention or contingent contracts may be the only tools of countries that due to high capital mobility and a fixed exchange rate or lack of credibility may not implement a countercyclical monetary policy. While there is a theoretical case for direct lending towards sectors and industries that generate positive externalities, critics of government intervention argue that this eventually leads to a situation in which credit allocation is dictated by political rather than economic considerations (Kane, 1977).18 It is however interesting to abstract from these considerations and ask whether, in the presence of a benevolent government, directed lending towards given industries or geographical areas would be better achieved by subsidizing private banks or by direct state ownership. Here, contract theory can help organize the ideas. Consider first the objective of providing banking services to isolated areas. This is a relatively straightforward case in which the government could write a contract and pay a private bank to open branches in a given location. In this case, contracting would seem to dominate direct ownership if the latter 16 The idea is similar to the argument that monetary policy has shorter implementation lags than fiscal policy. In addition, the marginal effect of subsidies on credit may be small if they are used in projects that, with or without the subsidy, would have been financed. 17 While limited to reducing the procyclical effect of prudential regulation, countercyclical provisioning requirements recently introduced in the regulatory framework of some developed countries are an example of this sort of contingent arrangement. See, e.g., Fernández de Lis et al. (2000). 18 Sapienza (forthcoming) looks at the case of Italy and finds strong support for this political view. 8 involves the de novo creation of a state-owned institution. A related concern usually voiced by public bank supporters is the lack of competition in underserved areas. As noted, however, optimal government intervention would entail in this case the enforcement of competition policies. Consider now the case in which a government wants to establish a development bank whose ultimate objective is to promote economic development by making loans to certain economic sectors at a subsidized interest rate, due to the presence of important externalities. The government could either establish a public development bank or contract a private provider. According to Hart et al. (1997), the private provider will have an incentive to reduce costs and innovate. However, the incentive to reduce costs may be in contradiction with the development objective. As economic development cannot be easily monitored (at least in the short term), the bank could take cost-saving actions that would reduce its long-term development impact. For instance, it could eliminate (or understaff) its research department and hence reduce its ability to identify and target sectors that generate large externalities.19 While this problem could be addressed by separating research from banking activities and by maintaining the former under state control, such a solution may be inefficient if there are important synergies between the two activities. This seems to suggest that development banking is a field in which there is a rationale for direct ownership. In fact, most development banks are indeed either public or have a mixed (public-private) structure (Figure 2). It is a common finding that state-owned banks tend to be less profitable than their private counterparts. Interestingly, the finding of profitable public banks would open the door for an alternative (also positive) failure story, namely, one in which the pressure for profitability induces public institutions to mimic private institutions, in what De la Torre (2003) called the Sisyphus syndrome. According to this story, if for prudential reasons retail public banks are subject to private regulatory standards and, in contrast with the social view, public recapitalizations are viewed as a sign of mismanagement and fuel pressures for privatization, public bank managers will deviate from their social mandate and mimic private banks in their credit allocation criteria. In this case, public banks, although efficient, would become redundant. Public Banks Around the World The industrial countries and Sub-Saharan Africa are the world’s regions with the lowest prevalence of state ownership of banks (around 30 percent in 1995, Figure 1). South Asia and the 9 Middle East are instead the regions with the largest share of state-ownership of banks (close to 90 percent in the former group of countries and above 60 percent in the latter). As one may expect, the transition economies of East and Central Europe are the countries that privatized the most during the 1990s, moving from almost full state ownership of banks (95 percent in 1985) to intermediate levels of state ownership in 1995 (data for 2001 indicate an even lower level of state ownership).20 Latin America has a level of state ownership of banks that is just below the developing countries average and similar to that of countries in the East Asia Pacific region (the data described here pull together commercial and development banks). La Porta et al. (2001) do an extremely careful analysis of the main characteristics of countries that have extensive state ownership of banks. We build on their work by separating the sample into developed and developing countries. This division allows us to check whether the relationship between the presence of public banks and a set of institutional and economic variables differs between these two group of countries. This is particularly important because some theories aimed at explaining the presence of public banks neatly apply to poor countries with underdeveloped institutions and financial markets but not to more developed countries. So, the reasons that may explain the presence of public banks in Central America may be very different from those that explain the existence of public banks in Western Europe. In particular, it is possible that public banks that were created with benevolent development objectives transformed themselves into an instrument of patronage that politicians refuse to privatize. If this is the case, pulling together developing and industrial countries would mix countries in the first stage (where the benefits of public banks may be greater than their costs) with countries in the second stage.21 When one restricts the analysis to developing countries there is a strong and statistically significant negative correlation between state ownership of banks and GDP per capita (indicating that poorer developing countries tend to have more state-owned banks). However, this correlation disappears if we control for an index of overall government intervention in the economy. In other words, for a given level of state interventionism, there seems to be no special relationship between GDP per capita and the propensity to intervene in the banking sector (column 1 of Table 19 Or just mimic the behavior of private commercial banks. (This is De la Torre’s Sisyphus syndrome.) For details of bank privatization in transition countries see Bonin, Hasan and Watchel (2003). 21 It is telling that La Porta et al find a negative and statistically significant relationship between state ownership of banks and subsequent growth in high income countries (where high income is defined as having initial GDP per capita income above the median), but fail to uncover any significant relationship between these two variables in low income countries (which, however, have a larger coefficient in the state-owned variable). 20 10 1). 22 23 At the same time, in the sample of industrial countries, state ownership of banks is never significantly correlated with GDP per capita. Finally, if all countries are pulled together, regressions analysis indicates that there is a negative correlation between GDP per capita and state ownership of banks (even after controlling for government intervention in the economy). These results suggest that the relationship between state ownership and the level of development is mostly driven by differences between developing and industrial countries rather than differences within these groups of countries.24 The next set of results shows that, once the state propensity to intervene in the economy is controlled for there is a significant negative relationship between financial development (measured as domestic credit over GDP) and state ownership of banks in developing countries. However, this result does not hold for industrial countries or for the joint sample of developing and industrial countries (Column 2, 5 and 7 of Table 1).25 Figure 3 shows that in the whole sample (if one does not control for initial GDP) there is a clear negative correlation (-0.41 and statistically significant at the 1 percent confidence level) between state ownership of banks (computed as a weighted average of the 1970-1985 and 1995 values) and financial development (averaged over the 1980-2001 period). Table 2 examines the correlation between state ownership of banks and several variables measuring institutional quality. It shows that if the sample is restricted to developing countries, there is no statistically significant correlation between state ownership of banks and each of corruption and democracy (surprisingly the correlation is positive), a positive and statistically significant correlation between state ownership of banks and rule of law, and a negative and statistically significant relationship between property rights and state ownership of banks. In the sub-sample of industrial countries there is no significant correlation between institutional variables and state ownership of banks either. The whole sample (including developing and 22 Table 1 focuses on the 1970s and Table 2 on the 1990s. Full regressions results are reported in the Appendix. The results reported in Tables 1and 2 and in the appendix differ from La Porta et al (2001) for several reasons. First of all, in order to emphasize that the results do not imply causality, the regressions reported here measure all the variables in the same period. So, state ownership in 1995 is regressed over average GDP per capita and financial development for the 1990s (and not in 1970) and state ownership in 1970 is regressed over average GDP per capita and financial development for the 1970s (La Porta et al. do not report regressions for state ownership in 1970). Rule of law and corruption are also measured in the 1990s. Finally, the GDP and financial development measures used here are from the World Development Indicator (WDI), while La Porta et al. complement WDI and IFS data with data from other sources and manage to obtain a slightly larger sample. 23 See also Column 1-2 of Table A1 in the appendix. 24 This is not due to the fact that there is limited variance in GDP per capita within the countries that are defined as developing. In fact this group of countries include both very poor countries (like Tanzania) and some high-income countries (like Singapore and Hong Kong). In fact, the coefficient of variation in the sample of developing countries (0.93) is not very different from that of the whole sample (1.19). 11 industrial countries) shows a negative and statistically significant relationship between property rights and state ownership of banks and no statistically significant relationship between state ownership of banks and the other institutional variables. Unfortunately, these results do not help clarify whether public banks’ existence is justified by development and social objectives or whether their existence is purely due to political reasons. In fact the correlation between state ownership of banks and each of poor institutional quality (as measured by lack of property rights), low financial development, government intervention in the economy, and low GDP per capita, is justified by all theories aimed at explaining state intervention in the banking sector. An alternative way to look at the issue is to use microeconomic data. But even in this case, the empirical implications of each theory are often difficult to distinguish given the available information because both the development and political view of public banks are consistent with low profitability of public banks, due to the financing of socially (but not privately) profitable investments, the dominance of agency costs, their exploitation for political patronage or their subordination to macroeconomic policy (Sapienza, 2002). Altunas, Evans and Molyneux (2001) investigate scale economies, inefficiencies, and technical progress for a sample of private, mutual, and publicly-owned banks in the German market. They find little evidence that privately-owned banks are more efficient than public and mutual banks. Indeed, inefficiency measures indicate that public and mutual banks have slight cost and profit advantages over their private commercial banking counterparts, a feature which may be explained by their lower cost of funds. On the other hand, their results suggest that public banks do not play the subsidizing role that the social view typically assigns to them. Sapienza (2002) studies the comparative performance of privately- and publicly-owned banks in Italy. She shows that: (i) state-owned banks charge lower interest rates than their private counterparts to similar firms, even if the latter have access to financing from private banks; (ii) state-owned banks’ lending behavior is affected by the electoral results of the party affiliated with the bank; (iii) state-owned banks favor mostly large firms; (iv) state-owned banks favor firms located in depressed areas. While the last finding is somewhat aligned with the development view, the first three findings provide strong evidence in support of the political view of stateowned banks. Although considerably more research needs to be done, there is no evidence that German and Italian public banks fulfill the role of financing rationed firms with large externalities. At best they mirror private banks (this is De la Torre’s Sisyphus syndrome) and at worst show signs of 25 The correlation is always significant for the 1990s see Table A2 in the appendix. 12 political bias. These previous studies do not focus on the potential effect of state-owned banks on volatility. As we already mentioned, state-owned banks could be used as a tool to smooth the economic cycle. Abstracting from this latter point, the existing evidence may justify some concern among policymakers in countries where state banks play a major role in credit allocation. Public Banks in Latin America: Evolution and Main Characteristics Latin America has a level of state ownership of banks similar to that of other developing countries. Most countries in the region privatized aggressively both in the 1970s (during the 1970-1985 period average state ownership of banks dropped from 64 to 55 percent) and 1990s (during the 1985-1995 period average state ownership of banks dropped from 55 to 40 percent).26 These averages mask large differences across countries in the region, with Costa Rica being the country with the largest share of government ownership of banks (90 percent in 1995, down from 100 percent in 1970, Figure 4) and Trinidad and Tobago the country with the smallest share of state ownership of banks (1.5 percent). Ecuador, Chile and Peru are the countries that privatized the most, moving from levels of state ownership that were above (or close to in the case of Peru) 90 percent, to state ownership below 40 percent (below 30 and 20 percent in the cases of Peru and Chile). Uruguay is the only country that increased state ownership of banks, moving from 42 percent in 1970 to 69 percent in 1995. Other countries experienced large swings in the bank privatization and nationalization process. Mexico, for instance, moved from 82 percent of state ownership in 1970 to 100 percent in 1985 and back to 35 percent in 1995. A similar pattern holds for Nicaragua, Colombia, El Salvador, and Bolivia.27 More recent data show that the pattern of bank privatization has continued in most countries. Over the 1995-2001 period, large bank privatizations raised US$ 5.5 billion in Brazil (with the privatization of BANESPA, Brazil’s largest bank raising US$ 3.6 billion) US$ 800 million in Mexico and more than US$ 500 million in both Colombia and Venezuela.28 Table 3 illustrates the recent evolution of state ownership of banks in 12 Latin American countries.29 It 26 Studies of bank privatization in Latin America include Beck, Crivelli and Summerhill (2003), Clarke and Cull (2002) and Haber and Kantor (2003). 27 In Nicaragua state ownership went from 90 (1970), to 100 (1985) to 63 percent. In Colombia, state ownership went from 57 (1970) to 75 (1985) percent and then back to 53 percent (1995). In El Salvador, state ownership went from 53 (1970) to 100 (1985) to 26 percent (1995). In Bolivia, state ownership went from 53 to 69 and then back to 18 percent (1995). 28 Source Megginson (2003). 29 The data are not directly comparable with that of figure 3 because the data in table 1 only include commercial banks and the data in figure 3 also include development banks. Furthermore, the data in figure 3 only include the assets of the ten largest banks, while table 1 includes all the banks operating in the 13 shows that Argentina, Brazil, Costa Rica and Nicaragua are the countries that privatized the most. The share of assets controlled by state-owned banks also dropped in Chile, El Salvador, and Guatemala while it remained more or less constant in Colombia. Figure 5 describes public bank performance indicators relative to that of private domestically owned banks.30 It shows that public banks charge lower interest rates than their private counterparts (at 30 basis points lower, this result is consistent with Sapienza’s findings discussed above) and also pay lower interest rates on their deposits (90 basis points less than private banks). It is also the case that public banks tend to lend more to the public sector (the difference between the share of public sector loans of private and public banks is 8 percentage points) and have a higher share of non-performing loans (about 8 percentage points). Finally, public banks have a lower profitability than their private counterparts (the difference in returns on assets is 0.4 percentage points). Table 2 illustrates the results country by country (it reports both the public and foreign coefficients) and shows that the relative profitability of public banks is particularly low in Colombia and Honduras (Costa Rica is the only case in which public banks are more profitable then their private counterparts). Brazil and Honduras are the countries where public banks pay and charge the lowest interest rates (again relative to domestically owned private sector banks), with a rate differential close to two percentage points in the case of loans in Brazil. Non-performing loans are particularly high for public banks in Costa Rica (this seems in contrast with their relatively high profitability), Guatemala and Honduras and public sector loans are particularly high in Chile and Costa Rica. Table 5 traces the evolution of public sector loans in public, private and foreign banks. There are three countries (Argentina, Brazil and Colombia) in which the share of public sector loans increased considerably over the 1995-2000 period, but only in Colombia do public sector banks seem to have absorbed a disproportionately large share of public sector debt. While these results should be taken with some caution because they are simple correlations that control only for bank size, they suggest that while public banks tend to be less efficient than their private counterparts (with higher non-performing loans, more loans to the public sector, higher overheads, and lower returns) they are also perceived to be safer and hence able to pay lower rates on their deposits and extend credit at a lower rate. An alternative country. Finally, the data of table 1 were computed by assigning 100 percent government ownership to banks that have at least 50 percent of assets owned by the government and 0 percent government ownership to others. 30 All the values were obtained by running a bank level regression, controlling for size (expressed as log of total assets) and including a dummy taking value one for public banks and a dummy taking value one for foreign-owned banks. The values plotted in Figure 4 are the coefficients of the public bank dummy. 14 explanation for this last result is that state-owned banks may benefit from indirect subsides coming from government deposits paying no or low interest rates.31 Finally, it is important to stress that state-owned banks may not maximize profits but social welfare. Therefore it could be the case that an efficient public bank loses money in projects with negative private present value but with positive externalities or social benefits. Development Banks Most of the literature on state ownership of banks either focuses exclusively on commercial banks or mixes commercial banks with development banks.32 However, these are very different types of institutions. Development banks are hybrid financial institutions that are better defined by the term “development finance institution” (Bruck, 1998) and are often primarily concerned with offering long-term capital finance to projects that are deemed to generate positive externalities and hence would be underfinanced by private creditors. Rather then working directly with the public they sometimes operate as second tier institutions (i.e., they operate through other banks) and often have a well-defined objective that is closely related to the economic development of either the country or a given sector. The last available survey (Bruck, 1998) indicates that there are 550 development banks worldwide with 152 development banks located in Latin America and the Caribbean. Figure 6 describes the relative importance of development banks in different regions of the world (expressed as a share of development bank assets over assets of the ten largest banks in each country). Latin America, together with South Asia and Sub-Saharan Africa, is characterized by a relatively large presence of development banks. There is some consensus that development banks played an important role in the industrialization of Continental Europe and Japan (Cameron, 1961, and Armedáriz de Aghion, 1999). Crédit Mobilier (a private institution with close government ties), for instance, played a very important role in financing the European railway system and through partnership with other banks contributed to overall European financial development.33 In Germany and Japan, development institutions were key to the post World War I and II reconstruction eras. According to Armedáriz de Aghion (1999) one key factor in the success of these financial institutions was their ability to disseminate their expertise due to their dispersed ownership (this is especially the 31 This is the case of Chile, where the Banco Del Estado de Chile manages the central government checking account. 32 Important exceptions include Armedáriz de Aghion (1999) Titelman (2003) and ALIDE (2003). 33 For a short history of Crédit Mobilier, see Rajan and Zingales (2003). Cameron (1961) provides a more detailed account. 15 case for institutions created before World War II) and charters that stated that these institutions should only provide supplementary finance (hence, leading to the necessity of cofinancing agreements). In comparing the experience of Crédit National de France with Nacional Financiera de Mexico, Armedáriz de Aghion (1999) suggests that the type of government involvement (with subsidized credit and loan guarantees in the first case and direct ownership in the second) and need for cofinancing agreements (strong in the first case and weak in the second) are among the factors that made the experience of Crédit National more successful than that of Nacional Financiera. She also argues that these findings are consistent with a theoretical model showing that well-targeted state intervention (via subsidies and credit guarantees) and the imposition of cofinancing restrictions are likely to maximize the positive spillover of development institutions. Not only may they lead to a better allocation of credit (co-financing may limit the opportunities for politically motivated credit allocation), but they also disseminate development expertise to the whole financial system. Latin America has a large number of institutions that define themselves as development banks and are part of ALIDE (Asociación Latinoamericana de Instituciones Financieras para el Desarrollo).34 Out of the 121 members of ALIDE, 75 are first tier banks, 21 second tier banks and the remaining have a mixed nature. The majority of these development banks are either stateowned or have mixed public-private ownership. In 2002 there were only 11 development banks with fully private ownership accounting for less than 2 percent of total assets of Latin American development banks (See Figure 2).35 Dominican Republic, Argentina, and Brazil have the largest number of development institutions (more than 10) and Paraguay, Nicaragua, Peru and Uruguay, the smallest (two or less). Development banks are particularly important in Uruguay, Brazil, Panama, the Dominican Republic, and Costa Rica (where, in 2001 loans totaled more than 15 percent of GDP) and relatively less important in Ecuador, Venezuela, Honduras, Peru, and El Salvador. BNDES in Brazil is the largest development bank with total net loans in 2002 of US$ 28.3 billion and annual disbursements of approximately US$ 11 billion (Figure 7). The second and third largest development banks are also Brazilian (Banco do Brasil and Caixa Economica Federal) followed by two Mexican (NAFIN and BANOBRAS) and an Argentine institution (Banco de la Nación Argentina). It should be clear from this classification that the sample of development banks also 34 The self-definition is adopted because it is difficult to define whether an institution is a development bank or not. 35 These were Banco Industrial S.A. (operating in Bolivia and Guatemala), Banco del Desarrollo (Chile), Banco BHD S.A., Banco Dominicano del Progreso S.A., Banco de Desarrollo Citicorp (the Dominican 16 includes institutions that originally functioned as development banks but now mostly engage in commercial banking activities. If these banks are dropped from the sample the share of development bank loans over GDP drops substantially. Brazil becomes the country with the largest presence of development banks followed by Mexico, Colombia and Chile. Development banks tend to have low profitability and their return on assets tends to be lower than that of private banks in the country (Figure 8).36 This is particularly true for Guatemala, Chile, Mexico and Colombia. It should be pointed out, however, that the figure for Guatemala is completely due to the disastrous returns of one institution which had an ROA of –26 percent! If this institution is dropped from the sample the return on assets of development banks increases to 0.6 percent. In the case of Chile the negative results are due to the crisis at CORFO (Corporación de Fomento de la Producción), which in 2001 had an ROA of –4.8 percent (ROA were mostly positive in the previous years). In the cases of Brazil and Peru, however, there is no large difference between the profitability of development and private commercial banks (this could be due to the fact that they get lower cost of funds) and in El Salvador and Bolivia, development banks seem to be more profitable than private commercial banks. Do Public Banks Play a Useful Role in Economic Development? The best way to test whether public banks satisfy some development objective or exist for purely political motives is to evaluate their effect on the economy. The key question here is whether public banks generate positive spillovers on financial development and economic performance. Do they promote growth and financial development? Do they help in smoothing shocks and limit credit and deposit volatility in times of crisis? In terms of their potential spillovers to the rest of the economy, the few existing empirical studies have tended to focus on the way in which public ownership of banks affects the evolution of the private banking sector and financial markets as a whole and, through them, the performance of the economy. Looking at the impact of public participation in the banking sector on financial development, Barth, Caprio and Levine (2002) argue that greater state ownership of banks tends to be associated with more non-performing loans but they find that, after controlling for bank regulation, government ownership of banks is not robustly linked with other indicators Republic) , Banco Empresarial S.A., Financiera Guatemalteca S.A., Financiera Industrial S.A. (Guatemala), FEDECREDITO (El Salvador) 36 The figure compares average return on assets (weighted by bank assets) for development bank (excluding first tier banks) and private commercial banks. 17 of bank development and performance.37 These results are somewhat in contrast with their previous work (Barth, Caprio and Levine, 2000) where, for a sample of 59 developed and developing countries, they found a negative association between state-ownership and financial depth as measured by the ratios of bank and non-bank credit to the private sector over GDP, and by the value of securities traded domestically, even after controlling for economic development and the quality of government. The interpretation of these findings in terms of causality is rather difficult, as the authors themselves recognize. Their correlations pair state-ownership ratios in 1997 with measures of financial development in previous years, suggesting that financial underdevelopment leads to public participation. This finding is in line with the development view that implies that public banks are likely to be more prevalent at earlier stages of financial development. Indeed, even if one is willing to accept that public shares are sufficiently stable to be used as measures of public participation in the past, no results are reported on its interaction with economic development and institutional quality, which may shed light on whether the presence of public banks is induced as a substitute for private institutions in a less than conducive environment. In addition, the fact that bank failures during a crisis tend to be followed by nationalization may generate a positive correlation between the propensity to face banking crises and the extent of ex-post state ownership, independently of whether or not state participation increases banking fragility. The results of Barth et al are consistent with those in La Porta et al (2001), who focus more specifically on the determinants and implications of state ownership of banks. Original data on public ownership (which comprises public shares for about 90 economies for the years 1970, 1985 and 1995) show that government ownership of banks at an earlier period is associated with a slower subsequent development of the financial system and slower economic growth. Their tests, while controlling for initial conditions (financial and economic development, state ownership ratio), are still limited to cross-section correlations and, as La Porta et al. note, “are not conclusive evidence of causality.” This is particularly true in light of the strong persistence of both credit shares and state-ownerships ratios.38 A negative link between government ownership and financial development is not at odds with Gerscherkron´s (1962) view since underdeveloped financial markets may call for a stronger public presence to foster economic growth. 37 They also study the relationship between banking crises and state ownership of banks, but they do not find a significant link. Some evidence for such a relationship is found by Caprio and Martinez Peira (2002). 38 The correlation between state ownership of banks is 0.77 (1970 vs 1995), 0.88 (1970 vs 1985) and 0.79 (1985 vs 1995). In turn, the correlations between private credit over GDP ratios are 0.68 (1960 vs 1986), 0.92 (1960 vs 1970), and 0.78 (1970 vs 1986). P-values are equal to 0.00 in all cases. 18 As the statistical analysis of La Porta et al groups together very different countries, including former socialist economies where state ownership was the rule and for which output data for earlier periods are less reliable, a revision of their results is warranted. Tables 6 and 7 revisit their findings using their own measures of state-owned (public) shares in the banking sector and updating and extending in time the private credit and GDP data following their definitions and sources.39 Table 6 focuses on the relationship between state ownership of banks and subsequent financial development. Column 1 reproduces the results in Table IV of La Porta et al. for ease of comparison. Columns 2 and 3 replicate the regression using the new data. Reassuringly, the original results remain virtually unchanged, indicating that state ownership of banks depresses subsequent financial development even after controlling for initial GDP and the initial level of financial development. This is also true when 1970 is used as the initial year (this dating is in line with the earlier measure of the state ownership ratio). While the negative association between public shares and private credit growth is robust, causality and omitted variable issues are more difficult to assess. In particular, according to the development view, public banks are more likely to arise in a context in which private financial intermediation is discouraged by institutional deficits. This suggests that the negative link between private financial intermediation and state ownership could be due to either reverse causality or to the omission of institutional variables. Results in columns 4 and 5 provide a robustness check for this potential simultaneity problem by implementing the state-ownership variable using an index of state-owned enterprises as a share of the economy.40 With these specifications the effect of state ownership of banks on subsequent financial development, while still negative, ceases to be significant.41 Columns (6) and (7) report additional robustness checks by focusing on the impact of state-ownership at shorter horizons by splitting the sample into two periods (1970-1985 and 39 Initial per capita GDP is expressed in current U.S. dollars (source: World Development Indicators). Credit to GDP ratios are computed as credit to the private sector (lines 22.d.f and 22zw, plus line 42d) over GDP (source: International Financial Statistics). The growth in the credit to GDP ratio is computed as the average of the log difference of the ratio over the period, for those countries for which a minimum of 10 observations is available. In order to maintain data homogeneity columns 2-10 only use data for which WDI and IFS information are available this reduces the sample to 70 observations 40 The variable, computed as the average of the index for 1975 and 1995 sourced from Gwartney et al. (1996), is shown in LLS to be highly correlated with state-ownership of banks. In addition, it is not significantly correlated with private credit growth once the share of public banks is included. 41 It should be pointed out, however, that the coefficient, while not statistically significant, does not chnage in value, which suggests that the change in significance may be due to the loss of efficiency typical of IV estimation. 19 1986-2002) in line with the available data on public shares.42 The link is still significant at 10 percent for the later period, but not for the first period. In sum, the evidence that the prevalence of state ownership in the banking sector conspires against its ultimate development appears to be weaker than hinted by previous studies. However, there is no clear indication that state ownership has the positive catalytic effect that some advocates have suggested. A more balanced reading of these results would indicate that public banks do not play much of a role in the development of their private counterparts. The same conclusion can be extracted from the more elusive question on the impact of public banks on long-run economic growth. While a direct nexus is difficult to construct, there are at least two indirect avenues through which one could envisage a link, either positive or negative: public banks may help develop projects with important externalities that would otherwise be shelved, or; they may inhibit financial development, which ultimately reflects in a poorer investment record and growth underperformance. Table 7 explores the link between state ownership of banks and economic growth. As before, it follows closely the work of La Porta et al., who report a negative association between state ownership and growth (column 1 reproduces their results for comparison). The first thing to note is that the negative link between state ownership of banks and growth does not disappear when the regression controls for the growth rate of financial development (column 2). This suggests that the relationship between bank ownership and growth is unrelated to changes in the amount of credit during the period, which is seemingly at odds with the view that financial underdevelopment (measured as total credit) is the main channel through which bank ownership may influence economic performance.43 This result may reflect the fact that total credit does not capture allocative efficiencies and is an intriguing finding that warrants further exploration. Column 3 interacts financial development with bank ownership (proxies of credit extended by public banks and credit extended by private banks, respectively) and shows that the two types of credit have an identical effect on growth. There is a possibility that the equation in column 3 is miss-specified because public ownership of banks may affect overall financial development generating non-linearities that are not controlled for in the specification of column 3. Column 4 addresses this issue by controlling for the main effect of public ownership. The results change dramatically. While the main coefficient on bank ownership is negative, high, and statistically significant (indicating that state ownership of banks is harmful for growth), the results now 42 Private credit growth is here computed only for countries with at least five observations within the period. 43 However, this result is consistent with la Porta et al.’s finding that the negative effect of state ownership of banks manifests itself in lower productivity rather than lower capital accumulation. 20 indicate that, conditional on a given share of public bank ownership, the positive impact of financial development on growth is larger in countries with a large share of state-owned banks.44 An alternative way to read this last result is that state ownership of banks has its strongest negative impact on growth in countries with low financial development. This is in line with the La Porta et al’s finding that state ownership of banks has a negative impact on growth in countries with low financial development but no statistically significant effect on growth in countries with high financial development. To make sense of this battery of results, it is useful to compare the total effect of state ownership in pairs of countries with similar levels of financial development but different levels of state ownership. Consider, for instance, the cases of Japan and Italy and South Africa and Turkey. In 1960, the first two countries had high and similar levels of financial development (around 58 percent of GDP, which put both countries above the 90th percentile of the distribution of financial development) and the other two countries had intermediate levels of financial development (17 percent and 16.8 percent respectively, just below the median value for the period). At the same time, Italy and Turkey had a large share of state ownership of banks (76 and 82 percent, respectively) and Japan and South Africa had a much smaller share (7 and 0 percent, respectively). According to the estimates in column 1 of Table 7, differences in state ownership of banks would have led Japan to grow at a rate that was 1.2 percentage points higher than Italy, and South Africa at a rate that was 1.5 percentage points higher than Turkey. The estimates in column 3 paint a very different picture, predicting that Italy would grow at the same rate as Japan and Turkey at the same rate as South Africa. Finally, the estimates in column 4 predict that Italy should have grown faster than Japan (by approximately 0.5 percentage points) and Turkey much slower than South Africa (by approximately 2 percentage points). These last results are somewhat difficult to interpret because they suggest that state ownership of banks promotes growth (or at least does not hurt growth) only in countries with highly developed financial systems, indicating that state-owned banks seem to be beneficial for countries that in principle should not need them. A possible explanation for this finding is that countries with well-developed financial systems are better equipped to deal with the distortions that arise from government ownership of banks (La Porta et al.). An alternative interpretation is that public bank ownership is proxying for some excluded variable which is correlated with both bank ownership and subsequent growth 44 The fact that the main coefficient for private credit is close to zero is probably due to non-linearities in the relationship between financial development and growth and should not be interpreted as saying that financial development has no growth effect in countries with no state-owned banks. Indeed, the point estimates suggest that for the average country (i.e. for the country with average level of state ownership of 21 (institutional quality, for instance) or that there is a causality issue with the relationship between state ownership of banks and financial development that is not fully addressed by the statistical models of Table 7. The remaining columns of the table show that the results are somewhat sensitive to the sample. For instance, column 5 uses data from La Porta et al but restricts the sample to countries for which World Bank and IMF data are available and finds no significant correlation between initial state ownership and subsequent growth. The same is true if data for the 1970-1995 period are used. While these findings somewhat mitigate the previous evidence for a negative effect of state ownership of banks they also fail to support the view that public banks mitigate market imperfections that lead to allocative inefficiencies. Indeed, the preliminary conclusion from this evidence suggests that, in terms of its impact on financial development and long-term growth, the average public bank does not appear to be any better than its private peers. Besides examining their effect on long-run growth, it is also interesting to check whether the presence of state-owned banks could play a role in smoothing the effect of external shocks. Appendix 1 (column 1 of Table A3) shows that external shocks have a large and statistically significant effect on GDP even after controlling for lagged values of GDP external demand and real credit (details of the estimation procedure are reported in the appendix). It also shows that financial development (measured as credit to the private sector divided by GDP) plays an important role in reducing the impact of the external shocks. In particular, the point estimates indicate that moving from the country in the 25th percentile of financial development to the country in the 75th percentile of financial development reduces GDP’s elasticity to external shocks by 20 percent. Column 3 of table A3 in the appendix checks whether credit to the private sector extended by public banks (measured as the interaction between the share of public banks and financial development) has a differential effect with respect to overall financial development. The main finding is that the coefficient is large and positive but not statistically significant and it is not robust to alternative specifications. While this does not provide convincing evidence that the presence of state-owned banks attenuates the smoothing effect of financial development, it clearly shows that state-owned banks play no role in amplifying the beneficial effect of financial development. The results are illustrated in Figure 9. The XX line measures how GDP responds to an external shock if all countries are grouped together. It shows that after a negative shock, GDP drops in period 1 and then recovers to reach its initial level after 3 periods. The YY line focuses banks) a one percentage point increase in financial development is associated with a half percentage point increase in growth. 22 on countries with low state ownership of banks (i.e., countries where state ownership of banks is below the sample median) and shows that the negative impact of the shock is somewhat smaller than that for the overall sample. Finally, the ZZ line focuses on countries with high state ownership of banks and shows that in this case the impact of the external shock is much larger and more persistent (GDP drops for two periods) and reaches a maximum effect of GG percent (versus HH percent in the full sample). It is also possible to check whether state ownership of banks affects how credit responds to external shock. The appendix shows (column 10 of table A3) that, in contrast with the analysis for GDP elasticity, state-owned banks may play a role in reducing the elasticity of credit to external shocks (although, as before, the coefficient is not statistically significant). Taken at face value, this seems to suggest that state-owned banks play a useful role in smoothing credit but that their credit is less effective in smoothing the overall output effect of external shocks. It is also possible to check whether state-owned banks allocate credit better than private banks by using industry level data to identify the role of bank ownership in explaining industry growth and volatility. Such a strategy was first employed by Galindo and Micco (2004) to check whether government-owned banking promotes growth by directing credit towards the industries that rely more on external finance and/or towards industries where informational asymmetries may be higher (the methodology and full results are discussed in the appendix). After having established that more developed financial systems tend to favor economic sectors that for technological reasons demand more credit (column 1 Table A4, replicating Rajan and Zinagales’s, 1998, result), it is possible to show that state ownership of banks reduces the beneficial effects of financial development. In particular, for a country whose median financial development rises from the 25th percentile of state ownership of banks to the 75th percentile of state ownership of banks, the beneficial effect of financial development declines by more than 50 percent. Interestingly, financial development is much more important for sector growth in developing than industrial countries and the negative effect of public banks is much stronger in industrial countries (state ownership of banks has no significant effect if the sample is restricted to developing countries). The appendix also checks if state ownership of banks has an effect on sectoral volatility but finds no evidence to support this hypothesis. While the finding that state ownership of banks reduces the beneficial effect of financial development may provide some evidence that state-owned banks allocate credit less efficiently than private banks, this is not equivalent to saying that state ownership reduces growth by itself; it just mitigates the growth effect of financial development due to allocative efficiency. Furthermore, the empirical strategy of the table is still plagued by a possible feedback from state- 23 owned banks to total financial development. In particular, if state ownership of banks plays a positive role in stimulating financial development (as claimed by the development view) the results discussed here would overestimate the negative role of state-owned banks on allocative efficiency which could, indeed, be positive once their effect on financial development is taken into account. If, on the other hand, state ownership of banks depresses financial development (as claimed by the political view) the results reported above would underestimate the negative role of state-owned banks. It is also interesting and worrisome that the exercise described above yields results which contrast with those of the cross-country regressions of Table 7 that found that state ownership of banks may amplify the positive effect of financial development. All this points to a deeper causality issue that the tests reported here cannot address. Conclusions Several prominent development economists writing in the 1960s and 1970s strongly supported government intervention in the banking sector and direct state ownership of banks. The more recent view is that state ownership of banks is not beneficial for economic development or, in the words of a recent World Bank report that: “…whatever its original objectives, state ownership of banks tends to stunt financial sector development, thereby contributing to slower growth.” (World Bank, 2001, P. 123). This paper reviews the existing evidence on the role of state ownership of banks, tests its robustness, and introduces new evidence. While the paper finds some evidence in support of the idea that state-owned banks do not allocate credit optimally, it also shows that the results demonstrating that state ownership inhibits financial development and growth are less robust than previously thought. On the other hand, the paper finds no evidence in support of the idea that state ownership of banks has beneficial effects in terms of spurring growth or reducing volatility.45 One argument that is often invoked against state ownership of banks is that private banks tend to be more profitable than public banks. There is in fact some robust evidence that this is the case (especially in developing countries). It should be pointed out, however, that whatever merit the development view has, it is unfair to judge it by using the profitability benchmark. In fact, it has been argued that having public banks that maximize profitability would generate an inherent 45 Paradoxically, some of the results seem to indicate that state-owned banks work best in countries with highly developed financial systems. 24 contradiction and a vicious circle in which public banks would start with a social policy mandate and concentrate on high risk and low private return activities. This would lead to recurrent losses and need for recapitalization that would soon be followed with a re-orientation towards profitable activities in competition with private banks. In turn, this would lead to insufficient attention to the social policy mandate and political pressure to restart the cycle (De La Torre, 2002).46 In this context, public banks should be judged on the basis of their development and stabilizing effect. The main problem in identifying whether state-owned banks play a positive role in economic development is that both the political view (which assumes that state-owned banks have a negative impact on the economy) and the development view (which assumes that public banks can play a beneficial role) are consistent with a negative relationship between state ownership of banks and both financial development and institutional quality. The main difference between these two interpretations lies in the fact that, according to the development view, state ownership helps promote financial development at initial stages and mitigates the negative effect of poor institutional quality (which would be even more damaging without public intervention) whereas, according to the political view, state ownership of banks depresses financial development and possibly promotes corruption. As both financial development and institutional quality are closely related with economic growth, it is very difficult to make a statement on the role of public banks without disentangling the causal relationship between these variables and state ownership of banks. Thus, a definitive answer on the development role of state-owned banks will require addressing this causality problem, one of the thorniest issues in economics.47 46 Some may find that the fact that state-owned banks have fiscal costs together with the fact that there is no convincing evidence that they play a beneficial role is sufficient to conclude that they should not exist. While it is difficult to argue with such a logic, it should be pointed out that this reasoning also applies to several other areas of government intervention (the IFIs,, inter alia). 47 A brief illustration of the reverse causality problem is useful. Suppose a social scientist wanted to test the hypothesis that going to the hospital makes people sick by looking at the health status of a randomly selected group of people. The social scientist would probably find a positive correlation between the probability of being sick and number of visits to the hospital. It would, however, be wrong to use this evidence that going to the hospital makes people sick. It is very likely that the causality goes in the opposite direction: sick people tend to go to the hospital more often! While there are statistical techniques that can address the causality issue (the instrumental variable method is such a technique) they are often difficult to apply because they require identifying a variable that is correlated with the variable of interest (in the previous example, the probability of visiting a hospital) without being directly correlated with the outcome variable (in the previous example, the probability of being sick). Such variables (called instruments) are 25 BOX 1: Bank Privatization: A Survey of the Empirical Evidence While there is some evidence that private banks outperform public banks in terms of profitability and operating efficiency (see Meggison, 2003) and hence privatization could involve fiscal benefits and increase microeconomic efficiency, the evidence on bank privatization indicates that, in developing countries, the benefits from privatization have been limited and that in some cases privatization had a negative effect. According to Haber and Kantor (2003) Mexico’s bank privatization “produced disastrous results.”48 Chile in the early 1980s is another example where a fast privatization led to a large financial crisis. Clarke and Cull (2002), instead, suggest that bank privatization in Argentina was highly beneficial and involved very large fiscal savings (up to half of a typical province total expenditure). There are a limited number of studies that try to measure the effect of bank privatization in developing and industrial countries.49 Their main findings are that in developed countries bank privatization leads to improvement in terms of profitability and stock performance but that these improvements are smaller than what is typically found in the case of non-financial companies. Studies that focus on non-transition developing countries tend to find that privatization has a positive impact on bank competition but no significant impact on profitability or operating efficiency and that poorly done privatizations (like the one in Mexico in the early 1990s) can carry very large costs. Studies focusing on transition countries found more beneficial effects of privatizations. The surveyed studies also found that privatization tends to be more beneficial if the state completely relinquishes its ownership (therefore full privatization tends to be better than partial privatization) and that privatization that involves (or, at least, allows for) the entry of foreign banks tends to be better than privatization that excludes foreign ownership. Privatization can be implemented either by directly selling the bank’s assets to a set of strategic investors or by selling equity shares in the capital markets (voucher privatization implemented by some transition countries shares many of the characteristics of this latter privatization method). There is some evidence that share-issues privatization tends to work better in countries with a strong institutional environment and well-developed capital market while often very hard to find, but the causality issue is very important because, in Rajan and Zingales’ (2003) words: “Correlation is the basis for superstition, while causality is the basis for science.” (p 109). 48 They do not criticize that idea of privatization but the way it was implemented in early 1990s. One of the major points of criticism was the exclusion of foreign banks from the privatization process. 49 This box is based on the surveys of William Megginson (2003) and George Clark, Robert Cull and Mary Shirley (2003). 26 direct asset sales (especially those involving foreign strategic investors) are preferable in countries with poor institutions and limited capital market development. 27 Appendix A. State ownership of banks and responses to external shocks Besides examining their effect on growth, it is also interesting to check whether the presence of state-owned banks could play a role in smoothing how external shocks affect the economy. The first step consists of building a measure of external demand obtained by computing a weighted average of trading partner’s GDP growth. Next, it is possible to regress this measure on its two lags and use the regression’s residuals as a measure of external shock. Column 1 of Table A3 uses a panel of 77 countries for the 1980-2001 period and shows that this external shock is positively and significantly correlated with GDP growth (expressed as a deviation from its long run trend) even after controlling for lagged value of GDP, lagged value of external demands and real credit. Column 2 shows that financial development plays an important role in reducing the impact of the external shock. In particular, the point estimates (1.96 and -0.919) indicate that moving from the country in the 25th percentile of financial development (0.22) to the country in the 75th percentile (0.62) the effect of an external shock on GDP goes from 1.78 to 1.39, equivalent to a 20 percent drop in the elasticity of GDP to external shock. Column 3 further interacts the effect of the external shock and financial development with the share of state-owned banks in the economy and finds that the presence of state-owned banks attenuates the smoothing effect of financial development; however, the coefficient is not statistically significant.50 Columns 6 and 7 split the sample according to the level of state ownership of banks. Column 6 restricts the analysis to countries with high state ownership of banks (i.e. state ownership above the median value of 38 percent) and column 7 restricts the analysis to countries with low state ownership of banks. Interestingly, external shocks seem to have a much larger effect on GDP growth in countries with large state ownership of banks and, in this subset of countries, financial development amplifies (even though the coefficient is not statistically significant) rather than smooth the effect of the external shocks. The opposite is true for the case of countries with low state ownership of banks. In this case, the effect of the external shock on GDP is much smaller (about half that of column 6) and financial development helps in smoothing the effect of external shocks (again the coefficient is not statistically significant). While these results do not provide definitive evidence that the presence of state-owned banks amplify the 50 For a country with a median level of financial development (0.42) located in the 25th percentile of the distribution of state ownership of banks (0.17), the estimated GDP elasticity of an external shock is 1.55; for a country with similar financial development but state ownership in the 75th percentile (0.61) the effect of the external shock would climb to 1.75. 28 effect of external shocks, they surely show that state-owned banks play no additional role in smoothing the effect of external shocks. It is also possible to check whether the presence of state-owned banks has an effect on how credit to the private sector responds to external shocks. Column 8 of table A3 shows that external shocks have a significant effect on private credit (expressed as a deviation from its longrun trend) and column 9 shows that financial development plays an important role in smoothing the effect of this shock. In particular, moving from the 25th to the 75th percentile of financial development reduces the elasticity of real credit from 6.0 to 4.8. Column 10 shows that the smoothing effect of financial development is even stronger in countries with a large presence of state-owned banks (even though the coefficient is not statistically significant). This result is confirmed by columns 11 and 12 that split the sample between countries with high and low state ownership of banks. In the first sub sample, external shocks have a strong effect on real credit but this effect is significantly smaller in countries with high levels of financial development. In the second sub sample there is no statistically significant correlation between external shocks and real credit. This seems to suggest that in countries with a high share of state-owned banks external shocks have a much larger impact on credit but that financial development plays a more important role in mitigating the effects of these shocks. B. State ownership of banks and sectoral growth and volatility An alternative empirical strategy consists of using industry level data to identify the role of private and government-owned banking in explaining industry growth and volatility. Such a strategy was first used by Galindo and Micco (2004) to check whether government-owned banking promotes growth by directing credit towards the industries that rely more on external finance and/or towards those industrial segments where informational asymmetries may be higher. Their empirical strategy is based on Rajan and Zingales (1998) who show that more developed financial systems, as measured by the ratio of credit to the private sector to GDP, are able to provide cheaper funds to firms that require more external finance. In particular, Galindo and Micco identify the requirements of external finance by using Rajan and Zingales' (1998) estimations of sectoral external financing needs and the ability to pledge assets as collateral by using the share of intangible assets with respect to total assets developed by Claessens and Laeven (2003). They use these measures to check whether state ownership of banks affects the borrowing ability of sectors that rely the most on external finance and have relatively fewer assets to pledge as collateral. The basic idea is that if the provision of credit by each type of bank is 29 efficient, larger amounts of credit supplied by it should have positive impacts on the relative growth rate of those industries that require external finance more and that have higher shares of intangible assets. Table A4 reports the basic results. The dependent variable measures the growth rate of real value added of sector j in country i, all the regressions control for country and industry fixed effects and for the share of industry j in country i of total value added in manufacturing at the beginning of the period (initial VA).51 The first column reproduces Rajan and Zinagales’s results showing that more developed financial systems tend to favor economic sectors that for technological reasons demand more credit. The second column shows that, after controlling for external financing needs, the ability to pledge collateral has no additional effect on sector growth. Column 3 further interacts financial development with the initial share of state-owned banks and shows that state ownership of banks reduces the beneficial effect of financial development. In particular, for a country with median financial development (0.415 in the sample of Table A4) going from the 25th percentile (0.18) of state ownership of banks (0.18) to the 75th percentile (0.69) of state ownership of banks reduces the beneficial effect of financial development by more than 50 percent (from 0.008 to 0.003). It should be clear that this is not equivalent to saying that state ownership reduces growth by itself; it just mitigates the growth effect of financial development. In particular, if state ownership of banks plays a positive (negative) role in stimulating financial development the results of table A4 would overestimate (underestimate) the negative role of state-owned banks. Column 4 shows that the results are robust to substituting the initial values of financial development and state ownership of banks with their period averages and column 5 shows that the results are robust to estimating the model for the 1985-1997 period. Interestingly, columns 6 through 9 show that financial development is much more important for sector growth in developing than industrial countries and that the negative effect of public banks is much stronger in industrial countries. In fact, the interaction term is never significant in the sub sample of developing countries of columns 9 and 10. Table A5 presents an experiment similar to that of table A4 but rather than looking at the effect of financial development on sectoral growth, it looks at how financial development affects sectoral volatility (so the dependent variable is the standard deviation of sectoral value added growth). Column 1 shows that financial development reduces volatility the most in sectors that 51 The sector value-added data comes from the United Nations Statistical Division and covers 20 industries in 33 countries. The measure of financial sector development is the standard ratio of credit to the private sector to GDP from the World Development Indicators of the World Bank. The measure of credit provided by government-owned institutions is from La Porta et al. (2001). 30 have a larger external financing need (this is the FDRZ interaction) and have a larger share of intangible assets (FDITG interaction) while there is no differential effect according to the level of inventory over assets (FDRD interaction). In all cases, the regressions show that the share of state-owned banks has no significant effect on volatility. 31 References Adrianova, S., D. Panicos and A. Shortland. 2002. “State Banks, Institutions and Financial Development.” Leicester, United Kingdom: University of Leicester. Mimeographed document. Asociación Latinoamericana de Instituciones Financieras para el Desarallo (ALIDE). 2003. Finanzas para el desarrollo: Nuevas soluciones para viejos problemas. Lima, Peru: ALIDE. Altunbas, Y., L. Evans and P. Molyneux. 2001. “Bank Ownership and Efficiency.” Journal of Money, Credit and Banking 33(4): 926-954. Atkinson, A.B., and J. Stiglitz. 1980. Lectures on Public Economics. London, United Kingdom: McGraw-Hill. Armendáriz de Aghion, B. 1999. “Development Banking” Journal of Development Economics 58: 83-100. Barth, J., G. Caprio and R. Levine. 2001. “The Regulation and Supervision of Banks Around the World.” In: R. Litan and R. Herring, editors. Integrating Emerging Market Countries into the Global Financial System. Brookings-Wharton Papers on Financial Services. Washington, DC, United States: Brookings Institution Press. Barth, J., G. Caprio and R. Levine. 2002. “Bank Regulation and Supervision: What Works Best?” NBER Working Paper 9323. Cambridge, United States: National Bureau of Economic Research. Beck, T., J.M. Crivelli and W. Summerhill. 2003. “State Bank Transformation in Brazil: Choices and Consequences.” Paper presented at the World Bank Conference on Bank Privatization in Low and Middle-Income Countries, November 23, Washington, DC. Bonin, J., I. Hasan and P. Watchel. 2003. “Privatization Matters: Bank Performance in Transition Countries.” Paper presented at the World Bank Conference on Bank Privatization in Low and Middle-Income Countries, November 23, Washington, DC. 32 Bruck, N. 1998. “The Role of Development Banks in the Twenty-First Century.” Journal of Emerging Markets 3: 39-67. Burgess, Robin & Pande, Rohini, 2004. "Do Rural Banks Matter? Evidence from the Indian Social Banking Experiment," CEPR Discussion Paper 4211, C.E.P.R. Discussion Papers Cameron, R. 1961. France and the Economic Development of Europe, 1800-1914. Princeton, United States: Princeton University Press. Caprio, G., and S. Martinez Peira. 2002. “Avoiding Disaster: Policy to Reduce Banking Crises. ” In: E. Cardoso and A. Galal, editors. Monetary Policy and Exchange Rate Regimes. Cairo, Egypt: Egyptian Center for Economic Studies. Cecchetti, S., and S. Krause. 2001. “Structure, Macroeconomic Stability and Monetary Policy.” NBER Working Paper 8354. Cambridge, United States: National Bureau of Economic Research. Claessens, S., and L. Laeven. 2003. “Financial Development, Property Rights, and Growth.” Journal of Finance 58(6): 2401-2436. Clarke, G., and R. Cull. 2002. “Political and Economic Determinants of the Likelihood of Privatizing Argentine Public Banks.” Journal of Law and Economics 45: 165-197. Clark, G., R. Cull and M. Shirley. 2003. “Empirical Studies of Bank Privatization: An Overview.” Paper presented at the World Bank Conference on Bank Privatization in Low and Middle-Income Countries, November 23, Washington, DC. Corrigan, G. 1982. “Are Banks Special?” 1982 Annual Report Essay. Minneapolis, United States: Federal Reserve Board. http://minneapolisfed.org/pubs/ar/ar1982a.cfm Demirugüç-Kunt, A., E. Detragiache. 2002. “Does Deposit Insurance Increase Banking System Stability? An Empirical Investigation.” Journal of Monetary Economics 49: 1337-1371. Diamond, D. 1984. “Financial Intermediation and Delegated Monitoring.” Review of Economic Studies 51: 393-414. Fama, E. 1985. “What’s different About Banks?” Journal of Monetary Economics 6: 3957. 33 Fernández de Lis, S., J. Martínez Pagés and J. Saurina. 2000. “Credit Growth, Problem Loans and Credit Risk Provisioning in Spain.” Madrid, Spain: Bank of Spain. Mimeographed document. Freixas, X., and J-C. Rochet. 1997. Microeconomics of Banking. Cambridge, United States: MIT Press. Haber, S., and S. Kantor. 2003. “Getting Privatization Wrong: The Mexican Banking System 1991-2003.” Paper presented at the World Bank Conference on Bank Privatization in Low and Middle-Income Countries, November 23, Washington, DC. Hart, O., A. Shleifer, and R. Vishny. 1997. “The Proper Scope of Government: Theory and Application to Prisons.” Quarterly Journal of Economics 112: 1127-1161. Jensen, M., and W. Meckling. 1976. “Theory of the Firm, Managerial Behavior, Agency Costs, and Ownership Structure.” Journal of Financial Economics 3: 305-360. Gerschenkron, A. 1962. Economic Backwardness in Historical Perspective. Cambridge, United States: Harvard University Press. Gwartney, J., R. Lawson and W. Black, editors. 1996. Economic Freedom of the World 1975-1995. Vancouver, Canada: Fraser Institute. Kane, E. 1977. “Good Intentions and Unintended Evil: The Case Against Selective Credit Allocation.” Journal of Money, Credit and Banking 9: 55-69. La Porta, R., F. López-de-Silanes, and A. Shleifer. 2002. “Government Ownership of Banks.” Journal of Finance 57: 265-301. La Porta, R.; F. Lopez-de-Silanes; A. Shleifer, and R.W. Vishny. 1999. “The Quality of Government.” Journal of Law, Economics, and Organization 15:222-279. Meggison, W. 2004. The Financial Economics of Privatization. New York, United States: Oxford University Press. ----. 2003. “The Economics of Bank Privatization.” Paper presented at the World Bank Conference on Bank Privatization in Low and Middle-Income Countries, November 23, Washington, DC. Rajan, R., and L. Zingales. 2003. Saving Capitalism from Capitalists. New York, United States: Crown Business. 34 ----. 1998. “Financial Dependence and Growth.” American Economic Review 88: 559586. Sapienza, P. “The Effects of Government Ownership on Bank Lending.” Journal of Financial Economics. Forthcoming. Shleifer, A. 1998. “State Versus Private Ownership.” Journal of Economic Perspectives 12: 133-150. Stigler, G. 1967. “Imperfection in the Capital Market.” Journal of Political Economy 75: 287- 292. Stiglitz, J. 1994. “The Role of the State in Financial Markets.” Proceedings of the World Bank Annual Conference on Economic Development 1993. Washington, DC, United States: World Bank. Titelman, D. 2003. “La banca de desarrollo y el financiamiento productivo.” CEPAL Serie Financiamiento del Desarrollo 137. Santiago, Chile: Economic Commission for Latin America and the Caribbean. 35 Table 1: State Ownership of Banks GDP per capita and Financial Development (1) (2) (4) (5) Developing Industrial 1970 1970 1970 1970 LGDP_PC -6.437 0.558 4.995 4.106 (1.24) (0.09) (0.37) (0.29) FD -1.075 0.088 (2.37)** (0.33) GOV_INT -3.736 -2.000 -11.846 -12.109 (1.84)* (0.85) (3.60)*** (3.50)*** Constant 125.712 92.599 45.823 50.989 (3.59)*** (2.54)** (0.38) (0.41) Observations 46 43 24 24 R-squared 0.14 0.19 0.42 0.42 Absolute value of t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% (6) (7) All Countries 1970 1970 -8.816 -5.172 (3.43)*** (1.47) -0.354 (1.51) -5.113 -3.953 (3.14)*** (2.15)** 148.061 125.089 (7.04)*** (5.13)*** 70 67 0.27 0.25 Table 2: State Ownership of Banks GDP per Capita and Institutional Quality (1) LGDP_PC RULE -15.976 (3.85)*** 7.479 (1.90)* CORR (2) (3) Developing Countries -13.975 -4.754 (3.51)*** (1.11) 11.542 (0.79) -8.931 (1.61) (6) (7) Industrial Countries 0.746 -1.604 (0.05) (0.12) -11.859 (2.45)** 144.293 (5.35)*** 55 0.22 133.667 (5.05)*** 55 0.19 (8) (9) 119.362 (4.63)*** 59 0.24 (10) -1.056 (0.08) -12.963 (4.14)*** 3.731 (1.21) -12.565 (4.20)*** -45.64 (0.35) 27 0.10 30.875 (0.23) 27 0.01 (12) -3.751 (1.18) -10.102 (3.78)*** 2.342 (1.19) -3.722 (0.39) 0.419 (0.32) 128.337 (4.56)*** 59 0.13 (11) All -1.759 (0.47) DEMO Observations R-squared -11.006 (2.82)*** (5) 3.502 (1.34) PRIGHT Constant (4) 56.995 (0.48) 27 0.01 -10.272 (2.43)** -0.711 (0.41) 40.169 (0.31) 27 0.01 133.212 (7.06)*** 82 0.26 129.184 (7.22)*** 82 0.25 107.548 (6.28)*** 86 0.28 0.230 (0.23) 122.329 (6.31)*** 86 0.22 36 Table 3: Share of Public Bank Assets Country Argentina Bolivia Brazil Chile Colombia Costa Rica Guatemala Honduras Nicaragua El Salvador 1995 1998 2000 2002 42.48% 0 52.77% 13.29% 19.60% 80.95% 6.36% NA 29.22% 0 49.56% 10.61% 16.32% 76.71% 3.84% 3.23% 25.70% 0 46.57% 9.49% 21.10% 73.23% 3.78% 2.28% NA 0 42.71% 10.34% 19.39% 68.02% 3.22% 1.78% 52.98% 9.05% 13.32% 6.99% 0.46% 5.73% NA 4.28% Source: own calculations based on Balance sheet data Table 4: Public and Foreign Bank Performance Indicators Relative to Private Domestic Banks Country Argentina Bolivia Brazil Chile Colombia Costa Rica Guatemala Honduras Mexico Nicaragua Peru El Salvador ROA Public -0.0037 -0.0026 -0.0001 -0.0098 0.0014 -0.0010 -0.0058 -0.0035 -0.0111 -0.0052 Foreign -0.0006 -0.0026 -0.0002 -0.0005 -0.0016 -0.0023 0.0058 0.0049 0.0010 -0.0003 -0.0013 Interest Rate (Loans) Public -0.0045 -0.0194 -0.0034 0.0078 0.0039 -0.0042 -0.0162 0.0013 0.0185 -0.0070 Foreign 0.0000 -0.0109 -0.0228 -0.0004 0.0094 -0.0101 -0.0098 -0.0096 0.0014 -0.0054 -0.0033 Interest Rate (Deposits) Public -0.0023 -0.0176 -0.0094 0.0001 -0.0013 -0.0021 -0.0147 0.0312 0.0056 -0.0041 Foreign -0.0002 -0.0060 0.0073 -0.0001 -0.0016 -0.0117 -0.0052 -0.0176 0.0035 -0.0013 -0.0005 NPL Public 0.0644 0.0090 0.0703 0.2337 0.2465 0.2620 0.0158 0.1163 0.1219 Foreign Loans to Public Sector Public 0.0876 0.0927 -0.0108 -0.0002 -0.0156 0.0590 -0.1051 -0.1216 -0.0360 0.0009 0.1725 0.0734 0.1661 -0.0031 0.0428 -0.0150 Foreign -0.0060 0.0806 -0.0013 -0.0205 0.0318 0.0100 0.1030 0.0296 0.0218 0.0003 0.0106 0.0636 0.0104 37 Table 5: Public Sector Loans Country Argentina Bolivia Brazil Chile Colombia Costa Rica Guatemala El Salvador Private Public Foreign Private Public Foreign Private Public Foreign 5.28% 0.97% 21.53% 0.14% 2.61% 6.25% 31.99% 16.48% 1995 16.65% 7.95% NA 10.09% 13.33% 18.96% 1.14% 0.80% 5.16% 2.02% 7.09% 7.44% 19.75% 27.40% 33.77% 9.13% 8.16% 6.46% 33.05% 0.11% 3.73% 3.29% 21.24% 25.05% 1998 14.42% 8.11% NA 9.80% 21.26% 33.68% 1.30% 0.56% 5.79% 4.06% 13.64% 2.70% 18.06% 34.51% 20.87% 17.40% 12.88% 6.30% 31.24% 0.15% 8.85% 4.36% 24.51% 30.26% 2000 21.64% 12.09% NA 6.70% 24.83% 33.13% 1.52% 0.89% 23.06% 13.11% 11.01% 2.17% 6.21% 36.76% 23.26% 20.39% Table 6: The Effect of State Ownership of Banks on Financial Development Dependent Variable: Average annual growth rate of private credit / GDP Source (periods) (1) LLS (60-99) (2) IPES (60-99) (3) IPES (70-02) (4) IPES (70-02) (5) LLS (60-99) (6) IPES (70-85) (7) IPES (86-02) Methodology (OLS) (OLS) (OLS) (IV) (IV) (OLS) (OLS) -0.056 (0.433) -0.056*** (0.019) -0.039*** (0.011) 6.681** (2.616) 82 0.21 -0.205* (0.122) -0.037*** (0.009) -0.021** (0.008) 6.651*** (1.225) 66 0.26 -0.176 (0.135) -0.036*** (0.009) -0.019** (0.009) 6.257*** (1.305) 70 0.20 -0.076 (0.152) -0.041** (0.019) -0.026 (0.027) 5.663*** (1.934) 65 0.17 -0.572 (0.487) -0.041** (0.178) -0.030 (0.024) 8.749** (3.744) 73 0.22 -0.030 (0.270) -0.083*** (0.025) -0.015 (0.015) 7.040*** (2.601) 66 -0.345 (0.212) -0.051*** (0.015) -0.039** (0.017) 9.411*** (2.276) 77 0.17 0.21 GDPPC (initial) Priv. Cred. (initial) Public share (initial) Constant Observations R-squared Robust standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% 38 Table 7: State Ownership and Output Growth Source (periods) GDPPC (initial) Public share (initial) School enroll. (avg.) Private credit (initial) (1) LLS (19601995) -1.749*** (0.300) -0.017** (0.007) 0.545*** (2) LLS (19601995) -1.740*** (0.308) -0.016** (0.008) 0.540*** (0.123) (0.126) 0.030*** (0.010) 0.031*** (0.011) 0.016 (0.073) Private credit (growth) Priv. Cred. * Public share (initial) Priv. Cred. * (1Public share) (initial) (3) LLS (19601995) -1.603*** (0.297) 0.586*** (0.126) (4) LLS (19601995) -1.922*** (0.277) -0.040*** (0.912) 0.586*** (0.113) (5) LLS (19601995) -1.872*** (0.384) -0.008 (0.008) 0.549*** (5) IPES (19702002) -1.604*** (0.376) 0.001 (0.007) 0.596*** (0.157) (0.140) 0.001 (0.012) 0.030*** (0.009) 0.020** (0.008) 9.817*** (1.917) 7.397*** 69 69 0.41 0.39 0.036** 0.082*** (0.016) (0.024) 0.031** (0.012) Constant Observations R-squared 9.417*** 9.292*** 7.338*** (1.628) (1.710) (1.415) 11.230*** (1.356) 82 82 0.42 0.42 82 0.36 82 0.49 (1.763) Robust standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% 39 Table A1 (1) (2) (3) Developing Countries 1970 1970 1995 LGDP_PC -10.510 -6.437 -11.315 (2.31)** (1.24) (3.31)*** gint_75 -3.736 (1.84)* Constant 136.161 125.712 131.712 (4.19)*** (3.59)*** (5.06)*** Observations 50 46 60 R-squared 0.10 0.14 0.16 Absolute value of t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% (4) (5) Industrial Countries 1970 1970 -11.616 4.995 (0.93) (0.37) -11.846 (3.60)*** 148.433 45.823 (1.25) (0.38) 27 24 0.03 0.42 (6) (7) (8) (9) All 1995 -4.030 (0.35) 1970 -10.087 (4.08)*** 63.886 (0.56) 27 0.00 133.411 (6.69)*** 77 0.18 1970 -8.816 (3.43)*** -5.113 (3.14)*** 148.061 (7.04)*** 70 0.27 1995 -9.906 (5.07)*** 121.571 (7.38)*** 87 0.23 Table A2 (1) (2) (3) (4) (5) (6) Developing Countries Industrial Countries 1970 1970 1995 1970 1970 1995 LGDP_PC -9.257 0.558 -6.359 -10.944 4.106 11.161 (1.98)* (0.09) (1.76)* (0.86) (0.29) (1.00) FD1 -0.422 -1.075 -0.345 -0.156 0.088 -0.374 (1.13) (2.37)** (2.98)*** (0.57) (0.33) (2.97)*** gint_75 -2.000 -12.109 (0.85) (3.50)*** Constant 135.357 92.599 108.784 149.504 50.989 -54.747 (4.25)*** (2.54)** (4.25)*** (1.24) (0.41) (0.51) Observations 46 43 60 27 24 27 R-squared 0.14 0.19 0.27 0.05 0.42 0.27 Absolute value of t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% (7) 1970 -8.022 (2.79)*** -0.207 (1.01) 122.544 (6.03)*** 73 0.20 (8) All 1970 -5.172 (1.47) -0.354 (1.51) -3.953 (2.15)** 125.089 (5.13)*** 67 0.25 (9) 1995 -4.017 (1.68)* -0.329 (3.81)*** 91.238 (5.29)*** 87 0.34 40 Table A3 (1) (2) (3) (4) (5) (6) (7) (8) (9) GDP Growth Ext. Demand (t-1) (10) (11) (12) Real credit growth (13) (14) Bank deposits Growth -0.094 -0.095 -0.081 -0.104 -0.089 -0.032 -0.173 0.427 0.426 0.411 -2.086 0.547 -0.579 -0.452 (0.153) (0.152) (0.153) (0.153) (0.154) (0.421) (0.163) (0.351) (0.348) (0.343) (1.594) (0.331)* (1.632) (0.410) Ext. Demand (t-2) -0.095 -0.059 -0.048 -0.050 -0.040 -0.392 0.015 -0.150 -0.036 -0.062 0.827 -0.131 1.359 0.431 (0.158) (0.157) (0.158) (0.164) (0.165) (0.379) (0.176) (0.360) (0.354) (0.356) (1.529) (0.341) (1.714) (0.373) GDP (t-1) 0.789 0.785 0.783 0.783 0.781 0.737 0.825 0.442 0.428 0.431 0.503 0.396 -0.087 0.257 (0.047)*** (0.048)*** (0.048)*** (0.048)*** (0.048)*** (0.052)*** (0.084)*** (0.120)*** (0.120)*** (0.121)*** (0.188)*** (0.162)** (0.165) (0.193) -0.246 -0.245 -0.245 -0.245 -0.245 -0.198 -0.272 -0.198 -0.196 -0.198 -0.273 -0.117 0.096 -0.353 (0.044)*** (0.043)*** (0.043)*** (0.044)*** (0.044)*** (0.060)*** (0.066)*** (0.118)* (0.118)* (0.118)* (0.192) (0.141) (0.179) (0.142)** -0.007 -0.007 -0.007 -0.012 -0.011 -0.010 -0.004 0.828 0.826 0.826 0.812 0.847 0.638 0.559 (0.009) (0.009) (0.009) (0.014) (0.014) (0.010) (0.020) (0.041)*** (0.041)*** (0.041)*** (0.055)*** (0.055)*** (0.058)*** (0.061)*** GDP (t-2) Real Cred (t-1) Real Cred (t-2) -0.015 -0.015 -0.014 -0.006 -0.007 -0.014 -0.018 -0.222 -0.221 -0.220 -0.225 -0.228 -0.159 -0.177 (0.008)* (0.008)* (0.008)* (0.012) (0.012) (0.009) (0.016) (0.036)*** (0.036)*** (0.036)*** (0.048)*** (0.045)*** (0.055)*** (0.046)*** 0.016 0.014 (0.040) (0.040) Real Cred (t-1) * FD Real Cred (t-2) * FD Ext Dda Shock -0.026 -0.023 (0.030) (0.030) 1.233 1.960 1.659 1.956 1.665 2.592 1.349 1.676 4.097 4.345 11.861 1.237 9.735 -0.999 (0.190)*** (0.372)*** (0.443)*** (0.373)*** (0.443)*** (0.857)*** (0.412)*** (0.457)*** (1.148)*** (1.281)*** (3.066)*** (1.027) (2.963)*** (1.405) Ext Dda Shock * FD -0.919 -0.815 -0.915 -0.816 0.499 -0.394 -3.058 -3.112 -11.642 -0.423 -8.759 0.839 (0.351)*** (0.390)** (0.358)** (0.395)** (1.741) (0.360) (1.202)** (1.203)*** (5.133)** (0.943) (5.716) (1.452) Ext Dda Shock * FD * SOB FD (AVG) 2.108 2.042 -1.979 (1.597) (1.600) (3.985) -0.000 -0.000 -0.000 -0.000 -0.001 0.000 -0.002 -0.004 -0.008 0.007 -0.014 0.005 (0.002) (0.002) (0.002) (0.002) (0.003) (0.003) (0.008) (0.008) (0.014) (0.009) (0.015) (0.012) FD SOB (AVG) 0.000 0.000 (0.004) 0.019 (0.004) (0.013) Observations 1340 1340 1340 1340 1340 670 670 1340 1340 1340 670 670 661 618 R-squared 0.48 0.48 0.49 0.49 0.49 0.49 0.50 0.53 0.53 0.53 0.52 0.58 0.35 0.29 High Low High Low High Low Sample Robust Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% GDP, Real Credit and the External Demand (in log) have been detrended using the HP filter (for annual frequency). FD is the time average of credit over GDP during 1980 and 2001. SOB is the time average share of SOB in the banking industry. We do not include the 2% country-year observations that have extreme changes in real credit (in t and t-1). We also exclude the county-year observation with inflation larger than 100% and lower than -10%. 41 Table A4: Growth (1) FD RZ (Initial Year) FD ITG (Initial Year) FD SOB RZ (Initial Year) FD RZ (AVG) FD SOB RZ (AVG) Initial VA (ln) Observations R-squared (2) (3) All Countries 0.015 0.015 0.025 (0.006)*** (0.006)*** (0.007)*** 0.004 (0.004) -0.027 (0.008)*** (4) (5) 1985-1997 0.021 (0.006)*** -0.016 (0.009)* (6) (7) Industrial Countries 0.009 (0.007) -0.021 (0.008)** (8) (9) Developing Countries 0.033 (0.016)** -0.017 (0.038) 0.018 -0.005 0.027 (0.004)*** (0.005) (0.007)*** -0.017 -0.018 -0.014 (0.007)** (0.007)*** (0.017) -0.010 -0.010 -0.011 -0.012 -0.013 -0.010 -0.013 -0.013 -0.014 (0.002)*** (0.002)*** (0.002)*** (0.001)*** (0.001)*** (0.003)*** (0.002)*** (0.002)*** (0.001)*** 767 767 754 1013 968 359 411 395 602 0.59 0.59 0.61 0.67 0.56 0.57 0.70 0.62 0.67 Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Average annual growth. All regressions include country and industry fixed effect. The sample includes 28 manufacturing industries, defined at 3 digits ISTC Rev2, for all middle and high income countries. RZ stands for external requirement from Rajan and Zingales, ITG for intangible assets (fraction), FD for financial development and SOB for state-owned banks (fraction). 42 Table A5: Volatility Sectoral Volatility: Whole Sample. (1) FD RZ (AVG) FD RD (AVG) FD ITG (AVG) FD SOB RZ (AVG) FD SOB RD (AVG) FD SOB ITG (AVG) (2) -0.016 -0.018 (0.005)*** (0.005)*** -0.006 (0.006) -0.014 (0.005)*** (3) -0.002 (0.006) (4) (5) (6) SD VA-Growth 1970-1997 -0.014 (0.005)*** -0.007 (0.006) -0.014 (0.005)*** -0.003 (0.009) -0.003 (0.011) (7) (8) (9) -0.013 (0.005)** -0.008 (0.004)** 0.003 (0.009) 1628 0.71 -0.007 (0.004)* -0.003 (0.009) 0.003 (0.009) 1628 0.71 -0.010 (0.006)* -0.011 (0.004)** -0.003 (0.012) 0.002 (0.009) 1628 0.71 Observations 1854 1854 1854 1854 1628 1628 R-squared 0.68 0.68 0.68 0.68 0.71 0.71 Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% SD of sectoral VA rate of growth during the specified period. All regressions include country and industry fixed effect. The sample includes 28 manufacturing industries, defined at 3 digits ISTC Rev2, that have 5 or more annual rate of growth to compute the SD. RZ stands for ext. req. from Rajan and Zingales, RD for inventory (over assets), ITG for intangible assets (fraction), FD for fin. Dev. and SOB for state-owned banks (fraction). 43