LAp$ a-7 (o POLICY RESEARCH WORKING PAPER 2760 Bank Lending to Small Do foreirgn banks lend less to small and medium enterprises Businesses in Latin America thandomesticbanksin J&e..elc'pirig countries? Does Bank Origin Matter? Analysis of data from four countries in Latin America suggests that although small George R G. Clarke foreign banks lend less than Robert Cull small domestic banks, the Maria Soledad Martinez Peria difference for large banks is Susana M. Sdnchez considerably less. The World Bank Development Research Group Finance January 2002 l POLICY RESEARCH WORKING PAPER 2760 Summary findings In recent years foreign bank participation has increased whether bank origin affects the share and growth rate of tremendously in Latin America. Some observers argue bank lending to small businesses. They find that although that foreign bank entry will benefit Latin American foreign banks generally lent less to small businesses (as banking systems by reducing the volatility of loans and share of total lending) than private domestic banks, the deposits and increasing efficiency. Others are concerned difference is due primarily to the behavior of small that foreign banks might choose to extend credit only to foreign banks. The difference was considerably smaller certain customers, leaving some sectors-such as small for large and medium-sized banks. And in Chile and businesses-unserved. Colombia, large foreign banks might actually lend Clarke and his coauthors examine this issue. Using slightly more (as share of total lending) than large bank-level data for Argentina, Chile, Colombia, and Peru domestic banks. during the mid-1990s, they empirically investigate This paper-a product of Finance, Development Research Group-is part of a larger effort in the group to understand the impact of entry by foreign banks on domestic banking systems in Latin America. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Paulina Sintim-Aboagye, room MC3-422, telephone 202-473-7644, fax 202-522-1155, email address psintimaboagye@worldbank.org. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at gclarke@worldbank.org, rcull@worldbank.org, mmartinezperia@worldbank.org, or ssanchez@worldbank.org. (36 pages) The Policy Research Working Paper Seoies disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polisbed. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclzssions expressed in this paper are entirely those of the autbors. They do not necessarily represent the vie.w of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center Bank Lending to Small Businesses in Latin America: Does Bank Origin Matter? George Clarke, Robert Cull, Maria Soledad Martinez Peria, and Susana M. Sanchez4 Keywords: foreign bank entry, small business lending JEL: G21, G3 ' Authors are World Bank staff members. We would like to thank the Central Bank of Argentina, the Argentinean Superintendency of Financial Institutions, the Chilean Superintendency of Banks and Financial Institutions, the Colombian Superintendency of Banks, and the Peruvian Superintendency of Banks for providing us data and advice. We are particularly grateful to the following individuals at these institutions: Juan Carlos Alfaro, Patricio Alvarez, Tamara Burdisso, Liliana Castineira, Jorge Cayazzo, Jaime C6rdova, Patricia Correa-Bonilla, Laura D'Amato, Monica Gabel, Francisco Gismondi, Socorro Heysen, Gonzalo Jimenez, Jorge Marshall, Javier Poggi, Luciana Rios- Benso, Luis Raul Romero, and Jairo Zubieta-Vela. We have benefited from comments from Jerry Caprio, Augusto de la Torre, Jim Hanson, and Fernando Montes-Negret. Augusto de la Torre also assisted us in obtaining the necessary data. Ivanna Vladkova and Ainur Bekturganova provided excellent research assistance. The Office of the Chief Economist for Latin America, the Financial Sector Board, and the Research Committee provided funding for this paper. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and do not necessarily represent the view of the World Bank, its Executive or the countries they represent. Contact information: Maria Soledad Martinez Peria, The World Bank, 1818 H Street, N.W., Washington D.C., 20433. Phone: (202)458-7341. Fax: (202)522-1155. E-mail: mmartinezperia@worldbank.org. Introduction The increasing participation of foreign banks has been one of the most striking structural changes experienced by banking systems in developing countries over the past decade. In Central Europe the percentage of assets controlled by foreign banks increased from 8 percent in 1994 to 56 percent in 1999. As of December 2000, foreign financial institutions controlled 38 percent of loans in the major countries of Latin America, up from 15 percent in 1996. In Asia, the numbers are less striking, but the trend is definitely visible. Foreign control in this region has increased from 2 to 13 percent.' Whether foreign bank entry will be beneficial for developing countries is an issue of controversy. Proponents of this process argue that foreign banks bring new capital, improve management expertise, and promote efficient and competitive banking practices. Also, because foreign banks have access to alternative sources of funding other than local deposits, many argue that foreign entry may lead to lower volatility and higher growth of lending. In contrast, those opposing opening up financial systems to foreign ownership argue that foreign banks may decrease the stability of aggregate domestic bank credit by facilitating capital flight during crises and simultaneously increasing countries' exposure to regional contagion and to shocks in the home countries of foreign banks. Another common argument against foreign bank entry is that these institutions might tend to "cherry pick" the most profitable customers, reducing financing to some sectors, increasing the risk exposure of local banks, and thus, affecting the overall distribution of credit. In particular, the main area of concern is the availability of credit to small businesses. In many developing countries, small businesses account for a very significant share of total value added and generate a large fraction of the total jobs in the economy. Banks are perceived as having a comparative advantage over other lending institutions in small business lending. This role is likely to be more important in less developed countries that are generally more heavily dependent on bank financing. In Argentina, for example, 79 percent of small industrial firms have bank debt (Llorens, van der Host, and Isusi, 1999). Moreover, small businesses tend to have exclusive dealings with a single bank with which they have a strong relationship. Given the paucity of information about small businesses, these relationships enable banks to generate information I on the risk characteristics of small firns. Therefore, access to credit by small businesses would be reduced if foreign banks were to neglect small businesses and/or drive domestic banks from the market, destroying the information generated through bank-borrower relationships. A number of recent studies have shown that foreign entry seems to improve banking system efficiency and to contribute to overall banking stability in developing countries (see Demirgiuc-Kunt, Levine, and Min (1998), Levine (1999), Barajas et al. (2000), Claessens et al. (2000), Clarke et al. (2000), and Dages et al. (2000)). On the other hand, the effect of foreign bank participation on access to credit in developing countries remains largely unexplored. Evidence from the United States indicates that large and organizationally complex institutions find it difficult to lend to informationally opaque small and medium-sized enterprises (see Berger and Udell (1995), Berger et al. (1995), Keeton (1995), Levonian and Soller (1995), Berger and Udell (1996), Peek and Rosengren (1996), and Strahan and Weston (1996)). These organizational diseconomies might explain why a number of studies have found that foreign banks, which as shown by Focarelli and Pozzolo are typically large, appear to allocate greater shares of their lending to activities and sectors dominated by large firms (see Goldberg (1992), Cho et al. (1987), and Clarke et al. (2000)) While large foreign banks are unlikely to replicate the lending methods of small domestic banks, technological innovation could offer them an avenue for increasing small business lending. Mester (1997) argues that advances in credit scoring methodologies, coupled with enhanced computer power and increased data availability, are likely to change the nature of small business lending. This suggests that there could be a nonlinear, and, in particular, U-shaped relation between bank size and lending to small businesses. On the left-hand side would be small domestic banks engaged in relationship lending for services not amenable to scoring. On the right-hand side would be large banks, many or most of them foreign owned, offering more standard products to small businesses based on credit scoring (Mester (1997) and Peek and Rosengren (1998)). The literature for the U.S. also suggests that the type of foreign entry may affect lending patterns. For example, the evidence from the U.S. indicates that de novo entrants seem to devote larger shares of 2 their assets to small businesses than other banks (See Goldberg and White (1998), DeYoung et al. (1999), and Jenkins (2000)). Foreign entry through merger and acquisitions (M&As) might have different effects on small business lending. Some studies find that M&As among small banks led to increased propensity to lend to small businesses following their consolidation (Peek and Rosengren (1998), Walraven, (1997), Strahan and Weston (1998), Berger et al. (1998)). However, when medium-sized and large banks are involved, a number of recent papers have reached conflicting conclusions. Keeton (1996) and Berger et al. (1998) uncovered that M&As tend to result in a reduction in small business lending when large banks are involved, but Strahan and Weston (1998) found no significant change in the ratio of small business loans to assets following large M&As. There is, however, very little literature that deals directly with the implications of foreign entry for lending to small businesses in developing countries. Argentina is the only country for which we found such studies. Bleger and Rozenwurcel (2000) indicate that foreign bank participation in Argentina is associated with a reduction of bank lending to small businesses from around 20 to 16 percent of total lending between 1996 and 1998. In contrast, Escude et al. (2001) find that despite their lower tendency to lend to small businesses, foreign banks have increased both their propensity and their market share of lending to the sector between 1998 and 2000. Finally, using a rich data set on Argentinean business debtors in December 1998, Berger, Klapper, and Udell (2000) find that large banks and foreign owned banks are less inclined to extend credit to smaller firms, which are likely to be informationally opaque. Given the paucity of research on the impact of foreign bank entry on access to credit and the importance of this issue from a policy standpoint, further investigation is clearly warranted. Using bank level data for Argentina, Chile, Colombia, and Peru during the mid-1990s, this study empirically investigates whether bank origin affects the share and growth rate of bank lending to small businesses. After Eastem Europe, Latin America has been the region that most rapidly and widely allowed foreign bank entry. In our study, we focus on four of the most important countries in the region that have witnessed the largest increase in foreign entry into their banking sectors. Having bank-level time series 3 data allows us to investigate the evolution of lending to small businesses over time, in particular, as foreign penetration increases. Because we analyze this question for a number of countries, we are able to examine whether foreign entry has a uniform impact on lending to small businesses across countries. This paper is organized as follows. Section II presents a brief overview of the Argentinean, Chilean, Colombian, and Peruvian banking sectors. Section III discusses the data used in this paper. Section IV presents descriptive statistics on the extent to which foreign banks lend to small businesses, relative to their domestic counterparts. Section V explains the econometric methodology, while section VI presents the empirical results. In Section VII we conduct a preliminary exploration of the cross-country differences in our results. Finally, section VIII concludes. II. Banks in Argentina, Chile, Colombia, and Peru: An Overview Both the number of banks and the distribution of their sizes vary across the four countries, and this could have implications for lending to small businesses (see Figure 1). Colombia with a population of 41.6 million had the second lowest number of banks (27) in 2000. Chile, which has a population of 15.0 million, or just over one-third that of Colombia, had 28 banks. Peru resembles Colombia more than Chile. Its 25.2 million residents are served by 19 banks. Finally, Argentina had the highest number of banks per resident - 91 banks serve a population of 36.6 million. Not only are there more banks per capita in Argentina and Chile, but also they are larger than those in Colombia and Peru. In both Chile and Argentina, the largest banks have between 15 and 20 billion dollars in assets. In Colombia, the largest banks have 3 to 4 billion dollars in assets. In Peru, this figure ranges between 4 to 6 billion dollars. Relatively small banks comprise an inordinate share of the total in all four countries. However, small banks in Chile are larger than those in Peru and Colombia. In Argentina, there are still a large number of very small banks. This is likely to diminish as the process of consolidation continues there. In the early 1990s, Argentina had more than 200 financial institutions. Less than half that number are now in operation, and some of the current banks are new entrants. While several banks have failed, many have been acquired by or merged with other entities. In the period covered by our data (June 1998 to March 2000), there were eighteen mergers or acquisitions, 4 ten of which involved foreign banks. The period between the Tequila Crisis in late 1994 and the beginning of our sample witnessed even greater consolidation (see World Bank 1998). Argentina experienced the most mergers and acquisitions in the late 1990s. However, Chile, Colombia, and Peru have also undergone a process of consolidation. In Colombia and Peru, there were seven mergers or acquisitions, four of which involved foreign banks. Chile had six such transactions, half involving foreign banks. Argentina also led the way with seven de novo entrants, compared to two for Peru and none in either Chile or Colombia. In our empirical analysis, we control for the effects of mergers, acquisitions, and de novo entry on the share and growth rate of lending to small businesses. The ratio of banking sector assets to GDP provides another indication that Colombia and Peru have a lower level of financial intermediation than Argentina and Chile (see Figure 2). Both for Colombia and Peru that ratio hovered near forty percent, and was declining at the end of our sample period. To some extent, these dips were due to Colombia's intemal conflict and banking crisis and to Peru's scandals at the end of the Fujimori administration. By way of comparison, in Argentina the ratio of banking sector assets to GDP went from 48 percent in 1998 to 56 percent in 2000. Owing to the hyperinflation of the late 1980s, Argentina started the 1990s with a very low ratio of banking sector assets to GDP. However, with the exception of the Tequila Crisis, that ratio has grown relatively steadily since the adoption of the Convertibility Plan in 1991, which brought about price stability. Chile's banking development occurred earlier. By 1996, banking sector assets were nearly 140 percent of GDP, and that figure had grown to over 180 percent by 2000. It is true that much of the Chilean banks' assets are in assets other than loans, and that banking assets relative to GDP may overstate their relative level of development. However, Chile holds a sizable advantage over the other three countries on other measures of financial development such as the ratio of private credit to GDP.2 The evidence on foreign penetration also indicates that Chile and Argentina differ from Colombia and Peru (see Figure 3). In Colombia, foreign banks comprise forty percent of the total number, but they account for only a quarter of banking sector loans. In Peru, over half the banks are foreign-owned, but they are responsible for forty percent of total system loans. There is a key difference between the two 5 countries. In Peru, the share of loans held by foreign banks grew throughout the late 1990s, while in Colombia it declined slightly. Still, foreign banks held only about a quarter of total loans in both countries throughout most of the period. By the end of the period, foreign banks held over 50 percent of Argentina's banking sector loans, and almost 45 percent of Chile's. Chile's foreign bank share climbed throughout our sample period. In Argentina, most of that growth occurred prior to our sample period. However, data on lending to small businesses for Argentina was only available since 1998, and thus our sample period is shorter than that for Chile. The key point, however, is that foreign banks comprised a substantially larger share of bank loans in Argentina and Chile than in Colombia or Peru. Judging from the U.S. empirical evidence one might expect that small borrowers would fare relatively worse in Argentina and Chile. Those countries experienced greater foreign penetration and had much larger banks than in Colombia or Peru. Moreover, in Argentina, foreign entry coincided with a massive consolidation in which many small banks left the market either through merger or failure. Section V tests these conjectures explicitly. III. The Data To analyze domestic and foreign bank lending to small businesses, we assembled a comprehensive bank level database for Argentina, Chile, Colombia, and Peru during the 1990s. These countries' central banks and bank superintendencies were our main sources of data. We gathered information on the origin (foreign or domestic) of each bank in these countries throughout the sample. Moreover, we identified whether foreign banks entered into the banking system through a process of mergers and acquisitions with previously existing institutions or by creating a new institution or de novo bank. In general, the first type of entry is likely to result in higher bank concentration, especially if foreign banks acquire large domestic institutions. Second, we gathered data on the distribution of bank lending by size. When information about small business loans was not available as a separate category, small business loans were defined by the size of the loan, instead of borrower size. Finally, we collected 6 balance sheets and income statements for all the financial institutions to control for the performance and health of domestic and foreign banks when comparing lending patterns across these institutions. Argentina According to a recent study on small business lending conducted by the central bank of Argentina, loans to this sector go to businesses with total debt between 50,000 and 2.5 million dollars (see Escude et al. (2001)). Since we do not have information on borrowers' total debt, we use bank level information on total lending stratified by loan size for the period 1998-2000. According to the aforementioned study, a small business with up to 2.5 million dollars in debt, on average borrows up to I million dollars from any individual bank. Thus, we define as loans to small businesses all loans up to this amount. Chile For Chile small business loans are identified based on the total debt of the business rather than the size of individual loans. Total debt is measured by the Chilean Banking Superintendency in 'unidades de fomento' (UF), which are equal to a fixed quantity of Chilean pesos indexed to inflation. Following the guidelines provided by the Superintendency, we define as small business lending all loans to borrowers with less than UF50,000 in total debt, which is roughly equal to 1.5 million dollars.3 Colombia The Superintendency of Banks in Colombia requires banks to keep a separate record of the amount of loans they give to small firms registered with the Superintendency of Businesses. According to Colombian regulations, small businesses are those with assets up to 1.7 million dollars. For each bank in Colombia, we obtained quarterly information on the loans to the 4,067 small businesses registered with the Superintendency of Businesses over the period 1997-1999. Because this might be a partial list of all small businesses in Colombia, we should be careful in trying to draw conclusions across countries. Peru Officials from the Peruvian Banking Superintendency estimate that small business loans range between $20,000 and $500,000. Thus, using a breakdown of bank lending by size for loans to businesses 7 prepared by the Peruvian Banking Superintendency, we defined small business loans as loans to businesses up to $500,000. IV. Profile of Small Business Lending in Latin America Table 1 and 2 provide some summary statistics on the share and growth rate of small business lending by domestic and foreign banks in Argentina, Chile, Colombia, and Peru during 1997-2000. Also, within these tables we distinguish between large and small, foreign and domestic banks. We consider large banks to be those in the top 5 percentile of the distribution of banks assets. According to Table 1, comparing all domestic banks with all foreign banks (i.e., without any regard for size), we find that foreign banks in all four countries in our sample devote a lower share of their total lending to small businesses. However, these shares are not significantly different in the case of Colombia. In fact, Colombia looks very different from the other countries not only because the shares of small business lending by domestic and foreign banks are not significantly different from each other, but also because the shares of lending to small businesses in Colombia are several orders of magnitude smaller than what we find for other countries. While for the other three countries the share of lending to small businesses ranges between 18 and 30 percent depending on the country and the origin of the bank, for Colombia this share does not exceed 2 percent. Again, this might be due to the fact that data on small business lending is available only for the 4,067 firms registered with the Superintendency of Businesses. Thus, while it seems acceptable to compare the share of small business loans for foreign and domestic banks in Colombia, it might not be fair to make comparisons between small business lending shares in Colombia and the remaining countries. If we separate banks according to size and origin, we find that regardless of bank origin small banks lend more to small businesses than large banks. Further, in all four countries, small foreign banks devote a lower share of their total lending to small businesses than small domestic banks. The difference between small foreign and domestic banks is statistically significant in all countries except Colombia. In contrast, the picture for large banks is quite different. Although large foreign banks appear to lend less to 8 small businesses than large domestic banks in Argentina and Peru, the reverse seems true in Chile and Colombia. This suggests that large and small foreign banks might behave differently in some countries. If lending to small businesses grows at a slower pace than other types of lending it is possible that the share of lending to this sector could be dropping, while the growth rate of the level could still be positive. Thus, to distinguish between the behavior of the share and the level of small business lending, Table 2 present tests of differences in mean growth rates of small business lending across domestic and foreign banks for all four countries, Table 2 shows that in Argentina, Chile, and Colombia, the annual small business lending growth is lower for foreign banks relative to domestic banks. For Argentina and Colombia, this result is coming from the behavior of small foreign banks, since the large foreign banks seem to outperform the large domestic banks. Peru is the one case where the growth rate of small business lending by foreign banks is positive and significantly different from that observed for domestic banks.4 V. Empirical Methodology Differences in the mean share or in the growth rate of lending to small businesses between domestic and foreign banks might be driven by factors other than bank origin that we are not controlling for. Thus, we tum to regression analysis to study the impact of bank origin on small business lending. For each country, we estimate equations (la) and (lb) below in order to examine the impact of bank origin on the share of bank lending to small businesses, controlling for other factors that might influence this ratio. (In( 1 )J =aj +aj'SIZEjI + a j'FCIAL'>I +a-1'PUBLICi +aj'ORGIGIN1 I-PI" 0 1 it-l 3 i't-1 4 ,,t 5 i, t +a1 'TREND1 +aJ'TREND*OR1GJN! +aj 'FDENOVOJit +aj'FDACQUIRE,R! +a1J'AGEFDENOVOJ 6 z,t 7 i, t 8 9, 9,t 1 ' + aj'AGEFACQUIRERIt 1 EGR2 'M'A GEM EG RGERS + J i,t (la) Iln ( ai ' a O +c a'SsiZEJ +ja'SIZE * ORIGINJ 1 +aSIZE2 * ORIGIN + a -'FCIAL't +aJ 'PUBLICJ +aJ 'ORIGINJ 1 +a8J'TRENDJ +aJ 'TREND*ORIGINJ +aJ .'FDENO vOJ + 5 I,t-1 6 ijt 7 O, , 10 , + aAJ'FDACQU1RERJ + aJ'AGEFDENOVOl2 + aJ'AGEFACQUIRERJ I +aJ'MERGERS+aI 5'AGEMERGERS + Ji, (lb) 9 Equations (1 a) and (1 b) are estimated in log-odds logit form where j= I.. .X represents the country identifier, i=l ...N captures each individual bank within a country, and t=l ...T refers to the time periods considered. Pit is the proportion of loans made by banks to small businesses. Focusing on the log odds ratio rather than the share gets around the problem that shares are by definition bounded between 0 and 1. Equation (la) and (lb) model the ratio of small business loans to total loans as a function of a number of bank indicators, including bank origin. We also include a general time trend (TREND) in the estimations to control for changes in the share of lending to small businesses associated with macroeconomic variables or any other factors that are common among banks over time. ORIGIN is a dummy variable for whether the bank is owned by domestic or foreign interests. This variable takes a value of one for foreign banks. The ORIGIN dummy is also interacted with the yearly trend to capture changes in foreign banks' share of lending to small businesses. By focusing on the sign and significance of the coefficients on ORIGIN and the interaction term, we can study whether we observe different lending patterns for foreign and domestic banks. In most countries in our sample, there are three kinds of foreign banks, namely, those that had been operating in the country for a number of years before our sample starts (e.g. Citibank in all four countries), de novo foreign banks that started operating at some point in the sample, and banks that within our sample acquired other domestic or foreign banks. As noted in the literature, there are reasons to expect that these types of banks might behave differently towards small businesses. In the U.S. literature for example, de novo banks, in particular, have been found to be more apt to lend to small businesses than other foreign entrants, while mergers among large banks have resulted in lower shares of small business lending. To allow for differences in the impact that the mode of entry by foreign banks has on the share of lending to small businesses, we include separate dummies for foreign de novo banks (FDENOVO) and for those foreign banks that either entered the system or increased in size by acquiring domestic institutions (FDACQUIRER). Both of these dummy variables are also interacted with a variable measuring the time since entry or acquisition ("AGE") to capture changes in their small business lending as they became better established in the market. 10 Because consolidation by domestic banks or between two foreign banks could have an impact on the share of small business lending, we include a dummy for mergers that did not involve a foreign bank merging with or acquiring a domestic institution (MERGER). We also interact this variable with a trend to control for the dynamic impact of bank consolidation (independent of ownership) and lending to small businesses. SIZE refers to the log of real total assets.5 In equation (lb), we include a size-squared term (SIZE2) to account for the possible non-linearity in the relation between bank size and lending to small businesses. Also, by interacting size and size squared with a dummy for bank origin, this equation allows for the possibility that the impact of size on the share of bank lending to small businesses is different between domestic and foreign banks. Publicly owned banks operate in three out of the four countries in our sample (Argentina, Chile, and Colombia). Because public bank lending decisions could be affected by politics and not by commercial factors, we include a dummy (PUBLIC) to control for this type of banks in our estimations.6 FCIAL refers to two financial health and performance indicators, return on assets and the ratio of administrative expenses to total assets. It is unclear what is the expected sign on these variables. On the one hand, banks with positive return on assets might be more able to grow over time and to expand into areas where it takes time and effort to acquire the know-how of the business, like lending to small borrowers. On the other hand, banks with low return on assets might be more willing to "gamble for resurrection" by venturing into higher risk segments, like lending to small businesses. Banks with higher ratios of administrative expenses to assets might be better suited to lend to small businesses, if their high expenses are associated with a more extensive branch network and a larger labor force that can provide the personalized attention and monitoring that is needed when lending to small businesses. On the other hand, banks with high administrative expenses might be at a disadvantage to compete with other banks in lending to small businesses. The share regressions provide important information about the impact of foreign entry on lending to small businesses, but they cannot tell us for certain whether such lending increased or decreased during 11 this period. For example, those regressions could indicate that the share of lending to small businesses by foreign banks was falling relative to shares at domestic banks while, at the same time, foreign banks' total lending to small businesses was actually increasing. The decline in the share of small business lending would occur because that line of business was not growing as fast as other lines of business at foreign banks, but there would be real growth in small business lending nonetheless. For these reasons, we also run regressions in which the dependent variable is the growth rate in real lending to small businesses. Both equations (2a) and (2b) examine the impact of bank origin on the growth rate of real lending to small businesses (SBLGROWTH). Equation (2b) allows for the possibility that the impact of size on lending to small businesses might depend on bank origin. SBLGROWTH i = aJ +ai'SIZE, -_ + 2a'FCIALj,_l + aJ'ORIGINJ + a-'PUBLIC3 +a 'FDENOVO6 SBLGROWTH 0 -a a 2IE~ a-FlA- 3 4 I't 5zFEOO + a] 'FDACQUIRER/J + aJ'MERGERS!,,J-is, (2a) SBLGROWTH'J =aa'S + IZEi +aJ'SIZE*ORIGIN! ' +aJ'FCIALJ +a i'OR"GIN!, i,1 - o SIZE- 2 ,1-1 3 ,,:-I 4 +Ja'PUBLIC! +al,'FDENOVOi +a'FDACQUIRER,i +ca'MERGERS!, +l c,, (2b) 5 I'j 6 ,,i 7 Once again, j=l ...X represents the country identifier, i=1 ...N captures each individual bank within a country, and t=l ...T refers to the time periods considered. All regressors are defined above. The lending growth specifications differ from those for shares of small business lending in one sense. Because the dependent variable already captures dynamic effects, there is no need to include interaction terms to measure trends over time.' VI. Empirical Results As described above, we estimate two types of regressions, one for shares of total lending devoted to small businesses (share regressions) and one for real growth rates in small business lending (growth regressions). In some of the share and growth regressions, we interact foreign ownership with bank size to explore whether large foreign banks behaved differently towards small businesses than did other banks. We discuss each of these four estimations in turn. 12 A. Share regressions - No bank size interactions Table 3 reports the results from equation (1 a) assuming the impact of size on the share of lending to small businesses is linear and is restricted to be the same for domestic and foreign banks. For our purposes, the key variable from this base regression is the foreign ownership dummy, which is negative in most cases, but significant only in the case of Chile. Therefore, controlling for other factors, a typical foreign bank did not devote a significantly lower share of its lending to small businesses in Argentina, Colombia, and Peru. In all countries except Argentina, the coefficient on the time trend is negative and statistically significant, indicating that lending to small businesses as a share of total lending was falling over time for domestically owned banks. The coefficient is positive and statistically insignificant in the regression for Argentina. The coefficient on the interaction between the time trend and the foreign ownership dummy is negative in the regressions for all four countries, but it is statistically significant at conventional levels only for Peru. Thus, except for the case of Peru, there is no consistent indication that the share of lending grew at a slower rate (or fell at a faster rate) for foreign banks vis-a-vis domestic institutions. In this specification, size has a negative impact on lending to small businesses. This effect is significant only in the case of Argentina and Peru. Private domestic banks generally appear to lend more to small businesses (as a share of total lending) than state-owned banks, after controlling for other factors that might affect lending. The coefficient on the dummy indicating govemment ownership is negative and statistically significant in the regressions for all three countries with state-owned banks (Argentina, Chile and Colombia). One frequently heard justification for state banks is that they resolve credit market failures by emphasizing lending to small businesses. These results appear to undercut that justification. Finally, the regressions also include a series of dummies to control for the effects of mergers, acquisitions, purchases of existing foreign banks by new foreign entrants, and de novo entry by foreign banks. These variables are included to control for the possibility that banks recently involved in mergers and banks that have only recently entered might take some time to reach their desired portfolio allocations. To allow banks to slowly adjust their portfolio allocations, variables representing the time 13 since the merger or entry occurred are also included. In Argentina, where the largest number of transactions took place, we are able to include more dummies than in the other countries. However, since there were very few mergers during our sample period in most of the countries, and because some observations are lost due to missing or incomplete data, the coefficient estimates are often based on only a few merger/entry observations. Consequently, these variables might best be thought of as controlling for temporary disequilibria following mergers and new entry, which thus enables us to estimate more reliably the coefficients for the banks not involved in such transactions. Given the limited number of observations, it is not surprising that the merger/entry variables do not tell a very consistent story across countries. We briefly mention only a couple of the results from those variables. The very limited evidence on de novo entry, which comes only from the share regressions for Argentina, appears to indicate that contrary to the U.S. experience, foreign de novo entrants do not concentrate on small business financing. Results for the rest of the merger/entry variables show little consistency across types of regression or across countries. For example, in equation (1), the impact of foreign acquisitions (i.e., existing foreign banks acquiring domestic banks) on lending to small businesses is negative and significant for Argentina, positive and significant for Chile, and insignificant in Colombia and Peru. Similar inconsistencies across countries are found for all the regression types discussed below. The effect of the mode of entry on the lending behavior of foreign banks is an important topic. We simply lack the number of observations necessary to provide consistent evidence on this issue.8 B. Share regressions - large vs. small banks Equation (lb) allows for non-linearities in the relationship between size and the share of small business lending and for interactions between size and size squared with origin, in order to explore whether bank size changes the impact of foreign ownership on lending to small businesses. Table 4 reports the results from this specification. The null hypothesis that the two coefficients on the interaction terms are jointly zero can be rejected at a 10 percent level or higher in the regressions for all four countries.9 Similarly, the null hypothesis that the two coefficients on the squared terms are jointly zero 14 can also be rejected at similar levels.'0 This strongly suggests that it is appropriate to include these additional terms. Once the interaction and squared terms are included, it becomes more difficult to see how foreign ownership affects the share of lending to small businesses based simply upon the coefficient values. To make it easier to interpret the results, Figures 4-7 show estimated lending to small businesses (as a share of total lending) for foreign and domestic banks."' Estimated shares are calculated for banks of different sizes using coefficients from Table 4 in the final period for each country, using median values of continuous variables (other than size) for banks of that type and assuming that the bank was not involved in any mergers or acquisitions. Although medium and large domestic banks generally lend less to small businesses (as a share of total lending) than small domestic banks, lending shares generally decline slowly for medium and large banks as size increases in some countries. In fact, in the two cases with the largest banks, Chile and Argentina, lending to small businesses appears to actually increase as size increases for medium and large domestic banks (see figures 4 and 5). In contrast to the results for the other three countries, lending to small businesses in Colombia increases as a share of total lending for very small banks before slowly declining (see figure 6). The pattern for foreign-owned banks appears quite different from the pattern for domestic banks. Although small foreign banks lend considerably less to small businesses than small domestic banks in all four countries, medium to large foreign banks generally appear similar to medium to large domestic banks. In fact, in two of the four countries, Chile and Colombia, estimated shares of small business lending for large foreign banks are larger than for large domestic banks (see Figures 5 and 6). In the other two countries, Argentina and Peru, although foreign banks of all sizes lend lower shares than similar domestic banks, the difference is smaller for large banks than it is for small banks (see Figures 4 and 7) 12 These results, and especially those for Chile, are consistent with the notion that large foreign banks, using credit scoring methodologies, enhanced computer power, and improved data availability, will increasingly enter small business lending. We seek further confirmation in the growth rate regressions below. 15 C. Growth rate regressions Table 5 includes regressions with the annual growth of small business lending as the dependent variable. 13 As noted previously, although dummy variables in the growth regressions might have a similar interpretation to the trend variables in the share regressions, their coefficients could differ. However, in the regressions following equation (2a) that omit the interaction between size and foreign ownership, the results for the foreign ownership dummy are broadly consistent with those for the interaction between the time trend variables and bank origin in the share regressions. The coefficient on the dummy indicating foreign ownership is negative and statistically significant in the regressions for Argentina and Colombia; positive and statistically significant in the regression for Peru; and positive and statistically insignificant in the regression for Chile (see columns (5.1), (5.3), (5.5) and (5.7)). The results for Argentina and Colombia are consistent with the results from the share regressions, with lending shares by foreign banks increasing more slowly than for similar domestically owned banks. The coefficients for Chile are statistically insignificant in both types of regressions. The one exception is Peru, where the coefficient on the origin dummy is positive and statistically significant in the growth regression, while the coefficient on the interaction between the trend term and the origin dummy in the share regressions is negative and statistically significant. Although this might seem contradictory, given the rapid growth of total assets held by foreign banks, from about 15 percent of total assets in December 1996 to nearly 40 percent of assets in March 2000, the two results might not be surprising - small business lending by foreign banks was growing, just not as fast as other types of lending by foreign banks. There is also broad consistency between the share regressions and the growth rate regressions with respect to state-owned banks. The coefficient on the dummy variable for state ownership in the growth regressions is negative and statistically significant in two of the countries with state-owned banks, Chile and Colombia, indicating that lending to small businesses by state-owned banks was growing slower than lending by private domestically owned banks. For the final country with state-owned banks, Argentina, the coefficient on the dummy variable indicating state ownership is positive, but statistically 16 insignificant. These results are robust to including the interaction between size and foreign ownership. The results provide further indication that small business lending might not be a primary objective of state-owned banks. D. Growth rate regressions - Large vs. small banks In the regressions omitting the interaction term between size and foreign ownership, the correlation between size and growth in small business lending is positive for Argentina and Chile (see columns (5.1) and (5.3)). This appears to be largely because growth was faster for large foreign banks than it was for small foreign banks. In the growth regressions that include the interaction between size and foreign ownership, the coefficient on size becomes smaller but remains positive for Argentina, and becomes negative and statistically insignificant for Chile. In contrast, the coefficients on the interaction term between size and origin are large, positive, and statistically significant in the regressions for both countries (see columns (5.2) and (5.4)). This suggests that lending to small businesses was growing considerably faster for large foreign- owned banks than it was for small foreign-owned banks, while the differences between large and small domestic banks were smaller. Figures 8 and 9 show estimated growth of lending to small businesses for foreign and domestically owned banks of different sizes for these two countries. In both Argentina and Chile, although lending to small businesses was growing more slowly for small foreign-owned banks than it was for small domestically owned banks, the reverse was true for large banks. This is our strongest evidence that large foreign banks were increasingly lending to small businesses. In the next section, we discuss why this occurred in Argentina and Chile, but not in Colombia or Peru. The statistically insignificant coefficient on the interaction term for both Colombia and Peru suggests that size affected growth of small business lending similarly for foreign and domestically owned banks in these countries. However, this does not imply that loan growth rates for foreign banks were less than those for domestic banks. The coefficient on the dummy variable indicating foreign ownership is also statistically insignificant in Colombia, which suggests that growth in lending was similar for similarly-sized foreign and domestically owned banks.'4 Moreover, when the interaction term is omitted 17 for Peru, the coefficient on the origin dummy variable becomes positive and statistically significant suggesting that lending to small businesses was growing faster there for both large and small foreign- owned banks than it was for similar domestic banks. The small business lending by foreign banks in Colombia and Peru may have been qualitatively different than that undertaken by large foreign banks in Argentina and Chile, but we do find evidence from the loan growth regressions from all four countries that suggests that foreign banks (or a subset of them) were expanding such lending at least as fast as both private domestic and state-owned banks during this period.'5 These results appear to undercut claims that foreign banks are unable or unwilling to enter the small business lending niche in developing countries. VII. Explaining differences in small business lending across countries: a preliminary exploration Our empirical results suggest that large foreign banks are more inclined to lending to small businesses in Argentina and Chile than in Colombia and Peru. The growth rate regressions for Argentina and Chile show that large foreign banks increased their lending at faster rates than large domestic banks. In Argentina, the large foreign banks had the highest estimated growth rates among all banks. In Chile, but not Argentina, large foreign banks devoted higher shares of their portfolios to small business lending than all banks, but the smallest domestic ones. Although many explanations might be consistent with these results, we briefly explore three possible causal factors: the quality of the contracting environment, the structure of the banking sector, and the macroeconomic environment. A. Contracting Environment Table 6 compares the countries using four indices that capture somewhat different aspects of the quality of the contracting environment. Figures for the United Kingdom and the United States are also included as reference points. These indices, whose constructions and sources are described in detail in the notes attached to Table 6, have become standard tools in cross-country empirical studies of financial development and economic growth. For three of the indices - the Index of Freedom's measure of the security of property rights, BERI's measure of contract enforcement, and the ICRG measure of the Rule of Law - the rankings are almost identical. Chile finishes first, followed relatively closely by Argentina. 18 Colombia and Peru finish third or fourth depending on the index. The gap between Argentina and Chile is typically much smaller than that between Argentina and either Colombia or Peru. The remaining index, developed by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (LLSV), measures the strength of the legal rights afforded to creditors in the event of bankruptcy or firm reorganization. It summarizes the types of laws that are on the books rather than the quality of the enforcerment of those laws. In that sense, the index differs from the other three, although, like the others, it indicates that laws in Chile and Argentina protect creditors better than in Colombia or Peru. In short, on a number of indicators, Chile and Argentina provide better contracting environments than either Colombia or Peru, and this could partially explain why large foreign banks are more inclined to lend to small businesses in those countries. We speculate that large foreign banks that use scoring methodologies to deliver relatively standardized loan products to a high number of small businesses or individuals need assurances that, in the event of default, recoveries can be made in a timely fashion. These indices might also help resolve why both the share and the growth rate regressions indicated that large foreign banks in Chile were emphasizing small business lending, while only the growth rate regressions provided similar evidence for Argentina. As noted, Chile held a slight edge over Argentina on these indices, which suggests that they enjoyed an adequate contracting environment for a longer period. Large foreign banks in Chile may have had enough time to reach their equilibrium shares of lending to small business. In Argentina, an adequate contracting environment arrived later than in Chile, and thus large foreign banks likely had still not achieved equilibrium shares of small business lending by the end of the period. B. Structure of Banking Sector The number and size of the banks in each country can also help explain the pattem of results. In Peru, for example, there were no foreign banks that approached the size of the largest domestic ones (refer to Figure 1). As a result, the type of small business lending undertaken by large foreign banks in Argentina and Chile was not a possibility in Peru. Of course, the fact that no large banks operated in Peru during this period is itself telling. It seems plausible that the weak contracting environment discouraged 19 their entry. Although some foreign banks were among the largest in Colombia, none of them compared to the largest banks in Chile or Argentina. Colombia's foreign banks, too, might have been too small to engage in small business lending through credit scoring. C. Macroeconomic environment The top panel in Figure 10 provides real growth rates for each of the four countries over this period. Because they exhibit almost the same growth patterns, it is unlikely that this factor can account for the differences in results across countries. One exception to the pattern is Argentina in 1995 and 1996, which, while recovering from the Tequila Crisis, was experiencing slower, but increasing growth, while the others were generally experiencing faster, but declining growth. However, our sample for Argentina starts in 1998. The bottom panel in Figure 10 does indicate, however, that Argentina and Chile both enjoyed lower inflation than either Colombia or Peru during this period. This, too, might have made it easier for large foreign banks to lend to small businesses. In short, it seems plausible that the contracting environment, the structure of the banking sector, and price stability all might have made it easier for large foreign banks in Argentina and Chile to lend to small businesses. However, further work is probably needed to confirm these suppositions. VIII. Conclusions As foreign participation increases in the banking sectors of developing countries, many questions remain unanswered. Key among them is the issue of whether foreign banks tend to shy away from lending to small businesses. Using bank level data for Argentina, Chile, Colombia, and Peru over the mid- 1 990s, this paper examines the impact of foreign bank entry on the share and growth rate of lending to small businesses, while controlling for other factors that might affect lending to this sector. Consistent with evidence from the U.S., we find that medium and large domestic banks in the four countries we study devote less of their lending (as share of total lending) to small businesses than small domestic banks. Also, in most countries, the share of lending devoted to small businesses by domestic banks was dropping in the late 1990s, as evidenced by the negative and significant trend in the share estimations. 20 Contrary to popular belief, public banks do not appear to surpass private banks in the extent to which they lend to small businesses. This would seem to indicate that the argument that privatization of public banks would hurt small businesses is, at best, weak. Regarding the impact of foreign bank entry on lending to small businesses we find, consistent with conventional wisdom, that, on average, foreign banks in the four countries generally lent less to small businesses (as share of total lending) than private domestic banks (at least by end of period). However, the difference appears to be primarily due to the behavior of small foreign banks. In all four cases, small foreign banks lent considerably less to small businesses than small domestic banks. In contrast, the difference was considerably smaller for large and medium-sized banks. In fact, after controlling for other factors that might affect lending, large foreign banks actually appear to lend more to small businesses (as share of total lending) than large domestic banks in two of the four case study countries, Chile and Colombia. Finally, in Argentina and Chile, the two cases where the ratio of banking sector assets to GDP consistently grew over the sample period, lending to small businesses by medium and large foreign banks was positive and was growing more quickly than for similar domestic banks. In contrast, lending by small foreign banks was shrinking and was growing far more slowly than for similar domestic banks. References Barajas, A., R. Steiner, and N. Salazar, "Foreign Investment in Colombia's Financial Sector," In The Internationalization of Financial Services: Issues and Lessons for Developing Countries, edited by Stjin Claessens and Marion Jansen, Boston, Mass.: Kluwer Academic Press, 2000. Berger, A., R. De Young, H. Genay, and G. Udell, "Globalization of Financial Institutions: Evidence from Cross-Border Banking Performance," Brookings-Wharton Papers on Financial Services, No. 3, 2000. Berger, A. N., L. Klapper, and G. F. Udell, "The Ability of Banks to Lend to Informationally Opaque Small Businesses", mimeo, World Bank, Washington DC, 2000. Berger, A., A. Saunders, J. Scalise, and G. Udell, "The Effects of Bank Mergers and Acquisitions on Small Business Lending", Journal of Financial Economics, 50(2), pp. 187-229, 1998. Berger, Allen N. and Gregory F. Udell, Relationship Lending and Lines of Credit in Small Firm finance, Journal of Business, 68, pp. 351-382, 1995. 21 Berger, A. and G. Udell, "Universal Banking and the Future of Small Business Lending," In Financial System Design: The Case for Universal Banking, edited by A. Saunders and I. Walter Burr Ridge, IL: Irwin Publishing, 1996. Bleger, L. and G. Rozenwurcel, "Financiamiento a las Pymes y Cambio Estructural en la Argentina: Un Estudio de Caso Sobre Fallas de Mercado y Problemas de Informacion," Desarrollo Economico, 40, pp. 45-71, 2000. Cho, Kang Rae, Suresh Krishnan, and Douglas Nigh, The State of Foreign Banking Presence in the United States, International Journal of Bank Marketing 5, pp. 59-75, 1987. Claessens, S., A. Demirguc-Kunt, and H. Huizinga, "The Role of Foreign Banks in Domestic Banking Systems" In The Internationalization of Financial Services: Issues and Lessons for Developing Countries, edited by S. Claessens and Marion Jansen Boston, Mass.: Kluwer Academic Press, 2000. Clarke, G., R. Cull, L. D'Amato, and A. Molinari, "On the Kindness of Strangers? The Impact of Foreign Entry on Domestic Banks in Argentina," in The Internationalization of Financial Services: Issues and Lessons for Developing Countries, edited by S. Claessens and M. Jansen, Boston, Mass.: Kluwer Academic Press, 2000. Dages, B. G., L. Goldberg, and D. Kinney, "Foreign and Domestic Bank Participation in Emerging Markets: Lessons from Mexico and Argentina," Federal Reserve Bank of New York Economic Policy Review, pp. 17-36, 2000. Demirgiic-Kunt, A., R. Levine, and H. G. Min, "Opening to Foreign Banks: Issues of Stability, Efficiency and Growth." In Proceedings of the Bank of Korea Conference on the Implications of Globalization of World Financial Markets, 1998. DeYoung, R., L. G. Goldberg, and L. J. White, "Youth, Adolescence, and Maturity of Banks: Credit availability to Small Business in an Era of Banking Consolidation," Journal of Banking and Finance, 23, pp. 463-492, 1999. Escude Guillermo, Tamara Burdisso, Marcelo Catena, Laura D'Amato, George Mc Candless, and Tomas Murphy, 2001, "Las MIPyMES y el Mercado de Credito en la Argentina," Documento de Trabajo Nro. 15, Banco Central de la Repuiblica Argentina. Focarelli, D. and A. Pozzolo, "The patterns of Cross-Border Bank Mergers and Shareholdings in OECD Countries," Journal of Banking and Finance, forthcoming, 2001. Goldberg, Lawrence G., The Competitive Impact of Foreign Commercial Banks in the United States, In The Changing Market in Financial Services, Proceedings of the Fifteenth Annual Economic Policy Conference of the Federal Reserve Bank of St. Louis, edited by in R. Alton Gilbert, Kluwer Academic: Norwell, MA, 1992. Goldberg, L. G., and L. J. White, "De Novo Banks and Lending to Small Businesses: An Empirical Analysis," Journal of Banking and Finance, 22, pp. 851-67, 1998 The Heritage Foundation and The Wall Street Journal, 1997 Index of Economic Freedom, Kim R. Holmes, Bryant T. Johnson, and Melanie Kirkpatrick, eds., Dow Jones and Co.: New York, NY. International Monetary Fund, "The Role of Foreign Banks in Emerging Markets," International Capital Markets: Developments, Prospects, and Key Policy Issues, September, 2000. Jenkins, Hatice, "Commercial Bank Behavior in Micro and Small Enterprise Finance." Development Discussion Paper No. 741. Harvard Institute for International Development, Harvard University, 2000. 22 Keeton, W., "Multi-office Bank Lending to Small Businesses: Some New Evidence," Federal Reserve Bank of Kansas City Economic Review 80, pp. 45-57, 1995. Keeton, W., "Do Bank Mergers Reduce Lending to Businesses and Farmers," Federal Reserve Bank of Kansas CityEconomicReview, 81, pp.63-75, 1996. Keeton, W., "The Effects of Mergers on Farm and Business Lending at Small Banks: New Evidence from the Tenth District States," Federal Reserve Bank of Kansas City, mimeo, 1997. Knack, S. and Keefer, P., "Institutions and Economic Performance: Cross Country Tests Using Alternative Institutional Measures," Economics and Politics, 7, pp. 207-227, 1995. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. W. Vishny, "Legal Determinants of External Finance," Journal of Finance, 1997. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. W. Vishny, "Law and Finance," Journal of Political Economy, 106, pp. 1113-55, 1998. Levine, R., "Foreign Bank Entry and Capital Control Liberalization: Effects on Growth and Stability," University of Minnesota, mimeo, 1999. Levonian, M. and J. Soller, "Small Banks, Small Loans, Small Businesses," Federal Reserve Bank of San Francisco, mimeo, 1995. Mester, Loretta J. , "What's the Point of Credit Scoring ?" Federal Reserve Bank of Philadelphia Business Review, pp.3-16, 1997. Peek, J. and E. Rosengren, "Small Business Credit Availability: How Important Is Size of Lender?," in Financial System Design: The Case of Universal Banking, edited by A. Saunders and I. Walter, Burr Ridge, IL: Irwin Publishing, 1996. Peek, J. and E. Rosengren, "Bank Consolidation and Small Lending: It's Not Just Bank Size That Matters," Journal of Banking and Finance, 22, pp. 799-819, 1998. Peek, J. and E. Rosengren, and F. Kasirye, "The Poor Performance of Foreign Bank Subsidiaries: Were the Problems Acquired or Created?" Journal of Banking and Finance, 23, pp. 579-604, 1999. Strahan, P. and J. Weston, "Small Business Lending and Bank Consolidation: Is There Cause for Concern?" Current Issues in Economics and Finance, Federal Reserve Bank of New York, 2, pp. 1-6, 1996. Strahan, P. and J. Weston, "Small Business Lending and the Changing Structure of the Banking Industry," Journal of Banking and Finance, 22, pp. 821-45, 1998. Walraven, N. "Small Business Lending by Banks Involved in Mergers," Finance and Economics Discussion Series 97-25, Board of Governors of the Federal Reserve, 1997. World Bank, Argentina Financial Sector Review, Washington, DC: The World Bank, 1998. 23 Table 1: Means Tests on Small Business Lending Shares Argentina Chile Colombia Peru Mean Mean Mean Mean (in %) (in °/O) (in %) (in °/O) All Domestic Banks 25.7 29.8 1.9 23.8 All Foreign Banks 19.7 24.1 1.8 17.9 (6.293) *** (5.574) *** (1.087) (4.608) *** Large Domestic Banks 16.9 19.3 1.0 15.8 Large Foreign Banks 9.9 24.3 1.3 10.2 (6.451) *** -(1.577) -(3.783) ** (3.774) *** Small Domestic Banks 26.4 30.5 2.0 24.6 Small Foreign Banks 21.8 24.1 1.8 18.2 (4.324) *** (6.132) *** (1.065) (4.905) *** Null hypothesis: Mean(domestic)-Mean(foreign)=0 t-stats are in parenthesis *, **, *** denote significance at the 10%, 5%, and 1% level, respectively Table 2: Means Tests on Small Business Lending Growth Rates Argentina Chile Colombia Peru Mean Mean Mean Mean (m %) (in °/O) (in %/O) (in %) All Domestic Banks 8.1 7.2 -12.9 -2.8 All Foreign Banks -3.8 0.0 -22.9 19.5 (3.326) *** (2.798) *** (1.604) -(4.281) ** Large Domestic Banks 14.9 13.8 -24.7 Large Foreign Banks 27.0 -0.2 -2.1 -(2.142) ** (1.399) -(1.603) Small Domestic Banks 7.6 6.8 -12.1 -1.9 Small Foreign Banks -6.5 0.0 -23.8 19.5 (3.659) * (2.569) ** (1.794) * -(3.886) * Null hypothesis: Mean(domestic)-Mean(foreign)=0 t-stats in parenthesis *, *, *** denote significance at the 10%, 5%, and 1% level, respectively 24 Table 3: Base Share of Small Business Lending Regressions Variables Argentina Chile Colombia Peru Log of real assets -0.115 -0.017 -0.040 -0.233 (Lagged I period) (-4.73) (-0.58) (-0.29) (-6.02) Return on assets -0.026 -6.255 -5.674 11.794 (Lagged 1 period) (-0.61) (-0.89) (-0.84) (0.88) Administrative expenses/assets 0.422 -5.519 -5.934 -41.651 (Lagged I period) -2.79 (-0.79) (-1.65) (-4.88) Public bank dummy -0.606 * -0.857 *** -0.466 ** (-8.21) (-7.95) (-1.97) Time trend 0.063 -0.044 * -0.248 ** -0.160 (0.96) (-1.8) (-2.29) (-3.80) Foreign bank dummy -0.129 -0.461 0.390 -0.150 (-0.86) (-2.99) (1.25) (-0.7) (Time trend)*(Foreign bank) -0.149 -0.025 -0.362 -0.207 ** (-1.42) (-0.65) (-1.59) (-2.27) Mergers dummy -0.265 *** -0.226 * -0.207 0.292 (-2.66) (-1.86) (-0.6) (1.48) Mergers* Age 0.255 * -0.063 0.651 0.377 -3.11 (-1.21) (1.58) (0.96) Foreign M&A dummy -0.200 0.392 -0.037 -0.010 (-2.61) (2.61) (-0.14) (-0.08) Foreign M&A*Age 0.111 -0.061 0.160 0.344 (1.01) (-0.72) (0.71) (3.00) De novo foreign entry dummy -8.417 (-69.33) (De novo foreign entry)*Age -1.138 (-3.64) Foreign purchase dummy 1.212 (8.97) Foreign purchase*Age 0.373 (1.63) Constant -0.018 0.038 -2.558 3.065 (-0.05) (0.04) (-0.93) (4.93) # of observations 1388 242 302 244 Adj. R-squared 0.32 0.28 0.08 0.38 Robust t-statistics within parentheses. * significant at 10, 5, and I percent, respectively. 25 Table 4: Share Regressions with Size and Foreign Interactions Variables Argentina Chile Colombia Peru Log of real assets -2.276 *** -9.287 *** 10.350 0.057 (Lagged I period) (-5.29) (-9.76) (1.52) (0.06) (Log of real assets)2 0.080*** 0.166*** -0.260 -0.014 (Lagged I period) (5.19) (9.58) (-1.57) (-0.43) (Log of real assets)*(Foreign bank) 2.156 *** 2.567 ** -0.496 ** -0.486 (Lagged I period) (3.22) (1.99) (-2.28) (-4.32) (Log of real assets) *(Foreign bank) -0.079 *** -0.037 0.024 ** 0.034 (Lagged I period) (-3.24) (-1.54) (2.28) (4.53) Return on assets -0.038 -15.641 ** -5.091 22.794 (Lagged I period) (-0.94) (-2.3) (-0.76) (1.72) Administrative expenses/assets 0.354 ** -5.600 -8.314 * -35.268 *** (Lagged I period) (2.54) (-0.8) (-1.79) (-4.20) Public bank dummy -0.652 *** -0.966 *** -0.172 (-8.31) (-9.28) (-0.74) Time trend 0.092 -0.068 *** -0.151 -0.103 (1.43) (-3.76) (-1.24) (-2.62) Foreignbankdummy -14.611 *** -43.018** 1.004** 0.146 (-3.2) (-2.47) (1.96) (0.518) (Time trend)*(Foreign bank) -0.192 * -0.034 -0.504 * -0.307 (-1.74) (-1.07) (-1.80) (-3.33) Mergers dummy -0.142 -0.342 *** 0.243 0.408 (-1.3) (-6.74) (0.42) (2.41) Mergers* Age 0.053 0.032 1.112 * -0.545 (0.54) (1.26) (1.95) (-1.29) Foreign M&A dummy -0.250 ** -0.413 * -0.504 -0.195 (-2.37) (-1.95) (-1.25) (-1.06) Foreign M&A*Age 0.152 -0.143 0.166 0.594 _______ ______ _______ _____ (1.15) (-1.46) (0.70) (3.12) De novo foreign entry dummy -8.327 * .__ _ _ __ - (-35.65) (De novo foreign entry)*Age -1.004 * (-3.48) Foreign purchase dummy 1.220 *** (8.96) Foreign purchase*Age 0.393 * (1.67) Constant 14.407 * 129.747 *** -105.999 1.715 (4.86) (9.9) (-1.53) (0.23) #of observations 1388 242 302 244 Adj. R-squared 0.34 0.55 0.16 0.47 Robust t-statistics within parentheses. * significant at 10, 5, and I percent, respectively. 26 Table 5: Real Small Business Lending Growth Regressions Variables Argentina Chile Colombia Peru (5.1) (5.2) (5.3) (5.4) (5.5) (5.6) (5.7) (5.8) Log of real assets 0.056 *** 0.020 * 0.038 *** -0.007 -0.039 -0.041 -0.044 ** -0.030 Lagged I year (5.81) (1.73) (3.46) (-0.54) (-1.00) (-1.12) (-2.53) (-1.81) (Log of real assets)*(Foreign bank) 0.108 *** 0.061 *** 0.006 -0.061 Lagged lyear (5.36) (2.97) (0.07) (-1.49) Return on assets 0.151 *** 0.131 ** 7.658 ** 8.311 ** 9.315 * 9.302 ** -2.769 -4.620 Lagged] year (2.84) (2.42) (2.01) (2.2) (2.02) (2.02) (-0.58) (-0.93) Administrative expenses/assets 0.104 * 0.098 * 0.372 -0.221 1.218 1.207 16.644 *** 15.647 Lagged lyear (1.81) (1.72) (0.23) (-0.13) (1.37) (1.36) (3.55) (3.31) Public bank dummy 0.004 0.048 -0.192 *** -0.135 ** -0.147 * -0.145 * (0.1) (1.23) (-2.94) (-2.04) (-1.90) (-1.91) Foreign bank dummy -0.131 *** -1.563 * 0.001 -1.675 *** -0.121 * -0.244 0.086 * 0.928 (-3.46) (-5.61) (0.04) (-2.95) (-1.71) (-0.14) (1.86) (1.58) Foreign M&A dummy -0.037 -0.057 -0.086 -0.127 -0.131 -0.138 0.173 0.171 (-0.72) (-1.51) (-1.13) (-1.6) (-0.94) (-0.85) (2.69) (2.65) Mergers dummy -0.206 *** -0.188 *** 0.026 0.046 -0.066 -0.061 (-3.95) (-4.01) (0.38) (0.69) (-0.63) (-0.62) Foreign purchase dummy -0.095 *** -0.084 ** (-2.74) (-2.53) Constant -0.699 *** -0.259 * -1.007 0.266 0.596 0.645 0.377 0.191 (-5.68) (-1.74) (-3.3) (0.65) (0.74) (0.83) (1.28) (0.68) #observations 548 548 212 212 183 183 168 168 Adj. R-squared 0.067 0.0958 0.14 0.1657 0.0837 0.0837 0.3567 0.3627 Robust t-statistics within parentheses. * significant at 10, 5, and 1 percent, respectively. 27 Table 6: Quality of Contracting Environment Security of Contract Rule of Lawc Creditor Property Enforcementb Rightsd Rights' (Index of Economic (BERI) (ICRG) (LLSV) Freedom) Argentina 4 2.07 3.21 1 Chile 5 2.42 4.21 2 Colombia 3 1.93 1.25 0 Peru 3 1.73 1.50 0 U.K. 5 3.42 5.14 4 U.S. 5 3.54 6.00 1 a Index from 1-5, from 1997 Index of Economic Freedom, The Heritage Foundation and the Wall Street Journal. A score of (5) indicates that "Private property guaranteed by the government, and efficient court system enforces contracts. Adequate justice system to punish those who unlawfully confiscate private property. Expropriation not likely"; (4) "Private property guaranteed by the government, but enforcement is lax. Expropriation unlikely"; (3) "Government recognizes some private property rights, such as land, but property can be nationalized. Expropriation possible"; (2) "Property Ownership is limited to personal items with little legal protection. Communal property is the rule. Expropriation likely, and government does not protect private property adequately. The legal system has collapsed"; (1) "Private property is outlawed. Everything belongs to the people or the state. Expropriation is certain, or the country is so corrupt and chaotic that property protection is nonexistent." b Index from 14, from Business Environmental Risk Intelligence (BERI). Measures the "relative degree to which contractual agreements are honored and complications presented by language and mentality differences." Higher scores indicate greater enforceability. Values averaged over 1980-95. As reported in Knack and Keefer (1995). ' Index from 1-6, from the International Country Risk Guide (ICRG), "reflects the degree to which the citizens of a country are willing to accept the established institutions to make and implement laws and adjudicate disputes." Higher scores indicate "sound political institutions, and strong court system, and provisions for an orderly succession of power." Lower scores indicate "a tradition of depending on physical force or illegal means to settle claims." Upon changes in government in countries scoring low on this measure, new leaders "may be less likely to accept the obligations of the previous regime." Values averaged over 1982-95. As reported in La Porta, Lopez-de- Silanes, Shleifer, and Vishny (1997). d Index from 04, that aggregates various creditor rights from La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998), Table 1. Index formed "by adding I when (1) the country imposes restrictions, such as creditors' consent or minimum dividends to file for reorganization; (2) secured creditors are able to gain possession of their security once the reorganization petition has been approved (no automatic stay); (3) secured creditors are ranked first in this distribution of the proceeds that result from the disposition of the assets of a bankrupt firm; and (4) the debtor does not retain the administration of its property pending the resolution of the reorganization." 28 Number of banks Number of banks 0 o * ' 00 -o o - o to to t o 8w 0 f I I I I I - I I I I = H H ( MI ° _~ t' I~~ ~ ~ ~~~~~~~~~~ @ I > ' .fo R~~~ ~~ ~ ~ ~~~~~~~~~ f Number of banks |Number of banks 0-o_ w tO * tA ON x0 O-ooto'--40 Mcn