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
The Impact of Sudden Stops on Bank Lending: Are there Cross-Sectional Differences? Michael Brei1 This draft: July 20, 2007 Abstract: Using annual financial statements of individual banks operating in 11 countries in East Asia and Latin America, we investigate whether domestic bank regulation and foreign bank entry can insulate emerging market economies from the effects of a sudden stop in foreign capital inflows. We use measures for the banks’ size, asset liquidity, capitalization, currency mismatch and the origin of the majority shareholder to identify the key determinants of domestic bank lending during sudden stops. To account for the different natures of sudden stops, the investigation discriminates not only between the two regions but also between sudden stops that are associated with major and minor devaluations. With regard to bank lending, we find evidence that in most cases sudden stops are associated with reductions in the domestic lending volume. Moreover, on average well-capitalized and foreign banks attenuate significantly more the adverse effects of sudden stops on the domestic lending volume playing an important stabilizing role. Keywords: Sudden stop, international capital markets JEL Classification: F34, F36, G21 1 Bonn Graduate School of Economics, e-mail: [email protected]. I would like to thank Joerg Breitung, Valeryia Dinger, Juergen von Hagen, Michael Schober, and the participants of the 8th Sir Arthur Lewis Institute of Social and Economic Studies Conference in Trinidad (2007) for useful comments and suggestions. The remaining errors are mine. The Impact of Sudden Stops on Bank Lending I Introduction In a sudden stop episode, an economy that has been the recipient of capital inflows, stops receiving such inflows, and instead faces unexpected withdrawals of foreign investments and the demand for the repayment of foreign loans that are falling due. Most of the considered episodes in this paper exhibited elements of a self-fulfilling crisis, in which capital withdrawals by some key investors resulted in a financial panic and unnecessarily deep recessions which most often affected particular regions entirely. The panic on its part may have been ’rational’ from the perspective of individual creditors, since each of them was trying to flee ahead of the others, even though the collective result was in some cases disastrous and the panic unnecessary, in the sense, that the countries’ fundamentals could have supported a much more favorable development. The particular outcome of the reversal in capital flows can be manifold ranging from outright default, to the rescheduling of debt payments, to bank runs, or to a rescue by a lender who provides new loans. Goldfajn (2001) states with regard to sudden stops: ’I define a sudden stop as a very large change in the supply of capital. Of course, this sudden stop is always in the negative direction. There are also problems with big booms of capital inflows in the sense that you need to know what you are doing with the big influx. But the real problem is when you get billions of dollars less from one year to the other - on the order of 10 percent of gross domestic product (GDP) or so. And most of the countries that had crises faced this challenge: Mexico, Asia, Turkey, Brazil, all of them.’ The present paper investigates recent sudden stop episodes from the perspective of individual banks that operate in the involved economies. Using information on the annual financial statements of banks, we address the question whether and in which way banks are affected. In particular, we explore a dataset comprising 945 1 The Impact of Sudden Stops on Bank Lending banks in 11 Asian and Latin American countries during 1991-2004 and test for crosssectional differences in the responses of banks to a total of 14 sudden stop episodes. To distinguish between banks with different financial positions and characteristics, we use measures for bank size, asset liquidity, capitalization, currency mismatch and the origin of the majority shareholder. The following sort of prediction is tested: a sudden stop has a disproportionately large impact on the lending volume of small domestic banks with vulnerable balance sheets.2 In emerging market economies, bank loans play an important role for firms, because domestic stock and bond markets are underdeveloped. For this reason, the banking sector plays a key role in determining the ability of an economy to attenuate the negative impact of large-scale capital withdrawals. The banking sector can cushion the impact, when it is in the position to grant additional loans to those sectors that found their credits cut. When banks themselves are in trouble, they can even be forced to liquidate outstanding loans and amplify the initial shock. With the underlying dataset, we can address these open issues and identify the main determinants of the banks’ vulnerability to external shocks. In an earlier work, Bernanke et al. (1991) compare the loan growth of large and well-capitalized banks with that of small and poorly-capitalized banks during the U.S. recession in 1990 and find evidence that poorly-capitalized banks contracted their lending volume by more than well-capitalized banks. Another strand of literature examines the impact of the rising foreign bank presence in emerging market economies.3 Several authors analyze the effects on competition and efficiency in the banking sector, see amongst others Claessens et al. (2001) and Martı́nez Perı́a et al. (2004). Most related to our work are Arena et al. (2006), Dages et al. (2000), Peek et al. (2000) and de Hass et al. (2002) who compare the lending behavior of foreign and domestic banks during crises. Arena et al. (2006) find evidence that 2 A bank is vulnerable when it is highly leveraged, especially, when the bank finances a domesticcurrency denominated, illiquid asset portfolio with foreign-currency denominated, short term debt. 3 See Clarke et al. (2003) for a review of this literature. 2 The Impact of Sudden Stops on Bank Lending the lending and deposit volume of foreign banks that operate in emerging market economies is less affected by financial crises than their domestic competitors. Peek et al. (2000) find that foreign banks in Latin America did not reduce their credit supply during economic recessions in the host country. More precisely, they argue that foreign banks view such situations as opportunities to expand by acquisition or by increasing their funding of existing subsidiaries. De Haas et al. (2002) find similar results for Central and Eastern European countries. The present paper focuses on related issues, however, our specific contribution is that we focus on bank lending in the context of a sudden stop, one of the most important sources of instability in emerging market economies. In particular, we find that the loan growth of well-capitalized banks is less affected by sudden stops, especially, when the sudden stops are associated with a large devaluation. In the case of foreign banks, we find similar results, but only during the sudden stops that are associated with minor devaluations. These findings indicate that a strong banking regulation, resulting in higher capitalization, and foreign bank entry are important determinants that can insulate an economy from the adverse effects of a sudden stop. The remainder of the paper is organized as follows. In Section II, we present a structural description of the adjustment mechanism within the banking sector in response to a sudden stop. In Section III, we describe the underlying dataset and present the empirical results on the impact of sudden stops on the banking sector. The final section concludes. 3 The Impact of Sudden Stops on Bank Lending II Sudden stops and the banking sector When foreign investors collectively refuse to roll-over existing debt and withdraw their investments, both the private and the public sector may be affected but not all parts in the same way. Initially, particular sectors face an unexpected reduction in the foreign supply of capital, later, potential effects on aggregate demand, prices, interest rates and the exchange rates affect the whole economy. As there are few papers on the effects of a sudden stop on banks, we intend to contribute with a structural description of the involved adjustment.4 Initial shocks Foreign creditors reduce their exposure to the economy and cut down credits and investments. This translates for banks to a reduction in external finance, in particular, in foreign funds (bonds, loans and deposits). At the same time, the deterioration in the economic conditions leads to an increased credit default risk from the part of the banks’ customers. In this sense, there a two adverse shocks, one to the asset and the other to the liability side of the banks’ balance sheets. Subsequent effects The impact on an individual bank depends strongly on its initial portfolio structure as well as on its clientele. The restricted access to global capital markets is likely to cause an overall increase in the costs of external finance, since the affected banks or firms increase their demand for domestic funding. In addition, the risky environment causes on average a higher risk premium on external finance. The increase in the costs of funding reduces, ceteris paribus, the banks’ equity via losses and profits. An increase in non-performing loans has the same effect. In particular cases, this can lead to a bank’s insolvency (liabilities exceed assets), or in expectation of a bankruptcy to a bank run. The sudden demand for the repayment of the foreign liabilities that are falling 4 A related theoretical investigation on a related issue, i.e. on aggregate liquidity shortages, is that of Diamond et al. (2005). The only empirical paper that focuses on sudden stops and the banking sector, however, from the macroeconomic perspective is that of Joyce et al. (2007). 4 The Impact of Sudden Stops on Bank Lending due can result in severe liquidity problems, especially, when too much assets are long-term and entail large costs in the case of liquidation (maturity mismatch). A sudden stop also involves pressures on the exchange rate. In the case of a pronounced depreciation, banks’ liabilities can rise extremely relative to assets (currency mismatch). In this case, the banks’ equity shrinks due to adverse stock effects of the devaluation. Domestic depositors may demand banks for the disbursement of their deposits. In particular cases, this can result in severe liquidity problems and trigger substantial uncertainty about the solvency of a bank which can trigger a bank run. Moreover, shifts of deposits towards banks with a sounder financial structure or a better reputation can occur. To underline the difficulties that a banking sector can face during a sudden stop, we quote here the statement of the IMF (1998) with respect to the situation in Indonesia: ’Following the closure of 16 insolvent banks in November last year, customers concerned about the safety of private banks have been shifting sizeable amounts of deposits to state and foreign banks, while some have been withdrawing funds from the banking system entirely... By mid November, a large number of banks was facing growing liquidity shortages, and were unable to obtain sufficient funds in the interbank market to cover this gap, even after paying interest rates ranging up to 75 percent. At the same time, another smaller group of banks [that is state and foreign owned banks] were becoming increasingly liquid, and were trading among themselves at a relatively low JIBOR (Jakarta Interbank Offer Rate) of about 15 percent... the Bank Indonesia was compelled to act. It provided banks in distress with liquidity support, while withdrawing funds from banks with excess liquidity...’. Determinants of vulnerability The following factors play a key role in de5 The Impact of Sudden Stops on Bank Lending termining the subsequent adjustment in the bank sector: the country-specific bank regulation (reserve requirements, solvability and liquidity rules, and deposit insurance arrangements); the central bank (can regulate domestic reserve and refinancing conditions, reallocate within the banking sector, and buy bonds issued from particular banks); and the banking sector’s average portfolio structure (currency and maturity composition of the banks’ balance sheets, the degree of capitalization and the amount of liquid assets). Responses of affected banks Banks can sell liquid assets which is probably the cheapest alternative to meet the demand for the repayment of foreign funds and make up the potential increase in the cost of external finance. Likewise, banks can securitize and sell parts of their outstanding loans. The third option is to call outstanding loans, e.g. forcing borrowers to discontinue long-lasting projects, and restructure them to harvest resources immediately. Alternatively, banks can attract new depositors or issue additional non-deposit funding (on the bond and interbank market, to the central bank). Here, the central bank plays an important role because it may buy bonds from the affected banks and, moreover, reallocate resources within the banking system. Foreign-owned banks can demand additional capital from their parent banks. Finally, banks can do a recapitalization (i.e. issue additional stocks). During sudden stops, however, only a few possibilities are a feasible option, since in many cases banks have to deal with large amounts of non-performing loans, deposit withdrawals and tense situations on capital and interbank markets. Sudden stops and the lending volume The lending volume of banks can be affected through several channels: an increase in the amount of non-performing loans reduces the lending volume; a reduction in a bank’s equity may push the bank below the limit of its minimal capital requirement and reduce the bank’s ability to grant new loans; adverse shocks to external finance and the cost of external finance have the same effect; and the conditions for central bank and interbank finance play an important role. 6 The Impact of Sudden Stops on Bank Lending Testable hypothesis The considerations above provide us several testable hypotheses which we will address in the following section. Unfortunately, we cannot address all questions because our database does not provide all necessary information, i.e. we do not have information on borrower and lender types, and the maturity and currency composition of assets and liabilities. The most important shortcoming is that we do not know which banks had a large exposure to foreign funding. We construct, however, a rough measure for the currency mismatch of the banks’ assets and liabilities. Our testable hypothesis is that the lending volume of large, well-capitalized and foreign banks with liquid asset portfolios and balance sheets that are not subject to a currency mismatch is less affected by a sudden stop than that of banks with the opposed characteristics. There are several reasons for this presumption: this group of banks is more reliable and may have a better reputation and, therewith, a more stable deposit base; a binding capital constraint is less likely for well-capitalized banks; liquid banks can build on a larger, better accessible stock of assets; foreign parent banks can allocate additional capital to their subsidiaries; banks without a currency mismatch are not subject to adverse stock effects on equity in the case of a devaluation; and these banks are more likely to have better access to external finance. III Empirical results Description of the database We use bank-level data from Bankscope and macroeconomic data from the International Financial Statistics (IFS).5 The unbalanced, annual dataset covers the period 1991-2004 and 11 countries from Latin America and East Asia.6 Overall, we have information on the annual financial state5 A more detailed description of the data is given in the Data Appendix. Latin America includes Argentina, Bolivia, Chile, Mexico, Peru and Uruguay, while East Asia includes Indonesia, South Korea, Malaysia, Thailand and the Philippines. 6 7 The Impact of Sudden Stops on Bank Lending ments of 945 individual banks that add up to about 5500 annual bank observations, distributed across time and countries as shown in Table 1. The data from Bankscope provides for each bank 87 balance sheet (sub-) categories, however, in most cases only 38 categories are available. In addition, the database provides information on the bank history, bank specialization, foreign shareholders and a detailed decomposition of losses and profits. In many cases, Bankscope reports consolidated and unconsolidated financial statements, and we used unconsolidated figures to the extend possible to reduce variations arising from changes in the subsidiaries’ ownership and to work with comparable accounting data. The specific dates of the banks’ financial years vary in some cases, however, for the majority of banks the financial year ends in December. To assign the macroeconomic variables to individual banks correctly, we use annualized quarterly data to the extend possible (in cases where quarterly data is not available, we use weighted averages of the variables of two subsequent years). To remove nominal variations, the balance sheet positions of each bank were transformed either to ratios over the bank’s total assets or expressed in real terms by dividing the balance sheet positions with the country-specific consumer price index. We excluded central banks and observations for which the real growth rate of particular balance sheet positions (loans, securities, deposits, non-deposit funding, equity, interest expenses, revenues and problem loans) exceeded 300% in absolute terms since these observation have an extraordinary high influence on the point estimates of the regressions. Average balance sheet positions Summary statistics for the average balance sheet positions as a ratio of total assets for the East Asian and Latin American economies are presented in Table 2. Latin American banks tend to have, on average, higher equity to asset ratios, moreover, these banks rely relatively more on money market funding (i.e. certificates of deposits and debt securities) than on other funding (convertible bonds and subordinated debt). On the asset side, East Asian banks 8 The Impact of Sudden Stops on Bank Lending tend to have larger fractions of problem loans and total other earning assets (i.e. bonds and deposits with banks). Measures for different bank types To distinguish between different groups of banks, we use measures for the bank size, asset liquidity, capitalization, currency mismatch and foreign ownership. As a measure for the size of an individual bank in a given year and country, we use the ratio of a bank’s total assets to the country average of total assets in a given year. Asset liquidity is defined by the ratio of the sum of marketable securities, government securities (including treasury bills) and cash holdings to total assets. Capitalization is defined by the ratio of total equity to total assets. For foreign banks, we include a dummy variable based on the Bankscope information on the origin of the majority shareholders (i.e. those who own at least 50% of the shares). As mentioned, there is no direct measure for the currency composition of assets and liabilities. We construct a rough measure for the degree of currency mismatch as follows: a percentage change in the difference between post tax profits and the change in equity that is similar to the periods’ devaluation indicates a currency mismatch.7 To be more precise, we consider all periods with devaluations larger than 10% and classify a bank as one with a currency mismatch when the percentage change in the difference between profits and the change in equity is between 0.5 and 1.5 times the devaluation. Identification of sudden stops To identify sudden stops, we follow Calvo et al. (2004) who identify a sudden stop episode by the following three criteria: (1) the episode includes at least one observation where the year-to-year fall in quarterly (net) capital inflows lies at least two standard deviations below its sample mean; 7 Consider a simple example: a bank grants all loans L in domestic currency and finances them with dollar debt D. Abstracting from other assets and external funds, the balance sheet can be written as L = eD +E, where e denotes the exchange rate and E equity. The resource constraint is given by L̇ = Π + eḊ, where Π denotes retained earnings and ẋ a change in x. Total differentiating the balance sheet identity and using the resource contraint leads to ėD = Π − Ė, i.e. the adverse stock effect due to a devaluation can be measured by the difference of retained earnings and the change in equity. 9 The Impact of Sudden Stops on Bank Lending (2) the sudden stop episode begins (ends) when the year-to-year fall (rise) in capital inflows falls (rises above) one standard deviation below its sample mean; (3) there is an associated output contraction. We calculated the first and second moments of capital inflows using a rolling window including the 12 previous quarters of each period to capture better the surprise element of sudden stops. Figure 1 in the Appendix shows potential sudden stop periods (i.e. those that satisfy the first criterion) of the 9 countries for which we were able to obtain quarterly macroeconomic data. To detect the sudden stops of Malaysia and Uruguay, we used annual information combined with quarterly information on the nominal exchange rate, i.e. we first identified the year of the sudden stop using the criteria from above and then identified the quarter in which a large nominal depreciation occurred. The exact dates and durations of the sudden stop episodes are reported in Table 1. Moreover, Figure 2 shows that the sudden stops in the considered economies bunch together, especially, in the East Asian region during the years 1997-98. The majority of sudden stops in Latin America occurred thereafter. In all, the dataset covers 14 episodes of sudden stops. Construction of the sudden stop dummy For the empirical investigation, we construct a bank-specific sudden stop dummy variable that equals to one during a sudden stop period and zero elsewhere. The particular dates and durations of sudden stops vary in most cases. We account for this fact as follows: we assign to the dummy variable a value of one, if the sudden stop took at least two quarters of the associated financial year, e.g. when a sudden stop occurred between Q2/97 and Q1/98, the dummy variable equals to one for those banks that published their financial statements in Q3/97, Q4/97, Q1/98, Q2/98 and Q3/98. As a consequence, the duration during which banks are affected by a sudden stop differs, i.e. in the previous example, banks that balance their accounts in Q4/97 are affected during 3 quarters, while those that balance their accounts in Q1/98 are affected the whole year. To account for this difference, we construct an additional dummy variable 10 The Impact of Sudden Stops on Bank Lending that takes a value of one if two quarters of the financial year are affected, a value of two if 3/4 of the year are affected, and a value of three in the case of the whole year.8 Sudden stops and the lending volume Preliminary investigation Figures 3 and 4 in the Appendix show the average net annual capital inflows as a share of GDP and the dollar value of the average lending and deposit volume of all banks for the Latin American (LA-6) and East Asian (A-5) countries. Most of the major sudden stop periods, i.e. 1994-95 and 2001-02 in Latin America and 1997 in East Asia, are associated with contractions in the average lending and deposit volume. In these cases, the lending volume lasts about 3 years below its level prior to the sudden stop. An exception is the series of sudden stops that occurred in Latin America in 1998. A similar pattern can be observed in the case of the average deposit volume. Figures 5 and 6 compare the dollar value of the average lending volume of all banks with that of well-capitalized banks, i.e. banks with a equity to asset ratio between 10% and 50%, and that of foreign banks. In Latin America, the lending volume of well-capitalized banks is less affected during the sudden stops compared to the average bank. Moreover, the lending volume of this group of banks increases on average by more in the aftermath of the sudden stops. In the case of foreign banks, the conclusions are less evident. Econometric investigation To test for cross-sectional differences in the effect of sudden stops on the lending volume, we estimate the following fixed effects regression Li,c,t = αi + p X βs xc,t−s + βus xt + γzi,c,t−1 + γf fi,c,t + δssi,c,t (1) s=0 + p X βs∗ xc,t−s ssi,c,t + γ ∗ zi,c,t−1 ssi,c,t + γf∗ fi,c,t ssi,c,t + ui,c,t s=0 8 The results using this dummy variable are not reported in this version of the paper and can be obtained from the author upon request. 11 The Impact of Sudden Stops on Bank Lending where i = 1, ..., N refers to individual banks, c = 1, ..., C to countries and t = 1, ..., Ti to the time dimension. The dependent variable Li,c,t denotes the banks’ growth rate of real total loans (net of problem loans) and αi bank-level fixed effects. The vector of country-specific variables xc,t−s controls for changes in the demand for loans caused by changes in economic conditions (real GDP growth) and monetary conditions (real money market rate). Moreover, we control for changes due to possible devaluations by including the change in the nominal exchange rate. The vector of bank-specific variables zi,c,t−1 includes the measures for bank size, asset liquidity, currency mismatch and capitalization. Due to potential endogeneity problems which would lead to inconsistent OLS estimates, these variables enter the regression with one lag.9 The dummy variables that measure the degree of currency mismatch and foreign ownership fi,c,t enter the regressions contemporaneously. In addition, we include an international interest rate, xt , measured by the real federal funds rate. The bank-specific sudden stop dummy variable is denoted by ssi,c,t . To test for cross-sectional differences among our bank categories each bankspecific variable is interacted with the sudden stop dummy. Moreover, to account for changes in loan demand particular for the sudden stops, we interact the control variables with the sudden stop dummy. The key issue is interpreting the coefficients γ ∗ and γf∗ associated with the interaction terms. In particular, our presumption is that the lending volume of vulnerable banks declines by more in response to a sudden stop. That is, we should find that γ ∗ and γf∗ are significantly larger than zero. Table 3 shows the results for Latin America and Table 4 for East Asia. To check for the robustness of the parameter estimates, we estimate the model in three specifications: first, the exchange rate and the measure on the currency mismatch are excluded; second, only the measure on the currency mismatch is excluded; and, 9 To be more precise, with regard to the size category there is a possible joint determination since a bank may become larger, precisely, because of a large loan growth. Similar problems arise with capitalization and asset liquidity. 12 The Impact of Sudden Stops on Bank Lending finally, the whole model is estimated. For Latin American banks, we find strong evidence for cross-sectional differences in the response of lending to sudden stops, however, only well-capitalized banks have a significantly higher growth rate of loans compared to the other banks. A 1% increase in capitalization implies a 0.76% higher growth rate of loans during sudden stops. For East Asian banks, we reach the same conclusion in the case of foreign banks, i.e. the growth rate of loans of foreign banks is 14% higher than that of the other banks during sudden stops. Finally, we find no evidence that the growth rate of loans of liquid and large banks and of those banks that are not subject to a currency mismatch is significantly higher during sudden stops. During tranquil times, the estimated coefficients on liquidity and capitalization are significantly positive in both regions, i.e. their growth rate of total loans is higher than that of the other banks. The coefficients associated with the size category are in both cases significantly negative. The same regressions are estimated for two distinct groups of sudden stops, i.e. Table 5 shows the results for sudden stops that were accompanied with devaluations larger than 30% and Table 6 shows the results for the other sudden stop episodes.10 Interestingly, we find in all specifications and cases that the loan growth of wellcapitalized banks is significantly larger than that of the other banks. The loan growth of foreign banks is significantly higher only in the case of sudden stops that are not associated with large devaluations. Sudden stops and other balance sheet positions To get a broader impression of the impact of a sudden stop on banks, we conduct a similar investigation using the other major balance sheet positions of banks, i.e. for the real growth rates of deposits, equity, problem loans, interest expenses, revenues, non-deposit funding and securities. The results for Latin America and East Asia are shown in Table 7.11 10 The sudden stops that were accompanied with devaluations larger than 30% are those of Mexico, Thailand, Indonesia, Uruguay, Philippines, Malaysia and Korea. 11 Note that the previous results obtained for the loan growth are reproduced in the first column of each Table. 13 The Impact of Sudden Stops on Bank Lending For Latin America, an interesting finding is that foreign banks receive additional deposits during a sudden stop suggesting that deposits were shifted towards these banks during the sudden stops. With regard to East Asia, we find that foreign banks have a significantly higher growth rate of revenues during the sudden stops and a lower growth rate of problem loans. Banks with a liquid asset portfolio have a significantly higher growth rate of interest expenses and non-deposit funding during the sudden stops, and lower growth rates of securities and problem loans. These findings might point to the fact that these banks used their buffer stock of liquid assets and that they were able to issue additional non-deposit funding. Finally, an interesting finding is that well-capitalized and foreign banks received additional deposits during the sudden stops which might be an explanation of our finding with regard to the higher loan growth of foreign banks. IV Conclusion In the present paper, we test for cross-sectional differences in the response of bank lending to sudden stops of capital inflows. As motivated in Section II, the lending volume of large, well-capitalized and foreign banks and those with a liquid asset portfolio and no currency mismatch between assets and liabilities should be less affected by a sudden stop than that of banks with the opposite features. Our empirical investigation reached a few results that are worth summarizing. For the banking sectors as a whole, we find evidence that sudden stops are in many cases associated with contractions in the aggregate lending and deposit volume. Moreover, we find evidence that well-capitalized and foreign banks increase their lending volume relative to the other banks even after controlling for changes in loan demand and the exchange rate. In addition, we find evidence that this finding is partly due to the fact that these banks had a more stable deposit base during sudden stops than the other types of banks. There might be another possible explanation 14 The Impact of Sudden Stops on Bank Lending for this finding: the other banks were unable to fulfill capital adequacy requirements, in the sense, that insufficient equity capital is the restricting force behind the banks’ lending activity. To answer this question, information on country-specific capital requirements should be taken into consideration. We leave this issue to future work. Other possible improvements could be: (i) construct additional variables for particular bank characteristics (e.g. state-owned banks), (ii) include country dummy variables, and (iii) take into account information on deposit insurance arrangements. 15 The Impact of Sudden Stops on Bank Lending V Literature Arena, Marco (2005), ’Bank Failures and Bank Fundamentals: A Comparative Analysis of Latin America and East Asia during the Nineties using Bank-Level Data’, Bank of Canada Working Paper, 05/19 Arena, Marco, Carmen Reinhart and Francisco Vázquez (2006), ’The Lending Channel in Emerging Economies: Are Foreign Banks different?’, NBER Working Paper Series, No. 12340 Barajas, Adolfo and Roberto Steiner (2002), ’Why Don’t They Lend? Credit Stagnation in Latin America’, IMF Staff Papers, Vol. 49, Special Issue Barajas, Adolfo, Ralph Chami and Thomas Cosimano (2005), ’Did the Basel Accord Cause a Credit Slowdown in Latin America?’, IMF Working Paper, No. 05/38 Bernanke Ben S., Cara S. Lown and Benjamin M. Friedman (1991), ’The Credit Crunch’, Brooking Papers on Economic Activity, Vol. 1991(2) Calvo, Guillermo, Alejandro Izquierdo and Luis-Fernando Meija (2004), ’On the Empirics of Sudden Stops: The Relevance of Balance Sheet Effects’, NBER Working Paper Series, No. 10520 Calvo, Guillermo and Ernesto Talvi (2005), ’Sudden Stop, Financial Factors and Economic Collapse in Latin America: Learning from Argentina and Chile’, NBER Working Paper Series, No. 11153 Claessens, S., Asli Demirguc-Kunt and H. Huizinga (2001), ’How does foreign bank entry affect domestic banking markets’, Journal of Banking and Finance, Vol. 25(5) Clarke, G., M. Martı́nez Perı́a and S. Sánchez (2003), ’Foreign Bank Entry: Experience, Implications for Developing Countries, and Agenda for Further Research’, World Bank Observer, Vol. 18(1) Dages, B.G., L.B. Goldberg and D. Kinney (2000), ’Foreign and Domestic Bank Participation in Emerging Markets: Lessons from Mexico and Argentina’, 16 The Impact of Sudden Stops on Bank Lending Federal Reserve Bank of New York Economic Policy Review, September De Haas R. and I. van Lelyveld (2002), ’Foreign Bank Penetration and Private Sector Credit in Central and Eastern Europe’, De Nederlandsche Bank Staff Reports, No. 91 Demirguc-Kunt, Asli and Tolga Sobaci (2000), ’Deposit Insurance Around the World: A Data Base’, Working Paper, World Bank Diamond, Douglas W. and Philip H. Dybvig (1983), ’Bank Runs, Deposit Insurance, and Liquidity’, The Journal of Political Economy, Vol. 91(3) Diamond, Douglas W. and Raghuram G. Rajan (2005), ’Liquidity Shortages and Banking Crises’, The Journal of Finance, Vol. 60(2) Galindo, Arturo and Fabio Schiantarelli (2002), ’Credit Constraints in Latin America: An Overview of the Micro Evidence’, Working Paper, Inter-American Development Bank Goldfajn, Ilan (2001), ’Roundtable Comments on Monetary and Regulatory Policy’, Domestic Finance and Global Capital in Latin America Conference, Federal Reserve Bank of Atlanta Guidotti Pablo E., Federico Sturzenegger and Agustin Villar (2004), ’On the Consequences of Sudden Stops’, Economia, Vol. 4(2) International Monetary Fund (1998), ’Indonesia – Memorandum on Economic and Financial Policies’, Washington (January 15) Jayaratne, Jith and Donald P. Morgan (2000), ’Capital Market Frictions and Deposit Constraints at Banks’, Journal of Money, Credit, and Banking, Vol. 32(1) Joyce, J. P. and M. Nabar (2007), ’Sudden Stops, Banking Crises and Investment Collapses in Emerging Markets’, Wellesley College Working Paper Kashyap, Anil K. and Jeremy C. Stein (1995), ’The Impact of Monetary Policy on Bank Balance Sheets’, Carnegie-Rochester Conference Series on Public Policy 42 17 The Impact of Sudden Stops on Bank Lending Loutskina, Elena (2005), ’Does Securitization Affect Bank Lending? Evidence from Bank Responses to Funding Shocks’, Manuscript, Caroll School of Management Martı́nez Perı́a, M. and A. Mody (2004), ’How foreign Participation and Market Concentration Impact Banking Spreads: Evidence from Latin America’, Journal of Money, Credit and Banking, Vol. 36(3) Peek, Joe and Eric Rosengren (1995), ’The Capital Crunch: Neither a Borrower Nor a Lender Be’, Journal of Money, Credit and Banking, Vol. 27(3) —– (2000), ’Implications of the Globalization of the Banking Sector: The Latin American Experience’, New England Economic Review, September/October Radelet, Steven and Jeffrey D. Sachs (1998), ’The East Asian Financial Crisis: Diagnostics, Remedies, Prospects’, Brookings Papers on Economic Activity, Vol. 1998 (1) Standard and Poors (2002a), ’Views on the Major Latin American Banking Systems’, http://www.standardandpoors.com —– (2002b), ’ The Argentine Crisis: A Chronology of Events After The Sovereign Default’, http://www.standardandpoors.com 18 The Impact of Sudden Stops on Bank Lending VI A Appendix Data Bank-level data: Source – Fitch IBCA/Bureau van Dijk’s Bankscope database from 2002 and 2004 Macroeconomic data: Source – International Financial Statistics 2006, International Monetary Fund Included variables: • GDP (series 99B) • GDP deflator (series 99BIP) • money market rate (series 60B)12 • financial (capital) account (series 78BJD) • consumer price index (series 64ZF) • federal funds rate (series 60B) • market exchange rate (series RF) 12 For the countries Bolivia, Chile, Peru and Uruguay deposit rates (seires 60L) were used instead, because of missing information on money market rates. Real interest rates were calculated by the difference in nominal interest rates and CPI inflation. 19 The Impact of Sudden Stops on Bank Lending B Tables and Figures Table 1: Overview of the dataset and sudden stop episodes per country13 Country Bank year obs. Number of banks Average no. of obs. per bank Sudden stops of capital inflows Assoc. annual % change in capital inflows per GDP Argentina 1080 161 7 9 7 8 5 Q4/94-Q1/95 Q2/98-Q3/99 Q4/00-Q3/01 Q2/99-Q3/99 Q4/02-Q3/03 Q1/98-Q4/98 Q4/94-Q1/95 Q3/97-Q1/98 Q2/02-Q3/03 -3.8 -1.5 -8.0 -5.1 -4.3 -6.0 -4.2 -6.4 -14.2 Bolivia 200 19 11 Chile Mexico Peru Uruguay 452 530 270 188 50 75 32 38 Indonesia Korea Malaysia Philippines Thailand 860 482 747 468 241 120 76 98 62 64 7 6 8 8 4 Q3/97-Q1/98 Q3/97-Q1/98 Q3/97-Q1/98 Q3/97-Q3/98 Q2/96-Q4/97 -5.1 -6.2 -7.3 -5.8 -17.5 Sum/average∗ 5518 945 5.84∗ 13 14 -6.3∗ The sample period is 1991 to 2004. Sudden stops of capital inflows are defined as in Calvo et al. (2004), see above. 20 The Impact of Sudden Stops on Bank Lending Table 2: Average composition of the major bank balance sheet positions per total assets in the period from 1991 to 2004 East Asia Latin America Customer Loans Problem Loans Loan Loss Reserves 61.1 10.3 5.3 59.7 8.1 5.2 Net Loans 55.8 54.5 Deposits with Banks Securities Equity Investments 12.1 16.7 3.4 8.4 17.1 2.4 Total Other Earning Assets 32.2 27.9 8.9 3.1 12.6 5.0 100.0 100.0 Total Deposits Money Market Funding Other Funding Other Liabilities 66.2 7.5 9.1 4.1 63.2 13.4 3.5 2.4 Total Liabilities 86.9 82.5 Total Equity 13.1 17.5 100.0 100.0 2676 412 2400 376 assets Total Non Earning Assets Fixed Assets Total Assets liabilities Total Equity+Total liabilities No. of Observations No. of Banks Asset side: Customer Loans include Private, Corporate, Government, Bank Loans, Mortgages and Leases; Problem Loans include Problem, Overdue, Restructured and Other Non-Performing Loans; Securities include Investment and Trading (Government) Securities; Total Other Earning Assets include Deposits with Banks, Securities, Equity Investments and Bonds; Total Non Earning Assets include Cash and Due from Banks, Intangible Assets and Deferred Tax Receivables; Fixed Assets include Land and Buildings and Other Tangible Assets. Liability side: Total Deposits include Customer and Bank Deposits; Money Market Funding include Certificates of Deposits, Commercial Papers, Securities and Other Negotiable Instruments; Other Funding include Bonds, Subordinated Debt, Hybrid Capital and Other Funding; Other Liabilities include Loan Loss and Non Equity Reserves; and Total Equity include Retained Earnings, Other Equity Reserves, Preference and Common Shares. 21 The Impact of Sudden Stops on Bank Lending Table 3: Estimation results for Latin America14 Dependent Variable: Growth rate of real total loans Method: Panel least squares, Sample: 1992 2004 White cross-section standard errors & covariance (d.f. corrected) Cross-section fixed (dummy variables) effects specification C real GDP growth real money market rate real GDP growth(-1) real money market rate(-1) real US money market rate devaluation real GDP growth*SS real money market rate*SS real GDP growth(-1)*SS real money market rate(-1)*SS devaluation*SS size(-1) capitalization(-1) liquidity(-1) foreign size(-1)*SS capitalization(-1)*SS liquidity(-1)*SS foreign*SS high $ debt SS Adjusted R2 Cross-sections Observations Model 1 -0.07 -0.14 -0.59∗∗ 0.32 -0.39∗∗∗ 5.12∗∗∗ -0.08∗∗∗ 0.89∗∗∗ 0.64∗∗∗ -0.17 0.00 0.49∗∗∗ -0.15 0.09 Model 2 -0.07 -0.23 -0.49∗ 0.30 -0.38∗∗∗ 4.74∗∗∗ -0.03 -0.30 0.53 1.66∗ -2.31∗∗ 0.22 -0.08∗∗∗ 0.88∗∗∗ 0.64∗∗∗ -0.17 0.00 0.51∗∗∗ -0.14 0.09 – – -0.02 -0.01 Model 3 -0.10 -0.39 -0.69∗∗ 0.38 -0.38∗∗ 5.15∗∗∗ -0.02 -0.29 1.57∗∗ -1.02 -5.21∗∗ -0.03 -0.07∗∗∗ 0.95∗∗∗ 0.64∗∗∗ -0.15 0.00 0.76∗∗∗ -0.10 0.09 0.05 -0.05 0.11 355 2094 0.11 355 2094 0.12 209 1383 – -0.22 0.85∗∗ -1.84∗∗ -2.22∗∗ – 14 SS indicates the sudden stop dummy variable, and (∗∗∗ ,∗∗ ,∗ ) indicate significance at the 1%, 5% and 10% level. The included countries are Argentina, Mexico, Peru, Uruguay, Bolivia and Chile. 22 The Impact of Sudden Stops on Bank Lending Table 4: Estimation results for East Asia15 Dependent Variable: Growth rate of real total loans Method: Panel least squares, Sample: 1992 2004 White cross-section standard errors & covariance (d.f. corrected) Cross-section fixed (dummy variables) effects specification C real GDP growth money market rate real GDP growth(-1) money market rate(-1) real US money market rate devaluation real GDP growth*SS money market rate*SS real GDP growth(-1)*SS money market rate(-1)*SS devaluation*SS size(-1) capitalization(-1) liquidity(-1) foreign size(-1)*SS capitalization(-1)*SS liquidity(-1)*SS foreign*SS high $ debt SS Adjusted R2 Cross-sections Observations Model 1 -0.01 1.77∗∗∗ 0.34 0.53∗∗∗ -2.24∗∗∗ 2.53∗∗∗ -0.33 – -0.27 Model 3 0.01 1.96∗∗∗ 0.47 0.46∗∗∗ -1.85∗∗∗ 2.41∗∗∗ -0.08 -0.86 1.38 3.03 -0.13 0.02 -0.09∗∗∗ 0.57∗∗∗ 0.33∗∗∗ 0.06 0.00 -0.58 -0.09 0.14∗∗ 0.05 -0.22 0.23 386 2202 0.23 386 2202 0.24 260 1773 – -1.42 1.44∗ 4.38 0.07 – -0.07∗∗∗ 0.63∗∗∗ 0.37∗∗∗ 0.06 0.00 -0.33 -0.06 0.17∗∗∗ – 15 Model 2 -0.01 1.83∗∗∗ 0.42 0.53∗∗∗ -2.09∗∗∗ 2.48∗∗∗ -0.05 -1.39 1.52 4.05 0.18 -0.01 -0.07∗∗∗ 0.63∗∗∗ 0.37∗∗∗ 0.07 0.00 -0.38 -0.01 0.17∗∗∗ SS indicates the sudden stop dummy variable, and (∗∗∗ ,∗∗ ,∗ ) indicate significance at the 1%, 5% and 10% level. The included countries are Thailand, Indonesia, Philippines, Malaysia and Korea. 23 The Impact of Sudden Stops on Bank Lending Table 5: Estimation results for Sudden Stops coupled with high devaluations 16 Dependent Variable: Growth rate of real total loans Method: Panel least squares, Sample: 1992 2004 White cross-section standard errors & covariance (d.f. corrected) Cross-section fixed (dummy variables) effects specification C real GDP growth real money market rate real GDP growth(-1) real money market rate(-1) real US money market rate devaluation real GDP growth*SS real money market rate*SS real GDP growth(-1)*SS real money market rate(-1)*SS devaluation*SS size(-1) capitalization(-1) liquidity(-1) foreign size(-1)*SS capitalization(-1)*SS liquidity(-1)*SS foreign*SS high $ debt SS Adjusted R2 Cross-sections Observations Model 1 -0.03 1.48∗∗∗ 1.35∗∗∗ -0.23∗ -0.46∗∗∗ 0.79 -0.08∗∗∗ 0.72∗∗∗ 0.44∗∗∗ 0.07 -0.01 0.54∗∗ 0.02 0.08 Model 2 -0.04 1.56∗∗∗ 1.57∗∗∗ -0.21 -0.38∗∗∗ 0.96 -0.14∗∗∗ 2.27∗∗ -1.39∗∗∗ -2.06 -0.87 0.19∗ -0.08∗∗∗ 0.73∗∗∗ 0.44∗∗∗ 0.07 -0.01 0.55∗∗ 0.03 0.08 – – 0.15 0.10 Model 3 -0.03 1.61∗∗∗ 1.60∗∗∗ -0.35∗∗ -0.39∗∗∗ 1.15∗ -0.16∗∗∗ 1.59 -1.26∗∗∗ 0.18 -0.80 0.22∗ -0.09∗∗∗ 0.75∗∗∗ 0.42∗∗∗ 0.04 0.01 0.47∗ 0.17 0.09 0.02 -0.08 0.14 504 2786 0.15 504 2786 0.16 340 2227 – 2.63∗∗∗ -1.09∗∗∗ -2.38∗ -1.10 – 16 SS indicates the sudden stop dummy variable, and (∗∗∗ ,∗∗ ,∗ ) indicate significance at the 1%, 5% and 10% level. The included countries are Mexico, Thailand, Indonesia, Uruguay, Philippines, Malaysia and Korea. 24 The Impact of Sudden Stops on Bank Lending Table 6: Estimation results for Sudden stops coupled with minor devaluations17 Dependent Variable: Growth rate of real total loans Method: Panel least squares, Sample: 1991 2004 White cross-section standard errors & covariance (d.f. corrected) Cross-section fixed (dummy variables) effects specification C real GDP growth money market rate real GDP growth(-1) money market rate(-1) real US money market rate devaluation real GDP growth*SS money market rate*SS real GDP growth(-1)*SS money market rate(-1)*SS devaluation*SS size(-1) capitalization(-1) liquidity(-1) foreign size(-1)*SS capitalization(-1)*SS liquidity(-1)*SS foreign*SS high $ debt SS Adjusted R2 Cross-sections Observations Model 1 -0.07 0.17 -1.02∗∗∗ 0.27 -0.54∗∗∗ 5.90∗∗∗ -0.09 – -0.27 Model 3 -0.05 0.19 -1.47∗∗∗ 0.16 -1.03∗∗ 7.09∗∗∗ 0.09∗ -0.54 3.71∗∗∗ 0.29 -6.64∗∗∗ -1.43 -0.09∗∗∗ 0.73∗∗∗ 0.56∗∗∗ -0.06 0.01 0.61∗∗∗ 0.06 0.14∗ 0.11 -0.16 0.13 237 1510 0.13 237 1510 0.15 129 929 – -0.06 2.27∗∗∗ -1.01 -3.29∗∗∗ – -0.09∗∗∗ 0.81∗∗∗ 0.59∗∗∗ -0.13 0.01 0.30∗ -0.14 0.17∗∗ – 17 Model 2 -0.08 0.34 -1.21∗∗∗ 0.23 -0.64∗∗∗ 6.59∗∗∗ 0.05 -0.20 2.56∗∗∗ -1.07 -3.34∗∗∗ 0.05 -0.09∗∗∗ 0.80∗∗∗ 0.60∗∗∗ -0.12 0.01 0.31∗ -0.14 0.17∗∗ SS indicates the sudden stop dummy variable, and (∗∗∗ ,∗∗ ,∗ ) indicate significance at the 1%, 5% and 10% level. The included countries are Argentina, Peru, Bolivia and Chile. 25 The Impact of Sudden Stops on Bank Lending Table 7: Estimation results for the real growth rate of the major balance sheet positions18 Latin America, Method: Panel Least Squares coef. size(-1) cap(-1) liq(-1) foreign size(-1)*SS cap(-1)*SS liq(-1)*SS foreign*SS SS loans -0.08∗∗ 0.88∗∗∗ 0.64∗∗∗ -0.17 0.00 0.51∗∗∗ -0.14 0.09 -0.04 deposits -0.10∗∗∗ 1.04∗∗ -0.08 -0.22 -0.01 -0.20 -0.12 0.15∗∗ -0.06 equity -0.07∗∗∗ -0.20 0.02 -0.31∗ -0.01 -0.30∗∗ -0.25∗ 0.11∗∗∗ 0.04 PL -0.04 0.40 -1.22∗∗ -0.54 -0.28∗∗∗ -0.83 2.96∗∗∗ -0.24 0.44∗∗ int. exp. -0.05∗∗∗ 0.44∗∗ -0.00 -0.26 -0.04∗∗ -0.59∗∗ 0.17 0.25∗∗∗ 0.12 reven. -0.08∗∗∗ 0.32 -0.03 -0.13 -0.04∗∗∗ -0.00 -0.11 0.21∗∗∗ -0.01 ND fund. -0.15∗∗∗ 2.87∗∗∗ -1.01∗∗∗ -0.02 0.01 -0.18 0.24 0.01 -0.12 secur. -0.15∗∗∗ 1.41∗∗∗ -2.14∗∗∗ -0.18 -0.02 0.25 -0.11 0.03 -0.20 R2 obs. 0.11 2094 0.11 2071 0.09 2131 0.17 888 0.18 357 0.15 2106 0.17 852 0.12 1729 East Asia, Method: Panel Least Squares coef. size(-1) cap(-1) liq(-1) foreign size(-1)*SS cap(-1)*SS liq(-1)*SS foreign*SS SS loans -0.07∗∗∗ 0.63∗∗∗ 0.37∗∗∗ 0.07 0.00 -0.38 -0.01 0.17∗∗∗ -0.31∗ deposits -0.05∗∗∗ 0.50∗ -0.22∗ 0.08 0.00 0.80∗∗∗ -0.16 0.20∗∗∗ -0.11∗ equity -0.04 0.37 -0.16 0.06 -0.01 -0.85∗∗ -0.19 -0.05 -0.74∗∗∗ PL -0.12 1.20 0.09 0.25∗∗∗ -27.6∗∗∗ -2.73∗∗ -0.60∗∗ 3.90∗∗∗ int. exp. -0.02 0.49∗∗∗ -0.58∗∗∗ 0.18∗∗∗ 0.01 0.60 1.02∗∗∗ 0.00 1.20∗∗∗ reven. -0.05∗∗ 0.21∗ -0.27∗ 0.04 0.00 0.01 -0.15 0.12∗∗ 0.08 ND fund. -0.07∗∗ 0.50∗ -0.37 -0.06 0.02 1.69∗ 3.42∗∗∗ -0.00 -0.39∗ secur. -0.05 0.10 -2.01∗∗∗ -0.29∗ -0.03 -0.05 -1.90∗∗ -0.02 0.64∗ R2 obs. 0.23 2213 0.22 2216 0.18 2242 0.22 369 0.28 2190 0.25 2137 0.24 1255 0.23 2047 18 Only the coefficients associated with the bank-specific variables are reported in this Table. As controls we use as before GDP growth, real interest rates and the devaluations. We did not yet include the measure for the currency mismatch. SS indicates the sudden stop dummy variable, and (∗∗∗ ,∗∗ ,∗ ) indicate significance at the 1%, 5% and 10% level. PL denotes problem loans, int. exp. interest expenses, reven. revenues, ND funding non-deposit funding (sum of money market and other funding), and secur. denotes securities. 26 Bolivia 1995 ← SS 2000 ← SS ← SS −0.5 −1 2005 ← SS 1995 ← SS ← SS 2000 0 ← SS −20 2005 −40 Peru ← SS 0 ← ←SS SS ← SS 1995 2000 2005 1995 2000 −20 −40 2005 ← SS ← SS 1995 2000 2005 Thailand 10 0 −5 2005 ← SS 0 ← SS ← SS ← SS ← SS −10 ← SS −4 2000 ← SS 5 −2 0 Philippines 2 ← SS ← SS 1995 ← SS← SS Mexico billion $ ← UY,BO 2004 ← BO 1995 ← SS 20 20 billion $ billion $ 0 −20 ← SS −2 −4 2005 40 −10 0 Korea 10 billion $ ↓ AR ↑ UY 2002 ↓ TH,KR,ID,PH,MY,PE 2000 ← AR ↓ PH,CL ← AR,BO 1998 year ← TH 1996 ↓ MX,AR 1994 Indonesia 2000 billion $ −20 ← SS 0 2 billion $ billion $ billion $ ← SS ← SS billion $ 8 7 6 5 4 3 2 1 0 1992 Sudden stops between 1992 and 2005 27 Potential sudden stops between 1992 and 2005 0 −10 Chile 0.5 1995 2000 2005 −20 1995 2000 2005 The Impact of Sudden Stops on Bank Lending Figure 1: Figure 2: number of sudden stops Argentina 10 The Impact of Sudden Stops on Bank Lending −.05 0 KI/GDP .05 .1 .15 million $ 600 800 1000 1200 1400 Capital inflows to LA−6 1995 2000 2005 1990 1995 Average lending volume in A−5 1500 Capital inflows to A−5 million $ 500 1000 KI/GDP 0 .05 0 1995 2000 2005 1990 1995 year 2000 2005 year Capital flows and the average banks’ lending volume Average deposit volume in LA−6 500 −.05 0 million $ 1000 1500 KI/GDP .05 .1 .15 2000 Capital inflows to LA−6 1995 2000 2005 1990 1995 2000 2005 year Capital inflows to A−5 Average deposit volume in A−5 million $ 500 1000 1500 2000 year 0 −.05 KI/GDP 0 .05 .1 1990 1990 1995 2000 2005 year Figure 4: 2005 year −.05 1990 Figure 3: 2000 year .1 1990 Average lending volume in LA−6 1990 1995 2000 2005 year Capital flows and the average banks’ deposit volume 28 All banks av. lending in LA−6 1995 2000 2005 1990 1995 2005 All banks av. lending in A−5 Capitalized banks’ av. lending in A−5 million $ 100 200 300 400 year 0 0 1990 1995 2000 2005 1990 1995 year 2000 2005 year 1000 1995 2000 2005 1990 1995 2000 2005 year All banks av. lending in A−5 Foreign banks’ av. lending in A−5 million $ 0 200 400 600 800 1000 year 0 million $ 500 1000 1500 1990 1990 1995 2000 2005 year Figure 6: Foreign banks’ av. lending in LA−6 million $ 1500 2000 All banks av. lending in LA−6 2500 Average lending volume of well-capitaized banks vs. all banks million $ 600 800 1000 1200 1400 Figure 5: 2000 year million $ 500 1000 1500 1990 Capitalized banks’ av. lending in LA−6 million $ 200 400 600 800 10001200 million $ 600 800 1000 1200 1400 The Impact of Sudden Stops on Bank Lending 1990 1995 2000 2005 year Average lending volume of foreign banks vs. all banks 29