IV. Bank-Lending Behavior in Uruguay
By Gaston Gelos and Marco Piñón
1. Five years after the crisis, the first signs of a recovery in bank credit are visible (Figure 1). In particular, lending in local currency has picked up recently. Still, lending levels remain distant from pre-crisis levels. In line with the post-crisis experience of other countries, output has so far been able to recover essentially without bank credit. However, the economy is entering in a phase where formal financial intermediation is likely to be needed to sustain growth into the medium term (Calvo, Izquierdo, and Talvi, 2005). It is therefore important to understand bank credit behavior both from a cyclical and medium-term growth perspective. What role will the new Uruguayan financial system play in the transmission of economic, including monetary, shocks? This question is particularly important given the current reassessment of the monetary policy framework in Uruguay. From a growth perspective, one relevant question is whether there are structural factors, including adequate competition, restricting financial deepening.
Figure 1:Bank Lending in Uruguay
2. This paper makes use of microeconomic bank data to draw inferences on bank behavior. In addition to the usual identification problems, understanding bank behavior from a time series perspective is challenging given severe structural changes since the 2002 crisis, including the failure of various banks, the creation of a new institution, and the subsequent substantial tightening in regulation and supervision. However, using detailed monthly balance sheet data, this paper exploits its cross-sectional as well as the time-series dimension. First, the paper tests for the existence of a bank lending channel of monetary transmission for domestic currency loans. Second, it assesses whether similar effects exist for the transmission of international shocks to foreign currency loans. Third, we examine the degree of competition in the Uruguayan banking system and its evolution since the crisis.
3. The paper finds evidence of the existence of a bank lending channel in local currency loans. In response to a monetary contraction (expansion), banks tend to reduce (expand) their loan supply; in line with the hypothesis on the role of financial frictions, this effect is stronger for less liquid and less capitalized banks. This transmission mechanism operates above and beyond other channels, such as the interest rate or exchange-rate channel. Its quantitative significance is still somewhat limited given the preponderance of foreign-currency lending, but its importance is growing.
4. Similarly, there is evidence of an equivalent transmission of foreign shocks. More capitalized and liquid banks reduce their lending in foreign currency less strongly in response to U.S. interest hikes. There is no statistically significant difference in the reaction of banks with different degrees of liability dollarization. Together with the findings on the bank lending channel, this implies that, as bank lending picks up and bank liquidity falls, the banking system will tend to amplify domestic and foreign shocks more in the future.
5. The banking system is characterized by relatively low competition intensity. Some indicators suggest that competition is low relative to other countries and that it has, if anything, been decreasing. This raises the challenge of strengthening competition intensity without endangering financial stability.
B. Bank Credit in Uruguay—Background
The importance of bank lending for the Uruguayan economy
6. Until before the crisis, credit levels relative to GDP in Uruguay were broadly in line with Latin American averages, but low by international standards. In the 1990's, bank credit to the private sector fluctuated around 30 percent, comparable to the levels in Argentina, Brazil, and Colombia—but below OECD levels of around 70 percent (Beck and Levine, 2004) and other emerging markets. Currently, the ratio of credit to GDP in Uruguay stands at 25 percent (Figure 2).
Figure 2.Bank Credit to the Private Sector in Selected Countries
7. Cross country empirical evidence suggest that growth prospects in Uruguay would be enhanced through deeper financial intermediation. A causal link from financial intermediation to growth has been empirically established in the literature. The evidence stems largely from cross-country regressions using proper instrumental variable procedures to avoid endogeneity problems (e.g., Levine, 2004). The results indicate that financial development affects growth by increasing productivity, rather than capital accumulation (see Beck, Levine, and Loayza, 2000, and WEO (2004)). Evidence from panel regressions presented in Beck, Levine, and Loayza (2000) suggests that if bank credit in Uruguay could be exogenously raised from 30 to 35 percent, average yearly GDP growth could be 0.4 percent higher. Nevertheless, these conceptual experiments are only illustrative, with the more relevant question being how to increase financial development (Levine, 2003).
8. Banks have traditionally been an important source of finance for Uruguayan companies. The importance of stock and corporate bond markets is minimal (Table 1). In 2007, 10 companies were listed on the stock market, of which two were not traded; 23 companies were listed on the corporate bond market. In 2005, only 4 companies issued bonds. Retained earnings remains the most important source of financing for Uruguayan firms. External financing is divided up in broadly similar shares among bank loans and suppliers' credit.
|Bonds||Bank Loans||Suppliers' Credit||Retained earnings|
9. Evidence suggests that the cyclical component of credit in Uruguay has been strongly correlated with economic activity. As expected, the correlation is stronger for those sectors that have less access to other sources of finance, such as households and construction (not shown). This illustrative evidence supports the notion that real activity and bank lending are linked in Uruguay; however, it does not allow to draw inferences on causality or bank lending behavior because it is unable to disentangle supply and demand effects. Moreover, the significant structural break linked to the 2002 banking crisis, with its associated bank closures and subsequent changes in supervision, hampers the possibility of extrapolating past bank behavior to the future (Figure 3).
Figure 3.Correlations between Leads and Lags of the Cyclical Components of Credit and GDP
Sample: 1988Q1 2002Q1
Included observations: 57
Correlations are asymptotically consistent approximations.
C. The Behavior of Bank Lending: Microeconomic Evidence
10. This paper uses detailed data available on Uruguayan banks, published by the central bank of Uruguay (BCU). The data contain monthly flow and stock variables; the paper focuses on the period 2003–06. Summary statistics are presented in Table 2. The Uruguayan banking system is characterized by a segmentation between a state bank, BROU (which represents close to 50 percent of total deposits) and a group of foreign-owned banks. The mortgage bank BHU has been banned from taking deposits and conducting lending since the crisis, and, therefore is not included in the sample.
|Liquid assets/total assets||5.3||9.5|
|Number of banks||14||1|
Is there a bank-lending channel?
11. The bank-lending channel assigns a specific role to banks in the transmission mechanisms of monetary policy. The basic hypothesis is that when there is a monetary contraction and banks loose deposits, financial frictions prevent banks from raising funds in an alternative manner (e.g., by issuing CDs). As a result, they reduce their lending. If firms, in turn, cannot easily substitute bank credit, this will lead to real effects.1
12. The paper exploits heterogeneity in bank characteristics to test for the existence of a lending channel. Since credit aggregates are determined jointly by supply and demand, examining correlations between credit and other economic variables does not identify a lending channel. Therefore, the strategy adopted in this literature is to focus on cross-sectional differences in bank behavior: banks that can a priori be expected to be less credit constrained should react less strongly to monetary policy shocks (see, for example, Kashyap and Stein, 2000). This paper differentiates banks according to their degree of liquidity and capitalization. It is presumed that firms cannot easily substitute bank loans with other sources of external finance, in light of the evidence presented earlier.
13. Loan growth regressions are estimated using quarterly data with short-term interest rates and changes in the announced growth rate of M0/M1 as monetary policy variables. To obtain a measure of truly exogenous monetary policy shocks, the paper first regresses the change in each of these domestic monetary variables on changes in U.S. interest rates and Latin American EMBI spreads and then use the predicted values of these regressions. Dynamic regressions are estimated using manufacturing growth, exchange rate change, the change in country risk, and the change in inflation expectations as control variables. Time dummies are also used as alternatives. The estimations are carried out with bank dummies, allowing for bank-specific heterogeneity, for correlation of contemporaneous errors, and correcting for first-order serial correlation.2
14. The results support the existence of a bank-lending channel for local currency loans. In line with the hypothesis, it is found that more liquid and more capitalized banks do react less strongly to monetary policy shocks (Table 3). For example, a 1 percent increase in the short-term interest rate yields an average drop in the quarterly growth rate in local currency credit of about 2 percent (compared to a median growth rate of 5 percent in the sample). A bank at the top 80 percentile liquidity ratio, however, would significantly change its lending. As expected, the bank-lending channel effects cannot be confirmed statistically for overall loans, given the large share of dollar loans. With the share of peso credit growing, the bank-lending channel will be gaining quantitative importance in Uruguay.
|Differential Impact by Bank Characteristic|
|Monetary Policy Variable||Liquidity||Capitalization|
|Δi x Bank characteristic||0.0015***||1.094***|
|ΔM x Bank characteristic||−0.003***||−1.604***|
15. A growing importance of the bank-lending channel brings opportunities and challenges. While the effectiveness of monetary policy is enhanced through this additional mechanism, the amplification of monetary shocks implies the need for a more fine-tuned policy. Moreover, as banks continue to lend more, their current high liquidity levels are likely to fall, further amplifying their responses.
The transmission of foreign shocks
16. Does a similar mechanism operate in the transmission of foreign shocks? To answer this question, the paper explores a similar specification with a focus on the transmission of foreign shocks. In particular, it examines the effects of changes of U.S. interest rates on foreign currency lending behavior, by type of bank.
17. More liquid banks and with fewer dollar liabilities reduce their lending less strongly in response to U.S. interest hikes. An increase in U.S. interest rates is associated with a drop in lending—a 100 basis points increase yields a drop in the quarterly growth rate of 9 percentage points (Table 4). This effect is milder for more liquid and better capitalized banks. There is no statistically significant difference in the reaction of banks with different degrees of capitalization.
|Differential Impact by Bank Characteristic|
|Foreign Shock||Dollarization of Liabilities||Liquidity Capitalization|
|Δi US x Bank Char||−0.34***||0.002***||−0.154|
D. What is the Degree of Competition Intensity in the Uruguayan Banking System?
18. To assess prospects for increased financial intermediation the paper examines competitive conditions in the banking sector. International evidence shows that higher banking competition is associated with lower spreads and easier access to finance by companies (Gelos, 2006 and Beck, Demirguc-Kunt, and Maksimovic, 2004). The degree of competition may also matter from a cyclical perspective, as it has been argued that imperfect competition in the banking sector may propagate external shocks and amplify swings (Mandelman, 2006).3
19. Concentration in the Uruguayan banking system has risen. The number of banks has declined during and after the crisis, and the market concentration has increased, as measured by Herfindahl-Hirschman (HH) indices (Figure 4). The index is now close to 1400. As a comparison, Gelos and Roldos (2002) report HH indices for 13 emerging markets; in their sample, the median HH index in 2000 was 900, and the median in Latin America 923.
Figure 4:Concentration Index
20. The paper uses a market structure test based on reduced form revenue functions suggested by Panzar and Rosse (1987). Panzar and Rosse show that the sum of the elasticities of a firm's revenue with respect to the firm's input prices (the so-called H statistic) can be used to identify the nature of the market structure in which the firm operates. In long-run competitive equilibrium, the H statistic should be equal to one, as increases in input prices should lead to a one-to-one increase in total revenues. By contrast, the H will be negative if the firm operates as a monopoly—an upward shift in the marginal cost curve will be associated with a reduction in revenue as a result of the optimality condition for the monopolist (Table 5). Under monopolistic competition, the H statistic will lie between zero and one. If the elasticity of demand is constant, then there is a monotone relationship between the mark-up over marginal costs—a measure of the degree of competition—and the H index. More formally, letting R denote a revenue function of input prices w and exogenous variables z:
|H<0||Monopoly [or conjectural variation oligopoly]|
|H = 1||Perfect competition or monopoly in a perfectly|
21. The paper treats banks as single product firms (De Bandt and Davis, 2000); banks produce intermediation services using labor and capital as inputs.4,5 The Panzar and Rosse (1987) approach has been applied widely to banking systems. Early studies examine competitive conditions in the U.S. and Canada (see Shaffer (1989) and Nathan and Neave (1989)), respectively.6
22. To derive the H statistic, the paper estimates the following reduced form revenue equation:
|IR||=||interest revenue (or interest revenue divided by total assets)|
|wL||=||unit price of labor|
|WF||=||unit price of funds|
|wK||=||unit price of capital|
|cap||=||capacity indicators, such as total fixed assets|
|oth||=||other factors potentially affecting interest revenues, such as the business mix of the bank, and the size of total assets (to control for scale effects), and nonperforming loans|
23. Equation (2) is estimated using semiannual data for 2003–06. The unit cost of labor is proxied by salary expenses over personnel, the unit cost of funds by interest payments over deposits, and the unit cost of capital by other expenses divided by total fixed assets. The paper also tests for changes in the H statistic between 2003–04 and 2005–06 (Table 6).
|Ln (cost of funds)||0.252||0.233||0.250||0.232|
|Lnw (cost of capital)||0.260||0.230||0.259||0.224|
|Ln (wages) x after||−0.038||−0.038||−0.039||−0.039|
|Ln (cost of capital) x after||−0.193||−0.186||−0.193||−0.188|
|Ln (cost of funds) x after||−0.023||−0.021||−0.023||−0.022|
|Ln (total assets)||0.977||1.065||0.972||1.062|
|No. of Obs.||88||88||84||84|
|Change in H||−0.25*||−0.25*||−0.25||−0.25*|
24. The estimates suggest that the banking system is characterized by monopolistic competition. With the H statistic around 0.5, the estimation indicates that the degree of competition intensity is low by international standards. In a study comprising 50 countries, Claessens and Laeven (2004) report H statistics around 0.6–0.8, with a median value close to 0.7. Although cross country comparisons of these statistics are problematic, the finding of low competition confirms Mello's (2006) results, who uses a different approach and does not look at changes over time. A caveat is that the standard errors are large, so that the measurement is not very precise.
25. Along with the rise in concentration, competition has tended to decrease since the crisis. The estimates show a drop in H since 2005, albeit only at low significance levels. Nevertheless, the more recent estimates may be more valid since the banking system is likely to be operating under conditions more closely resembling those of a long-run equilibrium.
26. The results suggest that increasing competition in the banking system without jeopardizing financial stability is an important challenge for Uruguay. While a discussion of concrete policy measures in this area would exceed the scope of the paper, leveling the playing field between private and public banks would likely contribute to this objective (see Mello, 2006, and IMF, 2006).
27. This paper assessed the role of and prospects for bank-lending in Uruguay from a cyclical and structural perspectives. Specifically, the paper tested the existence of bank-lending-channel transmission mechanisms in response to domestic and external shocks for local and foreign currency credit. It also looked at the competitive structure of the domestic banking system.
28. The results indicate that bank-lending in local currency reacts to changes in domestic policies. This suggests that monetary policy will become more effective over time, but also more challenging. With a growing share of peso-denominated debt, the bank-lending channel will be becoming more important. This means that monetary policy shocks will be amplified, which in turn implies that its fine-tuning will become more relevant.
29. The results also show that domestic credit in foreign currency is partly driven by foreign shocks. Moreover, the less liquid banks are, the more foreign and domestic shocks are amplified. With the ongoing resumption in lending, banks' liquidity will naturally decline from their current high levels. This will further amplify the effect of shocks in the future, implying the need for more vigilance in macroeconomic policies.
30. From a structural perspective, the study indicates that measures to promote a more competitive environment are likely to enhance sustained growth over the medium term. In particular, it finds that the degree of competition in the Uruguayan banking system is relatively low by international standards. Looking ahead, this may limit financial deepening and, thereby, economic growth. Thus, a challenge for the authorities is to further foster healthy competition while preserving the soundness of the financial system.
Arena,Marco,CarmenReinhart, and FranciscoVázquez,2006, “The Lending Channel in Emerging Economies: Are Foreign Banks Different?” NBER Working Paper No. 12340(Cambridge, Massachusetts: National Bureau of Research).
Arellano,Manuel, and StephenBond,1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations,” Review of Economic Studies, Vol. 58(2), pp. 277–97, April.
Arellano,Manuel, and OlympiaBover,1995, “Another look at the instrumental variable estimation of error-components models,” Journal of Econometrics,Elsevier, Vol. 68(1), pp. 29–51, July.
Beck,Thorsten, and RossLevine,2004, “Stock markets, banks, and growth: panel evidence,” Journal of Banking and Finance, Vol. 28(2004), pp.423–42.
Beck,Demirguc-Kunt,Asli, and VojislavMaksimovic,2004, “Bank Competition and Access to Finance: International Evidence,” Journal of Money, Credit and Banking, Vol. 36(3), pp. 627–48, June.
Claessens,Stijn, and LucLaeven,2004, “What Drives Bank Competition? Some International Evidence,” Journal of Money, Credit, and Banking, Vol. 36, No. 3(June)Part 2.
Chiesa,Pablo,PaulaGara, and María JoséZerbino,2004, “Efectos Reales de la Política Monetaria en Uruguay: Una aproximación al estudio de los canales de la tasa de interés y del crédito bancario” (Master's thesis; Montevideo: Universidad de la República, Uruguay).
De Bandt, O. andE.P.Davis,2000, “Competition, Contestability and Market Structure in European Banking Sectors on the Eve of EMU,” Journal of Banking and Finance, Vol. 24, No. 6, pp. 1045–66(22).
De Brun,Julio,NéstorGandelman, and EduardoBarbieri,2003, “Investment and Financial Restrictions at the Firm Level in Uruguay,” in Credit Constraints and Investment in Latin America,ed. byArturoGalindo and FabioSchiantarelli(Washington: Inter-American Development Bank).
De Brun,Néstor Gandelman,Herman Kamil, and ArturoPorzecanski,2006, “The Fixed Income Market in Uruguay,” (unpublished;Montevideo: Universidad ORT Uruguay).
Gelos,Gaston,2006, “Banking Spreads in Latin America,” IMF Working Paper 06/44(Washington: International Monetary Fund).
Gelos, and JorgeRoldós,2004, “Consolidation and Market Structure in Emerging Market Banking Systems,” Emerging Markets Review, Vol. 5, pp. 39–59.
Gourinchas,Pierre-Olivier,OscarLanderretche, and RodrigoValdés,2001, “Lending Booms: Latin America and the World,” Economía, Vol. 1(2001), pp.47–99.
International Monetary Fund, 2006, “Financial Sector Stability Assessment,” IMF Staff Country Report No. 06/439(Washington: International Monetary Fund).
Kashyap,AnilK., and JeremyC.Stein,2000, “What Do a Million Observations on Banks Say about the Transmission of Monetary Policy?” American Economic Review, Vol. 90(3), June2000, pp. 407–28.
Levine,Ross,2003, “More on Finance and Growth: More Finance, More Growth?Review, The Federal Reserve Bank of St. Louis(July/August), pp. 31–46.
Mandelman,FedericoS.,2006, “Business Cycles and Monetary Regimes in Emerging Economies: A Role for a Monopolistic Banking Sector,” Working Paper 2006–17,Federal Reserve Bank of Atlanta.
Martinez Peria, M.S.,A.Powell, and I.Vladkova-Hollar,2005, “Banking on Foreigners: The Behavior of International Bank Claims on Latin America, 1985–2000,” Staff Papers,International Monetary Fund, Vol. 52, No. 3.
Mello Costa,Miguel,2006, “Midiendo La Concentración y el Poder de Mercado en el Sector Bancario Uruguayo: 2003–05,” (unpublished;Montevideo: Central Bank of Uruguay).
Micco,Alejandro, and UgoPanizza,2006, “Bank Ownership and Lending Behavior,” Central Bank of Chile Working Paper No. 369(Santiago, Chile: Central Bank of Chile).
Mora,Nada,2003, “The Bank-lending Channel in a Partially Dollarized Economy” (unpublished;Beirut, Lebanon: The American University of Beirut)
Munyo,Ignacio,2006, “The Determinants of Capital Structure: Evidence from an Economy without Stock Market” (unpublished;CERES).
Panzar, J.C., and J.N.Rosse,1987, “Testing for ‘Monopoly' Equilibrium,” Journal of Industrial Economics, Vol. 35(4), pp. 443–56.
Schaeck,Klaus,MartinCihak, and SimnonWolfe,2006, “Are More Competitive Banking Systems More Stable?” IMF Working Paper 06/143(Washington: International Monetary Fund).
Takeda,Tony,FabianaRocha, and MarcioI.Nakane,2005, “The Reaction of Bank-Lending to Monetary Policy in Brazil,” Revista Brasileira de Economia, Vol. 59 No. 1.
Tornell,Aaron, and FrankWestermann,2003, “The Credit Channel in Middle Income Countries,” CESifo Working Paper No. 832.
Van den Heuvel,Skander,2001, “The Bank Capital Channel of Monetary Policy” (mimeo,Wharton School,University of Pennsylvania).
See Bernanke and Blinder (1998) and Bernanke and Gertler (1992). The bank lending channel hypothesis is a specific mechanism within the broader “credit channel hypothesis.” The latter states that due to financial frictions, monetary policy affects the external finance premium that firms and banks have to pay to borrow.
Given the presence of a lagged dependent variable, GMM estimation could have followed Arellano and Bond (1991) or Arellano and Bover (1995). We did not pursue this here because the cross-sectional dimension of the panel is small; moreover, we were not interested in the coefficient on the dependent variable.
Regarding the relationship between competition and financial stability, the evidence is not conclusive (see for example the discussion in Drummond, Maechler, and Marcelino, 2007).
However, product differentiation is allowed for in the monopolistic competition model.
See Freixas and Rochet (1997).
For studies of European countries, see Molyneux, Lloyd-Williams, and Thornton (1994), Bikker and Groeneveld, (2000), and De Bandt and Davies, (2000). For a study covering various emerging markets see Gelos and Roldós (2003). Individual country studies include, among others, Austria (Mooslechner and Schnitzer, 1995), Brazil (Belaisch, 2003), Bulgaria (Feyzioğlu and Gelos, 2000), Colombia (Barajas, Steiner, and Salazar), Italy (Coccorese, 1998), Switzerland (Rime, 1999), Germany (Lang, 1997, and Hempell, 2002), Japan (Molyneux, Lloyd-Williams, and Thornton, 1996), and Finland (Vesala, 1995).