Why Don t Banks Lend? The Mexican Financial System. Stephen Haber Stanford University

Transcription

1 Why Don t Banks Lend? The Mexican Financial System Stephen Haber Stanford University

2 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% Mexico s banking system is miniscule Figure 1: Ratio of Private Bank Credit to GDP, OECD 2005 Mexico Netherlands Switzerland Denmark Iceland United Kingdom Portugal Ireland New Zealand Spain Germany Luxembourg Sweden Austria Australia Japan Korea, Rep. France Italy Norway Belgium Greece Canada Finland United States Hungary Czech Republic Slovak Republic Poland Turkey Source: World Bank Financial Structure Database

3 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Mexico s banking system is small by Latin American standards Figure 2: Ratio of Private Credit from Deposit Money Banks to GDP, Latin America 2005 Panama Chile Belize El Salvador Honduras Uruguay Bolivia Guyana Costa Rica Brazil Nicaragua Ecuador Peru Colombia Guatemala Mexico Paraguay Argentina Venezuela Source: World Bank Financial Structure Ddatabase.

4 Mexico s current ratio of credit to GDP is at its 1950s level Figure 3: Mexico, Ratio of Private Credit by Deposit Money Banks to GDP, % 35% 30% 25% 20% 15% 10% 5% 0% Source: , IMF, International Finance Statistics database; , World Bank Financial Structure Database. 2004

5 Commercial credit has been hit harder than other types 9% Credit by type, as % of GDP 8% 7% IPAB 6% 5% Commercial 4% 3% Consumer 2% 1% 0% Housing SOFOLES Source: Haber and Musacchio 2006.

6 Three hypotheses 1. There are substitutes for the domestic banking system. 2. Banks constrain credit in order to raise rates of return. 3. Banks constrain credit because property rights are hard to enforce. We reject the first hypothesis, and suggest that the second and third are jointly determined by Mexico s pre-2000 political economy.

7 Hypothesis I Other forms of finance substitute for a domestic banking system

8 Mexico s securities markets are not a substitute for a banking system Ratio of Private Bond Market and Stock Market Capitalizations to GDP, % 40% 35% 30% Stock Market 25% 20% 15% 10% 5% Private Bond Market 0% Source: World Bank Financial Structure Database.

9 FDI is also not a substitute for a banking system. FDI Flows to Mexico, Billions U.S. $, Source: Secretaria de Economía (available at

10 Direct foreign lending is not a substitute for domestic banks Percent of Firms that use Bank Credit, (at year end) 80% 70% 60% Small Firms Medium Firms Large Firms AAA Firms 50% 40% 30% 20% 10% Small firms=sales of less than $12.5 million U.S. in 1997; medium=$12.5 to $65 milllion; large=$65 to $650 million; AAA=more than $650 million. Source: Banco de México.

11 % 26% 24% 22% 20% 18% 16% 14% In short, evidence not consistent with the substitution hypothesis Ratio of Gross Fixed Capital Formation to GDP Source: IMF, International Finance Statistics. 1980

12 Hypothesis II Banking market is oligopolized. Largest banks restrain credit in order to raise riskadjusted rates of return.

13 Mexico has an oligopolized market Market Share by Assets, Mexican Retail Banks, Averages 30% 25% 20% 15% 10% 5% BBVA BANCOMER BANAMEX BANORTE HSBC MEXICO (BITAL) SANTANDER SERFIN SANTANDER MEXICANO SCOTIABANK INVERLAT BANCO INBURSA BANCO DEL BAJIO BANCA AZTECA IXE BANCA AFIRME INTERACCIONES BANCO INVEX GE CAPITAL BANK BANREGIO BANCA MIFEL BANCEN BANSI 0% Source: CNBV.

14 Methods There is a difference between showing that a market has a structure consistent with oligopoly, and demonstrating that banks practice oligopoly pricing. Our methods are likely to underestimate the impact of oligopoly.

15 Largest banks do extend less credit Credit to Consumers and Private Firms, as Percent of Assets, Quarterly Averages % 80% 70% 60% 50% 40% 30% 20% 10% 0% BBVA BANCOMER BANAMEX BANORTE HSBC MEXICO (BITAL) SANTANDER SERFIN SANTANDER MEXICANO SCOTIABANK INVERLAT BANCO INBURSA BANCO DEL BAJIO BANCA AZTECA IXE BANCA AFIRME INTERACCIONES BANCO INVEX GE CAPITAL BANK BANREGIO BANCA MIFEL BANCEN BANSI Assets are net of IPAB bonds. Source: CNBV data, linked together in a panel in Haber and Musacchio (2006).

16 Larger banks make fewer loans, and assume less risk Lending Strategies of M exican Banks, OLS Regression Results Private Credit Non-Perform ing Loans a s % of A ssets 1 as % of Total Loans M arket Share (by Assets) ( )*** (4.69 )*** Equity Ratio (2.2 2)** (7.12 )*** C o m m e rc ia l C re d it/a s s e ts (6.54 )*** SO FO L Credit/Assets (3.20 )*** Consum er Credit/Assets '(0.66 ) Housing Credit/Assets (6.33 )*** Quarter Dum m ies Yes Yes R obust S td E rrors C lustered b Yes Yes Constant (65.57)*** (1.63) Observations R -sq ua re d Robust t statistics in parentheses * significant at 10% ; ** significant at 5% ; *** significant at 1% 1. Private Lending does not include IPAB Bonds. For that reason, we also net out IPAB bonds from Assets.

17 But larger banks do not appear to charge higher net interest margins Net Interest Margins, Mexican Retail Banks, Quarterly Averages % 11% 8% 5% 2% BBVA BANCOMER BANAMEX BANORTE HSBC MEXICO (BITAL) SANTANDER SERFIN SANTANDER MEXICANO SCOTIABANK INVERLAT BANCO INBURSA BANCO DEL BAJIO BANCA AZTECA IXE BANCA AFIRME INTERACCIONES BANCO INVEX GE CAPITAL BANK BANREGIO BANCA MIFEL BANCEN BANSI -1% Source: CNBV data, linked together in a panel in Haber and Musacchio (2006).

18 Larger banks charge lower risk adjusted NIMs, and do so because they have lower administrative costs Net Net In te re s t In te re s t M a rg in s M a rg in s M arket Share (by Assets) (2.55)** (1.57) Adm inistrative C osts/assets (11.13)*** C o m m e rc ia l C re d it/a s s e ts (3.77)*** (7.67)*** SO FO L Credit/Assets (1.7 1 ) (3.2 3 )*** Consum er Credit/Assets (7.52)*** (0.35) Housing Credit/Assets (6.58)*** (5.15)*** Non-Perform ing Loan Ratio (3.04)*** (2.12)* Q uarter Dum m ies Yes Yes Robust Std Errors Clustered by Q uarter Yes Yes Constant (6.17)*** (0.48) O bservations R-squared Robust t statistics in parentheses * significant at 10% ; ** significant at 5% ; *** significant at 1%

19 Larger banks have higher ROA, and do so because they make fewer commercial loans and earn more from fees Commissions Return on Equity Return on Return on and Fees (Net of Fees and Equity Equity as % of Equity Commissions) Market Share (by Assets) (2.76)** (0.98) (22.57)*** (1.73) Administrative Costs/Assets (0.15) (0.21) (5.10)*** (0.82) Commercial Credit/Assets (3.94)*** SO FO L C redit/assets (1.12) Consumer Credit/Assets (0.8) Housing Credit/Assets (0.22) Non-Performing Loan Ratio (0.45) (2.76)** (1.38) Quarter Dummies Yes Yes Yes Yes Robust Std Errors Clustered by Qtr Yes Yes Yes Yes Constant (1.07) (2.10)* (2.46)** (0.25) Observations R-squared Robust t statistics in parentheses * significant at 10% ; ** significant at 5% ; *** significant at 1%

20 BANSI 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Large banks earn more income from commissions and fees Commissions and Fees as Percentage of Income, Mexican Retail Banks, Quarterly Averages BANAMEX BANORTE HSBC MEXICO (BITAL) BANCO SANTANDER SERFIN SANTANDER MEXICANO SCOTIABANK INVERLAT BANCO INBURSA BANCO DEL BAJIO BANCA AZTECA IXE BANCA AFIRME BANCO INTERACCIONES BANCO INVEX GE CAPITAL BANK BANREGIO BANCA MIFEL BANCEN Source: CNBV data, linked together in a panel in Haber and Musacchio (2006). BBVA BANCOMER

21 BANSI 70% 60% 50% 40% 30% 20% 10% 0% Larger banks make fewer commercial loans Commercial Loans as Percent of Assets, Mexican Retail Banks, Quarterly Averages BANAMEX BANORTE HSBC MEXICO (BITAL) SANTANDER SERFIN SANTANDER MEXICANO SCOTIABANK INVERLAT BANCO INBURSA BANCO DEL BAJIO BANCA AZTECA IXE BANCA AFIRME INTERACCIONES BANCO INVEX GE CAPITAL BANK BANREGIO BANCA MIFEL BANCEN Assets are net of IPAB bonds. Source: CNBV data, linked together in a panel in Haber and Musacchio (2006). BBVA BANCOMER

22 Large banks make fewer commercial loans Lending Strategies of Mexican Banks, OLS Regression Results Commercial Credit Consumer Credit Housing Credit as %of Assets 1 as %of Assets 1 as %of Assets 1 Market Share (by Assets) (32.66)*** '(0.73) (18.34)*** Equity Ratio (0.69) (3.34)*** (3.12)*** Quarter Dummies Yes Yes Yes Robust Std Errors Clustered by Qtr Yes Yes Yes Constant (74.54)*** (6.92)*** (10.34)*** Observations R-squared Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% 1. Private Lending does not include IPAB Bonds. For that reason, we also net out IPAB bonds from Assets.

23 Hypothesis III: Property rights as a barrier to lending Higher ratios of commercial credit are associated with lower net interest margins and lower rates of return on equity. The trend over time has been a reduction in commercial lending.

24 Commercial credit has been hit harder than other types 9% Credit by type, as % of GDP 8% 7% IPAB 6% 5% Commercial 4% 3% Consumer 2% 1% 0% Housing SOFOLES Source: Haber and Musacchio 2006.

25 What makes commercial lending different? Answer: with commercial loan contracts it is harder to assess credit risk ex ante and enforce contracts ex post.

26 A brief history of property rights 1. Historically, courts, police, and property registers poorly developed. 2. Reforms to credit reporting (1995), mandatory use of credit reporting (1997), and bankruptcy laws (2002) have made consumer and housing lending more attractive. These were extremely clever reforms. 3. They have had less of an impact on the enforcement of commercial loan contracts, where assets more difficult to assign, where credit reporting harder, and where court system is crucial.

27 The challenge of reforming property rights: going beyond bankruptcy reform 1. Property registries. 2. The police 3. The courts. 4. All of these institutions are costly to reform, but the Mexican government has limited fiscal resources.

28 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% The limited resources of the Mexican Government Government Revenues as Percent of GDP, 2004 Guatemala El Salvador Sweden Costa Rica Colombia Mexico Argentina Nicaragua Peru Paraguay Honduras Spain Venezuela Canada Chile Panama Greece Bolvia Australia Brazil Uruguay Italy Poland Germany UK Portugal Denmark France Source: IMF (2005).

29 Options 1. Continue property rights reforms. A long and costly process. 2. Open banking system to more competition. This will increase access to credit, and also create interests who will push for property rights reform. Announcements of recent days suggests that this is underway.