Abstract
This Selected Issues paper considers the “informal sector” in Mexico. In Mexico and many other countries, the informal sector represents a large share of total employment. The paper reviews the literature on informality, with special focus on findings for Mexico, and develops a theoretical model that highlights the importance of externalities and the distortion associated with the informal sector. The analysis provides insight into the kinds of policy measures that might sustainably reduce the size of the informal sector. The paper analyzes financial saving in Mexico.
IV. Mexican Banks: Lending and Profitability In The Context of Reforms, 1998–20041
Abstract
This chapter characterizes bank lending and profitability in Mexico since 1998, a period of extensive reforms and structural change, including in bank balance sheets. This is done in two parts. First, we review bank reforms and other factors which may have influenced bank behavior in recent years. In particular, the nature and sequencing of reforms are described. Second, we estimate bank-level regressions which link bank lending to the private sector and profitability with balance sheet indicators for commercial banks. These regressions confirm the importance of sound balance sheets as a basis for bank lending growth, while refuting the hypothesis that bank credit to the public sector, mostly FOBAPROA notes extended to banks in exchange for their non-performing assets after the 1994–95 crisis, has crowded out private sector lending in recent years. However, the estimated relationships cannot explain the fast rebound of commercial bank lending to the private sector seen since end-2003—at an annualized rate of 26percent in real terms. Other determinants not captured in the regressions, most likely resulting from both structural reforms and demand forces, are at play. The chapter briefly reviews the experiences of Chile and Korea, to illustrate the range of factors that could drive a recovery of credit, as well as the potential benefits and risks involved.
A. Background and Motivation
1. The banking sector in Mexico has experienced tremendous changes in the past 15 years. In the early 1990s, the sector was quickly privatized and liberalized, and financial intermediation grew at a fast pace. The 1994 financial crisis caused massive losses for banks and the government put in place a rescue package over several years, with a cumulative cost in the order of US$ 64 billion, or 13 percent of 1999’s GDP (Gil Diaz (2000)). At the same time as banks were rescued through recapitalization and debtor and depositor support programs, extensive reforms began. Since 1996, reforms have been steadily implemented, improving most aspects of how banks operate and compete in Mexico.
2. Notwithstanding these reforms, the banking sector in Mexico has not yet regained a large financial intermediation role. The 1994 financial crisis abruptly reversed the fast expansion of bank credit that had followed the first liberalization of the sector. What followed was a protracted contraction of banking domestic assets and private sector bank financing—defined as including loans, security investments, and other forms of financing like changes in nonperforming private sector assets. This contraction was experienced across financial intermediaries (Figures 1 and 2), leading the non-bank private sector to turn to alternative financing sources such as supplier credits and foreign financing. The diminished role of Mexico’s banking sector is confirmed by looking at other country cases (Figure 3). Even as of March 2005, total commercial bank lending in Mexico as a share of GDP was less than half the Latin American average.
Mexico: Consolidated Financial Sector Assets, 1995—2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; covers the entire financial sector.*: Calculated as financial survey assets net of banking sector assetsMexico: Consolidated Financial Sector Assets, 1995—2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; covers the entire financial sector.*: Calculated as financial survey assets net of banking sector assetsMexico: Consolidated Financial Sector Assets, 1995—2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; covers the entire financial sector.*: Calculated as financial survey assets net of banking sector assets
Mexico: Nonfinancial Private Sector Financing from Consolidated Financial Sector, 1999–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; includes all forms of financing (e.g., lending, current and overdue, security holdings) provided by the entire financial sector.Mexico: Nonfinancial Private Sector Financing from Consolidated Financial Sector, 1999–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; includes all forms of financing (e.g., lending, current and overdue, security holdings) provided by the entire financial sector.Mexico: Nonfinancial Private Sector Financing from Consolidated Financial Sector, 1999–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology; includes all forms of financing (e.g., lending, current and overdue, security holdings) provided by the entire financial sector.
Commercial Banking Loans, as of March 2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
*: Ratio for 1999–2002. Source: IMF IDB and CNBV.Commercial Banking Loans, as of March 2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
*: Ratio for 1999–2002. Source: IMF IDB and CNBV.Commercial Banking Loans, as of March 2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
*: Ratio for 1999–2002. Source: IMF IDB and CNBV.3. That said, Mexico’s financial sector—including banks and non-banks—has shown signs of increasing vibrancy recently. While Figure 2 shows that private sector bank financing slightly declined in 2003–2004, measured as a share of GDP, this figure is affected by the fast write-off of bad loans. Looking more closely at commercial bank lending to the nonfinancial private sector, there is an important turning point at the end of 2003 (Figure 4). From then through mid-2005, commercial bank private lending increased at an annualized rate of about 26 percent, compared to an annual growth rate of 8½ percent in 2000–2003 (both in real terms). In addition, financing by other financial intermediaries, mainly specialized non-bank intermediaries—known as Sofoles—and pension funds (through the purchase of commercial paper), expanded at a fast pace.
Nonfinancial Private Sector Financing from Consolidated Financial Sector, Selected Components, 1998–2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology.Nonfinancial Private Sector Financing from Consolidated Financial Sector, Selected Components, 1998–2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology.Nonfinancial Private Sector Financing from Consolidated Financial Sector, Selected Components, 1998–2005
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Source: Staff calculations based on IFS methodology.4. The objective of this chapter is to analyze the behavior and performance of Mexican commercial banks in recent years, in light of the reforms that have taken place and other developments that may influence bank behavior. First, the chapter documents recent banking reforms and other factors likely to influence bank behavior, such as the recapitalization of banks after the 1994 crisis, and lays out preliminary hypotheses regarding the effects of these factors on banks. The overall direction and timing of these effects are in principle uncertain, motivating an empirical investigation. Second, the chapter seeks to characterize bank behavior over the past 7 years (1998–2004), documenting the evolution of the sector’s structure, consolidated balance sheets, lending portfolio, and overall profitability. The chapter then reports on bank-level panel regressions trying to relate bank lending and profitability to balance sheet indicators and other variables.
5. The regression analysis confirms the positive link between bank lending and basic balance sheet indicators of capital adequacy and credit quality, and finds a diminishing influence of bank size and capital adequacy on bank profitability. The regression analysis does not find evidence of a crowding out of private lending by public sector borrowing during the period studied. Statistical tests are also conducted finding evidence of significant changes of the estimated relationship during the period, including the significance of capital adequacy and interest rate level in explaining the growth of lending to the private sector.
6. However, balance sheet characteristics are found to explain only a small part of bank behavior including the expansion of bank lending since 2003. The relationships between balance sheets and bank lending estimated in the chapter cannot fully explain the very fast expansion of bank lending since 2003. This motivates the discussion of two country examples that suggest various lessons. The case of Chile shows that bank credit can indeed expand steadily, without an abrupt reversal, following comprehensive banking reforms that were broadly similar to those implemented in Mexico. On the other hand, the case of Korea suggests bank credit may remain particularly volatile, even in the context of reforms. These two country examples are briefly described with a view to highlighting the value of ongoing scrutiny of the determinants of the present credit growth in Mexico.
7. The rest of the chapter is organized as follows. Part B reviews bank reforms in Mexico and discusses the factors generally thought to influence bank behavior. Part C presents stylized facts on the consolidated balance sheet and performance of the commercial banking sector, and the results of panel regressions seeking to explain bank behavior. Part D looks for lessons in the evolution of bank lending in the context of reforms in Chile and Korea, and Part E concludes.
B. Bank Reforms in Mexico and Key Influences on Bank Behavior
8. Factors likely to have influenced bank behavior in Mexico over the last decade are banking sector reforms, placement of government debt with banks, and the improvement of the macroeconomic environment. Bank reforms since 1995 have touched most aspects of banking operations. Reforms have strengthened bank supervision, promoted competition from other intermediaries, and reduced credit-related information asymmetries. The recapitalization of banks with government notes (FOBAPROA notes) following the 1994–95 crisis had a major impact on the structure of banks. The stabilization of macro-financial conditions led to a sharp decline in inflation expectations, risk premia, and real interest rates with a significant impact on banks’ financial results. This section describes these factors and presents hypotheses regarding their effects on banks, including through a selective review of the literature.
Bank reforms I: stronger oversight and risk management
9. The collapse of bank credit in the mid-1990s has been seen in retrospect not only as a product of the 1994–95 macro-financial crisis but also as the bursting of a credit bubble. The fast credit expansion during 1988–942 coincided with a string of financial liberalization reforms, which consisted of financial deregulation, liberalization of the external capital account, and the privatization of most government-owned commercial banks.
Gil Diaz (2000) explains that a vast problem of nonperforming lending developed during that time, reflecting weak bank management before and after hasty privatizations, encouragement of excessive risk-taking by the unlimited government backing of bank liabilities and thin capitalization; the lack of proper credit evaluation skills compounded by the absence of a consumer loan credit bureau and the limited use of the business bureau that was available; inadequate bank supervision, unable to prevent the improper capitalization of privatized banks through cross-lending between banks as well as wholly fraudulent transactions; and improper accounting practices, which for instance defined nonperforming loans as the amount of overdue payments rather than the full value of the loans.
10. The first set of banking reforms following the crisis was designed to strengthen regulatory and supervisory frameworks and reduce sources of excessive risk-taking. Most of the reforms in these areas were implemented during 1995–2000, although regulations since then have been further revised and enhanced. The key changes compared to the pre-crisis framework are as follows:
Safeguards have been put in place to monitor and limit related-party lending. Disclosure requirements have been strengthened regarding loans to shareholders and related parties; and the ceiling on such operations has been tightened in terms of its base of computation, eligible credit operations, and percentage level (IMF (2001)).
Accounting rules were brought closer to best international practices, including by upgrading methods to classify loans and investments and to create loan-loss reserves and by aligning the accounting treatment of non-performing loans with generally-accepted standards.
Stricter capital adequacy requirements have been adopted. A new capital adequacy ratio, incorporating market risks, was adopted in 1996. The definition of regulatory capital was then improved in 1998 and was gradually implemented until 2003. Basic capital now has to account for more than 50 percent of net capital; deferred taxes can only account for 20 percent of basic capital (with some exceptions allowing for up to 30 percent); and rules for counting subordinated debt as capital conform to Basel principles.
Banks have been required to upgrade their risk management practices. This started with the new requirement for banks to have risk-management units which report daily to the Bank of Mexico (BoM). Banks also have been required to check with the credit bureau the credit score of potential borrowers before extending new loans.
Government insurance of bank liabilities has been overhauled. From 1999 to 2004, the coverage of that insurance was reduced, in yearly stages, from a broad range of bank liabilities to bank deposits, and the coverage of deposit insurance was reduced on a per depositor basis, from an unlimited amount to the equivalent of US$120,700 on a per depositor basis (in inflation-indexed terms).
11. The effects of these reforms on bank performance can be expected to change over time. In the short-term, the reforms create new constraints for supplying bank credit while shifting credit risk back to banks and this is likely to cause an increase in banks’ operating costs (e.g., through the need to back a given stock of loans with additional equity). In the longer-term, the restoration of safe balance sheets, together with proper credit screening and risk monitoring mechanisms, should help raise and sustain bank lending growth and profits while enabling a more efficient mediation of savings to investment.
Bank reform II: further liberalization and competition
12. A second group of reforms broadened competition, both within the banking sector and between the banking sector and other financial intermediaries. The latter reforms started in 1996 and still continue today, as legislative and regulatory improvements governing development banks, nonbank intermediaries, and capital markets are still being undertaken. Key measures so far include:
All restrictions on foreign ownership of Mexican banks were lifted by 1997. The participation of foreign banks has increased steadily ever since, reaching more than 80 percent of all banking assets in 2004. Haber (2004) finds evidence that foreign bank entry in Mexico has played a key role in recapitalizing the banking sector and has improved its overall profitability.
The role of state-owned banks has been streamlined. Regulatory frameworks for several public development banks have been strengthened, with a view to preventing forbearance and regulating them increasingly like private banks. A prominent example has been the rationalization of government trust funds for housing into a development bank, Sociedad Hipotecaria Federal, which is being supervised by the CNB V based on the same principles as for private commercial banks.
Banks have met greater competition on the asset side as a result of the development of domestic capital markets. This has enabled large businesses to finance themselves through issuing bonds. Although still relatively small compared to bank commercial credits, the stock of private bond capitalization increased at double digit rates annually during 2001-03. Reforms have also taken place allowing specialized non-bank financial intermediaries (Sofoles) to enter aggressively the business of consumer and housing loans.
Competition on the liability side has intensified more recently. The regulations of mutual funds and saving and loan institutions have been clarified and simplified, making them more attractive as investment vehicles. Further reforms are underway to grant them greater regulatory certainty and operational flexibility (i.e. proposals to allow them a broader range of activities or investments).
13. In Mexico’s context, these reforms can be expected to reduce profits and encourage new lending. A priori, greater competition within the financial sector can be expected to reduce banks’ profits, if it does not lead to further consolidation—the scope for which is small given Mexico’s relatively concentrated banking sector. And whereas the development of alternative financing vehicles may reduce at the margin the demand for bank credit, the level of credit is currently so low that the incentives for expanding lending as a result of reduced profit margins are likely to dominate.
Bank reform III: better control of ex-ante and ex-post credit risk
14. Lack of information about borrowers and weak contract enforceability have been perceived as major constraints on bank credit supply. A lack of effective infrastructure to follow debtors’ history and the uncertainties created by the 1994–95 crisis have exacerbated information asymmetries between lenders and borrowers. Weak contract enforceability is not only the result of poor legislation for repossessing collateral but is also related to aspects of a judicial system that has been perceived as ineffective, slow, and lacking reliability in terms of expected rulings. Difficulty in loan recovery may also have reflected a culture of non-repayment in the years following the 1994–95 crisis and subsequent rescue.
15. A third group of reforms has thus sought to improve creditor information availability and reduce the cost of resolving non-performing loans. Credit bureaus, bankruptcy procedures, and provisions for secured—backed with collateral—lending have been the main focus of these reforms:
The use of credit bureaus has been greatly expanded and their rules of operation enhanced. Risk management requirements set out by the government after 1995 gave incentives to use credit bureaus more proactively. The first credit bureau that monitored consumer lending was created in 1996. A new credit bureau law enacted in 2002 enhanced the coverage and quality of the credit information available, including by enhancing borrowers’ rights to access and rectify errors in their credit records. While most lending institutions are involved in Mexico’s two private Credit Bureaus (one for individuals and one for businesses), a number of limitations remain, such as the unavailability of Infonavit3 credit records and other information typically used in advanced economies, such as utility payment or tax payment records. Consistent with the low level of bank lending in Mexico compared to other countries, Credit Bureaus cover a smaller population share than other emerging market countries (Table 1).
A new bankruptcy law enacted in 2000 rebalanced bargaining power between creditors and debtors in case of loan default. Before the reform, bankruptcy procedures were widely viewed as excessively long and cumbersome and the likelihood of repossessing collateral for unpaid loans was low. The new law limits the time for good faith negotiations, after which liquidation has to start; allows for greater flexibility and value maximization in deciding possible reorganization plans; and clarifies the ordering of creditors in case of liquidation. However, the impact of the new law has likely been limited by the fact that businesses with debt below a certain threshold are not required to follow the new bankruptcy procedures and may then continue to rely on less efficient and certain mechanisms (Zuniga-Villasenor (2005)).
The legal framework for secured lending has been improved in 2000 and 2003. Hills (2004) explains that new legislation expands the scope for secured lending schemes while strengthening creditors’ rights in these schemes. It establishes clear rules for the pledging of collateral without actual transfer of possession to the creditor—something new—and for the use of trusts as a vehicle of secured lending. In particular, it clarifies secured lenders’ first preferences against the pledged collateral and sets forth clear provisions for nonjudicial foreclosure in the use of trust agreements.
Other steps have been taken to improve creditor rights, outside the court system. Creditor rights may have remained weak in practice as a result of inefficient state courts and public registries. Moody’s (2004) documents the wide heterogeneity of contract enforceability across Mexican states in terms of institutional quality, duration of judicial proceedings, quality and quantity of human and material resources, and efficiency in the enforcement of resolutions.4 The lack of efficiency of the judicial system in several states has encouraged the use of special arrangements that sidestep the court system. For instance, the reform of mortgage contracts in 2001 replaced liens on property with bilateral trusts, while lease-to-own contracts in automobile financing effectively place assets used as collateral outside an individual’s bankruptcy estate (Haber 2004).
Contract Enforceability and Access to Credit in Mexico and Comparator Countries, as of 2004
The Legal Rights Index ranges from 0–10, with higher scores indicating that those laws are better designed to expand access to credit.
The Credit Information Index measures the scope, access and quality of credit information available through public registries or private bureaus. It ranges from 0–6, with higher values indicating that more credit information is available from a public registry or private bureau.
Contract Enforceability and Access to Credit in Mexico and Comparator Countries, as of 2004
| Mexico | Chile | Korea | Latin America | East Asia | OECD | |||
|---|---|---|---|---|---|---|---|---|
| & Carribean | & Pacific | high income | ||||||
| Enforcing Contracts | ||||||||
| Number of procedures | 37 | 28 | 29 | 35 | 27 | 19 | ||
| Time (in days) | 421 | 305 | 75 | 462 | 316 | 229 | ||
| Cost (% of debt) | 20.0 | 10.4 | 5.4 | 23.3 | 56.9 | 10.7 | ||
| Registering Property | ||||||||
| Number of procedures | 5 | 6 | 7 | 6 | 4 | 4 | ||
| Time (in days) | 74 | 31 | 11 | 56 | 51 | 34 | ||
| Cost (% of property value) | 5.3 | 1.4 | 6.3 | 5.6 | 4.3 | 4.8 | ||
| Obtaining Credit | ||||||||
| Cost to Create Collateral (% of income per capita) | 25.7 | 5.3 | 8.1 | 19.4 | 2.0 | 5.2 | ||
| Legal Rights Index 1/ | 2 | 4 | 6 | 3 | 5 | 6 | ||
| Credit Information Index 2/ | 6 | 6 | 5 | 4 | 1 | 5 | ||
| Public Credit Registry Coverage (borrowers per 1000 adults) | 0 | 290 | 0 | 85 | 33 | 76 | ||
| Private Bureau Coverage (borrowers per 1000 adults) | 382 | 220 | 1000 | 325 | 67 | 577 | ||
The Legal Rights Index ranges from 0–10, with higher scores indicating that those laws are better designed to expand access to credit.
The Credit Information Index measures the scope, access and quality of credit information available through public registries or private bureaus. It ranges from 0–6, with higher values indicating that more credit information is available from a public registry or private bureau.
Contract Enforceability and Access to Credit in Mexico and Comparator Countries, as of 2004
| Mexico | Chile | Korea | Latin America | East Asia | OECD | |||
|---|---|---|---|---|---|---|---|---|
| & Carribean | & Pacific | high income | ||||||
| Enforcing Contracts | ||||||||
| Number of procedures | 37 | 28 | 29 | 35 | 27 | 19 | ||
| Time (in days) | 421 | 305 | 75 | 462 | 316 | 229 | ||
| Cost (% of debt) | 20.0 | 10.4 | 5.4 | 23.3 | 56.9 | 10.7 | ||
| Registering Property | ||||||||
| Number of procedures | 5 | 6 | 7 | 6 | 4 | 4 | ||
| Time (in days) | 74 | 31 | 11 | 56 | 51 | 34 | ||
| Cost (% of property value) | 5.3 | 1.4 | 6.3 | 5.6 | 4.3 | 4.8 | ||
| Obtaining Credit | ||||||||
| Cost to Create Collateral (% of income per capita) | 25.7 | 5.3 | 8.1 | 19.4 | 2.0 | 5.2 | ||
| Legal Rights Index 1/ | 2 | 4 | 6 | 3 | 5 | 6 | ||
| Credit Information Index 2/ | 6 | 6 | 5 | 4 | 1 | 5 | ||
| Public Credit Registry Coverage (borrowers per 1000 adults) | 0 | 290 | 0 | 85 | 33 | 76 | ||
| Private Bureau Coverage (borrowers per 1000 adults) | 382 | 220 | 1000 | 325 | 67 | 577 | ||
The Legal Rights Index ranges from 0–10, with higher scores indicating that those laws are better designed to expand access to credit.
The Credit Information Index measures the scope, access and quality of credit information available through public registries or private bureaus. It ranges from 0–6, with higher values indicating that more credit information is available from a public registry or private bureau.
16. Everything else equal, these reforms are expected to boost bank credit supply while reducing the cost of financial intermediation. Sound borrowers should become more easily identifiable and the recovery of nonperforming loans should become less costly. With credit risk easier to manage ex-ante as well as ex-post, private lending should become more profitable for banks, translating at the same time in higher average returns and additional private lending.
17. However, the adequacy of the above legal reforms without a broader overhaul of the judicial and enforcement system remains unclear. For instance, Haber (2004) interprets the current pattern of credit growth, with considerably faster growth for consumer credit than commercial credit, as evidence that these reforms have not been sufficient. He notes that when contract enforceability remains as a generalized problem, lending which is backed by assets that are easier to assign to creditors, such as consumer lending, would be expected to expand faster than other (more difficult to enforce) lending. It is important, however, to underscore alternative explanations for this lending pattern, including the higher profitability of consumer loans over business loans and the fact that fast growth of consumer lending is a worldwide phenomenon.
The placement of FOBAPROA notes with banks
18. The placement of government notes (FOBAPROA notes) following the 1994–95 crisis recapitalized distressed banks, and these notes remained a key feature of the balance sheet of the largest banks.5 The FOBAPROA notes were long-term nontradable bonds (most had a 10-year maturity) which the government issued to recapitalize distressed banks and thereby protect depositors. Because most bank liabilities were government-backed and the bank rescue process lasted several years, the amount of FOBAPROA notes issued was large, estimated at about 13 percent of 1999’s GDP. Partly as a result of bank mergers, all these notes ended up in the balance sheets of the six largest banks, which together managed more than 90 percent of total bank assets at the end of the 1990s.
19. While serving to restore a viable banking sector, the rescue operation may have at the same time affected bank lending behavior negatively. On the one hand, the recapitalization of distressed banks should have helped the resumption of bank lending to the private sector at the same time as restoring viable bank balance sheets. On the other hand, Gonzales-Anaya (2002) argues that banks with large holdings of FOBAPROA notes have had a disincentive to grant new loans to the private sector. First, FOBAPROA notes secure a better risk-return profile than loans to the private sector, with rates of returns indexed to short-term interest rates at a small risk and at virtually no cost.6 Second, the dominant presence of such assets in a bank balance sheets may in certain cases increase the marginal cost of new funding above the marginal return of new lending.7 Using bank-level data for 1998 to mid-2002, he finds evidence of a negative relationship between banks’ private lending and their FOBAPROA-related revenues.8 9
The macroeconomic environment
20. Changes in interest rates have immediate effects on bank profits and bank strategies to manage their assets and liabilities. These effects are well documented by Aguilar, Cabal, and De la Llera (2005) for the period 2000–2004. Using Mexican bank level data, they decompose annual changes in banks’ total net interest margin between the effects of interest rate changes, the growth of earning assets, changes in the structure of earning assets, and changes in funding sources. The results reveal that Mexican banks have broadly preserved their overall level of net interest margin over the past four years, despite a sharp decline in interest rates, thanks to the growth of earning assets and the steady restructuring of their asset-liability structure.
21. Quite naturally, banks also have responded to return and funding cost differentials. The average share of consumer credits in earning assets jumped from 2.3 percent at end-2000 to 6.9 percent at end-2004, while the share of commercial credits fell almost 3 percentage points, down to less than 13 percent, over the same period. This partly reflects a remarkable difference in returns between the two types of loans (31.2 percent for consumer loans versus 6.8 percent for commercial loans at end-2004). Similarly, the liability structure of banks has steadily shifted toward less expensive retail deposits, and away from money market funding and foreign currency deposits.
C. Bank Lending and Profitability, 1998–2004
22. The various mechanisms driving bank lending and profits described in the previous section are challenging to detect precisely with empirical methods. Helpfully for the purposes of empirical investigation, there is evidence that credit supply has remained constrained for most of the period since the mid-1990s (see Gonzales-Anaya (2002)). This ensures that the evolution of bank lending has generally reflected changes in credit supply rather than demand. Unfortunately, there are no simple quantitative indicators available to directly capture most of the structural changes that we described.
23. We choose therefore to document bank behavior in the past seven years focusing on available balance sheet and financial results information. Despite the limitation of this focus, this exercise allows us to better understand bank behavior in general, and also to detect the role of bank reforms, the government rescue of banks, and changes in macro-conditions in connection with bank balance sheets. By testing for structural changes during the period, we obtain some partial indication of whether bank behavior has structurally changed over the past seven years.
Consolidated banking sector
24. We first document stylized facts about the consolidated commercial banking sector, using CNBV quarterly bank-level data. Based on these data, the banking sector remained relatively concentrated throughout the period (Table 2 and Figure 5). In particular, the two largest commercial banks have consistently managed about half of the sector’s total assets.
Mexico: Distribution Among Commercial Banks of Banking Assets, Deposits, and Private Sector Lending
Mexico: Distribution Among Commercial Banks of Banking Assets, Deposits, and Private Sector Lending
| 1998Q4 | 2001Q4 | 2004Q4 | |
|---|---|---|---|
| Bank assets | (in percentage shares) | ||
| 50% percentile | 0.2 | 0.4 | 0.4 |
| 95% percentile | 20.1 | 23.1 | 22.4 |
| Bank deposits | |||
| 50% percentile | 0.2 | 0.3 | 0.3 |
| 95% percentile | 19.4 | 22.8 | 20.9 |
| Private sector lending | |||
| 50% percentile | 0.4 | 0.4 | 0.5 |
| 95% percentile | 24.6 | 27.2 | 21.3 |
| Number of commercial banks | 30 | 30 | 30 |
Mexico: Distribution Among Commercial Banks of Banking Assets, Deposits, and Private Sector Lending
| 1998Q4 | 2001Q4 | 2004Q4 | |
|---|---|---|---|
| Bank assets | (in percentage shares) | ||
| 50% percentile | 0.2 | 0.4 | 0.4 |
| 95% percentile | 20.1 | 23.1 | 22.4 |
| Bank deposits | |||
| 50% percentile | 0.2 | 0.3 | 0.3 |
| 95% percentile | 19.4 | 22.8 | 20.9 |
| Private sector lending | |||
| 50% percentile | 0.4 | 0.4 | 0.5 |
| 95% percentile | 24.6 | 27.2 | 21.3 |
| Number of commercial banks | 30 | 30 | 30 |
Mexico: Distribution of banks’ asset shares at end-1998
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Mexico: Distribution of banks’ asset shares at end-1998
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Mexico: Distribution of banks’ asset shares at end-1998
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
25. The main trends concerning bank performance and balance sheets are the following:
The level of commercial bank lending to the private sector declined steadily from 1998 to early 2000, when it bottomed out at around 7 percent of GDP. It started recovering significantly in 2003 (excluding non-performing loans, Figure 6). At end-2004, it reached a level equivalent to 8½ percent of GDP 10 and 33 percent of commercial bank assets, still significantly lower than the level observed in 1998. Looking at bank-level data, the amount of lending provided by individual commercial banks was generally volatile throughout the period, reflecting the short-term structure of most lending and the small size of most banks’ loan portfolios.
Bank profitability, as measured by pre-tax revenues divided by assets, edged up at the end of 2003 after being stable around 1 percent per annum since 2000. The main components of bank incomes and costs changed significantly (Figure 7): the gradual decline in net interest margins was more than offset by growing commissions and fees and a decline in administrative costs.
The consolidated balance sheet of commercial banks strengthened through improved capital adequacy and declining past-due loans. The ratio of bank capital to assets, the simplest proxy for capital adequacy, increased steadily from 7¾ percent to above 11 percent during the period studied (Figure 8). Loan loss reserves came down in tandem with past-due loans, liberating resources for additional lending or placements (Figure 10).
While its composition between loans and securities changed, credit to the public sector remained high throughout the period, at about 35 percent of combined commercial bank assets.11 Securities were gradually substituted for bank lending as a means of public sector financing (Figure 10). As a result, the share of loans to the public sector in banks’ total loan portfolio has steadily declined (Figure 12).
Against declining real interest rates, the liquidity of banks’ balance sheets is believed to have stayed broadly unchanged. Although recorded bank liquid assets rose, this would reflect a rise in regulatory deposits with the central bank and other frozen deposits, such as those associated with FOBAPROA liabilities (Figure 9) according to the BoM. Banks’ voluntary holdings of liquid assets would not have changed dramatically.
Commercial Banks’ Private Lending and Returns on Average Assets
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Private Lending and Returns on Average Assets
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Private Lending and Returns on Average Assets
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Profitability and Efficiency
(1-year moving averages)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Profitability and Efficiency
(1-year moving averages)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Profitability and Efficiency
(1-year moving averages)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Capital Adequacy, Credit to the Public Sector, and Lending to the Private Sector
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Capital Adequacy, Credit to the Public Sector, and Lending to the Private Sector
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Capital Adequacy, Credit to the Public Sector, and Lending to the Private Sector
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Bank Liquidity, Deposits, and Real Short-Term Rates
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Bank Liquidity, Deposits, and Real Short-Term Rates
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Bank Liquidity, Deposits, and Real Short-Term Rates
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Composition of Commercial Bank Credit to the Public Sector
(in percent of bank assets)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
1/ Computed as the sum of bank loans to the public sector and bank holdings of securities of the federal and local governments and IPAB.Composition of Commercial Bank Credit to the Public Sector
(in percent of bank assets)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
1/ Computed as the sum of bank loans to the public sector and bank holdings of securities of the federal and local governments and IPAB.Composition of Commercial Bank Credit to the Public Sector
(in percent of bank assets)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
1/ Computed as the sum of bank loans to the public sector and bank holdings of securities of the federal and local governments and IPAB.
Commercial Bank Loan Portfolio Quality
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Bank Loan Portfolio Quality
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Bank Loan Portfolio Quality
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Loan Portfolio Composition
(in percent of portfolio of current loans)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Loan Portfolio Composition
(in percent of portfolio of current loans)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Commercial Banks’ Loan Portfolio Composition
(in percent of portfolio of current loans)
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
26. These trends are broadly consistent with the intuitions expressed earlier about the effects of bank reforms, possible lending disincentives associated with FOBAPROA loans, and declining interest rates. The rebound of bank lending to the private sector (as a share of banking assets) coincides with a decline in non-performing loans (partly financed through FOBAPROA programs), the gradual retirement of the FOBAPROA debt, improved capital adequacy, and interest rate reductions. The following econometric investigation enables to measure the role of each of these factors with respect to the recovery of bank lending to the private sector.
Panel data regressions
27. We conduct a series of panel regressions using individual bank data, with the goal of linking growth of bank private lending, and also bank profitability, with balance sheet structures, changes in macroeconomic conditions, and (perhaps) banks’ holdings of FOBAPROA and other public lending.
Source data
28. We use the CNBV quarterly database on commercial banks, from 1998 to 2004. The sample covers 34 different banks but, as some disappeared or merged during the period, there are 30 in most periods. The structure of the database may create some small statistical biases. First, while the data cover most of the banks that operated during the period, some of the banks which were intervened by the CNBV or simply ceased their commercial banking operations during the period (e.g., Societe Generale) are not in the database. However, the missing banks are small. Second, the accounting of bank mergers is not uniform. In most cases, banks which merged generally appear to be accounted separately before and after the merger. In other cases, the database accounts for merged banks as a single entity throughout the period. This may bias the results—for instance by underestimating the role of capital adequacy if one bank had excess capital while the other did not.
29. We focus on the growth of bank lending to the private sector (excluding non-performing loans) and pre-tax financial revenues (scaled by their assets) as two summary indicators of bank performance to be explained empirically. Explanatory variables are the ratio of capital to assets (unweighted) as a measure of capital adequacy; the ratio of past-due loans to total current and past-due loans as a measure of credit quality; the ratio of FOBAPROA lending and the sum of all bank credit to the public sector (banking loans to the public sector plus all government domestic securities) as measures of crowding out risks; the ratio of a bank’s assets to total banking assets as a measure of bank size; and real short-term interest rates, year-on-year inflation, and real GDP growth to capture macroeconomic conditions.
Econometric method
30. Our econometric approach is inspired by the recent work of Baqir (2004) on commercial banks in the Philippines. We estimate panel regressions based on the specification of equation (1), where Pit is an indicator of the performance of bank i at time t—either private lending growth or returns on average assets. Bit-1 is the matrix of bank balance sheet and macro variables, lagged by one quarter to try to reduce endogeneity problems. In some specifications, we allow for bank fixed effects (the αi term) and time effects (the βt term). Time effects consist either of quarterly dummies to take account of seasonality, or a complete set of time dummies for the full period. We estimate each equation using ordinary least squares (OLS), and fixed and random effects panel models. We run additional specifications (to explore the robustness of the results) using a fixed or random effects specification on the basis of the Hausman test.
31. We then seek evidence of structural breaks in the relationships estimated. Linkages between bank lending behavior and profitability, and balance sheet structures and macroeconomic variables, may have changed as a result of bank reforms. For instance, bank reforms, by producing a more secure environment for lending, may have weakened the relationship between bank capital and private lending growth. Also, the relationship between interest rates and bank lending may change as credit supply catches up with unmet demand. To establish whether these various relationships have changed, we test the robustness of estimated coefficients and re-estimate the same regressions for sub-periods.
Results
32. The regressions explaining bank’s lending to the private sector confirm the intuitive links between bank lending and balance sheet indicators, while finding no evidence of crowding out by FOBAPROA lending or bank credit to the public sector. More specifically, Table 3 shows that:
Capital adequacy has a significant influence on private lending growth. This finding is robust across econometric specifications. Looking at estimations made for half-periods, the significance of this linkages lessens after mid-2001. While this evolution is consistent with the idea mentioned earlier that less insurance might be required following thorough banking reforms, it could also reflect the shortcoming of the capital adequacy indicator used (which is not risk-weighted).
Nonperforming loans tend to reduce private lending growth, with stronger evidence in the second half of the period. While the sign and magnitude of the effect are broadly the same across specifications, we do not obtain significance levels better than 15–20 percent for regressions covering the full period. We obtain significance at the 5 percent level for the second half of the period.
The amount of FOBAPROA loans or total credit to the public sector does not have a significant influence on banks’ private lending, refuting the possibility of a crowding out of private lending by bank rescue programs or high government financing needs. Hence, the positive effect of bank rescues following the 1996 crisis—through the absorption of banks’ non-performing loans—likely outweighs any possible negative effects. This contrasts with the result of Gonzales-Anaya (2002), which found evidence of a negative correlation between bank FOB APROA-related revenues and bank lending to the private sector. While the difference may be attributed to different statistical specifications (see footnote 7), we have also been able to use a longer time period.
Real short-term interest rates are found to be positively correlated with increases in bank lending to the private sector, although this relationship ceases to be significant in the second half of the period. One possible interpretation of this result is the presence of credit supply shortages and their gradual lessening over time. Based on standard models of credit rationing (Stiglitz and Weiss (1981)), credit can be expected to increase with rising interest rates below a certain threshold. This interpretation would also explain why activity variables do not appear to influence significantly bank lending. While this result may also reflect endogeneity between credit and interest rate variables, the result remains when longer lags are used for interest rates.
Mexico: Regressions for Bank Private Lending Growth
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
Mexico: Regressions for Bank Private Lending Growth
| Model 1/ | OLS | OLS | OLS | OLS | OLS | FE | RE | RE | RE | RE | RE |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Quarterly | Quarterly | Quarterly | Quarterly | Quarterly | |||||||
| Time effects | dummies | dummies | dummies | dummies | dummies | No | No | No | Yes | No | No |
| 1998Q1 - | 2003Q3 - | 1998Q1 - | 2003Q3 - | ||||||||
| Period | Full | Full | Full | 20001Q2 | 20004Q4 | Full | Full | Full | Full | 20001Q2 | 20004Q4 |
| Capital-to-asset ratio, lagged | 6.7 *** | 6.5*** | 6.0*** | 9.6* | 2.5** | 5.8* | 6.8** | 9.6** | 6.8*** | 9.6** | 1.9 |
| (2.7) | (2.7) | (2.7) | (5.2) | (2.8) | (2.8) | (5.2) | (5.2) | (5.2) | (5.2) | (1.7) | |
| Non-performing-loans-to-total-loans, lagged | -6.5* | -6.2* | -6.1 * | -2.2 | -4.8*** | -7.0 | -6.1*** | -6.3 * | -6.0 | -2.2 | -4.8 ** |
| (4.8) | (4.8) | (4.8) | (2.2) | (4.8) | (6.6) | (12.3) | (12.3) | (4.8) | (2.2) | (2.2) | |
| Public-sector-credit-to-assets, lagged 2/ | 1.2 | 1.1 | 1.5 | 0.7 | 0.7 | 1.2 | 1.1 | 1.1 | 1.5 | -0.1 | |
| (2.0) | (2.0) | (3.8) | (0.9) | (0.9) | (3.8) | (3.8) | (2.0) | (3.8) | (1.2) | ||
| FOBAPROA-notes-to-asset ratio, lagged | -0.1 | ||||||||||
| (2.7) | |||||||||||
| Real short-term interest rate, lagged | 31.8*** | 31.5*** | 49.8 ** | -27.2 | 30.2*** | 31.6*** | 76.7*** | 49.8 ** | -27.1 | ||
| (14.4) | (14.4) | (27.7) | (40.3) | (14.4) | (14.4) | 22.1) | 27.6) | (38.4) | |||
| Real GDP y-o-y growth, lagged | -13.6 | -14.9 | -13 | 20.7 | 5.8 | -14.1 | -13.7 | -15.0 | -21.3 | 20.7 | 5.6 |
| (15.0) | (14.9) | (14.8) | (46.9) | (9.1) | (15.0) | (14.9) | (14.9) | 28.5) | 46.9) | (8.7) | |
| CPI y-o-y inflation, lagged | -0.4 | 10.7* | -0.3 | 17.7 | 47.5 | -1.92 | -0.6 | 10.4* | 7.4 | 17.7 | 49.2 |
| (9.4) | (7.9) | (9.4) | (20.6) | (53.4) | (9.5) | (9.4) | (7.9) | (19.8) | (20.6) | (50.9) | |
| Observations | 777 | 777 | 780 | 387 | 390 | 777 | 777 | 777 | 777 | 387 | 390 |
| R-squared | 0.021 | 0.145 | 0.021 | 0.030 | 0.027 | … | … | … | … | … | … |
| Hausman test (prob>chi2) | … | … | … | … | 0.97 | 0.97 | … | … | … | … |
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
Mexico: Regressions for Bank Private Lending Growth
| Model 1/ | OLS | OLS | OLS | OLS | OLS | FE | RE | RE | RE | RE | RE |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Quarterly | Quarterly | Quarterly | Quarterly | Quarterly | |||||||
| Time effects | dummies | dummies | dummies | dummies | dummies | No | No | No | Yes | No | No |
| 1998Q1 - | 2003Q3 - | 1998Q1 - | 2003Q3 - | ||||||||
| Period | Full | Full | Full | 20001Q2 | 20004Q4 | Full | Full | Full | Full | 20001Q2 | 20004Q4 |
| Capital-to-asset ratio, lagged | 6.7 *** | 6.5*** | 6.0*** | 9.6* | 2.5** | 5.8* | 6.8** | 9.6** | 6.8*** | 9.6** | 1.9 |
| (2.7) | (2.7) | (2.7) | (5.2) | (2.8) | (2.8) | (5.2) | (5.2) | (5.2) | (5.2) | (1.7) | |
| Non-performing-loans-to-total-loans, lagged | -6.5* | -6.2* | -6.1 * | -2.2 | -4.8*** | -7.0 | -6.1*** | -6.3 * | -6.0 | -2.2 | -4.8 ** |
| (4.8) | (4.8) | (4.8) | (2.2) | (4.8) | (6.6) | (12.3) | (12.3) | (4.8) | (2.2) | (2.2) | |
| Public-sector-credit-to-assets, lagged 2/ | 1.2 | 1.1 | 1.5 | 0.7 | 0.7 | 1.2 | 1.1 | 1.1 | 1.5 | -0.1 | |
| (2.0) | (2.0) | (3.8) | (0.9) | (0.9) | (3.8) | (3.8) | (2.0) | (3.8) | (1.2) | ||
| FOBAPROA-notes-to-asset ratio, lagged | -0.1 | ||||||||||
| (2.7) | |||||||||||
| Real short-term interest rate, lagged | 31.8*** | 31.5*** | 49.8 ** | -27.2 | 30.2*** | 31.6*** | 76.7*** | 49.8 ** | -27.1 | ||
| (14.4) | (14.4) | (27.7) | (40.3) | (14.4) | (14.4) | 22.1) | 27.6) | (38.4) | |||
| Real GDP y-o-y growth, lagged | -13.6 | -14.9 | -13 | 20.7 | 5.8 | -14.1 | -13.7 | -15.0 | -21.3 | 20.7 | 5.6 |
| (15.0) | (14.9) | (14.8) | (46.9) | (9.1) | (15.0) | (14.9) | (14.9) | 28.5) | 46.9) | (8.7) | |
| CPI y-o-y inflation, lagged | -0.4 | 10.7* | -0.3 | 17.7 | 47.5 | -1.92 | -0.6 | 10.4* | 7.4 | 17.7 | 49.2 |
| (9.4) | (7.9) | (9.4) | (20.6) | (53.4) | (9.5) | (9.4) | (7.9) | (19.8) | (20.6) | (50.9) | |
| Observations | 777 | 777 | 780 | 387 | 390 | 777 | 777 | 777 | 777 | 387 | 390 |
| R-squared | 0.021 | 0.145 | 0.021 | 0.030 | 0.027 | … | … | … | … | … | … |
| Hausman test (prob>chi2) | … | … | … | … | 0.97 | 0.97 | … | … | … | … |
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
33. The regressions focusing on bank’s pre-tax revenues point to the negative impact of bank size, non-performing loans, and FOBAPROA lending. More specifically, Table 4 shows that:
The relative size of a bank’s assets weighs negatively on bank net revenues, although this effect loses significance in the second half of the period. This suggests that possible benefits from controlling large market shares, e.g., possible economies of scales or pricing power, are outweighed by greater inefficiencies in large banks compared to smaller ones.
Bank capital tends to lower bank profitability, although this effect was not found across statistical specifications. This suggests the cost of capital might have been higher than that of other bank liabilities.
A negative impact of non-performing loans is found, confirming the intuitive linkage between credit quality and bank profitability. While we do not find any significant effect of credit to the public sector as a whole, bank holdings of FOBAPROA loans have tended to reduce profitability. The latter finding may reflect the lower returns of these riskless loans or specific features of banks that required more assistance than others following the 1994 crisis, including for instance the restructuring costs not borne by the banks themselves, persistent operational weaknesses, or a lower appetite for risk.
No significant relationship is found between macro-variables and bank profitability across statistical specifications.
Mexico: Regressions for Return on Average Assets
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
Mexico: Regressions for Return on Average Assets
| Model 1/ | OLS | OLS | OLS | OLS | FE | FE | FE | FE | FE | FE | RE |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Quarterly | Quarterly | Quarterly | Quarterly | ||||||||
| Time effects | dummies | dummies | dummies | dummies | No | No | No | Yes | No | No | No |
| 1998Q1 - | 2003Q3 - | 1998Q1 - | 2003Q3 - | ||||||||
| Period | Full | Full | 20001Q2 | 20004Q4 | Full | Full | Full | Full | 20001Q2 | 20004Q4 | Full |
| Share of total banking assets | 0.06* | 0.07 ** | 0.04 | 0.06 | -0.5 *** | -0.43 *** | -0.46 *** | -0.45 *** | -0.8 *** | -0.11 | -0.01 |
| (0.05) | (0.05) | (0.07) | (0.07) | (0.3) | (0.1) | (0.1) | (0.1) | (0.3) | (0.48) | (0.1) | |
| Capital-to-asset ratio, lagged | 0.02 | 0.01 | -0.04 ** | 0.1 * | -0.07 *** | -0.07 *** | -0.07 *** | -0.07 *** | -0.05 * | -0.01 | -0.05 *** |
| (0.02) | (0.02) | (0.02) | (0.03) | (0.2) | (0.02) | (0.02) | (0.02) | (0.04) | (0.04) | (0.02) | |
| Liquid-to-total assets, lagged | 0.01 | ||||||||||
| (0.02) | |||||||||||
| Non-performing-loans-to-total-loans, lagged | -0.14*** | -0.14*** | -0.14*** | -0.19*** | -0.21 *** | -0.22 *** | -0.21 *** | -0.22 *** | -0.19*** | -0.24 *** | -0.19*** |
| (0.03) | (0.03) | (0.06) | (0.03) | (0.03) | (0.03) | (0.03) | (0.03) | (0.07) | (0.03) | (0.03) | |
| Public-sector-credit-to-assets, lagged 2/ | -0.02 | -0.03 | |||||||||
| (0.02) | (0.03) | ||||||||||
| FOBAPROA-notes-to-asset ratio, lagged | 0.04 ** | 0.06 *** | 0.03 | 0.14* | -0.16*** | -0.14*** | -0.16*** | -0.16*** | -0.24 *** | 0.07 | -0.13 *** |
| (0.02) | (0.03) | (0.02) | (0.06) | (0.03) | (0.03) | (0.03) | (0.03) | (0.05) | (0.07) | (0.03) | |
| Real short-term interest rate, lagged | 0.02 | 0.03 | -0.18* | -1.0 | 0.01 | 0.01 | 0.01 | -0.10 | -0.19** | -1.2** | 0.02 |
| (0.1) | (0.1) | (0.14) | (0.9) | (0.08) | (0.08) | (0.08) | (0.12) | (0.12) | (0.6) | (0.08) | |
| Real GDP y-o-y growth, lagged | -0.11 | -0.12 | -0.7 *** | 0.04 | -0.1 | -0.09 | -0.09 | -0.21 * | -0.64 *** | 0.07 | -0.1 |
| (0.10) | (0.10) | (0.2) | (0.2) | (0.1) | (0.1) | (0.1) | (0.1) | (0.19) | (0.15) | (0.1) | |
| CPI y-o-y inflation, lagged | -0.07 | -0.07 | 20.26*** | 0.57 | 0.01 | -0.01 | 0.02 | 0.26 *** | -0.12* | 0.77 | 0.01 |
| (0.06) | (0.06) | (0.10) | (1.2) | (0.05) | (0.05) | (0.06) | (0.11) | (0.08) | -0.85 | (0.05) | |
| Observations | 784 | 781 | 388 | 396 | 784 | 781 | 781 | 784 | 388 | 393 | 784 |
| R-squared | 0.046 | 0.047 | 0.055 | 0.100 | … | … | … | … | … | … | … |
| Hausman test (prob>chi2) | … | … | … | … | 0.00 | … | … | … | … | … | 0.00 |
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
Mexico: Regressions for Return on Average Assets
| Model 1/ | OLS | OLS | OLS | OLS | FE | FE | FE | FE | FE | FE | RE |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Quarterly | Quarterly | Quarterly | Quarterly | ||||||||
| Time effects | dummies | dummies | dummies | dummies | No | No | No | Yes | No | No | No |
| 1998Q1 - | 2003Q3 - | 1998Q1 - | 2003Q3 - | ||||||||
| Period | Full | Full | 20001Q2 | 20004Q4 | Full | Full | Full | Full | 20001Q2 | 20004Q4 | Full |
| Share of total banking assets | 0.06* | 0.07 ** | 0.04 | 0.06 | -0.5 *** | -0.43 *** | -0.46 *** | -0.45 *** | -0.8 *** | -0.11 | -0.01 |
| (0.05) | (0.05) | (0.07) | (0.07) | (0.3) | (0.1) | (0.1) | (0.1) | (0.3) | (0.48) | (0.1) | |
| Capital-to-asset ratio, lagged | 0.02 | 0.01 | -0.04 ** | 0.1 * | -0.07 *** | -0.07 *** | -0.07 *** | -0.07 *** | -0.05 * | -0.01 | -0.05 *** |
| (0.02) | (0.02) | (0.02) | (0.03) | (0.2) | (0.02) | (0.02) | (0.02) | (0.04) | (0.04) | (0.02) | |
| Liquid-to-total assets, lagged | 0.01 | ||||||||||
| (0.02) | |||||||||||
| Non-performing-loans-to-total-loans, lagged | -0.14*** | -0.14*** | -0.14*** | -0.19*** | -0.21 *** | -0.22 *** | -0.21 *** | -0.22 *** | -0.19*** | -0.24 *** | -0.19*** |
| (0.03) | (0.03) | (0.06) | (0.03) | (0.03) | (0.03) | (0.03) | (0.03) | (0.07) | (0.03) | (0.03) | |
| Public-sector-credit-to-assets, lagged 2/ | -0.02 | -0.03 | |||||||||
| (0.02) | (0.03) | ||||||||||
| FOBAPROA-notes-to-asset ratio, lagged | 0.04 ** | 0.06 *** | 0.03 | 0.14* | -0.16*** | -0.14*** | -0.16*** | -0.16*** | -0.24 *** | 0.07 | -0.13 *** |
| (0.02) | (0.03) | (0.02) | (0.06) | (0.03) | (0.03) | (0.03) | (0.03) | (0.05) | (0.07) | (0.03) | |
| Real short-term interest rate, lagged | 0.02 | 0.03 | -0.18* | -1.0 | 0.01 | 0.01 | 0.01 | -0.10 | -0.19** | -1.2** | 0.02 |
| (0.1) | (0.1) | (0.14) | (0.9) | (0.08) | (0.08) | (0.08) | (0.12) | (0.12) | (0.6) | (0.08) | |
| Real GDP y-o-y growth, lagged | -0.11 | -0.12 | -0.7 *** | 0.04 | -0.1 | -0.09 | -0.09 | -0.21 * | -0.64 *** | 0.07 | -0.1 |
| (0.10) | (0.10) | (0.2) | (0.2) | (0.1) | (0.1) | (0.1) | (0.1) | (0.19) | (0.15) | (0.1) | |
| CPI y-o-y inflation, lagged | -0.07 | -0.07 | 20.26*** | 0.57 | 0.01 | -0.01 | 0.02 | 0.26 *** | -0.12* | 0.77 | 0.01 |
| (0.06) | (0.06) | (0.10) | (1.2) | (0.05) | (0.05) | (0.06) | (0.11) | (0.08) | -0.85 | (0.05) | |
| Observations | 784 | 781 | 388 | 396 | 784 | 781 | 781 | 784 | 388 | 393 | 784 |
| R-squared | 0.046 | 0.047 | 0.055 | 0.100 | … | … | … | … | … | … | … |
| Hausman test (prob>chi2) | … | … | … | … | 0.00 | … | … | … | … | … | 0.00 |
FE indicates fixed effects and RE random effects. All models specified with constant. Standard errors are indicated in parenthesis.
Public sector credit comprises all bank loans to the government sector, including FOBAPROA notes, and all government sector securities, including IPAB securities.
significance at the 5 percent level
significance at the 10 percent level
significance at the 20 percent level
34. We find evidence that the relationships between bank lending and profitability on the one hand and balance sheet structure and interest rates on the other have changed with time. Statistical tests indicate significant changes in coefficient values before and after mid-2001 (the middle of the period) and before and after mid-2003 (Table 5). Separate estimations for each sub-period indicate a smaller role for capital adequacy and a greater role for past-due loans in explaining bank lending to the private sector after 2001. As mentioned earlier, the lessening of the positive correlation between the level of real interest rate and lending growth, also observed across the two sub-periods, is consistent with the phasing out of credit supply shortages. Turning to bank’s net results, bank size, capital adequacy, and holdings of FOBAPROA loans all stop having significant measurable effects in the period after mid-2001, whereas the estimated impact of outstanding past-due loans on net results increases.
Mexico: Wald Tests for Structural Break in Banks’ Lending
Mexico: Wald Tests for Structural Break in Banks’ Lending
| Cut-off date | Null hypothesis of no break |
|---|---|
| 2001Q2 | rejected at 58 percent level |
| 2003Q2 | rejected at 93 percent level |
Mexico: Wald Tests for Structural Break in Banks’ Lending
| Cut-off date | Null hypothesis of no break |
|---|---|
| 2001Q2 | rejected at 58 percent level |
| 2003Q2 | rejected at 93 percent level |
Mexico: Wald Tests for Structural Break in Banks’ Adjusted Results
Mexico: Wald Tests for Structural Break in Banks’ Adjusted Results
| Cut-off date | Null hypothesis of no break |
|---|---|
| 2001Q2 | rejected at 99 percent level |
| 2003Q2 | rejected at 100 percent level |
Mexico: Wald Tests for Structural Break in Banks’ Adjusted Results
| Cut-off date | Null hypothesis of no break |
|---|---|
| 2001Q2 | rejected at 99 percent level |
| 2003Q2 | rejected at 100 percent level |
Mexico: Bank Private Lending and Balance Sheet Indicators in 1998Q1 and 2004Q1
35. Overall, these results provide some indirect inference about the effects of banking reforms. First, banking reforms, insofar as they restored adequate levels of capital and brought a decline in past-due loans, supported the growth of bank lending to the private sector. While the evidence suggests that the role of better capitalization may have diminished—consistent with a more secure lending environment—the amount of remaining past-due loans discriminates even more between bank lending and profitability in the recent period than earlier. Second, the lessening of the positive relation between bank lending growth and real interest rates lends support to the idea that banking reforms alleviated credit rationing problems that have been diagnosed in the past. Third, the lesser negative influence of bank size and holdings of FOBAPROA loans on profitability in the second half of the period studied could indicate that banks that are large or have been rescued have restructured effectively. Finally, our findings do not find evidence that the bank rescue package implemented after the 1994 crisis, or the large domestic borrowing needs to the public sector, have discouraged banks from lending.
D. Post-Reform Credit Growth: Chile and Korea
36. The relationships between balance sheets and bank lending estimated in the chapter cannot explain fully the strong acceleration of bank lending that has been ongoing in Mexico since 2003. The latter must reflect other factors difficult to capture in a regression model. We therefore turn to two country cases, Chile and Korea, which both implemented banking reforms that were similar to Mexico’s post-1995 reforms following their own banking crises. These cases serve to illustrate the range of possible trajectories for bank credit in Mexico as well as potential risks.
37. Reforms in Chile and Korea following banking crises (respectively in the early 1980s and 1997–98) shared the same focus as post-1994 Mexican banking reforms, on limiting risk-taking, strengthening accounting and transparency, and further establishing market discipline. The milestone of post-crisis banking reforms in Chile was the enactment of a new banking law in 1986 that restrained related lending, strengthened capital adequacy rules, and required banks to rate the quality of their investments. Market discipline was also strengthened by establishing strong disclosure requirements, clarifying procedures for the resolution of bank insolvencies, rationalizing the deposit-insurance system, and prohibiting the use of last-lender-facilities to bail out banks (IMF (2004)). In Korea, legislation has been adopted starting in 1997 to expedite the resolution of insolvent financial institutions and corrective action with fragile ones, while gradually rationalizing the deposit-insurance system. Also, accounting, public disclosure, and corporate governance standards have been brought in line with generally observed international practices; and supervisory frameworks have been adjusted to Basel definitions, thus strengthening capital adequacy and loss provisioning requirements (Kim (2003)).
38. The evolution of bank lending following these wide-ranging reforms was very different in the two countries. As shown in Figure 13, bank lending in Chile continued to decline, as a share of GDP, for about four years after the start of the reforms before picking up and growing strongly over the next 10 years. In contrast, bank lending has grown faster as a share of GDP in Korea following its 1997–98 crisis and the above-mentioned reforms, and indeed Korea’s total level of credit is currently higher than Chile’s by more than 10 percentage points of GDP. Another difference is the significantly higher volatility of bank lending in Korea than Chile in recent years, largely on account of the recent boom-bust cycle of consumer lending in Korea.
Behavior of Bank Lending in Chile and South Korea, 1980–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Sources: Central Bank of Chile and Bank of Korea.Behavior of Bank Lending in Chile and South Korea, 1980–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Sources: Central Bank of Chile and Bank of Korea.Behavior of Bank Lending in Chile and South Korea, 1980–2004
Citation: IMF Staff Country Reports 2005, 428; 10.5089/9781451825657.002.A004
Sources: Central Bank of Chile and Bank of Korea.39. These different bank credit trajectories in Chile and Korea, against the backdrop of financial sector reforms with broadly similar objectives, point to the quality of reform implementation and the wide range of other determinants of credit growth. The unsustainable boom of consumer lending in Korea has been attributed to remaining shortcomings of credit screening methods of banks and non-banks, the deficiencies in the information made available by credit bureaus in terms of general quality and coverage, as well as government incentive schemes. As Mexico experiments with new guarantee schemes through development banks to encourage lending to special types of borrowers (e.g., small businesses), Korea’s experience, while it took place against a backdrop of a much higher level of bank lending, could suggest caution. Of course, the differences between Chile and Korea are also an evident reminder of the variety of credit determinants aside from policy reforms.
E. Conclusions
40. Mexico has undertaken a very broad banking reform effort which is still being pursued further. The latest initiatives, which focus on procedures for bank resolution and bank liquidation, will be essential to ensure fast and efficient bank interventions and minimize the cost of any bank rescues in the future.
41. This chapter has provided indirect evidence of the benefits of bank reforms since 1995, by linking the strengthening of banks’ balance sheets with lending growth and increasing profitability. We do detect evidence of structural change in bank behavior consistent with a more secure lending environment and, possibly, signs of diminishing credit rationing.
42. The recent fast growth of bank credit cannot be explained by stronger balance sheets alone. But the examples of Korea and Chile show that such an expansion may reflect factors other than bank reforms, some of which can carry risks. This underscores the need for further study of the determinants of the present credit growth in Mexico.
References
Abiad, A., and A. Mody (2003). “Financial Reform: What Shaskes It? What Shapes It?” IMF Working Paper.
Aguilar Ana Luisa, Cabal Roberto, and De la Llera Laura (2005), “Evolución del Margen Financiero de la Banca Mexicana 2000–2005,” Working Paper, Dirección de Análisis del Sistema Financiero, Banco de Mexico.
Bank of Mexico (2003). “G-20 Case Study: The Case of Mexico. Globalization: the Role of Institution Building in the Financial Sector.”
Gil Diaz, Francisco (2000). “The China Syndrome or the Tequila Crisis.” Working Paper No. 77, Center for Research on Economic Development and Policy Reform, Stanford University.
Gonzales-Anaya, Jose Antonio (2002). “Why Have Banks Stopped Lending in Mexico Since the Peso Crisis in 1995.” Center for Research on Economic Development and Policy Reform, Stanford University. Working Paper No. 118.
Baqir, Reza (2004). Philippines—Selected Issues. “Bank Balance Sheets and Performance in the Philippines.” IMF Country Report.
Frederick W. Hill (2004). “Creditors’ Rights in Secured Transactions Enhanced in Mexico.” The Los Angeles Lawyer Magazine, March 2004.
Gruben, William and Robert McComb (1997). “Privatization, Competition, and Supracompetition in the Mexican Commercial Banking System,” Federal Reserve Bank of Dallas.
IADB (2005). “Unlocking Credit: The Quest for Deep and Stable Bank Lending” Report.
IMF (2001). Mexico: Financial System Stability Assessment. IMF Country Report No 01/192.
IMF (2004).Stabilization and Reform in Latin America: A Macroeconomic Perspective on the Experience Since the Early 1990s. Occasional Paper No. 238.
Haber, Stephen (2004). “Mexico’s Experiments with Bank Privatization and Liberalization, 1991–2002.” Mimeo.
Haber, Stephen and Aldo Musacchio (2004). “Contracts Rights and Risk Aversion: Foreign Banks and the Mexican Economy, 1997–2004.” Mimeo.
Kim, Hyun E. (2003). “Institutional Development in the Korea Financial Sector Accompanying Globalization, and Its Economic Effects.” G20 conference paper.
Korean Ministry of Finance and Economy (2003). “Globalization: The Role of Institution Building in The Financial Sector.” G20 Case Study: Institution Building in the Financial Sector.
Moody’s Investors Service: Contract Enforceability Indicators, A Measure of Enforceability of Commercial Contracts and Mortgage Agreements in the Judicial Systems of the Federal Entities in Mexico, November 2004.
Negrin, Jose Louis (2000), “Mecanismos Para Compartir Informacion Crediticia: Evidencia Internacional y la Experiencia Mexicana.” Banco de Mexico, Dirección General de Investigación Económica Working Paper 2000-05.
Zuñiga-Villaseñor, Gerardo (2005) “Creditor Rights and Bank Lending in Mexico”, mimeo, University of California Berkeley.
Prepared by Vincent Moissinac (vmoissinac@imf.org). The author is indebted to the Mexican authorities at the Bank of Mexico, the National Banking and Securities Commission, and the Ministry of Finance for their helpful suggestions on an earlier draft of this paper.
During 1988–94, bank credit grew with at an annual average growth rate of 25 percent in real terms, with the result that bank financing to the private sector increased from 11 percent of GDP to 43 percent of GDP (Bank of Mexico (2003)).
Infonavit is a special public fund which finances housing development. Using a fixed share of social security contributions, it grants loans to housing developers and mortgages to workers registered with the social security system. It has been most active in low-income housing.
The study is based on perceptions of litigation attorneys practicing in each of the states.
In 2005, the FOBAPROA notes were exchanged for notes issued by IPAB, the deposit protection agency, as a result of an agreement reached in July 2004.
While guaranteed by the government, these notes were priced at a small discount to government debt. Costs were low because no capital requirement applied, and there was no need for credit screening, risk measurement, and monitoring.
Gonzales-Anaya adds that this crowding out risk was exacerbated by the open-ended nature of the government rescue operation, which encouraged banks to obtain additional bonds by using related lending to artificially generate more bad loans.
All existing empirical work on Mexican banks use the same data source, that is the quarterly bank reports of the CNBV.
Gonzales-Anaya (2002) regresses the level of bank credit on the size of banks’ deposits, the size of banks’ assets, the value of banks’ capital, an indicator of banks’ profitability, a measure of banks’ income derived from FOBAPROA notes, inflation, and Mexico’s global index of economic activity (most variables in logarithms and with appropriate lags). Using a similar specification, however, we found significant problems of residuals’ autocorrelation, suggesting bias in the regression coefficients, and we eventually opted for a different specification (see section C).
This compares to total commercial bank credit of 9.8 percent of GDP. The difference reflects credit vehicles other than lending (mainly outstanding past due loans and securities).
We calculate bank credit to the public sector as the sum of bank lending to the government sector and bank holdings of public sector securities—including securities of the federal and local governments and of IPAB, the deposit insurance agency.
