CAPÍTULO I: MARCO REFERENCIAL METODOLÓGICO
1.2 Delimitación del tema
Traditionally, bank soundness has been the concern of banking supervisors who supervise these institutions within nationally designed frameworks that are built around country-specific legislation, local market and business practices and domestic economic concerns.
The Basle Committee of Banking Supervision (BCBS), operating through the regional supervisory bodies, has been attempting to develop global minimum standards and harmonise regulatory and supervisory practices that are finding wide acceptance.
60
The justification for the collection and dissemination of banking sector data arises from the increased incidence of banking crises worldwide in terms of frequency and global spread as well as the economic consequences, not least in terms of resolution costs. A 1996 study by the IMF showed that three-quarters (133 out of 181) of it’s member countries had experienced significant problems in their banking sectors in the last 15 years.101 Banking crises have always been examined in great detail and attempts made after every crisis to determine whether or not the occurrence or the timing of the crisis was evident from the build up of events and whether or not the data available either in the public domain or with the national supervisory agencies would have given an indication thereof. With the increasing globalisation of financial systems, the concern for bank soundness has gone beyond the domestic agenda to being of importance to international financial stability.
Whereas macro-economic stability has always been seen as important precondition for banking soundness, the recent events in East Asia have shown that the health of the financial sector can have serious negative externalities destabilising the rest of the economy. Despite increasing disintermediation, banks continue to be seen as the most significant component of the financial sector because of their role in the transmission of monetary policy, ‘as a repository of liquidity, as the core payments system mechanism, and as a principal source of finance… ’. 102 The latter is especially true in many emerging economies where the capital markets are less developed and the banking sector continues to comprise most of the assets of the financial sector.
The main instruments of supervision are on-site and off-site supervision and the major concerns are the solvency and liquidity of the individual institutions. Increasingly, use is being made of rating models that use risk-based data collected from banks to give an indication of their financial condition. These models may be supplemented by, or even
101 Defining, Measuring and Predicting Soundness in Bank Soundness and Macroeconomic Policy, Carl Johan Lindgren, Gillian Garcia, Matthew I. Saal, IMF, 1996
102 Are Banks Still Special? By E.A.J George, Banking Soundness and Monetary Policy, Editors Charles Enoch and John H Green, IMF, 1997.
61
serve as, early warning systems (EWS) which would typically use data on capital, asset quality, earnings, liquidity and information on the management. EWS designed by supervisors are intended to reduce the large number of financial ratios and other information to a few key indicators which would, both individually and at the level of peer groups or other aggregates, give an indication of the incipient weaknesses in individual banks or bank groups.
Data available to supervisors are often not disclosed in the public domain and other analysts interested in bank soundness often mention this as a constraint. The extent to which data are made publicly available is, of course, a function of the level of disclosure and transparency in the published accounts of banks, for which country practices vary widely. On the basis of available data, the conclusions drawn by Fraser 103 from several studies on bank failures in the USA were that:
• financial ratios are useful in predicting bank financial distress and failure; and
• problem banks can be identified with a relatively simple set of financial ratios that were easily accessible, such as profitability and capital adequacy ratios.
Overall, these studies indicated that most bank failures in the early 1980’s were associated with either ‘malfeasance, sustained low performance or excessively uncontrolled growth.’ A study by Joseph Sinkey104 that compared bank failures in 1970 and 1980’s in the US found bad loans and a shift in corporate strategy that encouraged excessive risk taking to be the main cause of failures. The importance of bad loans as an indicator is confirmed by other studies (quoted in Fraser, ibid.) conducted in the USA, using data from the 1980s, which have suggested that asset quality appeared to be the dominant variable in the assessment of the financial condition of banks. Bank failures also followed periods of excessive rapid growth.
103 See Chapter 6, Commercial Bank Performance Guide to Financial Analysis by Donald Fraser and Lyn M. Fraser, Banker Publishing Co., Rolling Meadows, Illinois.
104 Fraser & Fraser ibid.
62
4.2.1 Some Possible Macro-prudential Indicators
Supervisors, with their privileged access to data on the financial condition of banks and the other information available to them from their off and on-site supervision, may feel comfortable about being able to gauge the condition of problem banks. However, those analysts concerned with the financial system as a whole and its linkages with the rest of the economy see the need for access to at least some of the individual institution data which has traditionally been in the supervisory domain and also for published aggregated information which supervisors may or may not produce. In the development of macro-economic models to serve as early warning systems for banking crises, data requirements often are a reflection of the perceived cause or symptom of the crisis. The requirements set out by three recent and comprehensive papers on the subject give an indication of the data needs to anticipate banking crisis.
Patrick Honohan (1997)105 classifies systemic failures in the financial system into three types by the syndrome that signifies the failure. These syndromes and the data requirements for early warning of crisis for each are:
Type of Syndrome Data Requirements (I) Macro-economic
epidemics (such as unusual asset price movements, rapid growth of lending especially for property prices and stock market positions):
(a) General Indicators
Aggregate balance sheet and operating account data suggest unsound banking. The assumption is that warning signs that are used to flag individual banks at risk can also be used at aggregate level to flag systems at risk. The suggested series are:
(i) growth in aggregate lending in real terms
(ii) loan to deposit ratios and indicators of reliance on foreign borrowing
(iii) gross interest margin as a percentage to total assets and share of non-interest to interest income (diversification)
105 Banking System Failures in Developing and Transition Countries: Diagnosis and Prediction, BIS, January 1997.
63
(b) Specific indicators (i) Rate of growth of aggregate bank lending to commercial property sector
(ii) Prices of real estate property (separately for commercial and residential) and equity prices
(iii) Aggregate inflows of portfolio capital (iv) Increase in enterprise indebtedness (II) Poor management and
(i) Balance sheet and other operating account information (ii) Market based indicators which would show
(a) Extent to which banks access inter-bank markets at high rates and the inter-quartile spreads offered by banks on large deposit rates. These would reflect on the cost of funding operations.
(b) Gap between Treasury bill rates and large bank deposits (or CDs), which would represent the price of the default rate risk
(iii) Regulatory information on the loan losses of banks and their capital adequacy
(III) Endemic crisis in a Government permeated banking system
(i) Proportion of Government ownership of the banking system
(ii) Proportion of lending that is at the free discretion of banks and not subject to compulsory deposits or sectoral allocations
(iii) Other indicators of dependency such as
• Persistent borrowing from the central bank
• High level of subsidy of the banking system
• Scale of implicit or explicit taxation of the banking system
Goldstein and Turner (1996)106 have looked at factors behind banking crises and suggested eight major factors. Though the paper does not suggest the specific data needs that would be required for diagnosis and prediction, the inferred data and information requirements are as follows:
106 Banking Crises in Emerging Economies: Origins and Policy Options by Morris Goldstein & Phillip Turner, BIS, 1996.
64
• Large fluctuations in terms of trade
• Real exchange rate volatility
• Relatively low export diversification Domestic:
• Excessive credit creation
• Asset prices declines
• Volatility in real estate and equity prices Increasing liabilities with
currency/ maturity mismatches
• Growth of bank liabilities relative to size of economy and stock of international reserves
• Difference between bank assets and liabilities with reference to liquidity, maturity and currency
• Growth in bank capital and loan loss provisions to compensate for volatility in assets
Inadequate preparation for financial liberalisation
• Proxies for financial liberalisation i.e. increases in real interest rates and the money multiplier
• Trends in regulation
• Increase in risk appetite and spread of risk management practices
• Entry of new competitors Heavy Government involvement
• Share of government owned banks in total assets and total NPLs
• Policies on sectoral allocation, directed lending
• Entry barriers for private banks
• High reserve requirements
• Exposure limits for connected lending and their implementation
• Disclosure practices especially for Non-Performing Loans
• Authority given to regulators
65
Philip Davis (1999)107 is also of the view that financial crises come in different types and that the study of the specific features of these types of instability together with the associated theoretical framework can pinpoint the data needs. The three main generic types of crises as identified by him are:
(i) bank failures following loan or trading losses, which can be further divided between those that are domestic and those that are ‘linked to cross border lending and indebtedness in foreign currencies;’
(ii) those involving extreme market price volatility after a shift in expectations; and (iii) those involving collapse of market liquidity and issuance.
A further set of distinctions is made in terms of broad causality between crises caused by financial deregulation, disintermediation of flows from banks to non-banks, failure of a large institution pivotal to the system, crises linked to international debt and those with an equity market linkage. Based on both theory and the stylised patterns of financial instability, he draws out a broad list of factors considered necessary for macro-prudential surveillance as follows:
(i) flow of funds balance sheet data to track overall patterns of corporate and household sector indebtedness;
(ii) financial prices together with benchmarks;
(iii) monetary data to assess monetary growth and stance;
(iv) detailed data on banks and other institutions with maturity mismatches;
(v) qualitative data reflecting changes in financial regulation and entry barriers;
(vi) external data to pinpoint the scope, maturity and distribution of international foreign currency lending; and
(vii) overall macro-economic data to assess the current stage of the economic cycle.
Thus, the data required by users for predicting banking crisis in general are a mix of macro-economic variables, asset prices and supervisory data. The macro-economic
107 Financial Data Needs for Macro-prudential Surveillance – What are the Key Indicators of Risks to Domestic Financial Stability? by E. Philip Davis (Attached to this report)
66
variables are covered by existing frameworks of collection and dissemination as explained elsewhere in this report, but there is no internationally agreed framework or source for the latter. The supervisory data requirements fall into three broad categories:
• data from Financial Statements which are normally available in the public domain and indicators based thereon
• data on risk profile of banks which are normally available in the supervisory domain (sometimes proxied by supervisory rating models);
• information on regulatory framework and supervisory practices within the country, which places other data in context.
4.2.2 Data from Financial Statements
It is standard practice for banks to publish their balance sheet and operating account statements. However, the periodicity and disclosure practices, and the underlying accounting standards, can differ markedly across countries. Further, average and aggregate data for the system as a whole based on the financial statements are sometimes compiled and disseminated by the private sector/supervisory agencies/central banks, but often may have to be done by the users themselves. The consolidated information for banks together with their subsidiaries/associates, or for conglomerates with the whole group including the banking arms taken together may not be available in all countries. As is borne out by the data requirements mentioned earlier, and as specified in the papers presented at the Workshop, users are particularly interested in data on the following indicators:
1. Capital Adequacy indicators 2. Asset Quality indicators 3. Earnings indicators 4. Liquidity indicators
67
Capital Adequacy indicators
Interest in bank capital as an indicator of bank soundness arises from its role as the final buffer against losses that a bank may incur. Davis (1999) also notes that declining capital adequacy of financial institutions was one of the features identified with 14 of 22 episodes of international financial episodes studied and that this could serve as one of the indicators in advance of crises.
The minimum amount of capital that a bank should have to meet future losses was specified differently by national regulators until the successful harmonisation of this by the Basle Accord of 1988. In realisation of the fact that capital, at a minimum, must be commensurate with the amount of risk that a bank took, a minimum Capital to Risk-weighted Asset Ratio (CRAR) at 8% was specified by the BCBS. However, the original framework was based on providing capital cushion against credit risk, and so in 1996 the accord was modified to incorporate capital cover against market risks in key portfolios.
The accord is currently in the process of being updated to take into account some of the deficiencies that have been pointed out in the framework regarding the risk weights. For instance, the framework suggests the same low risk weight for banks and high-risk weights for corporates irrespective of their size or condition. Since inter-bank exposure is an indicator in which the markets are interested, there is a need for separate data on inter-bank exposures to assess this. This is information that is normally collected by supervisors, for both assets and liabilities, from the point of view of both liquidity and counterparty risk.
As revealed by a survey conducted by the BCBS in 1996,108 capital requirements are now almost universally accepted and most countries use the Basle-like risk weighted approach. This degree of harmonisation has made the CRAR a useful indicator for analysts in making both inter-bank and inter-country comparisons of bank strength. What would be of interest to the user would be a decomposition of capital by Tier I and Tier II/III, a risk appetite ratio (such as of risk-weighted assets to total assets) and a coverage
108 Applying Basle Standards in Developing and Transition Economies, Frederic C Musch, in Banking Soundness and Monetary Policy, IMF, 1997.
68
ratio such as capital to non-performing loans. Once again, the availability of these data is linked to disclosure practices followed in different countries and users are interested in not only knowing the different ratios but also that they have been arrived at consistently with BCBS recommendations.
Asset Quality Indicators
Amongst the supervisory data in which the most interest has been shown are the asset quality indicators— that is, non-performing loans (NPL) carried by banks on their balance sheets and the provisions held by banks against loan losses. High levels of NPLs and inadequate provisioning can severely impact the profitability, and eventually the solvency, of the banking system.
Loan losses have been identified as one of the major causes of bank failures. Davis (1999) notes that bank failures following loan losses was the main feature in 12 of the 22 major international episodes of financial instability studied between 1970 and 1998. This is probably why ‘the one area in which international convergence on regulatory matters has received the highest degree of success has been the measurement of credit risk and the definition of minimum capital requirement against this risk i.e. the risk of bank failure due to loan losses.’109 The G22 Working Group report on Strengthening the Financial Systems (1998) mentions that ‘the lack of international consensus on sound practices for loan valuation, loan loss provisioning and credit risk disclosure seriously impairs the ability of market analysts as well as regulators to understand and assess the risk inherent in a financial institutions’ activities.’
Most assets of banks tend to be loans and these loans are either not normally traded or traded in very thin markets. It is therefore difficult to associate a market value with loans.
Due to the inherently opaque nature of credit data, which are best known only to the
109 Tax Treatment of Loan Losses of Banks, Julio Escalano, in Banking Soundness and Monetary Policy, IMF, 1997.
69
management, in times of trouble there may be an incentive to incorrectly value the loans on the balance sheet and to show a lower level of NPLs then actually present.
For those attempting a cross country comparison, this problem of valuation is compounded by the fact that accounting rules regarding loan valuation and income recognition vary from country to country. Also, the possibility of setting international standards remains distant, as loan accounting and provisioning for loan losses is part of country specific legislative and fiscal frameworks.
From a supervisory viewpoint, a NPL is essentially a credit facility given by a bank which is ‘past due’ over a certain minimum period. A NPL is further classified depending on the view taken on the chances of its realisability or period of delinquency. A typical classification may be ‘special mention, sub-standard, doubtful or loss’. This definition of NPL itself may vary widely from country to country. A survey of 9 major Asian economies after the recent financial crisis 110 revealed that the minimum period for classification as an NPL varied from 3 months to 6 months with some variation for the type of facility involved i.e. term loan, residential, overdraft etc. Also, sub-categorisation is different in different countries. ‘Special Mention’ loans were not a category in four countries, classification as ‘substandard’ loans varied between 3 to 6 months, classification as ‘doubtful’ loans varied between 6 months past due to 24 month past due, and classification as a ‘loss’ asset after being a doubtful loan, from 1 to 3 years. Along with differences in the categorisation of NPLs, provisioning for NPLs also varied. The provisioning for substandard loans varied from 10% to 25%, for doubtful loans from 20%
to 100%, with convergence being achieved for loss loans at 100%. The tax treatment for loan loss provisions, too, varied widely. Three countries allowed 100% tax deductibility after the crisis. In others, the tax deductibility varied.
These variations in practice are worldwide. Loan losses may be recorded either as write-offs or as provisions against loan losses, which in turn may either be shown as liabilities in the balance sheet or netted off from the gross assets. Some countries allow partial
110 Asian Banks Analyser Supplement, Warburg Dillon Read, September 1998.
70
write-offs ('charge-offs'). The Basle accord recognises the distinction between the general and specific provision by permitting the former to be considered as akin to regulatory capital. Similarly, the tax treatment for general and specific provisions for loan losses also varies across countries, varying from ‘full deduction of general and specific
write-offs ('charge-offs'). The Basle accord recognises the distinction between the general and specific provision by permitting the former to be considered as akin to regulatory capital. Similarly, the tax treatment for general and specific provisions for loan losses also varies across countries, varying from ‘full deduction of general and specific