September 17, 1999

Early Warning Indicators of Banking Sector Distress

Michael Hutchison

Pacific Basin Notes. This series appears on an occasional basis. It is prepared under the auspices of the Center for Pacific Basin Monetary and Economic Studies within the FRBSF’s Economic Research Department.

The financial crises in Japan and East Asia have been costly; they disrupted credit channels and curtailed economic activity not only in those countries but in other parts of the world as well. Such high costs make it desirable to have some form of early warning system of impending banking sector distress. If policymakers could identify the factors that lead to a higher likelihood of banking problems, they might be able to take steps to avert them. Of course, each banking crisis usually has idiosyncratic features that make it unique. But there may also be common features associated with episodes of banking sector distress identifiable across a large number of countries and cases. This Economic Letter discusses the common features associated with banking crises and whether distress signals may be identified before serious problems arise.

Frequent and costly banking sector distress

An episode of banking distress may be associated with depositor runs on banks, the closure or merger of financial institutions, or the extension of large-scale government assistance. Typical characteristics of a financial system under stress include a significant portfolio of nonperforming assets and a limited capital base. Using these measures, Glick and Hutchison (1999) find more than 94 episodes of banking sector distress in industrial and developing economies since the mid-1970s. And they find the frequency is rising–nine crises were marked in 1975-80, 34 during 1991-95 and, by 1997, there were seven new and 29 continuing episodes. Banking crises are commonplace regardless of development status, but they occur with somewhat greater frequency in developing or emerging market economies than in industrialized economies.

Banking crises generally impose significant costs on the economy. One cost is the loss of output. The great majority of countries have suffered recessions following episodes of banking sector distress, with the cumulative output loss associated with periods of banking sector distress averaging about 10% of GDP (Hutchison and McDill 1999). These costs range widely across countries, of course. Countries such as Chile and Thailand in the early 1980s suffered severe output losses, amounting to over 25% of GDP over several years. Recent experience indicates that the ultimate costs of the East Asia financial crises also will likely be very large.

Another cost is the fiscal cost associated with government efforts to solve the problems in the financial system. Direct government funds to recapitalize banks, shore up deposit insurance funds, and so on, amounted to a significant portion of output and a huge commitment of government budget resources–usually somewhere between 6-10% of GDP. Again, the range in fiscal costs is large. Estimates of the resolution costs of Argentina’s and Chile’s banking crises in the 1980s were over 40% of GDP (Caprio and Klingebiel 1996). This same study puts the cost of resolution cost of the U.S. thrift industry bailout in the 1980s at just over 3% of GDP. The commitment of additional funds by the Japanese Diet in 1998 to shore up the deposit insurance fund and recapitalize problem banks brings the cumulative fiscal cost to date (after seven years of banking problems) to about 12% of Japan’s GDP.

Why is the banking sector vulnerable?

Much of the theory on banks’ vulnerability to adverse shocks focuses on the special role of banks in asset maturity and currency transformation in an uncertain world with asymmetric information. The literature emphasizes that institutional features of economies, such as the existence of deposit insurance and a market-determined interest rate structure, affect the profitability of banks and the incentives of bank managers to take on risk in lending operations. Adverse macroeconomic shocks, such as a fall in asset prices (leading to a fall in bank capital or the collateral underlying loans) or economic activity (leading to more delinquent loans), may push an already vulnerable banking sector into a crisis situation. Adverse economic shocks may originate domestically (for example, via recession, inflation, budget deficits, or credit slowdowns) or externally (for example, via trade or exchange rate depreciation).

Banking systems may be particularly vulnerable in times of rapid financial liberalization and greater market competition, when banks are taking on new and unfamiliar risks on both the asset and liability side of balance sheets. Weak supervisory and regulatory policies under these circumstances also may give financial institutions with low capital ratios incentives to increase their risk positions in newly competitive environments and allow them to avoid full responsibility for mistakes in monitoring and evaluating risk. Further, deficiencies in accounting, disclosure, and legal frameworks contribute to the problem because they allow financial institutions (or financial regulators) to disguise the extent of their stresses. Governments frequently have failed to identify problem institutions quickly or to take prompt corrective action when a problem arises, resulting in larger and more difficult crisis situations.

Early warning indicators

Hutchison and McDill (1999) consider a number of early warning indicators of the onset of banking sector distress. Since focusing only on episodes of banking distress would introduce an inherent bias, a broad sample of developing, emerging-market, and industrial economies were considered. Important information may be gleaned from an environment that may have appeared ripe for the onset of banking sector distress but avoided it. In this way, a control group of countries that did not experience banking distress helps isolate those indicators that actually may predict banking problems

The study identifies two groups of indicators likely to be candidates signaling the onset of banking crises: macroeconomic and institutional. The choice of variables for each group is limited, of course, by the availability of data across a large number of countries and over a sustained period. But enough data are available to draw out some of the main statistical features common to most episodes of banking sector distress. Specifically, starting with a sample of over 130 countries over the 1975-97 period, with over 90 banking crises identified, the study examined 97 countries, of which 53 had severe banking problems at some point. Since several countries had multiple occurrences over the sample period, 65 episodes of banking distress are identified for detailed statistical analysis (using the probit model regression procedure on a panel data set).

On the macroeconomic side, a large number of variables were investigated to see if they either provided an early warning signal or were contemporaneously associated with the onset of banking problems. These variables include real output growth, exchange rate changes, real credit growth, real interest rates, inflation, changes in asset prices, and the ratio of international reserves to the money stock. Each factor has been offered, in different contexts, as an indicator of the likelihood of banking distress. Hutchison and McDill, however, find that only two of the macro variables are systematically correlated with the onset of banking sector distress: declines in real output and declines in asset values (reflected by equity values). This is not surprising. The likelihood of a serious banking problem would be expected to rise when recessions occur (associated with more bankruptcies and problem loans) and when asset prices fall (indicating a decline in collateral values, which is frequently associated with a depressed real estate sector).

What is surprising is that the other macroeconomic variables generally were not associated with the onset of banking crises. Simple accounting, for example, tells us that sharp exchange rate depreciation played an important role in bringing on or exacerbating banking crises in Korea and Indonesia in 1997 and many other countries. This was because bank subsidiaries (Korea) or firms (Indonesia) had large short-term net foreign liability positions at the time of the depreciation. But in many cases, collapsing currencies did not lead to banking distress. Japan and the United States, for example, had sharply appreciating exchange rates in the early 1990s and 1980s, respectively, when their bank and thrift problems were emerging to crisis proportions.

Another surprise is that macroeconomic variables are not very reliable predictors of the onset of banking sector distress, even though recessions and falling asset prices occur simultaneously with the onset of banking sector distress. This finding contrasts with two other studies that seem to find stronger evidence of a leading indicator relationship. Various permutations of the Hutchison-McDill study–such as data transformations, lead and lag checks, and so on–hold up the basic result: it is very difficult to find any macro variable that is a reliable predictor of the onset of future banking sector distress. The broader data set used in their study–including developing, emerging-market, and industrial countries as well as those countries not experiencing severe banking problems as a control group–appears to give a less rosy picture of our ability to pick out early warning signals of impending problems.

Institutional features

Institutional features were important in identifying those countries more likely to run into banking problems at some point. Because institutional features, such as the regulatory and supervisory structures, deposit insurance, financial liberalization, and so on, do not change much over time, including these variables mainly serves to pick up cross-country differences in institutional structures that may lead to a higher probability of banking problems. In this context, the most important institutional feature was a period of financial liberalization–a period of financial liberalization frequently is associated with, and often precedes, the onset of a serious banking problem. The particular measure of financial liberalization in Hutchison and McDill is the deregulation of interest rates. In many economies, this era usually heralded liberalization not only in the pricing of financial assets, but also in the types of assets allowed and the development of more open and competitive financial markets. In other words, banks faced greater market competition at a time when they also were taking on new and unfamiliar risks on both the asset and liability side of balance sheets. The evidence suggests that this is a particularly vulnerable time for banks in managing risk that sends a clear warning signal for financial supervisors.

Avoiding banking crises

An example of the model’s (modest) predictive power is Japan. Figure 1 shows the model’s forecasts of the probability of the onset of Japanese banking distress that did in fact emerge in the early 1990s. Only one-year lagged macroeconomic variables were used in the forecast, so the reported forecasts of the probability of the onset of an episode may be interpreted as an early warning indicator of impending problems. The forecast probability of a problem rises from 2% in 1990 to 13% in 1992, roughly tracking the emergence of serious banking distress in Japan. (The model with contemporaneous variables predicts a 20% probability of banking distress in 1992.) Though 13% may not seem a high probability, it should be seen in light of the magnitude of the event: the sharp rise warns that something is amiss.

Early warning systems of impending banking sector problems are not very accurate. Nonetheless, case studies and the empirical evidence point to some policy implications. In particular, the vulnerable transition period as financial markets liberalize appears to warrant a larger commitment of resources and a more vital role for the regulatory and supervisory authorities. Financial liberalization in many countries proceeded without these safeguards. Resources devoted to the supervision of U.S. thrifts, for example, were reduced as the industry was increasingly deregulated in the early 1980s. The supervisory authorities of Japan, Sweden, and elsewhere were similarly unprepared for the enhanced banking risks associated with the era of financial liberalization. Banking problems have provided an impetus to institutional reform of the supervisory agencies of the U.S. and elsewhere. For example, recently created financial supervisory agencies in Japan and Korea–allowing greater autonomy, emphasizing rules rather than discretion in dealing with problem institutions, and forcing greater disclosure and transparency in the supervisory process–appear to be appropriate responses to these countries’ banking problems.

Michael Hutchison
Professor of Economics, U.C. Santa Cruz,
and Visiting Scholar, FRBSF


Caprio, G., and D. Klingebiel. 1996. “Bank Insolvencies: Cross-Country Experience.” Working Paper 1620, World Bank (July).

Glick, R., and M. Hutchison. 1999. “Banking and Currency Crises: How Common Are Twins?” Mimeo. Federal Reserve Bank of San Francisco (June 18).

Hutchison, M., and K. McDill. 1999. “Are All Banking Crises Alike? The Japanese Experience in International Comparisons.” FRBSF Pacific Basin Working Paper No. PB99-02. Forthcoming in Journal of the Japanese and International Economies.

Pacific Basin Notes are published occasionally by the Center for Pacific Basin Studies. Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.

Please send editorial comments and requests for reprint permission to

Research Library
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120