Chapter

III Literature Survey

Author(s):
Paul Hilbers, Alfredo Leone, Mahinder Gill, and Owen Evens
Published Date:
April 2000
Share
  • ShareShare
Show Summary Details

This chapter reviews the theoretical and empirical literature, other than work done by the IMF,39 which would support the selection of a core set of MPIs. In general, these studies look at the features of crisis-prone systems, with a view to anticipating future crisis events. By attempting to identify leading indicators of crises, rather than contemporaneous indicators of financial soundness, much of the earlier literature did not specifically review the full range of potential MPIs. More recently, the focus of many studies has shifted toward contemporaneous indicators of financial health. No consensus has yet emerged, however, from this body of work on a set of indicators that is most relevant to assessing financial soundness, or to building effective early warning systems. The statistical significance of individual indicators is often found not to be strong, and some of the studies have produced conflicting results. This may be due to differences among crises, so that specific indicators may be more or less relevant to each case.

We present below the following:

  • A brief survey of the theoretical literature on the origins of financial crises. These theoretical studies are not used to derive MPIs directly, but they underpin the empirical studies discussed in subsequent sections.
  • A review of empirical evidence on macroeconomic factors that affect the health of the financial system. This literature has focused on leading indicators of financial crises.
  • A review of empirical evidence on prudential factors used to assess financial soundness. These studies suggest additional variables that can be used as contemporaneous indicators.

The literature provides some empirical justification for the use of most of the variables that have been identified as macroeconomic and prudential indicators of financial soundness. The variety of specifications, time periods, and objectives of the empirical studies, however, makes it difficult to prioritize the indicators, or to eliminate any of them on the basis of empirical evidence. The empirical results represent work in progress, and serve only to confirm the potential usefulness of the indicators.

Determinants of Financial System Soundness40

Over the years, researchers have developed a variety of economic theories to explain soundness in financial markets. While earlier researchers relied on movements in economic fundamentals as the origin of financial distress and crisis, recent studies have highlighted the role of the information available to, and the expectations of, investors in explaining the behavior of financial markets. The rest of this section reviews the literature on banking soundness, because historically, banks have been the most important financial intermediaries.

The classic explanation for financial fragility is given by Irving Fisher (1933). He argues that fragility is closely correlated with macroeconomic cycles, and highlights, in particular, debt liquidation. A downturn triggered by over-indebtedness in the real economy requires, at some point, liquidation of this debt in order to bring the economy back to equilibrium. Debt liquidation would result in a contraction of monetary liabilities and a slowdown of velocity. These changes have several economic implications—reductions in prices, output, and market confidence, and increases in bankruptcies and unemployment. According to Fisher, therefore, financial fragility is largely based on deterioration in economic fundamentals.

Other theories highlight factors affecting depositor confidence. Diamond and Dybvig (1983) discuss the potential existence of multiple equilibria in financial markets. Banks offer a mechanism of maturity transformation whereby deposits are often lent with longer maturities. It is possible that the “good” equilibrium prevailing in normal times is not the only equilibrium, and that the banking sector finds itself in a “bank run” equilibrium. Diamond and Dybvig assume that these equilibria are a function of random events known to all agents. Therefore, a bank run occurs when agents have deposited funds into a bank at a time of low probability of a bank run, and then later observe negative events that increase their anticipation of a bank run. This study points to the importance of a high level of confidence in banks as a source of banking sector stability.41

Some studies focus on information issues. Mishkin (1996) stresses that information asymmetries between creditors and borrowers result in an adverse selection problem.42 Borrowers often have more information than banks on the quality of the investment they wish to finance. Creditors insure themselves against this source of uncertainty by lending only at the average rate between nonrisky and risky investments. It follows that borrowers with high-quality investments (i.e., high-return investments with low risk) pay interest rates that are higher than in the absence of asymmetric information, while those with low-quality investments pay lower rates. This can lead to a situation where high-quality investments are displaced by low-quality investments, causing deterioration in the overall quality of bank portfolios.

Guttentag and Herring (1984) extend the argument on asymmetric information to the possible practice of credit rationing. In the presence of uncertainty about the true return on investment, there may be a discrepancy between return expectations on the part of creditor and borrower. When the creditor’s expected return on a project is less than the return on his alternative use of funds, the borrower may be rationed. Their argument suggests that credit rationing increases with the level of uncertainty, and thus of financial vulnerability. The introduction of a deposit insurance mechanism is often seen as one way to lessen this problem. But Keeley (1990) points out that the possible existence of moral hazard problems in a deposit insurance scheme can lead financial institutions to lake more risks than they would otherwise do—borrowing at the risk-free rate (i.e., the rate on the insured deposits) and investing in riskier assets.

Recent studies point to the existence of asymmetric information in financial markets as a source of contagion of financial crises from one country to another. This is a vital concern because more countries have liberalized their markets and are now highly linked with other countries’ markets. Through this channel, negative external shocks may be directly transmitted to countries that are healthy. Kodres and Pritsker (1998), for example, develop a theoretical, multiple-asset, rational expectations model of the determinants of contagion, in which adverse effects of contagion depend on the sensitivity of the affected country to common macroeconomic risks and to the level of asymmetric information prevailing in the economy. They also point out that in the presence of hedging mechanisms, contagion may occur without common macroeconomic risks in two countries if investors hedge by reducing their overall exposure to emerging markets. This seems to have been the experience of many Asian countries.

Davis (1996) argues that institutional investors may contribute to financial fragility because of principal-agent problems in the relationship between fund managers and their clients-that is, fund managers may not act to maximize the client’s profits without appropriate supervision. He argues that one way to reduce the principal-agent problem is to introduce more frequent monitoring and performance evaluation systems. From the perspective of the fund manager, when strict monitoring and evaluation are in place, one way to show the quality of his management is to imitate others (the so-called herd behavior) rather than trust his own judgment, since the initial financial asset information available to him often contains elements of uncertainty. In this way then, Scharfstein and Stein (1990) says mimicking other investors is likely to maintain the manager’s reputation by reducing his risk of underperformance relative to the average for the market. Therefore, an event perceived as adverse by just one investor may result in large movements in financial asset prices.

These theoretical studies have been the point of departure for much of the empirical work discussed below. Table 2 summarizes the main indicators identified in the empirical literature.

Table 2.Macroprudential Indicators in a Selection of Recent Studies
Studies by authors1C-KF-RS-T-VHGH-P-BB-GB-PE-LE-RFH-PKK-L-RR-SDK-D
Year of publication199619961996199719971999199919981998199819981999199819981998
Focus of studyBCCBBCCCBCBB&CCCB
B = banking crisis
C = currency crisis
Aggregated microprudential indicators
Foreign exchange exposure
Sectoral credit concentration
Nonperforming loans
Aggregate risk-based capital ratio
Central bank credits to financial
institutions2
Segmentation
Ratio of deposits to M2 (or GDP)
Stock exchange prices
Aggregate average returns
Macroeconomic indicators
Lending booms (e.g., credit/GDP)
Asset price booms
Contagion effects
External deficits
Aggregated growth rate
Volatility of interest and exchange rates
Terms of trade
Level of domestic interest rates
Exchange rate misalignments
Government recourse to banking system
Volatility in inflation

Caprio and Klingebiel (C-K);Frankel and Rose (F-R)Sachs. Tornell, and Velasco (S-T-V): Honohan (H):Gonzárez-HermosilloH Pazarbasioğlu. and Billings (GH-P-B); Baig and Goldfajn (B-G); Berg and Patillo (B-P). Esquivel and Larrain (E-L);Eichengreen and Rose (E-R); Fratzscher (F); Hardy and Pazarbasiog’lu (H-P); Kammsky (K); Kaminsky, Uzondo. and Reinhart (K-L-R); Radelet and Sachs (R-S);and Demirguc-Kunt and Deiragiache (DK-D)

As a proportion of their capital or liabilities.

Caprio and Klingebiel (C-K);Frankel and Rose (F-R)Sachs. Tornell, and Velasco (S-T-V): Honohan (H):Gonzárez-HermosilloH Pazarbasioğlu. and Billings (GH-P-B); Baig and Goldfajn (B-G); Berg and Patillo (B-P). Esquivel and Larrain (E-L);Eichengreen and Rose (E-R); Fratzscher (F); Hardy and Pazarbasiog’lu (H-P); Kammsky (K); Kaminsky, Uzondo. and Reinhart (K-L-R); Radelet and Sachs (R-S);and Demirguc-Kunt and Deiragiache (DK-D)

As a proportion of their capital or liabilities.

Studies of Macroeconomic Variables

Several studies of financial problems appeared in the wake of the Mexican crisis in 1994, and before the emergence of the Asian crisis in 1997.43 These studies investigate the vulnerability of financial institutions in the face of exogenous shocks. Financial intermediaries are generally highly leveraged, engage in maturity transformation, transact in markets with asymmetric information, and are subject to moral hazard through explicit or implicit deposit insurance. Sources of financial fragility explored in the studies include a falling growth rate, deterioration in the balance of payments, high inflation, volatile exchange rates, surges in stock market activity and prices, credit booms, weakening performance of export sectors, and deterioration in the terms of trade. In addition, these studies highlight nonquantifiable indicators of financial fragility, such as deficient banking supervision, inadequate instruments of monetary control, overly generous deposit insurance, inadequacies in the operation of the legal system, overexposure in international financial markets, lack of adequate accounting standards and practices, insufficient financial disclosure, and perverse incentive structures.

The Asian crisis has provoked a new wave of financial sector studies, which confirm that macroeconomic shocks to output, exports, prices and the terms of trade, asset price booms, and inappropriate monetary and exchange rate policies, all result in financial pressure and contribute to crises in financial systems that are inherently fragile.44 In addition, this research points to the destabilizing effects of market overreaction, the feedback effects of crises that weaken corporate balance sheets, and the impact of unexpected shocks, such as the rapid change in the yen-dollar exchange rate and the swift emergence of new competition from Mexico (in the wake of the deep devaluation in 1994–95) and from China. These are important factors that must be evaluated in cases of economic instability, Except for the impact of third-party exchange rate changes on the domestic economy, however, these developments generally do not appear in advance of financial weakening and, therefore, do not offer additional early warning indicators of financial health.

The contagion of financial crises from one country to another has been the focus of several empirical studies.45 The factors that appear to expose a country’s financial system to contagion include close correlation in the past behavior of currency and equity markets, export and import ties (or competition in trade), cross-market banking links, low levels of foreign reserves, the extent of exchange rate overvaluation, and the inherent weakness of the financial system. In addition, Kaminsky and Reinhart (1998) find evidence that sharing a common creditor with a crisis country creates a high risk of contagion.

Given that currency and financial crises often occur simultaneously, the factors underlying currency crises have the potential to contribute to an assessment of the health of financial institutions. Causation between exchange rates and financial variables, however, may go in either or both directions. This relationship has been the subject of several studies, including Dornbusch, Goldfajn, and Valdes (1995); Kaminsky and Reinhart 1999Kaminsky (1999); and Kaminsky, Lizondo, and Reinhart (1998). Their results suggest that exchange rate crises provoke financial crises when the banking sector is vulnerable, that is, when the impact of a devaluation on the quality of bank assets is large enough to wipe out the banks’ net worth. Therefore, simulations (stress tests) of the impact of a devaluation of various magnitudes on banks’ capital adequacy can be useful as an additional indicator of financial robustness. Kaminsky and Reinhart (1999), however, point out that in about half of the crises in the 1980s and early 1990s that they examined, financial crises preceded currency crises.

Studies of Aggregated Microprudential Indicators

Much of the earlier literature on aggregated microprudential indicators follows the categorization of the CAMELS rating—see Altman (1968), Sinkey (1978), and Thomson (1991). This portfolio-based assessment method is broadly consistent with the list of MPIs identified in Section II. These variables are used in empirical research less frequently than macroeconomic indicators, due to the availability of higher frequency data for the latter. A classic study by Altman (1968) uses the so-called Z-score model, which is based on several financial ratios capturing asset quality, earnings performance, and liquidity, but this analysis is at the level of the individual firm.46

More recent literature—including Frankel and Rose (1996); Sachs, Tornell, and Velasco (1996); and Honohan (1997)—emphasizes the important role of foreign borrowing, particularly short-term liabilities denominated in foreign currency, to measure the degree of exposure to currency and inflation risks. Recent literature also focuses on the level of nonperforming loans-such as González-Hermosillo, Pazarbasioğlu and Billings (1997). González-Hermosillo (1999) shows empirical evidence that the CAMELS-type assessment is statistically significant only if nonperforming loans and capital adequacy are simultaneously considered.47 This is consistent with theoretical explanations for the eruption of the Asian financial crisis, which posit financial institutions’ weaknesses as a major cause of the crisis.

Other indicators to capture financial vulnerability include a measure of segmentation (often proxied by an interbank interest rate differential), the deposits-to-M2 ratio, and aggregate stock indices. In surveying literature on these indicators, Demirgüc-Kunt and Detragiache (1999) point to criticisms on the use of CAMELS-based criteria to measure bank strength. A comprehensive study by Kaminsky, Lizondo, and Reinhart (1998) concludes that these indicators are less able to explain currency crises than is exchange rate misalignment.

Many of these studies use logit/probit models to capture banking fragility or to differentiate healthy banks from unhealthy banks. But their ability to detect future events in the out-of-sample forecasting context is limited. Lane, Looney, and Wansley (1986) and Whalen (1991) use the Cox proportional hazards model, which is capable of providing information on the expected time of failure, but the overall conclusion on the poor performance of the CAMELS-type model remains unchanged. Consequently, González-Hermosillo (1999) combines both micro- and macro-factors in explaining banking fragility, and concludes that the introduction of macroeconomic variables significantly improves the explanatory power of models based on microprudenlial indicators only.

Indicators of capital adequacy provide important information about financial fragility. Minimum standards for risk-weighted capital adequacy have been agreed to by the Basel Committee on Banking Supervision, but there remain reservations about these standards, which are currently under review.48 The literature points to some of these limitations. For example, the Basel Committee on Banking Supervision (BIS (1999b)) shows that the improvement in the ratio for Group of Ten (G–10) countries from 9.3 in 1988 to 11.2 percent in 1996 did not reflect a significant improvement in the overall health of the system.49 Proposals currently under discussion at the Basel Committee would supplement capital adequacy measures with supervisory reviews that could require higher levels of capitalization and use different measures of risk exposure, such as the increasingly popular Value-at-Risk (VaR) models.50

    Other Resources Citing This Publication