This second article in Part 12 deals with banking disasters. Author Richard Herring starts by wondering why the US banking and finance scene was so prone to disasters during the 1980s.
He suggests that one of the main reasons was disaster myopia a tendency to underestimate the probability of infrequent shocks (such as bank failure). This was coupled with institutional factors such as accounting practices, managerial bonuses, the tendency of bankers to herd (all follow the same policy) and inactivity by regulators. Herring also points out that the very safety nets designed to cushion bank failures can actually help bring them about by increasing the incentives for banking executives to take excessive risk, especially in go-for-broke behaviour.
He concludes that the lessons to be drawn from the US banking disasters of the 1980s is that US banks should be allowed greater diversity and that banks and other institutions should not be allowed to operate without risking large amounts of shareholder funds.
More From This Section
As recently as 1990 it was reasonable to ask why the US, alone among advanced industrial economies, had been plagued by banking disasters.
The eighties was indeed a troubled era for depository institutions in the US.
Over the decade more of them failed in the US than at any time since the Depression. By 1995, however, banks and Saving and Loan institutions (S&Ls) were earning record profits and had greatly strengthened their capital positions. Moreover, by then it was clear that banking disasters were not confined to the US. Serious banking problems had swept over most of Scandinavia, France and Japan.
Disasters in the banking system are often preceded by a lengthy period of relative calm. In the US during the first 20 years following the second world war, for example, bank failures were extremely rare and usually the result of idiosyncratic circumstances, often involving fraud. (See Figure 1 showing bank and S&L failures since 1934.)
After a shock to the banking system, it becomes clear that the banks damaged by the shock had assumed excessive insolvency exposure.
It is important, however, to focus on the period before the shock occurs and pose the question: why do banks become increasingly vulnerable to shock?
Consider two possible answers. First, managers may underestimate the probability that a shock will occur. Second, they may perceive the probability of a shock correctly but willingly assume greater insolvency exposure because they expect it to be profitable.
In earlier work with Jack Guttentag, I argued that underestimation of such shocks may be a plausible consequence of the way in which decisions are made within an environment of uncertainty.
Our ability to estimate the probability of a shock depends on two key factors.
First is the frequency with which the shock occurs relative to the frequency of changes in the underlying causal structure. If the structure changes every time a shock occurs, then events do not generate useful evidence regarding probabilities.
On the other hand, if the shock occurs many times while the structure is stable, probabilities may be estimated with considerable confidence, such as default rates on portfolios of car loans and credit card receivables.
In general, high-frequency shocks are not a source of insolvency exposure for banks. Banks have both the knowledge and incentive to price high-frequency shocks properly and to make adequate provisions to serve as a buffer against loss.
Second, our ability to predict the probability of a shock depends on our understanding of the causal structure. Banks often lack the knowledge to price low-frequency shocks with uncertain probabilities and are likely to rely on subjective probability estimates.
Disaster myopia hypothesis
Researchers in cognitive psychology have found that decision-makers tend to formulate subjective probabilities on the basis of the availability heuristic, that is, the ease with which the decision maker can imagine that the event will occur.
Since the ease of imagining an event is highly correlated with the frequency that the event occurs, this rule of thumb provides a reasonably accurate, although imperfect, estimate of high-frequency events.
At some point, the tendency to underestimate shock probabilities is exacerbated by the threshold heuristic. This is the rule of thumb by which busy decision makers allocate their scarcest resource: managerial attention.
The availability and threshold heuristics together cause disaster myopia, the tendency over time to underestimate the probability of low-frequency shocks.
Disaster myopia can lead banks to become more vulnerable to a disaster without anyone having made a decision to increase insolvency exposure. It is likely to be shared by a large number of banks because uncertainty may also be conducive to herding, in which banks take on largely similar exposures to some shock.
Being part of a group provides an apparent vindication of the individual bankers judgment and some defence against recriminations if the shock occurs.
John Maynard Keynes perceived this clearly, writing in 1931:
A sound banker, alas, is not one who foresees danger and avoids it, but one who, when he is ruined, is ruined in a conventional way along with his fellows so that no one can really blame him.
Perhaps even more important, the banker knows that the supervisory authorities cannot terminate all the banks or discipline them harshly.
Indeed the authorities may be obliged to soften the impact of the shock on individual banks in order to protect the banking system.
Disaster myopia may also afflict the regulators and supervisors who should constrain the increasing vulnerability of banks. Regulators and supervisors, after all, are likely to be subject to the same perceptual biases as bankers.
Institutional factors
The tendency toward disaster myopia may be reinforced by several institutional factors.
Managerial accounting systems, for example, may inadvertently favour activities subject to low-frequency shocks. Although generally accepted accounting principles are extremely helpful in monitoring, pricing and provisioning for high-frequency shocks, they are not very useful in controlling exposure to a low-frequency hazard because the shock occurs so infrequently that it will not be captured in the usual reporting period.
Moreover, in the absence of appropriate provisions for potential losses, an activity subject to low-probability shocks will appear misleadingly profitable.
The illusion of high profitability creates additional problems. To the extent that salaries and bonuses are based on reported short-term profits without adjustment for reserves against shocks, the line officers who are in the best position to assess such dangers will be rewarded for disregarding them.
The appearance of high profitability may also impede the effectiveness of the supervisory authorities, which find it very difficult to discipline banks that appear to be highly profitable.
In addition, competition may interact with disaster myopia in two related ways to increase vulnerability.
First, competitive markets make it impossible for banks that are not disaster myopic to price transactions as if there were a finite probability of a big shock when banks and other competitors that are disaster myopic price them as if that probability were zero.
Second, if banks are apparently earning returns above the competitive level, equally myopic banks will be encouraged to enter the market, thereby eroding returns. Banks may respond by increasing their leverage to protect their returns on equity, accelerating the process through which banks make themselves more vulnerable to a major shock.
Disaster myopia in the US
Trends that are consistent with, and conditions that are conducive to, disaster myopia are apparent in the evolution of the S&L industry over the 1970s. The sector prospered in the post-second world war era by making fixed-rate, long-term, residential mortgages backed by short-term savings deposits.
The S&Ls charter gave them a structural asset/ liability mismatch, with the duration of their assets greatly exceeding that of their liabilities. For more than 20 years following the war, their exposure to interest rate risk generated substantial profits. The yield curve was generally upward sloping so that interest rates on 30-year residential mortgages exceeded short-term rates on deposits.
A series of increasingly severe interest-rate spikes beginning in 1966, however, led to temporary yield curve inversions and an erosion of the net worth of S&Ls. More important, this erosion was greatly exacerbated by the upward trend in interest rates. This forced S&Ls to reprice their liabilities at higher interest rates more rapidly than they could reprice their assets. During the 1970s interest rates became more volatile and the upward trend accelerated.
Instead of reducing their vulnerability to an interest rate shock or reinforcing their capacity to withstand it by increasing their capital, S&Ls let the book value of their capital positions decline steadily throughout the 1970s.
Of course, the economic value of their capital positions declined even more sharply since the book value measures did not reflect the decline in the market values of mortgage portfolios.
The tendency to disaster myopia was undoubtedly reinforced by the accounting system. It took no account of the impact on balance sheets of changing market values in response to rising interest rates and created no reserves to deal with an interest rate shock.
This obscured the erosion of economic net worth from owners, managers, creditors and regulators. It led to an overstatement of profits that sustained the salaries and bonuses of managers who might otherwise have reduced exposures.
Increasing competition from other financial institutions may have hastened the process of increasing vulnerability. S&Ls were subject to disintermediation as depositors shifted to money market mutual funds in search of higher returns and as banks and insurance companies became more active in the mortgage market.
In the face of shrinking margins, it may have seemed tempting to increase leverage in order to maintain returns on equity. Any residual doubts about the prudence of such a policy may have been eased by herding.
Because commercial banks were better diversified and had a much more balanced asset/liability mix, they were less exposed to the interest rate shock and generally weathered it with less difficulty than the S&Ls.
This was not true, however, of some important bank borrowers. When borrowers found that they could not meet sharply higher interest payment obligations, banks were confronted with higher credit and transfer risks.
Financial institutions differ substantially from most other companies because they are regulated and protected by a safety net. This fact makes disaster myopia in banks a public policy concern. In addition, regulation and the safety net may lead to increased vulnerability to disaster.
US depository institutions are subject to relatively tight restrictions on what they may do and where they may do it. In general, these restrictions constrain their ability to diversify and make them more accident prone than less-constrained counterparts.
The vulnerability of S&Ls to an interest rate shock was implicit in their charter. Moreover, their enforced specialisation in residential mortgage lending left them vulnerable to the vicissitudes of just one industry. Banks were also subject to activity restrictions such as prohibitions on underwriting corporate securities and selling insurance, which impeded their ability to evolve as the needs of their customers changed.
In addition, both banks and S&Ls were subject to geographic restrictions that limited their ability to diversify deposits and loan portfolios and made them more vulnerable to local economic conditions. These restrictions are implicated in a large proportion of bank failures.
Although rigid constraints on the activity and branching powers of depository institutions undoubtedly made them more vulnerable to shocks, this need not have resulted in the enormous costs borne by US taxpayers and prudently managed institutions. The magnitude of the losses is directly attributable to the design and implementation of the safety net.
Virtually every industrial country has erected a safety net for depository institutions to guard against a banking disaster that might ignite a financial crisis. The safety net includes the chartering function, prudential supervision, authority to terminate a failing institution, deposit insurance, lender of last resort and policy control over bank reserves.
Moral hazard and the safety net
In most industrialised countries, the safety net has been successful in preventing a disaster at one bank from escalating into a financial crisis that damages the nations economy.
In an important sense, however, the safety net has been too successful in protecting depositors. Because depositors are confident they will be protected against loss, they have little or no incentive to monitor and discipline risk-taking by their banks. This is the classic moral hazard problem in which insurance may undermine the incentive for depositors to be concerned with preventing the insured risk in this case, the risk of insolvency from occurring.
In addition, depositors at larger banks in the US have benefited from implicit deposit insurance that arose from the way in which the other two components of the safety net were deployed. First, the Federal Reserve, as lender of last resort, lent routinely to banks long after they became insolvent rather than lending solely to solvent, but illiquid, banks.
This gave anxious creditors who were not covered by explicit deposit insurance the opportunity to withdraw their deposits before a bank was terminated.
Second, the termination authorities generally did not intervene to terminate a bank promptly, that is before it became insolvent. Instead they usually delayed until the bank was deeply insolvent. Then, rather than liquidating the bank and imposing loss on uninsured depositors and creditors, the authorities provided assistance.
They kept the bank open or arranged a purchase and assumption transaction in which all commitments to uninsured depositors and other creditors were honoured by the acquiring bank.
This policy applied to all banks with more than $500 million in assets and gave credibility to the notion that such banks were too big to fail.
In sum, the safety net increased incentives for bank and S&L executives to take excessive risks, which then put greater pressure on the supervisory function to constrain risk-taking.
But at the same time, it reduced incentives for regulators and supervisors to take prompt corrective measures. Without the market pressure of a bank run, the supervisory authorities were free to engage in forbearance, which often exacerbated bank disasters.
Go-for-broke behaviour
Managers, who may exercise a constraining influence while an institution has positive net worth, find that when capital is exhausted their main hope for survival is in very high-risk ventures. Although some institutions in the 1980s rode out the interest rate cycle and were restored to healthy condition, many others played go-for-broke.
At mid-year 1983 almost all insolvencies were attributable to the interest rate shock and could have been resolved for about $25 billion. But authorities chose to forbear. S&Ls played go-for-broke and many of the gambles failed. The result was an increase in losses due to asset quality problems that totalled about $140 billion.
The rolling regional recessions that caused losses at S&Ls damaged smaller banks as well. Banks regulators practised forbearance just as their S&L counterparts did. The number of failures and losses increased sharply and, by 1990, led to concerns that insurance reserves were inadequate to pay off insured deposits at all banks expected to fail.
The prospect of yet another taxpayer bail-out of financial institutions led to the adoption of fundamental reforms in the Federal Deposit Insurance Improvement Act (FDICIA).
Just as the act was being implemented, banks benefited from favourable macro-economic developments. The overall level of interest rates fell and the yield curve took on a much steeper positive slope, giving most banks a more favourable interest rate spread.
The economy began to recover from recession and commercial and residential real estate markets rebounded. By the mid-1990s US banks had fully recovered and are now generally in stronger financial condition that at any time since the 1960s.
Lessons from the US
What can we learn from the US experience about preventing future disasters? The remedy depends on the diagnosis and as we have seen several pathogens are suspected. The traditional bank-supervisory process is not well designed to deal with disaster myopia and exposure to large shocks of unknown probability.
While the identification of weak banks is useful for managing crises, it is inadequate for their prevention. To prevent crises, the central concern of prudential supervision should be to identify banks that are becoming heavily exposed to a major shock.
The identification of emerging sources of systemic vulnerability involves a continuing interplay between assessing the exposure of banks to particular shocks, and assessing the probability that a particular shock may occur.
Once exposure data measuring vulnerab-ility are collected, the supervisory authorities have three basic policy options.
First, they can return the information to the individual banks, perhaps accompanied by supervisory commentary, but permit each bank to determine whether its exposure is prudent. This is a measure and confront approach. It has the merit of forcing the bank to face the issue of whether its exposure is prudent.
To the extent that excessive exposure is inadvertent the result of inattention or poor communications among operating officers, senior management and directors this approach may be sufficient to prevent excessive vulnerability.
But the bank may already be aware of its exposure to the shock, having made a deliberate choice to accept a larger exposure in the belief that shock probabilities are low and in anticipation of higher expected profits.
Indeed, the bank may take comfort in the knowledge that its peers are equally exposed. Under such conditions, the policy of measure and confront may prove wholly inadequate to prevent systemic vulnerability to a major shock and may actually facilitate herding behaviour.
Second, the supervisory agency may release exposure data to the public in the hope that markets will discipline banks regarded as excessively exposed.
The disclosure of exposure data, however, may reveal proprietary information, abrogate confidential relationships and, if disclosure occurs only after the shock, undermine confidence.
Moreover, it may not be sufficient to constrain exposures because creditors may also suffer from disaster myopia or because they anticipate being protected by the safety net in the event of trouble.
Third, the supervisor can specify stress tests that banks should be prepared to meet. This would constitute specifying the minimum shock magnitude that a bank should be able to sustain.
Banks using the internal models approach to market risk sanctioned by the Basle Committee, for example, must conduct regular stress tests to gauge their vulnerability to low-probability events in several types of risks.
The basic problem with this approach is that judgments about whether vulnerability to a particular shock is excessive and what minimum shock magnitudes should be are inherently subjective.
Moreover, if the supervisory authorities suffer from disaster myopia to the same degree as bankers, they may not identify the appropriate shocks to stress test.
From a regulatory perspective, perhaps the most important reform to counter vulnerability to disaster myopia is to reduce regulatory restrictions on diversification.
Liberalisation of powers for solvent, well-capitalised banks should help reduce vulnerability to future shocks. The greater the degree of diversification across activities and geographic regions, the lower the vulnerability to any particular shock, even if disaster myopia cannot be corrected.
In addition to these direct supervisory measures to counter disaster myopia, it is also important to deal with factors that encourage it. Accounting practices that mask deteriorat-ion in the market value of exposures to hazards are a fundamental source of vulnerability.
They impede the ability of managers, owners, creditors and supervisors to monitor insolvency exposure and they may also make a risky activity appear misleadingly profitable.
In contrast to the measures for countering disaster myopia, the measures for countering moral hazard are quite straightforward. The first principle is to refrain from providing full protection for all creditors especially large creditors such as corporations, other banks and institutional investors.
This is largely a matter of ending implicit deposit insurance. A policy of too big to fail places the entire burden of monitoring risk-taking on the supervisory authorities.
The other channel through which implicit insurance is extended is lender-of-last-resort assistance to insolvent banks. FDICIA attempts to deter such practices by depriving the Federal Reserve Bank of the protection of collateral for extended advances to banks near insolvency, except when such advances are necessary to prevent a severe adverse effect on ... the national economy.
The second principle to counter moral hazard is to prevent banks from operating without substantial amounts of shareholders funds at risk.
One clear lesson from the S&L debacle is that losses surge as institutions find their capital depleted and shareholders and managers are tempted to play go-for-broke.
FDICIA tries to reduce the scope for forbearance by replacing supervisory discretion with rules designed to stimulate prompt corrective action as soon as a banks capital position deteriorates.
Although FDICIA calls for accounting reforms that would move regulatory measures of capital closer to market values, little progress has been made.
This is a crucial omission. The rules for prompt corrective action will be effective only to the extent they capture the deterioration in the economic value of capital.
Without more transparent accounting practices, it will be difficult for supervisors to monitor the moral hazard incentives of banks and difficult for taxpayers to monitor the performance of their agents, the supervisors.
Although this article has focused on illustrations of disaster myopia, moral hazard and go-for-broke behaviour drawn from the S&L disaster in the US, examples could have been readily based on Scandinavian and Spanish banking crises, banking problems in France and Italy, and the ongoing financial debacle in Japan.
Looking beyond the rich industrial nations, indications of disaster myopia and go-for-broke behaviour can be discerned in many of the developing countries that have sustained banking crises over the past 20 years.
History may not repeat itself, but with regard to banking disasters, as Mark Twain once observed, it almost certainly rhymes. Failure to learn from these painful experiences can be costly.
At least a dozen countries over the past 20 years have sustained banking losses or government bailouts of the banking sector amounting to 10 per cent or more of gross domestic product.