ByR. Alton Gilbert
A bank supervisor's chief task is to prevent banks from failing. When a bank experiences serious problems that could lead to failure, the supervisor intervenes to prevent failure or to limit damage to the deposit insurance fund, bank customers and the community where the failed bank had offices. To be effective in preventing or limiting such damage, supervisors need to act as quickly as possible. Thus, supervisors value any tool that can give them a reliable indication of impending bank problems.
The most important way supervisors assess the condition of a bank is to conduct an on-site examination. Each of you has been visited by examiners employed by your supervisory authority, who review your institution's safety and soundness. Because examiner judgment is a valuable but scarce resource, supervisors supplement the on-site exams with off-site surveillance to help schedule and plan future examinations. Supervisors rely primarily on two analytical tools to conduct off-site surveillance: supervisory screens and economic models. Both tools help identify banks with emerging problems, as well as the sources of the problems, as early as possible.
Supervisory screens are financial ratios that, in the past, have given forewarning of safety and soundness problems. For instance, supervisors may use the ratio of net income divided by total assets as a screen. Banks with relatively low income ratios would be subject to closer scrutiny than others. Because some ratios used as screens are more important than others, supervisors draw on their experience to establish various rules of thumb for combining the information in the ratios.
Econometric models, on the other hand, combine various financial ratios to derive a single measure of bank condition. In some models, the measure is an estimate of the probability that a bank will fail. In others, the measure is an estimate of the probability that a bank's supervisory rating will be downgraded to problem status.
Prior studies report that econometric models are better predictors of emerging problems in banks than supervisory screens. Yet supervisors tend to rely on screens, probably because they are more intuitively appealing than the single numerical rating that an econometric model provides.
In the article cited below, my co-authors and I attempt to develop an econometric model that addresses supervisors' concerns about using econometric models for surveillance. Like some other models, our model is designed to forecast downgrades in supervisory ratings. Our model of downgrades, however, is more forward-looking than existing models. In addition, the model is able to pinpoint specific areas of developing weakness in a bank, instead of just providing an overall rating. For example, the model may highlight a relatively high ratio of nonperforming loans to total loans as the chief reason for a bank's high probability of being downgraded to problem status. For a different bank with the identical overall rating, the model may point to a relatively low liquidity ratio as the primary concern. In each case, subsequent examinations of these two banks could be tailored to the aspects of their operations highlighted by the model.
Our research does not suggest that screens should be dropped from the surveillance toolbox. When abrupt changes occur in the cause of bank failures and downgrades of supervisory ratings, supervisors can use their first-hand knowledge to modify screens long before models can be revised to reflect these new conditions. We conclude that both screens and models add value in off-site surveillance, but that supervisors should rely more heavily on models in the future than they have in the past.