PREDICTION OF BANK FAILURES

Abstract

IN A FULL-EMPLOYMENT economy a firm's bankruptcy is prima facie evidence of resource misallocation. This is not to say, however, that the authorities should adopt every policy which minimizes bankruptcy. Fostering monopolies to reduce failures, for example, only increases the misallocation of resources. In one sense bankruptcy is desirable for it prevents further losses. But it is nevertheless true that a change of policy, voluntary liquidation, or any other method of eliminating losses at an early date would be preferable to bankruptcy from both a social and private viewpoint. Predicting failures, due to whatever cause, reduces the length of time losses are incurred and the misallocation of resources. The following section briefly discusses the causes of bankruptcies. Section II presents the empirical results. To screen potential failures from viable firms, we calculated regression equations in which the regressand is dichotomous and the regressors are the value and trend of selected balance sheet and income variables. The regressions differ in the number of independent variables and in the lead time of the data to failure. The regressions are significant though, as expected, their explanatory power is negatively related to the time period the data lead bankruptcies. Section III demonstrates the ability of these functions to classify solvent and closed firms and to predict the proper classification of a randomly selected holdout sample. We also illustrate how efficient regulatory policy should be formulated. While the techniques used in this paper could be applied to an analysis of bankrupt firms in any industry, our sample industry is banking.' The sample is particularly appropriate for two reasons. First, since defalcation explains a substantial proportion of bank failures in the post-World War II period, the banking industry provides an especially strict test of the predictive value of financial ratios, as will be explained more fully in Section I.2, 3 Second, federal