Bankruptcy Prediction Using Quarterly Financial Statement Data

Abstract

The purposes of this study were to assess the usefulness of quarterly data for predicting bankruptcy and to determine if the earlier prediction by quarterly bankruptcy models can be obtained without the sacrifice of accuracy achieved by annual bankruptcy models. A sample of 40 public firms entering bankruptcy from 1977 to 1983 was matched on the basis of fiscal year, industry, and asset size with 40 nonbankrupt firms. Quarterly financial data were obtained from the firms' 10-Q reports filed with the Securities and Exchange Commission (SEC), whereas annual data were obtained from the 10-K reports.

Multiple discriminant analysis was used to derive quarterly bankruptcy prediction models for each of the three quarters before and after the last annual period preceding bankruptcy and for the last annual period preceding bankruptcy. Twenty-four financial ratios that were identified in previous studies as being useful for bankruptcy prediction were selected as the independent variables in the stepwise discriminant process.

The classification accuracy, using alternative assumptions regarding prior probability of bankruptcy and cost of misclassification and the statistical significance of the quarterly models for each of the six quarters tested, indicated that quarterly data are useful for predicting bankruptcy. There was no statistical evidence to suggest that the classification accuracy of the annual model was superior to that of the quarterly model. This finding suggests that more timely bankruptcy predictions can be provided to investors, creditors, and auditors by quarterly models without the loss of accuracy provided by annual models.