Statistical forecasting of bankruptcy of defense contractors

the long version

Declining revenues in the defense industry may eventually result in bankruptcies. This report considers whether a model could reliably predict defense contractors' bankruptcy. The authors find that models based on multivariate discriminant analysis or multiple linear regression analysis using financial data from nondefense firms are unlikely to perform well for defense firms. An approach using securities market information also appears unlikely to predict accurately. The authors conclude that subjective methods of pro forma analysis and expert financial judgment remain more promising. However, they suggest that methods unreliable for estimating bankruptcy probabilities might be reliable enough to rank companies according to these probabilities. They also recommend that those attempting to develop objective models consider using bond yields and accounting data as predictor variables.