A note on failure prediction models as aids to the credit evaluation procedure in a South African Commercial Bank

Abstract

Errors by bank lending officers in assessing a loan application are costly to the bank. The literature suggests that the credit evaluation process can be enhanced by the use of failure prediction models.

A sample of companies identified as problematic by a South African bank, was selected for study. Three failure prediction models, one univariate and two multivariate, were used to test whether the models were able to signal distress earlier than the bank was able to. The univariate and one multivariate model signalled distress on a significant number of occasions before the bank did. It can therefore be concluded that the use of selected models in conjunction with the bank’s usual procedures can enhance the credit evaluation process.

Keywords:

• Failure prediction
• credit evaluation