Prior Probability Disclosure and Loan Officers' Judgments: Some Evidence of the Impact

Abstract

The findings of recent studies on commercial loan officers' predictive accuracy of corporate failure have been inconsistent. Using only limited sets of data consisting primarily of financial ratio profiles, subjects in studies by Libby [1975] and Zimmer [1980] predicted corporate failure for samples of real-life firms with 74 and 77 percent accuracy, respectively. Participants in studies by Abdel-khalik and El-Sheshai [1980] and Casey [1980], however, achieved only 62 and 57 percent accuracy, respectively.' In earlier studies, Abdel-khalik [1973] and Kennedy [1975] found that approximately one-half of the subjects' probability judgments were greater than 50 percent for the incorrect alternative. Commenting on the poorer performance of loan officers in Casey's study relative to his own and Libby's, Zimmer remarked that the inconsistency appears to be attributable to the fact that Casey declined to reset subjects' priors [1980, p. 635]. Libby and Lewis [in press] also believe Zimmer's results support the notion that different performance results can occur depending on whether subjects know the probabilities of failure for their samples of firms. The current study was undertaken to provide some empirical evidence