On bankruptcy information systems

Abstract

Bankruptcy prediction information systems provide a noisy signal. Thus, these information systems are not perfectly reliable. This paper investigates a means for including the reliability of the signal into a basic bankruptcy prediction model. Then the implications of integrating reliability into the model are explored. The results indicate that the model is highly sensitive to reliability. As a result, the bankruptcy decision models should include information system reliability or else decision outcomes can be suboptimal (for example, loans made when they should not be or not made when they should be). The results also indicate that some heuristics that may be used by decision makers to account for reliability are likely to lead to quite misleading results.