Does recasted data improve the predictive accuracy of bankruptcy prediction models?

This paper examines conventional bankruptcy prediction models under the thesis that such models perform better when applied to adjusted financial statement data, rather than to reported data. This hypothesis is tested by revisiting already existing bankruptcy models as well as developing two new models for predicting imminent corporate failures. The result of the study reveals that financial adjustments are not necessary for conventional bankruptcy prediction models, whereas such adjustments lead to improved predictive accuracy for our models. So even though the necessity of using recasted financials in other areas of finance is generally accepted, we find that this necessity is not as apparent within the field of bankruptcy prediction