Corporate Bankruptcy Prognosis : An Attempt at a Combined Prediction of the Bankruptcy Event and Time Interval of its Occurrence

We consider the problem of predicting a firm's bankruptcy with a simultaneous assessment of the time interval within which the bankruptcy must occur. We carry out anew a comparative statistical analysis of various financial indices and identify four relatively independent factors, which have remarkable potential in bankruptcy prognosis. Two of these characterize the quantity and quality of corporate debt, while the other two characterize the ability to pay the debt. We consider the prognostic problem as a multi-alternative problem of Statistical Decision Theory. The synthesized prognostic rules involve calculation of posterior probabilities of firm bankruptcy within the first, second, third etc. year, while a special approximation of likelihood functions avoids restrictive assumptions of multivariate normality of the prognostic factors. We have found that optimal rules for prediction a firm's bankruptcy within the next year, within the next two years, or say, within the second or the fifth year must be different. This means that correct formulation of the prognostic problem must specify the time interval of the prognosis. We assess the efficiency of the variants of the prognostic rules developed in the paper and show that in comparable conditions they are more efficient than Altman's familiar Z-score rule and tested logistic rules