Business failure results not only in economical loss to stockholders and financial institutions but also results in mis-allocation of social resources, and further reduces social wealth. Predicting potential business failure accurately and expediently, therefore, is crucial for the stockholders, financial institutions and the society.
The purpose of this paper could be summarized as follows.
First, finding better financial variables that describe business failure by using multivariate analysis(M])A) and logistic regression analysis after the financial crisis in Korea in 1997. As a result of empirical study, six significant financial ratios were found in the MDA and five ratios in the logistic regression model. The hit-ratio of predicting the business failure in multi-variate discriminant model, moreover, is higher than that of logistic regression model when predicting the corporate defaults.
Secondly, testing the prediction power of the KMV model. Empirical study shows that the time series distributions of Expected Default Frequency(EDF) differ significantly between default companies and non-default companies. The EDF of the default companies steeply increased many months prior to, or at least one year before, and continuingly remained at high levels. Even in the case of the non-default companies, furthermore, the EDF is shown to be sensitive depending upon the economic situations.
As a result of empirical studies on how well default potential is detected in advance, the EDF of the KMV model is shown to be very strong in terms of its prediction power. The average EDF had increasingly and steeply upclimbed from approximately nine and a half months prior to the actual arisal of default.