The minimizing prediction error on corporate financial distress forecasting model: An application of dynamic distress threshold value

Abstract

In this study, we have adopted factors such as intellectual capital, financial ratios and corporate governance variables to construct a financial distress forecasting model by logistic regression. Furthermore, we employ the criteria of minimizing the sum of the two error probability in models I and II to determine the optimal threshold value, so as to increase the forecasting ability of a financial crisis forecasting model. We have taken 54 electronics companies listed in Taiwan Stock Exchange (TSE) and Over the Counter (OTC) during the periods from 2012 to 2015 to be our observation. 18 companies out of the 54 has been financially distressed in 2015. The results show that we could effectively construct a lower threshold value on the basis of the dynamic threshold value to carry out early warning (such as p = 0.32 ∼ 0.43 < p = 0.5) than those in terms of the traditional one half rule. The total error prediction probability could be reduced by 8.33% to 30.56%. In addition, the empirical evidence shows that after adding the intellectual capital variables, it could enhance the forecasting power.