Optimization of the cut-off points in bankruptcy prediction in Turkish companies

Prediction of bankruptcy using financial ratios is a popular area of research for academicians as well as the managers, creditors and investors. Most of the research about bankruptcy prediction aim to maximize the number of companies correctly classified, without considering the costs related with the errors of misclassification. The purpose of this study is to minimize the expected costs of bankruptcy by considering different costs in the Type I and Type II error regions. The research question - if the optimal cut-off point considering to minimize the costs in Type I and Type II Error Regions in Bankruptcy Prediction is equal to optimal cut-off point considering to minimize the number of companies in Type I and Type II Error Regions- is answered. Non-financial companies listed in Borsa Istanbul between 1998 to 2015 are categorized according to their bankruptcy status and a logistic prediction model is constructed. After the validation of the logistic model, the effect of the cut-off point change on the expected cost of misclassification is measured. The logistic model is proved to be successful in predicting the bankruptcy 2 years before it occurred. And it is also observed that cut-off point is important in minimizing the expected costs of bankruptcy and it is different than the cut-off point to maximize the number of correctly classified companies.