기업부도예측을 위한 통계모형의 비교분석

Comparative Analysis of the Statistical Models for Predicting Corporate Default

In general there are two models for predicting corporate default. One is based on accounting informations, and the other one is based on market informations using the movements of stock prices. In this paper, we are concerned about the default prediction based on the accounting informations. There are various statistical models to predict the default probability based on the accounting informations. The purpose of this paper is to compare the statistical models such as logistic regression model, probit model, decision tree model and Cox regression model to predict the default probability. We have compared the above mentioned statistical models empirically by the model assessments such as misclassification table, lift chart and ROC curve. Overall, we conclude the Cox regression model is the most powerful among the above statistical models for predicting corporate default.