기업도산예측모형 설정에 관한 실증적 연구

Since the Asian financial crisis in 1997, the importance of measuring each firm's solvency and predicting the possibility of bankruptcy has been increased in the view of well-structured national economy and the valuation of each individual firm. Bankruptcy prediction techniques, which has been a hot issue for a long period among investing companies, is getting more sophisticated and approached in various ways. From late 1960s - when researchers started to test their bankruptcy prediction models - to 1970s, the main stream of prediction technique was multiple discriminant analysis and from early 1980s to late 1980s, the major used tool was logistic regression. Currently, artificial neural network is introduced and widely used for default prediction in the industry. Therefore, many different methods has been introduced and their effectiveness has been tested by the researchers; however, optimal model selection is difficult, as the used data, processes and definitions of bankruptcy vary. The study evaluated prediction accuracies of discriminant analysis.