시장 위험 관리를 위한 조건부 왜도 모형의 유용성에 관한 실증 연구

This paper examined the conditional skewness model as a tool of the market risk management in Korean stock return data. First, it specified the estimation problems and issues about the dynamics settings of 3rd and 4th moments. Second, it estimated the several different type models and selected the most fitted model for performance testing. In order to choose the model, not only the conventional parametric approach but also non-parametric approach (run test) was used. It suggested that the run test was modified for the conditional skewness model. For empirical performance test of the conditional skewness model comparing to the conventional models of stock return, violation numbers of VaR was used as a measure. The VaR model was calibrated daily for testing the power of out-of-sample periods. The empirical results show that there is the conditional skewness in KOPSI index and the Fama-French style portfolio as a sub-portfolio of KOSPI. Fama-French style portfolio is more adapted to the conditional skewness model than KOSPI index. The results of the VaR one-step-ahead performance show that the conditional skewness model is better than any other models that are be able to describe to the skewness of the stock return including the asymmetry GARCH class models.