??? ???? ?????: ?? ??? ?? (The Relationship between the Multifactor Volatility and the Excess Returns: Using the Kalman Filter)

?? ???? Fama-French(FF) 3??? ???? ????? ??? ??? ?????, ?? ??? ?? ??? ?? ??? ???? ??. ??? ??? ???? ?? ?? ??? FF 3??? ??? ??? ????. ?? ??? ? ??? ?? ?? ???? ???? ???, ?? ?? ??? ??? ??? ???? ??? ???? ??? ??. ? ??? ??????? FF 3??? ??? ? ?? ??? ???? ?? ??? ???? ??. ?? ??? ??? ??? ?? ???? ??? ?? ?? ??? ? ???? ??? ??? ????? ???? ????? ??? ?? ??? ??? ? ??. ??? ?? ??? ???? ? ??? ?????? ?? ?? ?? ? ??? ?? ????? ??? ??? ???? ? ??? ??? ??? ? ??.
?? ?? ?????? ?? ??? GARCH(1,1) ??? ?? ???, ???? ????? ?? ??? ? ???? ?????. ??, ??? ??????? ????? ? ?? ?? ????? ???? ??? ????. 81? ???? ? ???? ?????? ?? 2?? ????? ??? ?? ?? ?? ???? ???? ?? ??? ???? ?? ? ???. ??? ?? ? ??? ?? ??? ?? ??? ?? ?? ?? ????? ?? ??? ?????, ??? ?? ??? ?? ?? ????? ?? ??? ?????. ??, ??? ?? ?? ?? ????? ?? ??? ?? ????? ??? ?? ? ?? ?? ?? ???, ? ??? ?? ????? ??? ?? ?? ??? ??? ??? ? ??? ??? ? ???.

Many researchers have analyzed the relationship between the volatility and the excess return using the Fama-French (FF) three factors, but the empirical results are shown differently depending on the method of analysis. One of the reasons for the difference in results arises from the measurement of FF 3 factors. Previous studies used each factor mainly as an observational rate of return, and for this reason, the time series characteristics of the factors are not accurately reflected in the analysis. In this study, the dynamic factor is extracted using the Kalman filter after applying FF 3 factors to the state-space model. Analysis using dynamic factors can reduce the uncertainty of estimates and improve forecasting of portfolio performance because it shows the relationship of financial time series over time well. Therefore, this study using a dynamic factor model can provide more accurate results in various analyses related to characterization of stock returns and the volatility risk premium.
As a result of the analysis, it was found that the variance errors in the state equation has a GARCH(1,1) process, and that small firms are more sensitive to size factors than large firms. In addition, the Korean stock market was found to require a higher risk premium in a bearish market. As a result of applying a two-pass regression analysis for 81 equal-weighted and value-weighted portfolios, the explanatory power of the model was improved when the volatility of error variance was included. It was found that the ex-post size and value factor had significant positive and negative risk premiums, respectively, and the variance factor of the value had a negative risk premium. On the other hand, it was confirmed that the risk premium of the value variance factor has a very large negative value compared to the risk premium of the value factor, and the risk premium of each factor can vary according to the method of estimating the average return.