An Empirical Study For Dynamic Relation Between Stock Returns And Idiosyncratic Volatility

Risk and return relationship in Security market is always research focus. In traditional financial theory field, the base of it——the Portfolio Theory by Markowitz says that idiosyncratic risk (unsystematic risk) can be dispersed plenty by constructing portfolio, return only need reflecting systematic risk assumed by the investors. So as regards asset pricing, traditional financial theory focuses on systematic risk, such as CAPM,APT and so on. But Merton (1987) thinks that idiosyncratic risk cannot be dispersed plenty due to many investors holding few stocks because of kinds of restrictions, investors require corresponding risk premium for not only systematic risk but also idiosyncratic risk. Thereafter, numerous foreign scholars researched idiosyncratic risk and they found idiosyncratic risk and stock return not only had a positive relationship but also found that existed a negative relationship between them. This is called "idiosyncratic volatility puzzle".We begin to do research on idiosyncratic risk in Chinese A-share market on such setting. We measured idiosyncratic volatility based on average stock idiosyncratic, CAPM and Fama-French model. In the many past literatures, they got the results based on the regress model on excess return and idiosyncratic risk with its lagged values. But because the idiosyncratic volatility is highly persistent, it could result an inefficient estimate of predictive regressions for return. To avoid this problem we analyze the repose of stock returns to the typical random volatility shocks.It means that we regress idiosyncratic volatility on its own lagged values and take the residuals (innovations) without serial correlation as explanatory variable analyzing idiosyncratic risk and stock return. The advantage is that we needn't consider the high persistent and self-correlation of idiosyncratic volatility, and the accuracy of parameters estimated by the model is increased.In the thesis we find whatever we create market portfolios based on equal-weighted method or value-weighted method that the dynamic relation between idiosyncratic volatility and stock return is positive. To ensure the robustness of our results we do lots of analysis. At first, we created the market portfolio based on the whole A-share market, so now we created the market portfolio based on the Shanghai and Shenzhen A-share market separately and do the same analytic work. Then we consider the liquidity factor would affect excess return, so we add the liquidity factor into the regress model to eliminate the effect. Thirdly relation we analyze is a dynamic relation, so the economic change has an effect on our results. We also add it into the model. Finally we create idiosyncratic volatility based on Fama-French model and compare the results to the findings found by the idiosyncratic volatility created based on CAPM.we not only construct market portfolios based on equal-weighted and value-weighted method, but also construct idiosyncratic volatility form high frequency data (day degree data) and low frequency data (monthly data). We find that the relationship of idiosyncratic volatility and stocks return in this thesis is significant positive correlation, and this relation don't change when we change the market portfolios from equal-weighted to value-weighted or from Shanghai A share market, Shenzhen A share market. Even if adding multi-variables as controlling factors, we still get the same results。...