An Empirical Study Of Conditional Factor Model In China’s A-share Market

Empirical asset pricing is among the predominant topics in finance literature.Marked with the rapid growth of China’s financial market in recent years;empirical asset pricing research based on China’s stock market became the focus of domestic finance scholars.Majority of the literature describing the empirical research regarding asset pricing in China’s A-share market primarily focus on the applicability of the theoretical asset pricing models.The purpose behind is to test whether the proposed pricing factors can provide an explanation in respect to variation in the cross-sectional returns of certain sets of portfolios.The researchers had been fact finding new factors based on the characteristics of China’s Ashare market in contemplation of explaining the numerous financial anomalies in the market.When exploring the available literature material,researchers analyze the validity of a model or a single factor established on strong assumption that is the factor loading(Beta)is constant at different time t.This model is also called the “unconditional factor model”.Alongside the development of asset pricing theory and econometric theory;some researchers put forward an assumption that the factor loadings are constant in the traditional factor model.Considering the factor loading changes with time or certain economic variables,this model is also called the “conditional factor model”.Given that the Beta in asset pricing model is not directly observable,there are two methods utilized to deal with the varying beta.The first approach is using a parametric method,i.e.assuming beta is a certain function of time or economic variables.If the functional form is correctly specified,the model can be estimated unbiased and efficiently.Otherwise,the estimates would come out critically biased which is caused by the inaccurate functional form.The Second practice is using a nonparametric method which avoids the estimation error for the reason that nonparametric models are robust on model misspecification.This dissertation employs the nonparametric time-varying model following Cai(2007),Li and Yang(2011),Ang and Kristensen(2012).It can be safely assumed that the factor loadings are time-varying and estimates the conditional factors model using a nonparametric method.The aforementioned is the fundamental econometric method used in this dissertation.The prime motive and contribution is to combine China’s A-share market data with the nonparametric time-varying coefficient conditional model to systematically test the validity of the conditional factor model in China’s A-share market.At hand four models were tested using time-varying coefficient conditional models,out of which three are the most distinguished classic models of asset pricing theory.Three models are CAPM model,FamaFrench three-factor model and Fama-French five-factor model;Other being the “China version” three-factor model(Liu et al.,2019).The stated method is currently the most effective model on China’s A-share market.The key improvement here is about relaxing the assumption of constant factor loadings,changes can be made regarding the unconditional model to conditional model.Then nonparametric methods are used to estimate and test the models.Foremost,the econometric model based on the nonparametric time-varying conditional factors demonstrated the basic asymptotic properties of the estimates as well as the method of optimal bandwidth selection.It is also proposed to use the joint test based on the long run alpha.At last,confirming the validity of the joint test statistics by a Monte-Carlo simulation utilizing the characteristics of the actual China’s A-share market data.In the empirical study;there is primary execution of an empirical test of the three classical unconditional models on China’s A-share market and compared to the conditional factor model in the following article.The test results validates the applicability of the classic asset pricing model in China’s A-share market;confirming an existence of size effect and value effect in the market.In terms of factor redundancy test,it contradicts the classic empirical findings of Fama and French(2015)on the U.S.stock market.It is found that value factor(HML)are effective in the A-share market,while the investment factor(CMA)is a “redundant factor” that can be thorougly explained by the value factor(HML)and profitability factor(RMW).The empirical test results under numerous conditional factor model show that the factor loadings exhibit obvious time-varying characteristics.The conditional CAPM model,the conditional three-factor model and the conditional five-factor model with time-varying coefficients describe the cross-sectional excess return variation of the 25 asset portfolios constructed by Size-B/M in China’s A-share market.In terms of the difference in an explanatory power,the conditional models are superior to the unconditional models in a general aspect.Based on the characteristics of “Split Share Structure” in China’s A-share market,the weighting of circulated market value and total market value are used respectively.According to the time point of the “Non-tradable Share Reform”,the whole sample can be divided into two sub-samples before and after the reform from July 2007.The empirical results of the conditional five-factor model show no evidence to reject the null hypothesis that is stated as “all of the long-term alphas of 25 portfolios are zeros”.The empirical results hold in different subsamples and hold on different weighting modes,which proves the robustness of the conditional factors model.Simultaneously it has been observed that the time-varying characteristics of factor loading in the conditional five-factor model are quite different in the two different subsamples before and after the reform.After reforming subsamples,the standard of fit of the model has been greatly enhanced.This represents that after the Non-tradable Share Reform,the pricing efficiency of China’s A-share market has significantly progressed.Besides the gradual improvement of the A-share market system influence of some irrational disturbances affected the return of A-share portfolio were gradually diminished.Finally,extension of the CH three-factor model,currently leading and most effective in China’s A-share market,to the conditional CH three-factor model and compare its explanatory power to the unconditional CH three-factor model,also the conditional fivefactor model.The results depict that the unconditional model and conditional model based on CH three-factors are consistent in the explanatory power of different benchmark portfolios.The conditional CH three-factor model slightly improves the explanatory power in the full sample conditional CH three-factor model and conditional Fama-French fivefactor model show significant differences in the explanatory power of different portfolios;but the overall explanatory power of the two conditional models were not exceedingly different.It was confirmed that some of the information about the CMA and RMW in the Fama-French five-factor model had been captured by the conditional CH three-factor model.