Font Size: a A A

The Empirical Research Of Risk-return Trade-off In China's Stock Market

Posted on:2020-11-26Degree:DoctorType:Dissertation
Country:ChinaCandidate:H ChengFull Text:PDF
GTID:1360330602455054Subject:Financial engineering
Abstract/Summary:PDF Full Text Request
The time-varying conditional equity premium is considered to be the most important problem in current asset pricing research(Cochrane(2011)),and the positive relationship between conditional equity premium and conditional volatility is also regarded as the first law of financial theory(Ghysels,Santa-Clara and Valkanov(2005)).However,due to the unobservability of conditional volatility and the simple belief that conditional equity premium is only a positive linear function of conditional volatility,there are different conclusions on the relationship between conditional equity premium and conditional volatility in the empirical research of developed stock markets represented by the United States.In the cross-sectional empirical research,both the individual stock beta and idiosyncratic volatility are negative with expected return,which is also inconsistent with the classical financial theory.China,s stock market is now the world's second largest stock market by market value.and due to the relevant policies of China's capital market,it is relatively closed with the international market.This paper mainly focuses on the relationship between "risk and retum " relationship of finance theory,and provides new empirical evidence for China's stock market.However,due to the particularity of China's stock market,foreign academic research methods cannot be directly applied,which needs to be reasonably adjusted according to the special situation of China's market.The main empirical research logic and results of this paper are as follows:Firstly,we reasonably adjust the main variables used in the asset pricing research of this paper to the Chinese market.Scaled price ratios(pd,pb,pe,etc.)are discounted by the part of floating market value in according to the viewpoint of Hu etc.(2019).Based on French,Schwert and Stambaugh(1987),the measurement of realized volatility takes into account the adjustment of higher order autocorrelation.The method of Amihud(2002)is changed to use Brennan,Huh and Subrahmanyam(2013)which measure is calculated by turnover ratio,and Pastor and Stambaugh(2003)illiquidity measure is changed as the return reversal coefficient of compound daily return because of price limit.Secondly,due to the unobservability of conditional volatility,this study attempts to measure conditional volatility in the China's market.However,the relationship between macroeconomic activities and stock market volatility has always been controversial in previous literature,and there are many potential predictors of market volatility.This paper uses the method of variable selection for the first time to select the best prediction variables and models of China's stock mearket volatility among many potential variables.We use Bayesian model averaging,LASSO and in-sample regression,sub-sample analysis,and out-of-sample test and the sample period is from 1995 and 2018.The empirical results indicate that in the quarterly data of China's stock market,the lagged market volatility,turnover.Pastor and Stambaugh(2003)illiquidity measure can significantly predict the next period stock market volatility.Different with other previous conclusions,scaled price ratios have no significant predictive power.None of them are chosen in the two variable selection method:Bayesian model averaging and LASSO while its in-sample regression is not significant,and the out-of-sample show they have no additional forecast power.The relationship between the scaled price ratios and the realized volatility is time-varying,sometimes is positive and sometimes is negative.Their relationship in the two sub-sample are opposite,and they are both significant at 1%confidence level.This finding also provides the empirical basis for the subsequent study of the time-varying equity premium.In the China's stock market,the macroeconomic variables used in this paper cannot predict stock market volatility.Subsequently,under the framework of Merton(1973)'s ICAPM model and Guo and Whitelaw(2006)'s empirical research,this paper conducts an empirical study on the predictability of the China's stock market return and the relationship between conditional volatility and expected return.This paper selects quarterly data from 1995 to 2018.We jointly estimate the conditional volatility equation(prediction variables include:the lagged volatility,turnover rate and Pastor and Stambaugh(2003)'s illiquidity measure)and conditional equity premium equation(it contains two components:the conditional volatility and the hedge risk factor5 which is measured by scaled price ratio in our paper)using GMM method.The empirical results show that after controlling the scaled price ratio as the hedge risk factor,the conditional volatility has a signifieant positive correlation with the exante return,and there i5 a positive risk-returm trade off.This indicates that hedging factor should not be ignored in ICAPM model,and multi-factor model is also needed in time series asset pricing.In the limited participation model of Guo(2004),the relationship between conditional volatility and discounted stock prices is time-varying which is consistent with our finding of China's data.In the subsample results,the first sample from 1995 to 2007,the conditional volatility is negatively related to the scaled price ratio,consistent with the theoretical model of Campbell and Cochrane(1999)and Bansal and Yaron(2004).The forecasting power of these two variables is the same.Due to the collinearity problem,there is only one variable significant in the multi-regression of stock market return.But in the second sample of positive correlation in 2007 and 2018,this conforms to the Guo(2004)limited participation in weak liquidity situation,the prediction power of these two variables are different,then there is a omitted variable problem.None of them is significant,but while they are included in the forecasting regression together they are both significant at 1%confidence level.In the robustness test,this paper uses different measures of market volatility adjusted by different orders,different measures of scaled stock price ratios,additional prediction variables,out of sample prediction and monthly data.The empirical results are very robust.Chinese investors are risk averse,and there is a positive risk-return trade off relationship in the Chinese market.Although there are many differenees between the China's market and US market,the determinants of conditional equity premium are the same.In terms of cross section research,CAPM model also means that there is a positive relationshiP between individual beta and return,and it can fully explain the variance of cross-section return,which means that alpha is not significant.However,in this thesis,according to the method of Bali,Engle and Murray(2017),this paper systematically study the statistical properties of beta of individual stocks in China's stock market from 1995 to 2018.The empirical results showed that the longer the data used to calculate the beta is,the more accurate the estimation of beta is.In addition,the daily data is better than the monthly data.However,in combination analysis and cross section regression of Fama and MacBeth(1973),there is no significant positive correlation between beta and return of individual stock,most of which are negative,and significant alpha exists.This empirical anomaly needs further study.Empirical methods to solve cross-sectional risk-return relationship include multi-factors model,classification test and conditional CAPM model.However,in many multifactor models,the main concern is the elimination of alpha without concern for the significant relationship between beta and return(Liu,Stambaugh and Yuan(2019a),Hu et al.(2019)).In this thesis,starting with the conditional CAPM model,a conditional asset pricing model based on the use of IPO first-day return rate is constructed,and economic uncertainty index is added.Under the two-step regression analysis framework of Avramov and Chordia(2006),the detection effect of multiple pricing models on the return anomalies of China's stock market was compared and tested.It is shown that time-varying risk measurement beta can well explain the change of return rate in cross section,and it is found that the conditional asset pricing model including IPO first-day return rate can more significantly explain the size and value anomalies in China's stock market.As for idiosyncratic volatility,the risk-adjusted return after adding IPO first day return into the conditional asset pricing model is signifieantly positively correlated with idiosyncratic volatility,which is also in line with the theoretical results of Merton(1987),and the result is still stable after adding economic uncertainty index as conditional information.For liquidity premium,only the conditional CAPM after the economic uncertainty index is taken as additional conditional information can be explained,which indicates that economic uncertainty is the main cause of liquidity premium in China's stock market.The empirical results are robust for different measures of idiosyncratic volatility and illiquidity.In the models when controlling the first three months of the composite benefits,but the traditional asset pricing model is the reversal effect,but after joining the investor sentiment and political uncertainty,are converted to the momentum effect,thus further confirmed that investor sentiment plays a very important factor in asset pricing model,the role of,of course,the steady momentum effect empirical results still needs further research.According to the hypothesis of Merton(1987),if the investment portfolio held by investors is not sufficiently diversified,then risk-averse investors need a positive relationship between idiosyncratic risk and return rate.This is in line with market conditions in China where investors are mainly retail investors and mutual funds are still in their infancy(Carpenter and Whitelaw(2017)).The cross-sectional relationship between idiosyncratic risk and return is the subject of further research in this thesis.Use Wang,Yan and Yu(2017)of capital gains index CGO,used in cross section and samples from 2000 to 2018,and through the portfolio risk analysis find that the trait of significant relationship between return and idiosyncratic volatility is only in higher CGO group,in the lower CGO group,there is no significant relationship.In the cross-sectional regression of Fama and MacBeth(1973),after controlling the intersection of CGO and CGO and idiosyncratic volatility,the relationship between idiosyncratic volatility and the next period return becomes positive and significant at the confidence level of 1%.In the unreported results,the calculation methods of various idiosyncratic volatility are significant,and the same as are the total volatility of individual stocks.The most likely empirical interpretation of this chapter is based on the perspective of RDP(reference-dependent-preferences,such as PT).The existence of reference-dependent investors changes the traditional positive risk-return relationship implied by standard preferences.However,whether investors' risk aversion level really varies among different CGO groups also needs to be further tested.In a word,based on the empirical study of China's stock market,which is now the second largest in market value,this paper provides new evidence for financial theory,and finds that there is indeed a positive risk-return relationship in China's stock market,and the hedging factor in ICAPM model cannot be ignored.On the cross-section,conditional CAPM model can solve the failure of static CAPM without positive risk-return relationship,and the CGO group of capital gains can obtain the positive relationship between idiosyncratic risk and return of idiosyncratic volatility measurement.
Keywords/Search Tags:risk-return trade off, conditional volatility, conditional equity premium, ICAPM model, conditional CAPM
PDF Full Text Request
Related items