Research On The Nonlinear And Time-varying Characteristics Of Riskreturn Trade-off In China Stock Market

Currently,China financial market is in the stage of rapid development and quickly changing where financial innovations and financial instruments are changing at a breathless pace,and the financial system is becoming more and more complex.And the price mechanism is seriously distorted.The complex characteristics such as highdimensional,high-frequency,non-stationary,nonlinear,time-varying and so on bring great challenges to traditional financial theories and modeling methods.More accurately risk measuring and risk pricing is an urgent problem to be solved in price discovery and risk management,which is also a key scientific problem in modern financial data modeling theory and method.This dissertation focuses on the most fundamental problem in the financial field: the relationship between return and risk,which is called risk-return trade-off.In fact,there are many contradictories among theoretical and empirical literature,with no consistent conclusion.In view of these problems and contradictions as well as the limitations of traditional financial econometric and models,this dissertation,under the background of complex and dynamic financial system,measures and models the complex characteristics of the increasingly complex financial system by using frontier econometrical methods and model systems to capture the confused relationship between risk and return.This dissertation studies the nonlinear and time-varying characteristics of risk premia in China stock market from four aspects:the influence of information flow,uncertainty measurement and its influence,nonstationary and non-normal.The mainly contents and findings are as follows:First,facing the complex financial law implied in the modern complex financial system,this dissertation proposes a new risk pricing model with endogenous variables based on ICAPM model and functional coefficient model with endogenous variables proposed by Cai,et al(2019).The new model is mainly used to model the nonlinear and time-varying characteristics of risk-return trade-off and market efficiency under the influence of important variables.It has good flexibility and generality.It not only avoids the assumption on the dynamic process of risk-return trade-off,but also considers the endogenous problem of risk.In the dynamic changing market environment,this dissertation studies the nonlinear and time-varying characteristics of risk compensation and market efficiency in China stock market under the influence of information flow level.This dissertation uses high-frequency data and calculates the realized volatility to avoid additional assumptions on the dynamic process of volatility model.The empirical results show that: under the influence of new information flow,the relationship between realized volatility and return has a stable "flat-up-down" fixed weaving pattern with the increase of new information flow in China stock market.And its value can be positive or negative.Unexpectedly,this pattern is opposite to the "up-down-up" weaving pattern of the US market.China stock market is weak inefficiency,especially during the trading day when the new information flow changes rapidly.Second,as quantifying researches on uncertainty in China stock market is still really rare,this dissertation constructs an uncertainty index for China stock market by econometric model as well as makes a detailed analysis and research on its statistical features and compensation.Although uncertainty is the main problem faced by any decision-making process and risk management,the direct measurement of uncertainty is still in its infancy.This uncertainty index constructed in this dissertation can provide data support for the research of uncertainty in China stock market.The empirical results show that the uncertainty index has strong long-memory with statistical characteristics of non-stationary but mean reversion.At the same time,it has local trend.The uncertainty also has periodicity with seven different main periods from 32 days to 107days;Although the uncertainty will reduce the current return,there is a significant risk compensation.Its risk compensation is found to be a U-type nonlinear relationship with the changes of uncertainty.Third,for the first time,this dissertation studies the dynamic correlation,internal correlation and long-term equilibrium relationship among return,realized volatility and uncertainty in the same model system under the framework of fractional differential.In view of the long memory(non-stationary)characteristics of realized volatility and uncertainty,this dissertation analyzes the fractional cointegration relationship and longterm equilibrium relationship between RV and uncertainty as well as their endogenous problems by using the fractional difference method.Furthermore,this dissertation constructs a FCVAR model system for return,realized volatility and uncertainty.At the same time,the nonlinear and time-varying risk compensation of realized volatility and uncertainty are studied.The results show that: there is a stable fractional cointegration and long-term equilibrium relationship between uncertainty and realized volatility,and the uncertainty has a positive impact on the volatility and is weak exogenous;In FCVAR model system,there is a stable long-term equilibrium relationship among return,realized volatility and uncertainty with weak exogenous for uncertainty;With the increase of uncertainty,the realized volatility risk compensation has a downward hyperbolic curve changing pattern.The negative impact of uncertainty on return decreases with the decrease of realized volatility level.Finally,in view of the contradiction among theoretical and empirical researches on risk compensation,this dissertation studies the impacts of skewness and kurtosis of return on volatility risk compensation in China stock market from the perspective of non-normality of return,distinguishing the skewness risk compensation and volatility risk compensation.It is found that the negative skewness and peak kurtosis of return will reduce the "pure" volatility risk compensation estimators,resulting in its value can be positive or negative;Although the sharp ratios are similar,there are big differences between the "I" type and the "II" type risk compensations and volatility risk compensations among different industries.The three type risk compensations have respectively downward,upward and slight upward trend with the growth of investment period.In a word,this dissertation shows that in the dynamic market environment,the riskreturn trade-off is non-linear and time-varying,and its value can be positive or negative.This dissertation measures and describes the risk and its pricing more accurately,which not only expands the research content of risk compensation,but also helps to explain the contradictions among existing related research.