Study On The Price Dependence,Risk Spillover Effect And Portfolio Management Between China And International Crude Oil Markets

The purpose of this paper is to explain and describe the dynamic correlation between the oil prices in both China and international markets so that improve the performance of risk management and assets portfolio.Firstly,we proposed three partial equilibrium models to analysis the mechanism of correlation between different oil markets.Then we used modified TVP-VAR to certify theoretical results.Secondly,this dissertation proposed SV-MIDAS and Stochastic Factor Copula(SFC)model to describe the marginal and joint distribution of oil prices.SV-MIDAS taken the advantage of MIDAS technique to incorporate the time-varying impact of low-frequency factors on high-frequency returns.SFC taken the advantage of factor structure and dynamic loading to describe the dynamic correlations among several oil prices.In order to solve the problem of high-dimension of parameters and hidden variables,this dissertation used MCMC algorithm to estimate models.Thirdly,this dissertation made use of dynamic correlation to measure CoVaR and construct risk budgeting portfolio.Related research was carried out based on daily data in 2005~2018 of Brent?WTI?Dubai?Oman?Minas and Daqing.The important results as follow:Firstly,pricing mechanism,financialization and exchange rate play important role in correlation of prices between China and international markets.This dissertation made use of decay factor and EWMA to improve TVP-VAR and calculate the impulse function based on improved TVP-VAR.The results shown that: the asymmetric impact between China and international markets was caused by pricing mechanism;the international oil markets can not only impact China market directly but also indirectly with the intermediary of exchange rate.When the demand is high and elasticity is low,international oil prices exert a positive impact on China oil price,conversely international prices play a negative role;financialization strengthen not only the direct interaction between China and international oil markets but also the intermediary effect of exchange rate.Secondly,the correlation between China and international oil prices was not static but varied severely around certain level.Estimation of SFC model shown that the persistence parameters of loading range from 0.95 to 0.98 so that the impact of past correlation impact significantly on current correlation.The standard error of loading was bigger than 0 so that constant loading in static models is not reasonable.From the perspective of individual market,the correlation among Daqing,Oman and Dubai were strong,sometimes the correlation over 0.9.The correlation among Daqing,WTI and Brent is low,correlation varied around 0.2.From perspective of period,when oil markets in “Bear”,investors held pessimistic view,the correlation of Daqing with Dubai and Oman were high,conversely the correlation were low.Thirdly,the risk spillover effect between China and international oil markets were time-varying and asymmetric.In period of 2008 financial crisis and 2018 oil slump,the spillover in return,volatility and tail risk between China and international oil markets were amplified.From the perspective of interaction,China was in peripheral position in global oil market system,impacted by external shocks.Dubai and Minas were the main resources of external risk of China,70%~90% varying of returns and volatility in Daqing oil were caused by these markets.Sometimes,the tail risk spillover from Dubai and Minas to Daqing were over 20%.Fourthly,SV-MIDAS and SFC model can successfully describe the variation of volatility and correlation in oil markets and dynamic budgeting portfolio based on these models was a reasonable choice.The estimation results of SV-MIDAS shown that the impact of GPR index and long-term volatility on oil returns was a mean-reverting process and volatility series characterized with clustering.Q-Q plotting and SC information criterion certified the better performance of SV-MIDAS model in fitting marginal distribution of oil returns than other SV-type and Garch-type models.Factor structure in SFC model solve the problem of “dimension curse” when it was used to describe the dynamic correlation.The log-likelihood and SC information criterion certified the better fitting of SFC than static model.Based on estimation of SV-MIDAS and SFC model,this dissertation construct ERC,MV,MDP and EW portfolio dynamically.The dynamic portfolios performed much better than static portfolio in value,sharpe-ratio and risk-diversification and ERC was the best dynamic portfolio.