The Study Of The Risk Correlation Between Chinese Growth Enterprise Market And Main-board Market With Copula

The first28listed companies of China’s Growth Enterprise Market were traded on theShenzhen Stock Exchange at October30,2009. China securities market GEM era is uponus. Compared with the main-board of the stock market, there is a big differencebetween the GEM market in listed enterprise size, industry distribution, structure ofinvestors, the delisting mechanism etc. But both markets operate in some same factors as apart of China’s multi-level capital market, then whether or not there is correlation betweenthem. And how they correlated under different situations. For example, whether there is ahigh correlations under soaring or crashing situations. This kind of ‘tail correlation’ is avery important issue for investors and risk managers. Therefore this paper has carried outintensive research into the correlation structure between China’s main-board and GEMmarket. This paper studies the correlation structure with copula technique, and analysis theability of risk prediction with VaR.Firstly, this paper analysis the market return distribution characteristics of China’smain-board and GEM market separately. Based on the Q-Q tests, found GARCH-T modelcan capture the tail structure of GEM market then GARCH-Normal model through thecomparison of the ability to predict the risk value VaR. But GARCH-Normal modelperformance better the GARCH-T model in main-board market. These means that GEMmarket has a thick tail then main-borad market.Secondly, based on the effectively distribution estimation of market return ofmain-board and GEM market, this paper computes four kinds of different copulas.Through the estimation and inspection, found that the mixed copula is more suitable fordescribing the correlation structure between main-board and GEM market. Overall, there isa positive correlation between the two markets. And there is also a asymmetric tailcorrelation structure, which means that correlation increased significantly under soaring orcrashing situation. And the correlation is larger under crashing situation then soaringsituation. Then simulate the portfolio’s VaR with copula distributions, founding that themixed copula can predict the portfolio’s risk effectively.