基于不同分布假设下波动模型估计效果比较——以上海股票市场为例

本文基于不同分布假设,即正态分布、Student-t分布以及EGB2分布,使用2005年1月4日至2011年6月30日上证综指日收益率数据对GARCH模型和GJR GARCH模型估计效果进行实证比较。实证结果显示：（1）基于非对称EGB2分布的GJR GARCH模型更适合中国证券市场;（2）中国股票市场存在波动不对称性,且好消息引发的波动大于坏消息引发的波动,这可能与中国股票市场特有的市场结构和交易制度有关;（3）波动的不对称特性可能部分来自于对分布偏度特性考虑的欠缺,验证了合理的分布假设在波动行为分析过程中的重要性。

Comparison of Estimation Efficiency of Volatility Models Based on Different Distribution Hypothese——Taking Shanghai Stock Market as an Example

Using the data of daily return rate of Shanghai Stock Exchange from 1/4/2005 to 6/30/2010,GARCH and GJR GARCH models are compared for the ability to capture the asymmetry of volatility in China’s stock market based on different distribution hypotheses,which are symmetric Normal Distribution,Student-t Distribution and asymmetric EGB2 Distribution.The empirical results indicate that GJR GARCH model is more efficient to capture the volatility of China’s stock market and the history that good news induces larger volatility than bad news.The reason might be the specific and unique market structure and trading system of China’s stock market.Besides,the volatility of China’s stock market could partly draw from the misuse of distribution hypotheses.This indicates the importance of proper distribution on the analysis of financial market volatility behaviour.