改进的GARCH-M模型在股市风险分析中的应用

Application of an Improved GARCH- M Model in Risk Analysis on Stock Markets

首先对上海和深圳的收益序列进行自相关检验，发现两市均不具有明显的自相关特征．其次对两市收益进行异方差检验，发现它们的异方差特征显著．在前两步基础上提出改进的GARCH—M（1，1）模型进行风险分析，结果表明：沪深两市的风险都具有时变、正偏、高峰、波动聚集性和长记忆性等特点；中国股市的收益与风险没有必然联系，投资者还处于不完全理性阶段；风险对市场的正负面消息的反应存在显著差别，负面消息在一定程度上加剧了市场波动．

Firstly, autocorrelation test is used to investigate Shanghai and Shenzhen stock markets＇ returns. It proves the returns have weak autocorrelation. Secondly, heteroscedasticity test is used to check the two markets＇ returns. It can be found that their heteroscedasticities are significant. Thirdly, an improved GARCH - M model is developed to analyze the two markets＇ risks. We get the conclusion that the risks have time- varying, positive skewness,high kurtosis, volatility - collective and long memory characters ; that the returns and risks are not correlated statistically, which implies the investors in Chinese markets are irrational ; and that there exists big difference between the reflection by the risks of the market＇ s positive and negative information and the negative information increases the market＇ s volatilities in a sense.