【摘要】 本文探讨了广义自回归条件异方差模型(generalized autoregressive conditional heteroscedasticity model, GARCH)在金融风险度量中的应用，特别是针对上证指数。研究发现，GARCH模型能有效地捕捉市场波动性，刻画收益率的时变性、非对称性及波动聚集效应。然而，在不同分布假设下，风险度量的结果存在差异。而风险价值(value at risk, VaR)的风险度量结果显著依赖于分布假设。进一步的分析表明，EGARCH模型(exponential generalized autoregressive conditional heteroskedasticity model)和PARCH模型(power autoregressive conditional heteroskedasticity model)的VaR在计算误差上更小，且能更好地适应厚尾特征，更能反映市场风险。更多还原

【Abstract】 This paper explored the application of the generalized autoregressive conditional heteroskedasticity(GARCH) model in financial risk measurement, with particular reference to the Shanghai Composite Index. The study found that the GARCH model can effectively capture market volatility and characterize the time-varying nature, asymmetry, and volatility clustering effect of return series. However, under different distributional assumptions, the results of risk measurement show discrepancies. Moreover, the value at risk(VaR) risk measurement results significantly depend on distributional assumptions. Further analysis indicates that the VaR of the exponential GARCH(EGARCH) and power ARCH(PARCH) models has smaller calculation errors, can better accommodate fat-tailed characteristics, and thus more effectively reflects market risks.更多还原