中国股市风险CAViaR方法的稳定性分析及其时变建模

The Stability Analysis and Time-varying Modeling of CAViaR for Chinese Stock Markets

在介绍诺贝尔经济学奖得主 Engle 及合作者 Manganelli 于 1999 年提出的 CAViaR 模型理论及实际意义的基础上,采用 Hansen 检验方法重点探讨了中国股市风险 CAViaR 建模的稳定性问题.通过对上证综合指数与深圳成指的实证研究, 认为 Engle 及 Manganelli 所力荐的四个参数 CAViaR 模型用于中国股市风险建模是相当不稳定的,不能很好地适合现实的中国股票市场.鉴于此,提出了具有时变参数的 CAViaR,模型通过失败率检验法得到了相对较好的结果,明显地提高了模型的度量与防范市场风险能力.

In this paper, we introduce the CAViaR(Conditional Autoregressive Value at Risk by Regression Quantiles)model, proposed by Engle and Manganelli(1999) to compute VaR(Value at Risk), and its practical contributions for risk management. Given the stability of VaR modeling is usually important for prediction and econometric inference in practice, the Hansen’s test for parameter instability is carried on to explore the stability of CAViaR modeling of the risk of the Chinese stock markets. The empirical results show that the 4 popular CAViaR models due to Engle and Manganelli do not fit well the real situation of the Chinese stock markets. Thus, we propose a reasonable time-varying parameter CAViaR model, which is strongly supported by the empirical results.