SETAR模型在中国股市波动率研究中的应用

文章对上证指数2006年1月6日-2011年5月23日收盘价的波动率进行了研究,介绍并使用随机系数SETAR模型与ARCH族模型进行对比拟合,根据数据的特点,文章构建了一种新型的SETAR模型,即AR(r)-SETAR(l,p1,p1)模型,模型利用ADF检验和AIC准则进行识别和估计。结果表明:可用AR(4)-SETAR(2,1,1)模型来拟合中国股市中的上证指数,研究其波动率特点,上证指数波动率呈不对称的响应,而且"负"响应比"正"响应高出约1.3倍。用ARCH族模型也证明了这种不对称响应的特征,但无法度量波动的强度,预测效果也没有SETAR模型精确。说明上证指数波动率不对称响应明显且呈现非线性的趋势,这种非线性的趋势更适合用SETAR模型来拟合。

SETAR Model in the Chinese Stock Market Volatility

The volatility of closing price in the Shanghai Securities Composite Index from January 6, 2006 to May 23, 2011 is studied. Random coefficient SETAR model and ARCH model are used to do comparison and fitting. According to the characteristics of data, a new SETAR model is constructed, i.e. AR（r）-SETAR（1,p1,p1） model, which is identified and estimated by ADF test and AIC criterion. The results show that the Shanghai Composite Securities Index of the Chinese stock market can been fitted by the AR（4）-SETAR（2,1,1） model. It is shown that the volatility of Shanghai Securities Composite Index admits asymmetrical response, and neg-ative response ratio was about 1.3 times higher than positive one. The ARCH family model also proves the asymmetric response, but it can not measure the intensity fluctuations, and prediction is less accurate than the SETAR model. The volatility of Shanghai index admits asymmetry response and exhibits nonlinear trend. This nonlinear trend is more suitable for SETAR model.