条件自回归极差模型与波动率估计

Conditional Autoregressive Range Model And Estimation of Volatilities

无论是在理论研究领域还是在应用领域,波动率的预测已经成为现代金融经济学和金融工程的重要课题。Chou(2005)针对极差提出了条件自回归极差模型(CARR)。本文在Parkinson(1980)的基础上,对极差作出了一个简单的修正,使得相应的CARR模型成为标准差的动态模型;然后以上证指数2001年4月27日至2005年12月5日的周收益率数据为样本,采用滚动样本的方法,利用CARR模型和GARCH模型分析了样本数据,作出了上证指数波动率样本外1至8周的预测,在多种事后波动率的测度下比较了修正后的CARR模型与GARCH模型对上证指数波动率的预测能力,证实了CARR模型在理论上的有效性。

Both in the financial economics and financial engineering , forecasting the market volatility has been an important task in the last twenty years. The conditional autoregressive range model （CARR） which was firstly proposed in Chou （2005） is a model for range. Based on the result of Parkinson （1980）, this paper shows that after a modification the CARR model can also serve as a standard error model. With the modified CARR model, the statistical properties of Shanghai composed index from April 27, 2000 to December 25, 2005 are examined. To compare the forecasting power of CARR and GARCH, we have performed forecasts of sample out from 1 week ahead to 8 week ahead with the rolling sample and make several comparisons using different measured volatilities which verifies the efficien- cy of CARR model.