在GARCH模型框架下发现的波动率“周内效应”可信吗?

Is True the Day-of-the-Week Effect on Volatility within GARCH Framework?

通过分析模型中的自回归结构与"周内效应"之间的相互影响关系发现:基于GARCH模型框架以考察收益率波动"周内效应"的计量方法极易产生误判。随后的一系列Monte Carlo模拟实验不仅印证了上述结论,而且还发现,使用绝对值收益率作为波动率代理对哑变量做回归的方法虽然发现"周内效应"的能力稍逊于上前者,但可以避免误判问题。

This paper investigates legitimacy of the econometric method that was widely used to test the day‐of‐the‐week effect on volatility within GARCH framework . After analyzing the interaction between the autoregression relationship and the day‐of‐the‐week effect of conditional variance within GARCH framework ,we find that the methods are apt to result in misestimate for the day‐of‐the‐week effect on volatility .Then ,the results from a series of Monte Carlo simulations not only confirm above argument ,but also suggest that the regression of proxy of volatility on dummies can avoid the problem of misestimate ,although it performs imperfectly .