基于小波多尺度分析的GARCH建模方法的拓展

发布时间：2018-03-28 20:20

本文选题：GARCH　切入点：增广FIGARCH-M　出处：《系统工程理论与实践》2011年11期

【摘要】：考虑到交易周期对资产价格波动特征的重要影响,将小波多尺度分析引入广义自回归条件异方差(GARCH)建模理论,提出了多尺度广义自回归条件异方差模型和多尺度增广分整广义自回归条件异方差均值模型,同时通过改进迭代的步长参数,得到了收敛速度快于BHHH算法的数值优化方法.对上证综合指数进行实证分析,结果表明:该模型克服了GARCH理论无法同时揭示蕴含在资产价格内部的多时间尺度信息的缺陷,还能够捕获到资产收益率在不同时间尺度上的局部波动特征;改进后的算法对模型参数估值效果十分明显.这类模型有助于探究资产价格伴随交易周期演化的微观动力学机制.
[Abstract]:Considering the important influence of transaction cycle on asset price volatility, wavelet multiscale analysis is introduced into generalized autoregressive conditional heteroscedasticity (GARCH) modeling theory. The multiscale generalized autoregressive conditional heteroscedasticity model and the multiscale augmented integral generalized autoregressive conditional heteroscedasticity mean model are proposed. A numerical optimization method which converges faster than the BHHH algorithm is obtained. The empirical analysis of Shanghai Composite Index shows that the model overcomes the defects of GARCH theory which can not reveal the multi-time scale information contained in the asset price at the same time. It can also capture the local volatility characteristics of asset return on different time scales. The improved algorithm is very effective in estimating the parameters of the model. This kind of model is helpful to explore the microdynamic mechanism of asset price with the evolution of transaction cycle.
【作者单位】：重庆大学经济与工商管理学院;
【基金】：高等学校博士学科点专项科研基金(20100191110033) 国家自然科学基金(70501015)
【分类号】：F224;F832.51

【参考文献】