指数条件异方差过程中检验跳跃现象的方法

Testing for Jumps in Jump-EGARCH Processes

由于存在边界检验和不可定义参数检验问题,检验跳跃现象的沃尔德、尤度比统计量都不再渐进服从正态分布(或卡方分布)。本文从Jump-EGARCH(N)和Jump-EGARCH(t)过程两个侧面,应用迪拉克·得鲁塔函数解决退化参数的微、积分计算的基础上,提议与不可定义参数无关的拉格朗日乘数检验统计量,并实施蒙特卡罗仿真实验和实证分析验证提议统计量的检验能力。

The Wald test and the Likelihood ratio test containing nuisance pa- rameters cannot follow normal distributions or chi-squared distributions, and then cannot be applied to test jumps. Based on Jump-EGARCH （N） or Jump-EGARCH （t） processes, dealing with differential and integral calculation for degenerative parameters by the Dirac＇s Delta function, this paper proposes Lagrange Multiplier Tests which are free of nuisance parameters, and test their power by Monte Carlo experiments and illustrations for daily stock-return series of S＆P 500 index.