报告人：张术林
报告地点：数学与统计学院五楼科学报告厅（501室）
报告时间：2015年12月09日星期三10:45-11:30
主讲人简介：
张术林，西南财经大学统计学院副教授。2000年-2005年在武汉大学概率统计系学习，师从胡迪鹤先生，研究方向为随机环境下的Markov过程。2005年-2010年， 在广发证券股份有限公司从事债券和衍生产品的研究和投资管理方面的工作。 2010年至今在西南财经大学统计学院工作。目前的研究兴趣主要在金融计量学和社会、经济网络分析。
报告摘要：
The test of time series models is challenging due to the temporal dependence structure of time series. In this paper we propose a novel class of test statistics named modified “in-and-out- of-sample" pseudo likelihoods ratio test (MPIOS) to check a specification of copula function for semiparametric time series models. Performing the proposed test faces a computational challenge due to repeatedly estimating the parameters. In the paper, we propose an approximation which is shown to be asymptotically equivalent to the proposed MPIOS. The approximation requires the computation of only one dependence parameter and hence significantly reduces computing bur- den. Furthermore, due to dependent structure in time series, derivation of the asymptotic prop- erties of the MPIOS test statistics is considerable challenging compared to that for independent observations. Under the regular conditions, we also show that the proposed test is consistent and asymptotically normal.
