【主讲】宋晓军 (北京大学)
【主题】A Nonparametric Test for Serial Independence of Errors in Nonparametric Models
【时间】2016年3月18日 (周五) 15：30-17：00
【地点】上海财经大学经济学院楼801室
【语言】英文
【摘要】We propose a nonparametric test for the serial independence of unobservable errors in nonparametric time series regression models. A Hoeffding-Blum-Kiefer-Rosenblatt type empirical process based on the nonparametric residuals is considered and its null limiting distribution is established. Especially, it converges weakly to the same limit as the process based on the true errors. Our test statistic is a generalized spectral type test applied to the nonparametric residuals. It is asymptotically distribution-free and powerful against any type of pairwise dependence at all lags. Extensive Monte Carlo experiments are conducted to evaluate its finite sample performance.
