A fixed-bandwidth view of the pre-asymptotic inference for kernel smoothing with time series data.pdfVIP

Journal of Econometrics 197 (2017) 298–322 Contents lists available at ScienceDirect Journal of Econometrics journal homepage: /locate/jeconom A fixed-bandwidth view of the pre-asymptotic inference for kernel smoothing with time series data? Min Seong Kim a, Yixiao Sun b,?, Jingjing Yang c a Department of Economics, Ryerson University, Canada b Department of Economics, UC San Diego, USA c Department of Economics, University of Nevada, Reno, USA article info Article history: Received 23 September 2015 Received in revised form 29 May 2016 Accepted 12 November 2016 Available online 16 January 2017 JEL classification: C12 C14 C22 Keywords: Heteroskedasticity and autocorrelation robust variance Calibration Fixed-smoothing asymptotics Fixed-bandwidth asymptotics Kernel density estimator Local polynomial estimator t -approximation Testing-optimal smoothing-parameter choice Temporal dependence abstract This paper develops robust testing procedures for nonparametric kernel methods in the presence of temporal dependence of unknown forms. Based on the fixed-bandwidth asymptotic variance and the pre-asymptotic variance, we propose a heteroskedasticity and autocorrelation robust (HAR) variance estimator that achieves double robustness — it is asymptotically valid regardless of whether the temporal dependence is present or not, and whether the kernel smoothing bandwidth is held constant or allowed to decay with the sample size. Using the HAR variance estimator, we construct the studentized test statistic and examine its asymptotic properties under both the fixed-smoothing and increasing-smoothing asymptotics. The fixed-smoothing approximation and the associated convenient t-approximation achieve extra robustness — it is asymptotically valid regardless of whether the truncation lag parameter governing the covariance weighting grows at the same rate as or at a slower rate than the sample size. Finally, we suggest a simulation-based calibration approach to choose smoothing parameter