Department of Economics

Faculty of Arts and Social Sciences


Seminar - Upcoming

2013 Asia Summer Institute in Behavioral Economics
Asian Meeting of the Econometric Society 2013
Past Events

Department Seminar


Panel Autoregressive Inference Using Averaging Estimators


John C. Chao
University of Maryland, College Park

Date: May 28, Thursday

Time: 12:30 - 2:00 pm


Lim Tay Boh Seminar Room (AS2 03-12)
National University of Singapore
1 Arts Link, Singapore 117570


John Ham


This paper considers problems of estimation and inference for the autoregressive coefficient (ρ) in a panel autoregression for which the degree of persistence in the time dimension is unknown. The estimator studied is an average of the Anderson-Hsiao IV estimator and the pooled ordinary least squares (POLS) estimator and is an improved version of an averaging estimator analysed earlier in Chao, Kim, and Sul (2014). The rationale behind averaging these two estimators is to exploit the
differential strengths of these estimators, since the POLS estimator is consistent with a faster rate of convergence than the IV estimator when the underlying process exhibits unit root or near-unitroot behavior but is inconsistent when the process is stable whereas in that case the IV estimator is consistent. A primary contribution of this paper is to show that a statistic obtained by appropriately
standardizing the proposed averaging estimator converges uniformly over the parameter space ρ ∈ [0, 1] to a standard normal distribution, so that asymptotically valid confidence intervals can be constructed by inverting this statistic. Simulations show that this averaging procedure performs well
infinite samples and is convenient to implement in practice.