【主讲】Zhenlin Yang (Singapore Management University)
【主题】Unified M-Estimation of Fixed-Effects Spatial Dynamic Models with Short Panels
【时间】2015年5月8日 (周五) 15：30-17：00
【地点】上海财经大学经济学院楼602室
【语言】英文
【摘要】It is well known that quasi maximum likelihood (QML) estimation of dynamic panel data (DPD) models with short panels depends on the assumptions on the initial values, and a wrong treatment of the initial values will result in inconsistency and serious bias. The same issue applies to spatial DPD (SDPD) models with short panels. In this paper, we propose a unified QML-type approach (M-estimation) to estimate fixed-effects SDPD models that is free from the specification of the distribution of the initial observations. The approach starts from the conditional quasi-likelihood, given the initial observations, and then makes corrections on the quasi-score function. Consistency and asymptotic normality of the proposed estimators are established. A simple bootstrap method for robust standard error estimation is introduced. Monte Carlo results show that the proposed methods have excellent finite sample performance. Compared with the traditional methods where the initial observations are modeled and thus the full likelihood is used, there is a small loss in efficiency, but there a significant gain in applicability.