IDENTIFICATION ROBUST INFERENCE FOR MOMENTS-BASED ANALYSIS OF LINEAR DYNAMIC PANEL DATA MODELS

We use identification robust tests to show that difference (Dif), level (Lev), and nonlinear (NL) moment conditions, as proposed by Arellano Bond (1991, Review of Economic Studies 58, 277–297), Ahn Schmidt (1995, Journal Econometrics 68, 5–27), Bover 29–51), Blundell (1998, 87, 115–143) for the linear dynamic panel data model, do not separately identify autoregressive parameter when its true value is close one variance initial observations large. prove combinations these however, so there are more than three time series observations. This then solely results from a set of, so-called, conditions. These moments spanned combined Dif, Lev, NL conditions only depend on differenced data. that, contain identifying information parameter, discriminatory power Kleibergen (2005, Econometrica 73, 1103–1124) Lagrange multiplier (KLM) test using identical largest rejection frequencies can be obtained moments. shows KLM implicitly uses they parameter.