Robust estimation of dynamic fixed-effects panel data models (Q2442685)

The paper examines a family of robust estimators for the dynamic panel data model with fixed effects, \(y_{i,t} = \alpha y_{i, t-1} + \eta_i + \varepsilon_{i,t},\) where \(t= 1,\dots, T\) and \(i=1,\dots, n\). The estimators are based on the median of the ratio of pairwise differences, where the differences may be at lags other than one. Conditions ensuring the asymptotic normality of the estimators are established. The paper also includes a study of the breakdown behaviour of the estimators under contamination by independent additive outliers and by patches of additive outliers. A simulation study is reported which compares the new estimators with established estimators associated with panel data models.