Panel Data Models: Some Recent Developments

摘要: Abstract This chapter focuses on two of the developments in panel data econometrics since Handbook by Chamberlain (1984). The first objective this is to provide a review linear models with predetermined variables. We discuss implications assuming that explanatory variables are as opposed strictly exogenous dynamic structural equations unobserved heterogeneity. compare identification from moment conditions each case, and alternative feedback schemes for time series properties errors. next consider autoregressive error component under various auxiliary assumptions. There trade-off between robustness efficiency assumptions stationary initial or homoskedasticity can be very informative, but estimators not robust their violation. also problems arise multiple effects. Concerning inference variables, we form optimal instruments, sampling GMM LIML-analogue drawing Monte Carlo results asymptotic approximations. A number limited dependent variable fixed effects available literature, well some consistent asymptotically normal estimation such models. type including lags variable, although even less known nonlinear Reviewing recent work discrete choice selectivity second chapter. feature parametric fragility distributional situation prompted development large literature dealing semiparametric alternatives (reviewed Powell, 1994’s chapter). part thus at intersection cross-sectional