Inference on trending panel data

Semiparametric panel data modelling and statistical inference with fractional stochastic trends, nonparametrically time-trending individual effects, and general cross-sectional correlation and heteroscedasticity in innovations are developed. The fractional stochastic trends allow for a wide range of nonstationarity, indexed by a memory parameter, nesting the familiar I(1) case and allowing for parametric short-memory. The individual effects can nonparametrically vary simultaneously across time and across units. The cross-sectional covariance matrix is also nonparametric. The main focus is on estimation of the time series parameters. Two methods are considered, both of which entail an only approximate differencing out of the individual effects, leaving an error which has to be taken account of in our theory. In both cases we obtain standard asymptotics, with a central limit theorem, over a wide range of possible parameter values, unlike the nonstandard asymptotics for autoregressive parameter estimates at a unit root. For statistical inference, consistent estimation of the limiting covariance matrix of the parameter estimates requires consistent estimation of a functional of the cross-sectional covariance matrix. We examine efficiency loss due to cross-sectional correlation in a spatial model example. A Monte Carlo study of finite-sample performance is included.