Applying recent developments in time series econometrics to the spatial domain

Abstract

This paper surveys some recent developments in time series econometrics and examines to what degree they might have useful analogs in spatial econometrics. Spatial analogs of stationary vector autoregression models might be useful in modeling groups of spatial series, but the literature on non-stationarity and cointegration does not have a useful purely spatial analog. With the exception of some special cases, pure spatial series cannot be integrated processes. However, cointegration might apply to space-time processes. Space-time cointegration and Granger causality methods are developed and applied to explaining reductions in sulfur emissions in Europe.