Fixed Effects Spatial Panel Data Model with Time-Varying Spatial Dependence: An Application to Regional Expenditure Competition in China

Abstract

To analyze the variation of spatial spillover effects over time as well as control unobserved individual-specific features, we extend the fixed effects spatial panel data model by introducing time-varying spatial dependence. For estimation, we propose a two-stage least squares method and the quasi-maximum likelihood approach. Finite sample properties of these two methods are investigated through Monte Carlo simulations. We apply the model to study the expenditure competition among county-level governments in China from 2002 to 2016. Our empirical findings show that the downward pressure of the economy in 2007-2008 and 2015 may lead to stronger competition. Central government regulation on official assessment after 2013 may cause weaker competition.