A0318
Title: Testing endogeneity of a spatial weight matrix in a weak spatial dynamic panel data model
Authors: Jieun Lee - Emory University (United States) [presenting]
Abstract: A stochastic or non-predetermined spillover framework is very general and broadly applicable to account for the effects of economic interactions. However, in this case, a spatial weight matrix might be endogenous and thus needs to be tested since a valid and optimal spatial model depends on its endogeneity. To this end, the robust score test (or equivalently Lagrange multiplier test) is developed to determine the endogeneity of a spatial weight matrix in a weak spatial dynamic panel data (SDPD) model in the sense that parameters associated with the stability condition are weakly identified from zero. First, the score function biases are analytically corrected to resolve the incidental parameters problem. Second, the score functions are orthogonalized so that the score test statistic is robust to local parametric misspecification in the weakly identified parameters. A Monte Carlo simulation performs in favour of the theory and shows nice finite sample properties in terms of size and power. Finally, an empirical illustration using the PennWorld Table version 7.1 describes how this testing helps researchers select the valid and optimal spatial model at their early research stage.