Modeling international diffusion: Inferential benefits and methodological challenges, with an application to international tax competition

Although scholars recognize that time-series-cross-section data typically correlate across both time and space, they tend to model temporal dependence directly, often by lags of dependent variables, but to address spatial interdependence solely as a nuisance to be “corrected” by FGLS or to which to be “robust” in standard-error estimation (by PCSE). We explore the inferential benefits and methodological challenges of directly modeling international diffusion, one form of spatial dependence. To this end, we first identify two substantive classes of modern comparative-and-international-political-economy (C&IPE) theoretical models—(context-conditional) open-economy comparative political-economy (CPE) models and international political-economy (IPE) models, which imply diffusion (along with predecessors, closed-economy CPE and orthogonal open-economy CPE)—and then we evaluate the relative performance of three estimators—non-spatial OLS, spatial OLS, and spatial 2SLS—for analyzing empirical models corresponding to these two modern alternative theoretical visions from spatially interdependent data. Finally, we offer a substantive application of the spatial 2SLS approach in what we call a spatial error-correction model of international tax competition.