Modeling Spatial Relationships in International and Comparative Political Economy: An Application to Globalization and Capital Taxation in Developed Democracies

Social scientists have long recognized that time-series-cross-section (TSCS) data typically correlate across time and space. Today, standard social-science practice is to model dynamics (i.e., temporal dependence) directly, typically with lags of the dependent variable, but to address spatial dependence solely by applying panel-corrected (robust) standard-errors (PCSE), thereby treating spatial dependence as a nuisance (Beck and Katz’s 1996 terminology). However, regardless of the analyst’s substantive interest in these relationships, direct modeling of spatial dependence (plus PCSE perhaps) enhances efficiency and is required for unbiased and consistent estimates of the coefficients on non-spatial regressors and any associated hypothesis tests or confidence intervals. Whereas analysts less interested in spatial relations per se will understandably wish to employ simple proxies for more complicated diffusion processes, those directly interested will prefer more sophisticated modeling techniques for estimating the dyadic patterns of diffusion in their data. Our broad project uses Monte Carlo experiments to evaluate the performance of several simple and sophisticated estimators under three important types of spatial correlation (the more-complex being spatial analogues to estimators devised for dynamic panel models: e.g., Hsiao 1986; Baltagi 1995) and develops a set of techniques and guidelines to help analysts diagnose, characterize, and gauge various sorts of spatial correlation, and to choose and interpret appropriate estimators based on their objectives. This paper leverages globalization and capital taxation as substantive venue to start such methodological explorations. Many academic and casual observers argue that the dramatic post1970s rise in international capital mobility and the steadily upward postwar trend in trade integration, by sharpening capital’s threat against domestic governments to flee purportedly excessive and inefficient taxation, has forced and will continue to force welfare-/tax-state retrenchment and tax-burden shifts away from more-mobile capital (especially financial capital) and toward less-mobile labor (especially manual labor). Several important recent studies of the international and comparative political economy of policy change over this period challenge such claims. We offer a preliminary comparison of several such arguments, using specifications that reflect the spatial relationships central to such diffusion processes.