Empirical Modeling of Interdependence in the Political & Social Sciences

Spatial interdependence is ubiquitous and central substantively and theoretically across social science. Empirically, clustering of outcomes on some dimension(s), spatial association, is also obvious. However, outcomes may exhibit spatial association for three reasons. Units may respond similarly to similar exposure to similar exogenous internal or external stimuli (common exposure), units’ responses may depend on others’ responses (interdependence, or contagion), or the putative outcome may affect the variable along which clustering manifests (selection). We may find states’ adoptions of some economic treaty, e.g., to cluster geographically or along other dimensions of proximity, e.g., bilateral trade-volume, because proximate states experience similar exogenous domestic or foreign conditions or because each state’s decision to sign depends on whether proximate others sign or because signing the treaty spurs bilateral trade. The theories and policy implications that these alternative sources of spatial association support differ starkly. We discuss how to specify and estimate empirical models that can distinguish these alternative sources of spatial association, using spatial lags to reflect interdependence, and how to interpret and present the spatial and spatiotemporal effects, response paths, and long-run steady-states that such models imply (along with their associated standard errors). We illustrate such spatial-econometric modeling of interdependence with replications of previous studies and applications of our own to a wide range of much-studied topics in the political and social sciences, such as: globalization and international tax-competition, active labor-market policies in the EU, children's welfare and health-care policies in the U.S. states, and international trade, alliance-formation, and conflict behavior.