Political interaction in the senate: estimating a political “spatial” weights matrix and an application to lobbying behavior

Many motivations exist that cause legislators to behave strategically when voting. These include logrolling, towing the party line, or political favor trading. However, it has proven difficult in the existing literature to uncover the magnitude of the interaction among politicians. This paper takes a “spatial” approach to the problem, using a spatial autoregressive model to not only uncover the magnitude of interactions, but also the direction of the interactions. In contrast to most applications of spatial econometrics, I allow the elements of the spatial weights matrix to be estimated as parameters. The political spatial weights matrix is calculated for 96 senators in the 110th U.S. Congress. Furthermore, in a marginal effects simulation, I calculate the overall effect on voting from “flipping” a senator’s vote, allowing flipped votes to have a cascading effect. I apply these measures to study political fundraising, mildly suggesting that political interest groups direct donations to senators with the most influence. Copyright Springer Science+Business Media New York 2014