A Distributive Model of Legislative Voting
Despite the prominence of distributive theories of the legislative process (e.g., Baron and Ferejohn 1989; Morelli 1999; Diermeier, Eraslan and Merlo 2003), no statistical model of legislative voting consistent with distributive micro-foundations exists in the literature. Instead, all the major alternatives for scaling roll call votes (e.g., Poole and Rosenthal 1985; Heckman and Snyder 1997; Clinton, Jackman and Rivers 2004) are built on spatial micro-foundations. In this paper, we propose a distributive model of legislative voting estimable from roll call voting data and consistent with a one-shot distributive bargaining game. The gist of the model is that legislators can be ranked in terms of “clout,” and members with higher clout are more likely to secure distributive benefits for their districts and, hence, to vote in favor of distributive bills. While the model is mathematically very similar to standard spatial models of legislative voting, the interpretation of the parameters is quite different. Thus, we do not know how to interpret scaling results unless we can establish what kind of data generating process they come from; but the mathematical similarity of the models makes it difficult to determine whether votes come from a spatial or distributive process without additional information (beyond the roll call voting matrix).