Theory and Evidence for Counteractive Lobbying

Abstract

Introduction In Austen-Smith and Wright (1992), we developed a formal model to study the pattern of lobbying and influence when two rival groups contend to affect the voting behavior of a legislator over a given pair of alternatives. Subsequently, in Austen-Smith and Wright (1994), we analyzed comparative static results derived from that model. Our analysis focused on the relationship between the decision to lobby by two groups, A and B, and the prior probability, p, that the legislator favors a particular alternative from the given pair of alternatives. The comparative statics imply that, ceteris paribus, if p is less than .5, so that the legislator is expected to vote against A's interests and in favor of B's, then: (1) the decision of A to lobby the legislator depends on p but is independent of the decision of B; (2) if A does not lobby the legislator, then neither will B; and (3) if A does lobby, then B lobbies when p is sufficiently high. We termed (3) counteractive lobbying. Intuitively the idea is that when legislators' prior positions are changeable (i.e., when p is sufficiently greater than zero or less than one), then we should observe interest groups actively lobbying to change the positions of legislators aligned against them; and under these conditions, we should also observe groups aligned with the same legislators lobbying to counter the potential influence of their rivals. We tested this prediction, along with results (1) and (2), in AustenSmith and Wright (1994) with data drawn from the lobbying efforts of organizations on Robert Bork's nomination to the United States Supreme Court. Since different legislators can be associated with different values of p-the probability of voting for or against Bork prior to any lobbying activity-our three hypotheses are cross-sectional statements about the pattern of lobbying across legislators. We emphasize that (1)-(3), along with the conditions under which they occur, are implications, not assumptions, of the model. Baumgartner and Leech (1996) contend that we specified our theoretical and empirical models incorrectly, coded our variables improperly, based