Congressional Voting on Term Limits

Between 1990 and 1995, twenty-three states unilaterally imposed term limits on their own delegations to Congress. In 1995 the House of Representatives defeated a constitutional amendment that would have limited the terms for all of Congress. Only weeks later, the Supreme Court struck down the individual state laws. In 1997 the House again brought the issue to a vote, which also failed. This paper models congressional voting on term limits with a simple game within an interest-group theory with legislators as imperfect agents of constituents. The game foremost predicts that members from term-limited states would be more likely to support term limits in the first vote but no more likely on the second vote. The empirical section employs probit, multinomial logit, and ordered probit maximum likelihood estimations to confirm the stated hypotheses. Among other results, in particular both the joint and conditional probability of a "yea" on the first vote and a subsequent "nay" on the second vote is higher for members from states that had unilaterally self-imposed term limits. The results are robust to model specification, estimator, and alternative sampling. Implications are proposed in the concluding comments. Copyright 2002 by Kluwer Academic Publishers