Rational gridlock

Abstract We examine the design of lawmaking institutions when advocates have agenda setting power and there is randomness in the status quo laws eligible for reform. The institutional designer maximizes voter welfare. We find that the optimal arrangement consists of two lawmaking institutions that must agree to enact any reforms. The institutions do not share preferences with one another or with the median voter. As a result, gridlock arises: the institutions reject some reforms that the median voter favors. However, when reform succeeds, it tends to be modest in scope and to more closely track what the median voter prefers. The optimal institutional design trades off the cost of failing to change law due to gridlock against the benefit of forcing advocates to moderate their proposals and offer more centrist reforms. Surprisingly, voters are best served by a pair of polarized and unrepresentative institutions.