RETHINKING WARD AND AT-LARGE ELECTIONS IN CITIES - TOTAL SPENDING, THE NUMBER OF LOCATIONS OF SELECTED CITY SERVICES, AND POLICY TYPES

This study shows why the conventional wisdom that cities with ward ele ctions will spend more than cities with at-large elections is too simp le and explains why the empirical findings have been so mixed. Ward vs . at-large elections will only affect the policy choices of city counc ils when the policy choice is one that is decided by the median legisl ator. When the policy is one that is decided by the rule of universali sm, the relevant institutional determinant of choice is the number of legislators, and not whether they are elected at-large or by wards. Un iversalism is politically rational for divisible policies that all con stituents deisre (''pork''); the majority rule equilibrium (at the med ian) is more rational for divisible policies that are generally desire d, but only when they are not located too close to any one constituent 's home (''LULUs''). The expectation then is that larger city councils will provide more parks than smaller ones, and that election by wards or at-large will be irrelevant for these policy choices. By contrast, community centers and libraries are thought to be generally desireabl e, but not when they are in one's backyward. The size of the council i s not predicted to be relevant for these decisions, but councils elect ed by wards are more likely to have a median legislator who represents geographically concentrated constituents, such as the minority poor; their preferences will have a bigger impact on councils elected by war ds than on councils elected at large. Data from a sample of council-ma nager cities with weak mayors who have no veto uphold these hypotheses .