). In such cases, parties can depend on community organizers, ethnic chiefs, local bureaucrats, neighborhood leaders, patrons, caciques, local politicians, and other notables to act as brokers and close the gulf between party politicians and voters.Brokers, however, can be disloyal, acting more as free agents in a market with a bloc of voters to "sell" than as committed activists. From the perspective of the party, these brokers are useful because they control a private constituency or a module of nonpartisan voters who can readily help party candidates win elections. However, when multiple parties need brokerage to win elections, only the one offering the most attractive incentive package will gain the support of a broker. Since parties' relative ability to hire brokers may fluctuate over time, ongoing brokerage deals may suddenly break apart as brokers receive counteroffers and decide to place their modules elsewhere. Hence, these modular parties need brokers to mobilize voters, but broker disloyalty makes them fragile.This article identifies how brokers' disloyalty affects the organization and the electoral performance of political parties through an empirical analysis of Brazil. The research design in this article takes advantage of the fact that Brazilian parties use local politicians, especially mayoral candidates, as agents for their congressional candidates, making the measurement of broker affiliation less elusive. In addition, the article employs an empirical strategy that counters the endogeneity problem that weak electoral support stimulates disloyalty, or in the case of mayoral candidates, party switching (Desposato 2006). In 2007, a Supreme Court decision greatly inhibited party switching. This decision by an independent actor increased exit costs for those brokers who had won mayoral elections and would then have to ponder whether party switching was worth losing the mayoral seat. Thus, the court increased the survival odds of some party-broker relations.