<p>By what means can legislative committees exercise influence on policy outputs? How and why do committees in different countries differ in their abilities to do so? This dissertation argues that legislative committee power is a multidimensional concept. Committee procedures can be distinguished into three analytic dimensions: 1) committees' positive agenda power, their power to ensure the placement of legislative versions preferred by them on the floor; 2) committees' negative agenda power, their power to delay or block the progress of legislation; and 3) committees' information capacity, institutional incentives granted to them to gather and transmit information. These distinct dimensions benefit different legislative actors. Therefore, they reflect different features of a political system, and may not be consistently strong or week.</p><p>Based on an original cross-national data set, the dissertation shows that committee procedures cluster empirically in these three distinct dimensions. Furthermore, the dissertation also demonstrates how legislators' electoral incentives, the composition of multiparty governments, preexisting authoritarian incumbents' uncertainty and bargaining power, and the changes in legislative memberships affect different dimensions of committee power.</p> / Dissertation

[pt] Projetos de infraestrutura pública, como estradas e escolas, são considerados propulsores do desenvolvimento. No entanto, democracias em desenvolvimento falham sistematicamente em fornecer esses investimentos, e projetos semi-acabados são um problema comum. Usando um novo banco de dados com mais de 75.000 pequenos projetos de desenvolvimento no Brasil, estimamos que mais de 40 por cento dos projetos iniciados nunca são concluídos. Empregando um design de regressão em descontinuidade em eleições acirradas de prefeituras brasileiras, descobrimos que a alternância partidária afeta negativamente a entrega de projetos herdados em uma fase de construção, enquanto causa respostas positivas na entrega de projetos mais recentes. Argumentamos que nossos resultados são consistentes com uma teoria que vincula a não conclusão de projetos a incentivos eleitorais, em que as ineficiências na entrega de projetos são motivadas por uma dinâmica de reivindicação de crédito que desestimula a conclusão de obras herdadas da oposição. Nossas resultados destacam a importância de isolar políticas públicas do processo eleitoral na política local. / [en] Public infrastructure projects like roads and schools have been regarded as drivers of development, yet developing democracies systematically fail to deliver such investments, and half-finished projects are a common issue. Using a novel database of over 75,000 small development projects in Brazil, we estimate that more than 40 per cent of projects that start are never completed. Employing a close races regression discontinuity design on Brazilian mayoral elections, we find that turnover negatively impacts the delivery of projects inherited in a construction stage, while causes positive responses on the delivery of more recent projects. We argue that our results are consistent with a theory linking project non-conclusion to electoral incentives, where inefficiencies on project procurement are driven by a credit-claim dynamics that disincentives the conclusion of works inherited from the opposition. Our findings highlight the importance of insulating policies from the electoral process in local politics.