Articulação intrapartidária e desempenho eleitoral no Brasil

We address the issue of whether electing a mayor affects the same party performance at subsequent proportional elections in the same municipality. We employ a Regression Discontinuity Design (RDD) model to identify the causal effect of elected mayors over their party performance in elections for federal and state representatives. RDD allows us to compare municipalities where a party has barely won to municipalities where the same party has barely lost. These two municipalities can be considered as an estimate of the counterfactual for each other, as the definition of winning candidates is related to the slight random variation in party vote shares due to weather or other exogenous causes. Our results suggest electing a mayor (even with a 1% margin) increases votes for the same party for both federal and state representatives. We interpret this result as evidence of party integration, as elected mayors would work systematically to favor their own party candidates in subsequent proportional elections.