Julieta Peveri*, Zheng Wang**
IBD Salle 21
AMU - AMSE
5-9 boulevard Maurice Bourdet
13001 Marseille
Kenza Elass: kenza.elass[at]univ-amu.fr
Camille Hainnaux: camille.hainnaux[at]univ-amu.fr
Daniela Horta Saenz: daniela.horta-saenz[at]univ-amu.fr
Jade Ponsard: jade.ponsard[at]univ-amu.fr
*While the role of interest groups' contributions in shaping policies has been well-studied, little is known about their effects on electoral outcomes. This paper exploits a Brazilian reform that banned firms' contributions to provide evidence in this regard. I use a weighted differences-in-differences strategy exploiting variation in municipalities' and candidates' dependence on firms' funding before the reform. I find that the ban had little effect on the composition of the pool of candidates. However, it deteriorated the electoral advantage of low-educated incumbents and of incumbents from traditional political parties. The negative effects for incumbents are concentrated in oil-dependent municipalities, where rent-seeking is more likely, and in localities with low economic growth and high mortality rates. These results are consistent with the reform crowding out incumbents who heavily relied on the financial advantage to be re-elected rather than on a good performance while in power. Yet, the ban also discouraged women from running for office, suggesting adverse effects for candidates with higher fund-raising costs.
**Endogeneity of network formation is a major obstacle for the causal identification of peer effect in non-experimental studies. In this paper I first propose a causal framework to analyze contextual peer effect where novel peer effects are defined in terms of causal estimands instead of the usual linear-in-means regression coefficients. Then I develop a new propensity score based identification strategy for endogenously formed networks. The causal peer effect estimators proposed in this paper are straightforward to implement with existing statistical packages, but do not suffer from the usual criticisms that propensity score methods face due to the nature of network data. Finally, with AddHealth data, I apply the methodology to study the causal effect of having second generation high school friends, those who have at least one parent with college education, on one’s own probability of going on to pursue a college degree. Preliminary analysis shows that having more second eneration friends is beneficial to first generation students, even after controlling for friends’ ability.