Information as an Incentive: Experimental Evidence from Delhi

Rohini Pande (Harvard University)

7 décembre 2018, 11h30–12h30


Salle MF 323

IAST General Seminar


Do anticipated performance disclosures influence politician behavior and, in turn, party and voter behavior? Two years prior to Delhi's 2012 municipal elections, we informed a random sample of councilors that a leading daily newspaper would report on their performance just before elections. To enhance our credibility, that newspaper immediately reported on the performance of a sub-sample of those councilors. To check whether improved information directly influences councilor performance, we implemented a cross-cutting experiment wherein treated councilors received ``for your eyes only'' audit reports on the condition of toilets and garbage dumps in slums present in their wards. A final source of variation arises from the unanticipated expansion of the gender quota four months before the elections -- the fraction of councilor positions reserved for women was raised from 33% to 50%. We have four findings. First, in high-slum-density wards disclosures caused councilors to move spending closer to slum-dweller preferences. Second, the public nature of disclosures matter: the ``for your eyes only'' audit report card intervention led to a limited and, arguably, perverse effect wherein the incidence of closed toilets rose in treatment wards. Third, newspaper disclosures influenced party ticket allocation. Specifically, treated incumbents who undertook more pro-poor spending but were unable to re-contest from own ward due to gender quotas were more likely to receive a party ticket for a different ward. Fourth, councilors who were subject to anticipated performance disclosures benefited electorally in the 2012 elections.