Quarterly Journal of Political Science > Vol 17 > Issue 2

The Unintended Effects of Bottom-Up Accountability: Evidence from a Field Experiment in Peru

Renard Sexton, Emory University, USA, renard.sexton@emory.edu
Suggested Citation
Renard Sexton (2022), "The Unintended Effects of Bottom-Up Accountability: Evidence from a Field Experiment in Peru", Quarterly Journal of Political Science: Vol. 17: No. 2, pp 223-257. http://dx.doi.org/10.1561/100.00020079

Publication Date: 28 Apr 2022
© 2022 R. Sexton
Comparative political economy,  Democracy,  Elections,  Political participation
Comparative political economydemocracypolitical participationelections


Download article
In this article:
Conceptual Framework 
Intervention and Hypotheses 
Research Design 
Discussion and Additional Findings 


Numerous recent studies show that sharing politician or bureaucrat performance information with voters seldom succeeds at generating bottom-up pressure. One explanation is that local politics in developing democracies are typically mediated through local "development broker" elites, who often lack critical procedural information. Does supplying local leaders with information about decentralized political processes allow them to participate more effectively, generate accountability pressure, and improve public goods provision? A randomized field experiment in Peru demonstrates that information can indeed provoke attitudinal and behavioral change among village elites regarding participation, government performance and protest, but often in unanticipated ways. The study finds that training workshops in fact reduce participation in local participatory budgeting processes, reduce satisfaction with mayors, and increase support for protest and recall elections. Aggregate behavioral measures find a reduction in participatory budgeting attendance, but no changes in the probability of protest or recall initiation. Although village leaders are frustrated with mayoral performance, they struggle to generate collective action and thus have to settle for withdrawing from processes they can leave unilaterally. A test of mechanisms indicates this is not driven primarily by rent-seeking. There are no overall effects on district government performance in the following year. The evidence indicates that information can prompt local leaders to make political moves, but the high costs of collection action makes it difficult to produce meaningful change in politician behavior.