At least 65 million Indians (and likely more) live in urban slums, where services and infrastructure are lacking, and property rights are often nonexistent. Nevertheless, slum dwellers vote: in Delhi’s 2017 municipal elections, informal or slum areas saw some of the highest rates of voter turnout. So why is it that constituents continue to receive poor (if any) services and investment? 

Since 2008, Rohini Pande and Abhijit Banerjee of MIT have collaborated with coauthors to explore whether better information to slum dwellers about the performance of their elected politicians would change electoral and spending outcomes. 

A few days before an election, voters assigned to 200 ‘treatment’ polling stations received pamphlets about the importance of voting and reminders of how and when to do so. These households then received a newspaper that provided report cards on their current representative’s performance. The report cards included information about each incumbent’s legislative activity (attending sessions and asking questions), attendance of committee meetings, and how they allocated over a million dollars-equivalent of discretionary spending. It also included information about the wealth, education, and the criminal record of the incumbent and two challengers. The results showed that in the areas that received report cards, voter turnout was 3.5 percent higher, and voters evidently compared incumbents to their challengers: they punished incumbents who performed poorly or had better qualified challengers, and gave a higher vote share to incumbents who performed well and were more qualified. The researchers also observed a 19 percent reduction in vote buying with cash.

In a second wave of research, the team worked with a local transparency NGO – Satark Nagrik Sangathan (SNS) – and a local newspaper (Dainik Hindustan) to publish information about how local councillors allocated funds as well as their attendance and participation in committee meetings. Voters in 168 wards had report cards about their representatives published in the run-up to the 2012 elections. One portion (110 wards) received this information first in 2010 and again in 2012, while voters in the remaining 58 wards received the information only before the 2012 elections. Seventy-two wards served as control and did not have access to report cards. 

The results from this second intervention showed that politicians who represented high slum areas allocated a greater share of their discretionary spending to reflect slum dweller priorities and improved their attendance at committee meetings when they knew these decisions were publicly disclosed. 

However, researchers wanted to know whether politicians were responding to the new information alone, or whether revealing information publicly was important. Therefore, a subset of politicians representing Delhi wards also received personalized information on the status of infrastructure (in this case public toilets and garbage dumps) within slum areas of their wards. This information, derived from independent audits of the facilities, was provided eight months prior to the 2012 election and updated two months prior to the election. Unlike the public disclosures, the situation of public toilets and garbage dumps did not improve as a result of information shared privately.

Principal Investigators:
Rohini Pande
Abhijit Banerjee, MIT
Michael Walton, Harvard University
Nils Enevolsden, Harvard University

This research is supported by:
USAID
National Science Foundation
International Initiative for Impact Evaluation (3ie)
International Growth Center
William and Flora Hewlett Foundation
Empowerment Lab at CID