Can Access to Information Improve Governance?

For the past two months, I have had the pleasure of coordinating a project examining the impact of information on voting behavior among poor residents of Delhi. This project is a randomized trial run by J-PAL Professors Abhijit Banerjee and Rohini Pande.

The main goal of this project is to evaluate the impact of providing objective information on MLA (MLAs are state-level legislators) candidates to low-income voters on voting behavior and public perception of politicians and the political process in general. Collaborating with consultants and NGO experts, we designed a three-phase objective information campaign. The first phase of the campaign was a door-to-door module that aimed at educating voters about the responsibilities of their MLA (such as allocating development funds, overseeing municipal agencies, etc). The door to door campaign also educated voters on the actual mechanics of voting, such as what types of identity proof are acceptable and how to determine voter eligibility. In the second phase of the campaign, we collaborated with the Dainik Hindustan (a major Hindi newspaper) to publish a series of articles delivering objective information on sitting MLAs and major party candidates. For sitting MLAs, we provided data on how they allocated their development funds (each MLA receives 2 crore every year to spend on the development of their constituency), as well as information on their attendance and participation in Vidhan Sabha (legislative assembly) meetings. This data was compiled by Satark Nagarik Sangatham (SNS), a Delhi-based NGO. For all major party candidates, we also provided information on their criminal history, personal assets, and education, which we compiled from affidavits available under the Right to Information Act. We then coordinated a mass distribution of these newspapers featuring objective information in randomly selected treatment areas. In the third and final phase of the campaign, our NGO workers went around treatment neighborhoods and ran community discussions focusing on the data presented in the newspaper to and door-to-door campaign to reinforce and provide context for the information.

The treatment for this project was done at a polling station level: each individual polling station corresponds to a neighborhood of about 1000 voters (roughly 400 households). We identified a sample of 400 slum-area polling stations across ten assembly constituencies of Delhi, and randomly selected half to receive the information campaign (all three phases). The other half served as a control group. Our impact evaluation will compare outcomes between voting patterns in treatment and control polling stations. To help tease out differential program effects across demographic groups and generally provide some context to the voting data, we are also running a post-election survey in treatment and control polling stations. This survey is currently underway and votes are being counted, and we are looking forward to the arrival of this data for analysis.

Although we cannot make any rigorous claims about program impact until we get the data, the time I have spent in the field has been fascinating and informative in its own right. I would like to take the time to discuss some of the insights and mysteries this project has elucidated with the readers of this blog.

One of the things that struck me the most when I went around with our fieldworkers was the passionate and emotional response that many slum-dwellers had to the information that we were providing. In some areas where the condition of local development was particularly abysmal, residents went so far as to say that “If that bastard MLA comes around asking for votes this time were going to thrash him.” What was particularly interesting for us was that the information did not generate a universally anti-incumbent sentiment – people’s perceptions of their representatives seemed to be highly correlated with the amount of development that had occurred in the locality. It was a general consensus of the field staff that they were providing information that the public really wants to have. This suggests that there is tremendous potential for “pro-active” information delivery campaigns. SNS, one of the NGOs we worked with, has been a strong local advocate of information delivery through regular public hearings and meetings. Delhi Election Watch, a nonpartisan think tank, has developed a SMS based initiative that disseminates candidate-specific information through mobile phones, an idea that seems to have great potential. Jaago Re, perhaps the most prominent voter awareness movement in India, is also pursuing the active delivery of information.

Another interesting feature of slum voting in India is the prevalence of vote-buying. All of the major parties are actively involved in the distribution of goods (cash, alcohol, and clothing being the most popular) in exchange for votes. This is such a common feature of elections that slum-dwellers tend to speak quite openly about it. I had many interesting discussions with people on this topic, and the general consensus seemed to be that although people generally understand that their own well-being is correlated with the quality of their politicians, they lack the discipline and self-control to resist the temptation of these bribes when they are offered. In economic speak, voters are dynamically inconsistent: they are willing to accept bribes in the present even when they are in the bigger picture acting against their own self-interest. The fact that many voters are conscious of this inconsistency suggests that triggering behavioral change by simply trying to educate them about the real costs of bad governance would probably be difficult. One potential way to mitigate the effects of vote-buying is to emphasize to voters that their individual ballot is a secret, and that however many parties give them liquor, sarees, and cash, they should still vote for the one that they feel would be best for local development. This is however complicated by the fact that many elected MLA’s have a tendency to “punish” localities that do not support them in elections by denying them development funds. I would be interested in hearing readers suggestions of how to mitigate the negative impacts of vote buying on governance outcomes.

The other main thing that struck me about slum voters was their overwhelming commitment to vote, in spite of what seemed like a deep disillusionment with the political process. A very large proportion of people I interacted with expressed the notion that all politicians are generally corrupt and uninterested in the problems of the poor, yet literally all of them voiced a commitment to vote. When I asked them why they felt compelled to vote, most simply replied that “It’s my right to vote and I’m going to.” The perceived lack of benefit to political participation in general is therefore quite perplexing in light of this. The notion that the disenfranchised are more active political participants is widely accepted borne out by statistics – poor areas almost always have dramatically higher voter turnout than rich ones in Delhi.

This post has already turned out much longer than I anticipated, so I hope readers will continue this discussion from here on. I look forward to presenting our empirical results once they are ready…