ABLE AND MOSTLY WILLING:AN EMPIRICAL ANATOMY OF INFORMATION’S EFFECT ON VOTER-DRIVEN ACCOUNTABILITY IN SENEGAL∗ ABHIT BHANDARIy HORACIO LARREGUYz JOHN MARSHALLx APRIL 2019 Political accountability may be constrained by the reach and relevance of informa- tion campaigns in developing democracies and—upon receiving information—voters’ ability and will to hold politicians to account. To illuminate voter-level constraints without dissemination constraints, we conducted a field experiment around Senegal’s 2017 parliamentary elections to examine the core theoretical steps linking personal delivery and explanation of different types of incumbent performance information to electoral and non-electoral accountability. Voters immediately processed information as Bayesians, found temporally benchmarked local performance outcomes most in- formative, and updated their beliefs for at least a month. However, information about incumbent duties had little independent or complementary effect. Learning that more projects than expected reached their districts, voters durably increased non-electoral accountability—making costly requests of incumbents—but only increased incumbent vote choice among likely-voters and voters heavily weighting performance in their voting calculus. Voters are thus able and mostly willing to use relevant information to support political accountability. ∗We thank Fode´ Sarr and his team of enumerators for invaluable research assistance, and Elimane Kane, Thierno Niang, and LEGS-Africa for partnering with us to implement this project. We thank Antonella Bandiera, Nilesh Fernando, Matthew Gichohi, Lakshmi Iyer, Kate Orkin, Julia Payson, Amanda Robinson, Arturas Rozenas, Cyrus Samii, Moses Shayo, Jay Shon, Alberto Simpser, and participants at talks at APSA, MPSA, Notre Dame, NYU, NYU CESS Experimental Political Science Conference, and WGAPE-NYU Abu Dhabi for excellent comments. This project received financial support from the Spencer Foundation, and was approved by the Columbia Institutional Review Board (IRB-AAAR3724) and the Harvard Committee on the Use of Human Subjects (IRB17-0880). Our pre- analysis plan was registered with the Social Science Registry, and is available at socialscienceregistry.org/trials/2324. yDepartment of Political Science, Columbia University. Email: [email protected]. zDepartment of Government, Harvard University. Email: [email protected]. xDepartment of Political Science, Columbia University. Email: [email protected]. 1 1 Introduction Providing voters with information about their incumbent’s performance in office is thought to help voters retain high-quality politicians (Fearon 1999) and hold politicians to account beyond the ballot box (Aker, Collier and Vicente 2017; Gottlieb 2016). In practice, however, recent studies identifying the effects of informational campaigns on electoral accountability (Banerjee et al. 2011; Chong et al. 2015; Cruz, Keefer and Labonne 2018; Dunning et al. forthcoming; Ferraz and Finan 2008; Humphreys and Weinstein 2012) and individual or community action (Bjorkman¨ Nyqvist, de Walque and Svensson 2017; Casey 2018; Lieberman, Posner and Tsai 2014; Olken 2007) yield mixed findings. Given the complex chain of conditions linking the provision of information to better gover- nance (Dunning et al. forthcoming; Lieberman, Posner and Tsai 2014), it is often hard to know where accountability breaks down. For example, the limited effects of providing incumbent per- formance information on self-reported vote choice in the six-country Metaketa study (Dunning et al. forthcoming) could reflect low levels of information retention arising from the difficulties of disseminating information, failures to provide sufficiently relevant information, or voters’ low capacity or willingness to electorally reward (punish) better(worse)-performing incumbents. Fur- thermore, while community empowerment interventions have received significant attention (Casey 2018), little is yet known about whether incumbent performance information can influence non- electoral accountability-seeking political engagement—the communication of information, prefer- ences, or requests by citizens to politicians through non-electoral means. This article dissects voters’ ability and will to use different types of incumbent performance information to hold legislative deputies to account in Senegal. By personally distributing and explaining such information, we abstract from dissemination challenges to focus on three links be- tween providing incumbent performance information and voter engagement in electoral and non- electoral accountability-seeking political behavior. First, we examine the extent to which voters update their beliefs in a Bayesian manner, both immediately and a month later. Second, we vary the 2 information’s content to understand what information voters regard as relevant. Specifically, we combine indicators of the current incumbent’s performance with: (i) information about deputy du- ties; and (ii) a temporal benchmark against previous incumbents’ performance that helps voters to abstract from district-specific factors influencing all incumbents within the same district. Third, we study whether changes in beliefs translate into electoral and non-electoral accountability-seeking behavior, and the extent to which this behavior varies with the information’s relevance to individual voters. Together with a local civil association, we designed a field experiment in Senegal around the 2017 parliamentary elections to examine these voter-level mechanisms underpinning political ac- countability among deputies seeking re-election for a second five-year term. Across 450 rural villages from five of Senegal’s 45 districts, we trained enumerators to personally distribute and explain informational leaflets to nine voters aged 20-38 in treated villages in the month preceding the election. Our factorial design varied whether respondents were informed about: (1) parlia- mentary deputies’ duties; and (2) their current deputy’s participation in legislative affairs and the projects and transfers received by their district, either with or withouta comparison with their dis- trict’s previous deputy. To separate immediate effects of these informational treatments from any subsequent interactions with other voters or political operators, our panel survey measured vot- ers’ beliefs and accountability-seeking behavior immediately before and after treatment, and again after the election. Our findings first demonstrate that rural Senegalese citizens processed incumbent performance information as Bayesians. Immediately after receiving the information, voters favorably updated their beliefs in line with their relatively pessimistic prior beliefs and the fact that current incum- bents generally outperformed previous deputies. Such updating further indicates that voters care principally about local outcomes (projects and transfers), rather than legislative efforts within par- liament. Moreover, while information about deputy duties did not affect beliefs, temporally bench- marked information substantially shaped the extent of voter updating and increased the precision of posterior beliefs. We find similar—albeit somewhat smaller—effects on beliefs around a month 3 after treatments were administered. Immediately after receiving the information, voters also sought to hold politicians to account on the basis of the information provided. The average treated voter—who updated more favorably about current incumbents than challengers—became three percentage points more likely to intend to vote for the incumbent. Heterogeneity in such rewards reflected the extent to which voters up- dated their beliefs and to which performance information is the most important factor determining vote choices. Alongside electoral intentions, non-electoral accountability-seeking behaviors also increased, with treated voters being more likely to request a visit from, or an opportunity to express their views to, their incumbent deputy. While voters persistently updated their beliefs and demonstrated an initial willingness to hold politicians to account, lasting electoral accountability was only sustained among likely-voters on issues relating to performance on local outcomes. While our treatments did not affect self-reported vote choices on average, the treated respondents that cared most about incumbents lobbying for local development projects or had turned out at the last election did reward the incumbents over- seeing more local projects and transfers. Voters even penalized incumbents for nationally-oriented parliamentary activity. Consistent with substantial within-village diffusion of our information—by voters and political parties—to the older and more experienced voters most likely to respond to it, we further find greater incumbent vote shares at polling stations encompassing treated villages that received information revealing higher rates of local projects and transfers. In contrast, non-electoral accountability-seeking behavior increased more uniformly, even a month after receiving treatment. The average respondent continued to make more requests of the winning candidates, especially in districts that received more projects and transfers. This increase in requests reflects not only relatively costless requests for the winning candidates to call respon- dents or visit their village, but also citizens paying to send SMS or voicemail messages to winning candidates using a hotline that we created. This effect was also most pronounced among those receiving benchmarked