Peasants into democrats: Evaluating the determinants of democratic failure in Mali

Peasants into Democrats: Evaluating the determinants of democratic failure in Mali investigates the puzzle of poor government performance in the understudied francophone West African country of Mali, with implications for other young and weakly consolidated democracies. Motivated by the failure of elections to generate accountability across many newly democratized countries, this three-part dissertation proposes and tests related arguments that, together, help explain how a bad governance equilibrium can persist in spite of de jure democratic institutions. I argue that low levels of information about what voters can expect of their government decrease voter control over politicians and reduce accountability. I then posit that in societies governed principally by social than legal norms, marginalized groups face greater social costs to mobilization so closing information asymmetries is not sufficient to improve more visible forms of participation such as civic engagement. Finally, I show that information asymmetries are not only a cause of poor governance but also a symptom in that political elites have a strategic incentive to withhold information from voters. I test these arguments in the context of a single country case using both quantitative and qualitative methods. A within-country strategy takes advantage of rich local variation on social and political dimensions permitting a more precise evaluation of the micro-level mechanisms I attempt to uncover.;In the first part of the dissertation, I argue that if citizens systematically underestimate what their government can and should do for them, then voters will hold politicians to a lower standard of performance, sanction poor providers of public goods less often, and more often rely on kin or ethnic cues signaling better access to patronage. To identify causal impacts of raising voter expectations of government performance, I conduct a large-scale field experiment in 95 municipalities. The randomly-assigned treatment consists of a brief civics course informing voters about the capacity and responsibility of local government. I find evidence that the intervention effectively raises voter expectations of government. Voting simulations from a household survey show that people in treated villages are indeed more likely to vote based on performance. A behavioral measure---that voters are more likely to challenge local leaders at a town hall meeting---adds external validity to survey results.;Second, I argue that women face discrimination in the public sphere, impeding their participation in civic life. Examining effects of the same treatment on civic participation reveals striking heterogeneity among genders. The civics course causes men to participate significantly more in civic activity, and women less. Follow-up visits suggest the course made civic participation more salient, generating increased attention in the village. In a context where the woman's role is confined to the home, costs to civic participation increased for women. Following the course, women who wished to engage in civic activity reported both implicit and explicit threats of sanctions from male relatives and village elders. Conversely, the intervention succeeded in improving women's knowledge of and ability to evaluate government performance, and increased male civic participators and the airing of grievances.;In the third and final part of the dissertation, I address a general equilibrium problem posed by the first two: if voters are lacking information that would lead them to sanction incumbents, why are opposition parties failing to provide it? I argue that even in a multiparty democracy such as Mali's, a credible opposition may fail to emerge because political elites from different parties have incentives to collude rather than compete. Specifically, I show that illicit collusion among parties on representative local councils is more likely when all viable parties win seats because parties are better able to commit to a cartel that engages in rent-seeking at the expense of citizens. Conversely, when at least one party fails to win representation on the council, a credible opposition is more likely. I identify the effect of a party losing a seat on the town council using a regression discontinuity design and find that, indeed, places with a losing party provide more public goods on average than places without.