"In this paper we assess the electoral consequences of candidate selection into the supply of widely-disseminated programmatic information in the setting of Liberia, where clientelism is pervasive and the media sector is weak. We partnered with USAID and the NGO Internews to study the impact of randomized elements of a nationwide initiative to hold debates for all 73 House of Representatives seats ahead of the Liberian election of October 2017. Beyond lawmaking, House members in Liberia control access to development funds as well as play key roles in the allocation and implementation of public goods, and thus voters have incentives to care about the policy priorities of the candidates and vote accordingly. However, historically, votes have been bought as often as won (Bowles et al., 2017). In an effort to improve democratic accountability, Internews organized 129 standardized debates, with at least one in each electoral district, to solicit the policy promises of the participating candidates. In the debates, the 59% of candidates who participated were asked a series of questions by moderating journalists on particular issues of local policy relevance, most often relating to district schools, primary healthcare facilities, and infrastructural investments. Rather than large townhall-style debates, the emphasis was on soliciting concrete policy platforms and promises from the candidates that would then be rebroadcast by community radio stations. To shock the supply of policy promises, we randomly varied the intensity of invitation efforts to persuade candidates to participate in the debates. The decision to participate is risky, particularly in clientelistic settings where the returns to programmatic competition can be both limited and highly uncertain. Candidates who ‘win’ a debate may enjoy greater publicity and net electoral gains, but ex ante they risk performing poorly, revealing their policy priorities to be disconnected from their constituents and restricting their ability to target campaign promises to small groups of influential voters. These risks are especially pronounced for the leading candidates (incumbents and their challengers), who enjoy greater resources for campaigning, are adapted to the existing clientelistic equilibrium, and are more likely to be attacked by opponents to gain publicity." (Pages 2-3)