Essays in Development Economics

Abstract

This dissertation consists of three essays in development economics. The first chapter studies how political dynasties affect economic development in India, using variation from three distinct natural experiments. We develop a theoretical framework to show that dynastic politics has theoretically ambiguous effects: bequest motives may lengthen politicians’ time horizons (founder effects), but heritable political capital may make elections less effective at holding dynastic heirs accountable (descendant effects). We compile detailed biographical data on all Indian legislators since 1862, and present three main empirical findings. First, we identify descendant effects using a close elections regression discontinuity design, and find that descendants worsen poverty and public good provision in villages they represent. Descendants underperform partly due to moral hazard: they inherit voters loyal to their family, which dampens their performance incentives. Second, we estimate founder effects by examining constituency boundary changes. Founders have positive effects on economic development, and this is partly driven by bequest motives: politicians with a son are twice as likely to establish a dynasty and exert more effort while in office. Third, we identify the overall effects of a dynastic political environment using an instrumental variables strategy based on the gender composition of past incumbents’ children. Dynastic politics generates a “reversal of fortune” development pattern, where places develop faster in the short run (because of positive founder effects), but are poorer in the long run (because negative descendant effects outweigh positive founder effects). A simple model that can explain these empirical findings is a political agency model with overlapping generations where descendants inherit both ability and political capital. Chapter Two, based on work with Sarika Gupta, Manoj Kumar and Yusuf Neggers, studies how information frictions and coordination failure result in adverse selection to political office in India, where 9% of legislators face charges for murder, kidnapping, rape or armed robbery. We ran a voter information campaign experiment around the Uttar Pradesh state assembly elections in partnership with 3 telecom providers and involving over 600,000 voters. Voters in treated villages received a voice call and text message informing them about the criminal charges of all candidates in their constituency. Our campaign increased vote shares of clean candidates and decreased vote shares of violent criminal candidates by 2.2pp. Effects were strongest for the coordination treatment arm, in which individuals were informed that many other voters had also received the message. Turnout increased very marginally, suggesting that the campaign caused voters to support other candidates. The results suggest that information may sometimes be insufficient to break bad political equilibria but can help voters coordinate on good candidates. Chapter Three, co-authored with A. Nilesh Fernando, studies how the presence of a neutral colleague can debias prejudiced evaluators. We compile data on all international cricket matches since 1893, and analyse a series of reforms that introduced neutral umpires in international cricket matches. We present four results. First, prior to the reforms, both on-field umpires shared the nationality of the home team and make 9pp more discretionary calls against the foreign team. Requiring one of the two on-field umpires to be from a neutral country eliminates this bias. Second, half of this bias reduction is due to home umpires being less biased when paired with a neutral umpire. The de-biasing effects of neutral umpires are largest when an experienced neutral umpire is paired with an inexperienced home umpire. Third, we find, consistent with this, that a further reform requiring both on-field umpires to be from a neutral country had no additional bias reduction effect. Fourth, a “career concerns” reform that introduced TV referees and match executives to monitor and assess on-field umpires has no effect on bias. Collectively, these results suggest that social pressure from colleagues can discipline discriminators.