Voter Rationality and Politician Incentives:Exploiting Luck in Indian and Pakistani Elections1Madiha Afzal, Yale UniversityOctober 20, 2007Preliminary DraftAbstractRecent empirical work on US elections argues that any relationship between exogenous shocksand election outcomes is evidence of voter irrationality. In contrast, I develop a theoretical frame-work which highlights two di¤erent mechanisms underlying politician behavior, e¤ort and corrup-tion, which relate exogenou s shocks to election outcomes under the assumption of complete voterrationality. Although corruption is generally considered the motivation underlying politician be-havior, I show that the more benign e¤ort story also leads to the same predictions on electoraloutcomes. In addition, the model shows that the relationship between politician behavior andthus probabilities of re-election and rainfall can vary with the underlying political environment ofincumbency advantage or disadvantage. I empirically examine the relationship between politicianre-election and rainfall using data from both Indian and Pakistani parliamentary elections. Theresults show that rainfall is signi…cantly positively related to re-election in times of incumbencyadvantage and negatively related to re-election in times of incumbency disadvantage in both coun-tries, consistent with the theoretical framework. Furthermore, evidence using de velopment fundspending and politician occupations shows that the results are consistent with the e¤ort story.JEL Classi…cation: D72, O11Email: [email protected] thank Mark Rosenzweig, Chris Udry, Rohini Pande, Ebonya Washingto n, as wel l as Prashant Bhara dwaj, Bjoer nBruegemann, A V Chari, Rahul Deb, Seema Jayachandran, Fabian Lange, Jeremy Magruder, Ted Miguel, and SalikNur for helpful comments and suggestions. I als o thank pa rticipants at the Yale De velopment Lunch, UC BerkeleyDevelopment Lunch, the Lei tner Political Economy Workshop , and th e Paci…c Development Conference for feedbackand encouragement. All errors are my own.11 IntroductionClassical democratic theory states that voters collect information on party p latforms, policy pro-nouncements, legislative voting, etc, on all candidates and cast their vote for the candidate whoo¤ers them the best package. However, collecting such information on each candidate is costly,and voters know that policy pronouncements are non-binding. The alternative approach proposedby the retrospective voting model places less of an information demand on voters and argues thattheir decision rule is based on readily available information2. This information is simply the per-formance of the incumbent while in power. If his performance is satisfactory, voters will re-electthe incumbent; if it is unsatisfactory, they will vote for the opponent.This leads us to the classic political agency problem: voters want to re-elect competent leaders,but cannot directly observe e¤ort and ability. In voting retrospectively, they should respond toeconomic indicators that re‡ect performance. However, they should also …lter out exogenous factors,such as rainfall, which a¤ect economic outc omes but lie beyond the politician’s control. An inabilityto do so raises the probab ility of re-electing lucky but incompetent incumbents and results ina lack of political accountability. A number of recent working papers, notably Wolfers (2007),Achen and Bartels (2004), and Healy (2006), study data from US elections to show that politiciansare re-elected in lucky times and voted out in unlucky times. Wolfers relates oil price shocks toUS governors’probability of re-election, while Achen and Bartels show that there is an electoralresponse to droughts and ‡oods in US presidential elections, and Healy uses individual voting datato show that weather a¤ects voting behavior through income3. All three papers argue that anyrelationship between exogenous shocks and electoral outcomes is evidence of voter irrationality.However, this literature relies on the critical assumption that there is no response by politiciansto these exogenous shocks. As soon as there is any interaction of an exogenous shock with politicianbehavior, we can no longer conclude that a relationship between shocks and politician re-electionis evidence of voter irrationality. I develop a theoretical framework which highlights two possiblemechanisms, e¤ort and corruption, which relate politician behavior to exogenous shocks, speci…callyrainfall shocks in the South Asian context, under the assumption of complete voter rationality. Ialso formalize the notion that voter behavior only changes because politician behavior changes;speci…cally, voters vote out the politician if he behaves badly enough that a randomly selected2See Key (1966) for seminal work on this topic; also see Fiorina (198 1).3In similar vein, Bertrand and Mullainathan (2001) show that CEO pay responds to exogenous factors that a¤ecthis company’s industry. Social psychology experiments also show that subjects in experiments aiming to asse sscompetence systematically fail to take su¢ cient account of background or environmental factors.2challenger is deemed better than him4. In the e¤ort story, rainfall a¤ects the p olitician-landowner’sown farm, and therefore changes his incentives to put in farm labor versus political e¤ort. Inthe corruption story, rain a¤ects the constituents’aggregate income, and changes the politician’sincentives to steal from them. The corruption and e¤ort stories yield analogous results for thedirection of the e¤ect on politician incentives but the e¤ort story is more benign in terms of itse¤ect on constituents’welfare. In my empirical work, I test whether the e¤ort story is consistentwith politician behavior in the South Asian context, but cannot directly tes t for the corruptionstory.The theoretical model shows that the e¤ect of exogenous sh ocks on both e¤ort and corruptioncan vary with the underlying p olitical environment. Speci…cally, I show that politicians increasee¤ort or lower corruption when there is higher rainfall (a good shock), if there is a high enoughincumbency advantage; and they lower e¤ort or raise c orruption when there is higher rainfall intimes of incumb e ncy disadvantage. A good rainfall shock in times of incumbency advantage servesas an "income" shock and politicians are able to undertake better actions in the political arena.However, in times of incumbe ncy