Campaign Advertising and Election Outcomes: Quasi-natural Experiment Evidence from Gubernatorial Elections in Brazil
利用巴西选举法产生的准自然实验，发现电视和广播广告对选举结果的影响远大于以往基于美国数据的研究结论。
Despite the “minimal effects ” conventional wisdom, whether and how campaign advertising influence elections outcome remains an open question. This is paradoxical because in the absence of a causal link from advertising to candidate performance, it is difficult to rationalize the amounts spent on campaigns in general and on TV advertising in particular. In this paper we argue that this “absence of documentation ” is due to the usage of US data, which suffer from omitted variable bias and reverse causality problems caused by the decentralized market-based method of allocating campaign spending and TV advertising. In contrast with received literature, we explore a quasi-natural experiment produced by the Brazilian electoral legislation to show that TV and radio advertising has a much larger impact on election outcomes than previously found in the literature. In Brazil, by law, campaign advertising is free of charge and allocated among candidates in a centralized manner. Gubernatorial elections work in a runoff system. While in the first round, candidates ’ TV and radio time shares are determined by their coalitions ’ share of seats in the national parliament, the two most voted candidates split equally TV time if a second round is necessary. Differences in TV and radio advertising time between the first and second rounds are a source of exogenous variation to evaluate the impact of TV