Essays on applied political economics
MetadataShow full item record
This dissertation consists of three essays on the determinants of voting behavior. In Chapter 1, I empirically examine why candidates who are listed first on voting ballots enjoy substantial advantages such as winning 10% more elections. I use Californian election data where ballot order is randomized but identical for every voter. With these data, I provide new empirical regularities on how such ballot order effects change with the number of votes available to voters and candidate popularity. I show that these patterns are difficult to reconcile with existing models in which ballot order directly affects a voter's choices. In Chapter 2, I propose a novel theory of ballot order effects where rational voters respond to behavioral voters and then amplify the advantage of candidates listed first due to the inherent strategic complementarity in voting. My model is an extension of a standard voting model and allows me to explicitly model the interaction of different types of voters. I estimate my model using a simulated method of moments and find that the interaction between voters is empirically important: rational order effects account for around half of total ballot order effects in terms of vote shares (votes gained just for being listed first), while also reducing the number of behavioral voters necessary to explain the data in other dimensions as well. Motivated by these findings, I suggest new policies to address ballot order effects. In Chapter 3, I investigate how newspaper consumption affects political engagement. To circumvent potential endogeneity issues, I use variation in European languages as an instrument for newspaper consumption. Specifically, I consider variation in how much physical space languages require to express some given information content. I first estimate such language efficiency from large bilingual text compilations. Using a European-wide survey that spans 18 different languages, I find that respondents who speak efficient languages are more likely to read newspapers, as is consistent with this mechanism. This finding is robust to a large variety of alternative specifications. Using language efficiency as an instrument for newspaper consumption, I find that newspaper consumption increases turnout and political interest of immigrants.