PhD thesis defence by Gustav Axén
This thesis contains three independent chapters on applied empirical economics and econometrics, focused on the area of political economy and education.
In the first chapter, I study how important national swings are in determining election outcomes in the U.S. Senate. Developing a model and identification strategy that takes prediction market prices as inputs, I estimate the relative importance of local and national uncertainty in determining the election outcomes. I find that national uncertainty has approximately 40-60% the importance of local uncertainty, suggesting that national swings are important to explain election dynamics under simultaneous local elections.
The second chapter is jointly co-authored with Andrea Ichino, Fabrizia Mealli, and Javier Viviens Martín. We show how principal stratification analysis can be used to evaluate whether test scores help teachers make better track recommendations for students. Additionally, we show that it can be used to measure the 'fairness' of these recommendations with respect to protected attributes such as gender, socioeconomic status, or immigration background.
In the third chapter, I describe the campaign spending patterns of political parties in the Canadian first-past-the-post setting. The key theoretical prediction is that marginal districts should be targeted, as they more easily influence the election results. I find that this holds in practice and that the parties invest systematically more in districts that were competitive in the last election. This is true for most parties in most elections, despite the fundamentals differing substantially. Incumbency is also predictive of spending, and marginal incumbents receive the most funds from their party on average. However, the most relevant competing party in a district provides only limited predictive power over spending when correcting for how competitive the district is.