We study how county-level migration in the US responds to political state-level shocks. To that end, we examine how the state-level outcomes of toss-up gubernatorial elections affect county-level out-migration. For these toss-up elections, we distinguish counties with a majority for the losing candidate (treatment) and the winning candidate (control). We then show that there are no significant differences in the number of emigrants between treated and control counties pre-election. Post-election, emigration is five percent higher in treated counties. We find that this is mainly driven by increased out-migration from Democratic counties after a Republican win. We rationalize these results in a Roy-Borjas-type selection model with individuals who differ in political views. The model predicts bilateral migration decreases in political distance, which we confirm empirically.

Geopolitical tensions increase worldwide, and some foresee a decoupling of international trade from politically distant countries. Still, we know little about the role of political distance in international trade. I introduce a novel trade cost to the gravity model: political distance computed from countries’ voting behavior at the United Nations General Assembly. On average and over the last 50 years, I find an increase in political distance within a country pair by one standard deviation to predict a significant decrease in trade by 4 percent. To zoom in on different sectors, I construct a novel trade panel. I find the predicted decrease in trade to be more than twice as large for trade involving the US, the EU, or the UK and trade in strategic sectors. In a counterfactual decoupling scenario comparable to a New Cold War in 2018, the median absolute change of a trade flow is 56 percent of its actual value in 2018, suggesting substantial trade diversion.

Slides on previous version titled "Regionalization of Global Value Chains - Evidence from detailed import data"

In this study, we investigate how firm expectations about their own developments respond to different types of news. We classify news as either micro or macro, with micro news being information about firm-specific developments and macro news being information about the aggregate economy. Our analysis of firm surveys from Germany and Italy shows that both types of news consistently predict forecast errors, contradicting the idea of full-information rational expectations. Yet while firm expectations overreact to micro news, they underreact to macro news. We propose a model in which firms suffer from "island illusion" to explain these patterns in the data. 

I presented this paper at the 16th Dynare Conference (Slides) and the online seminar of DFG Priority Program "Experience and Expectation"

This chapter revisits survey evidence about firm expectations, with a particular focus on firms’ production and prices. We aim at synthesizing the evidence established on the basis of various firm surveys from different countries. We complement our discussion of existing work with new evidence based on the ifo Survey of German firms. This allows us, first, to put together five stylized facts regarding firm expectations and expectation errors. In addition, we present new evidence regarding the stickiness of firm expectations. Second, we use the same data set to revisit key results regarding the formation of firm expectations. Firm expectations react strongly to firm-specific developments, whereas aggregate variables are less important. Third, we summarize the evidence on how firm expectations drive firm decisions.