Work in progress
Restrictions to Civil Liberties in a Pandemic and Satisfaction with Democracy
With Daniel Graeber and Panu Poutvaara.In times of crises, democracies face the challenge of balancing effective interventions with civil liberties. This study examines Germany’s response during the earl stages of the COVID-19 pandemic, focusing on the interplay between civil libertie and public health goals. Using state-level variation in mobility restrictions, we emplo a difference-in-differences design to show that stay-at-home orders notably increase satisfaction with democracy and shifted political support towards centrist parties. Individuals that were exposed to the authoritarian regime of the German Democrati Republic show the largest reactions, underscoring the endogeneity of preferences for state intervention.
The Effect of Terror on Risk Attitudes
With Daniel Graeber and Neil Murray.Terrorism imposes high economic costs on the affected population. In this paper, we argue that conventional studies on the cost of terrorism are missing one important channel: costly behavioral responses. We show that terrorism impacts the risk attitudes of individuals that were in the same region at the time of the attack. Based on a representative sample of the German population, we employ a staggered difference-in-differences (DiD) design that compares individuals living in counties that were affected by a terrorist attack and those that were not. Our results reveal an immediate and significant decline in individuals’ risk propensity following a terrorist attack. Moreover, we find that the emotion of happiness serves as potential mediator in the relationship between exposure to terror attacks and risk attitudes.
Remote Work, Stock Market Participation and Inequality
With Lukas Menkhoff and Carsten Schröder. [Poster] Stock market participation jumped upwards in Germany in the year 2020 by abou 30%. A major cause for this was the enforced use of remote work. We show this by repeating a benchmark study with demanding data requests and adding remote work to the explanatory variables. Moreover, we implement an instrumental variables estimation based on work-from-home capacity. The transmission channel seems to wor via relaxing time constraints for those working from home. Finally, we show that remote work makes the stock market accessible to a broader population, including lower income groups, which contributes to lower income inequality.
Interpretable Machine Learning Methods in Empirical Social Science
With Daniel Graeber, Carsten Schröder and Sabine Zinn.This paper presents interpretable machine learning as a promising alternative to traditional regression models in empirical social science research, particularly for exploratory studies with unclear outcome-generating mechanisms. We compare decision trees and random forests, which offer potential solutions but face interpretability challenges, to classical logistic regression analysis in a case study examining the 2020 Corona pandemic's impact on short-time work. Our analysis highlights the strengths and weaknesses of each approach and offers guidelines for employing interpretable machine learning in similar social science studies. Ultimately, interpretable machine learning emerges as a valuable tool for understanding correlations and uncovering driving mechanisms in situations where conventional regression models may fall short.