Research

Job Market Paper

PDF · Please check for regular updates

The Effect of Public Sector Relocations on Regional Development in Germany

Combining newly collected relocation data with administrative employer–employee records, I estimate the consequences of public-job relocations for both the places gaining and losing employment. I then develop a quantitative spatial model in which public and private employment generate different productivity spillovers, linking the reduced-form evidence to the spatial mechanism.

Selected presentations: NBER, CEPR, EALE, UEA, IAB

Book chapter · 2026

The Potential of Public Employment Reallocation as a Place-Based Policy

In The Economics of Place-Based Policies, ed. Cécile Gaubert, Gordon H. Hanson, and David Neumark. Chicago: University of Chicago Press, 2026.

Reviews international evidence on capital-city relocations and public-agency decentralization. Moving government jobs can generate employment and population gains in receiving places, with effects shaped by the scale, skill mix, and permanence of the relocation; the chapter also examines consequences for the regions that lose public employment.

Work in progress

The Growth–Affordability Trade-off in Zoning

With Gabriel Ahlfeldt, Jan Bakker, and Nina Gläser

We combine matched employer–employee data with an original postcode-level zoning dataset covering all of Germany and discipline a spatial general-equilibrium model. The project quantifies how allocating land between residential and commercial uses affects housing affordability, firm agglomeration, productivity, distributional outcomes, and welfare.

Policy paper

35 cents less for gasoline and 17 cents less for diesel — The fuel tax reduction was passed on to consumers — on average

With S. Syga · ifo Dresden berichtet 29 (2022), 13–18

We exploit Germany’s temporary 2022 fuel-tax reduction and comparable reforms in other European countries to estimate pass-through using dynamic difference-in-differences. The analysis shows how quickly and how fully temporary tax relief reaches consumers, and how pass-through varies across countries and over time.

Coverage: German Federal Antitrust Agency (Bundeskartellamt) · Monopolies Commission of the German Bundestag · DNN

Working paper

Do place-based policies matter at the poll?

With Luisa Doerr and Florian Dorn (ifo Munich)

We exploit quasi-experimental variation from Bavaria’s rule-based public-agency relocations to ask whether place-based economic policy changes voting behavior. The design links a large regional-development intervention to electoral outcomes across treated municipalities and elections.

Working paper

The Drivers of Far-Right Populism in Western Democracies

With Patrick Zwerschke

We assemble a cross-country dataset covering more than 26 candidate drivers of far-right populism in over 30 Western democracies and explicitly confront model uncertainty. Bayesian Model Averaging identifies robust determinants, while Bayesian Additive Regression Trees uncover nonlinearities that standard linear models can miss.

Working paper

A comment on Xu (2022) “Reshaping Global Trade: The Immediate and Long-Term Effects of Bank Failures”

With Guillaume Bérard (Luxembourg Institute of Socio-Economic Research) and Priyam Verma (Aix-Marseille Univ, CNRS, EHESS, AMSE)

We reproduce Xu’s estimates of the long-run trade effects of the 1866 banking crisis, identify implementation issues, and reassess the results using alternative weighting, spatial-HAC inference, and staggered difference-in-differences. The paper’s central result remains robust.

Working paper

Reproducible and Robust by Design: The Reproducibility and Robustness of Social Science Research by Method and Field

With Nikolai Cook, Derek Mikola, Juan P. Aparicio, Lenka Fiala, Ghina Abdul Baki, Abel Brodeur, and many co-authors (Institute for Replication)

We document the reproducibility and robustness of 274 published articles in economics and political science by identification method and field. Results can be reproduced from the provided data and code for 89% of articles, 32% contain a coding error or research inconsistency, and 70% of originally statistically significant results remain significant under reasonable alternative research choices. Experimental methods prove more robust than observational ones.