I’m the Deputy Head of the Digital Economy Department and Head of the Junior Research Group Digital Market Design at ZEW Mannheim. We conduct academic research, give advice to policymakers, industry practitioners, and the public based on our research, and develop talent.

You can find more details below and in my curriculum vitae.

Dominik Rehse

Research

Digital & financial markets Market design Artificial intelligence Data economy

Published

Fostering participation in digital contact tracing With Felix Tremöhlen

Digital contact tracing creates a quid-pro-quo data market, effective only if participation is broad on the extensive margin and sustained on the intensive margin. Both margins lagged during SARS-CoV-2, limiting effectiveness. We review the literature on how to bolster adoption and use.

Information Economics and Policy (2022)
Policy brief Podcast
Wirtschaftsdienst Tagesspiegel
Bild Web.de
Digital contact tracing Digital contact tracing The effects of uncertainty on market liquidity: Evidence from Hurricane Sandy With Ryan Riordan, Nico Rottke and Joachim Zietz

We test how uncertainty affects market liquidity using Hurricane Sandy as a natural experiment. The storm's unprecedented strength and course made damages in New York City highly uncertain. We compare stock liquidity of REITs with and without properties in the NYC evacuation zone before landfall using geolocation data. Affected REITs show reduced trading volume and wider bid-ask spreads, confirming theory on uncertainty's negative impact on liquidity.

Journal of Financial Economics (2019) Award Hurricane Sandy impact Hurricane Sandy impact

Working papers

Competition among digital services: Evidence from the 2021 Meta outage With Sebastian Valet

We study behavioral responses to a global outage of Meta's services with high-frequency tracking data. Key findings: Users primarily switched to other social media and messaging apps, some crossed service categories, substitution varied by demographics, multi-platform users switched more readily, switching increased as the outage continued, patterns differed between countries, and non-Meta service usage remained elevated post-outage.

Working paper
Tagesspiegel Der Standard
Meta outage Meta outage Designing incentive and coordination schemes for red teaming generative AI With Sebastian Valet and Johannes Walter

Red teaming is frequently used to uncover unwanted behavior in generative AI models and systems. We provide a formal framework for designing red teaming markets. The framework addresses core operational questions: defining goals, incentivizing and coordinating red teamers, verifying proper conduct, and comparing results across models/systems. We explore these questions through theory, simulations, and experiments.

Paper in preparation Policy brief Academic workshop Berlin policy workshop Brussels lunch debate Op-ed Red teaming Red teaming Estimating utility functions for machine learning models With Sebastian Valet and Johannes Walter

The double-descent phenomenon and the application context of machine learning models often yield non-standard utility functions, such as non-convex and non-monotonic function segments. We develop and illustrate a neural network architecture that can address such complex utility functions in interpretable ways. The architecture also allows us to estimate utility and demand functions simultaneously, which we illustrate in simulations.

Paper in preparation Code in preparation Industry brief of research consortium Non-standard utility functions Non-standard utility functions

Work in progress

Early consumer adoption patterns of generative AI With Chiara Farronato and Sebastian Valet

We analyze early consumer adoption of generative AI services using large-scale, individual-level tracking data. The high-frequency and granular nature of the data allows us to model consumer choices at the finest temporal resolution, capturing precisely when users first engage with the technology, how their usage evolves over time and how it substitutes or complements other services.

Work in progress Generative AI adoption Generative AI adoption Designing pseudo-markets for digital services With Sebastian Valet

We develop a method to experimentally induce non-usage of digital services by paying consumers to forgo individual services and service bundles in order to identify complements and substitutes from behavioral responses and approximate consumer welfare. A pilot study with a representative German sample revealed that consumers cannot reliably self-assess their willingness-to-accept, and that income effects from overly generous offers could distort measurements. We are currently looking for sources of funding to run a larger-scale experiment.

Work in progress Pseudo-markets Pseudo-markets Testing and correcting for measurement order bias With Sebastian Valet

We define measurement order bias as arising from sampling units of observation in a non-random sequence, such as when geographic clustering occurs during data collection. This distorts the interpretation of intermediate outcomes. To address this, we propose a statistical test to detect the bias and a method to correct it in cases of repeatedly sampling the same or similar units of observation.

Work in progress Measurement order Measurement order Algorithmic coercion with faster pricing: Evidence from German gas stations With Jacob Schildknecht

We test theoretical implications of Brown and MacKay (2025) on the effects of pricing speed differentials using high-frequency data on the population of German gas stations. We develop empirical measures for speed and commitment, the two core concepts of their model, and find evidence consistent with their predictions.

Work in progress Algorithmic pricing speed Algorithmic pricing speed

Advice

Policy makers Industry practitioners General public

AI policy

Competition policy

Data economy

Financial markets

Talent

Advising Mentoring Teaching

Current advisees and mentees

Past advisees and mentees

Teaching