Research

Collaboration in Technology and Multinational Production  (JMP, new draft coming soon)

Abstract: How flexible are multinational production networks when assembly technologies are specialized and costly to reconfigure? I develop and estimate a structural multinational production model  incorporating discrete production technology (platforms) choices and inter-firm licensing, using detailed global automotive industry from 2000–2023. Exploiting the 2011 Tōhoku earthquake as a quasi-natural experiment, I show that platform-specific production constraints significantly limit firms’ ability to reallocate output across space. In the model, platform specialization restricts the feasible assembly set, raising marginal costs relative to standard frameworks that assume full technological substitutability of locations, while licensing relaxes these constraints at the expense of fixed collaboration costs. Quantitatively, accounting for technology increases median costs of firms by about 28%, whereas licensing reduces expected costs by 4%. The model allows, in addition to analyzing trade policy shocks, to simulate industrial policies that explicitly target the access to production technologies. While a unilateral export or import ban of specific technologies raise production costs and induce reallocations comparable to tariff shocks, the distributional consequences are different and largely borne by individual firms.

The U.S.-China Trade War and the Geography of Global Production, with Harald Fadinger, Lei Li & Jan Schymik

Abstract:  We study how the U.S.-China trade war affected manufacturing activity in third countries using a novel plant-level dataset covering millions of establishments in 50 major economies, including affiliates of more than 200,000 multinational enterprises (MNEs). Combining establishment-level data with detailed tariff information, we estimate the effects of U.S. and Chinese punitive bilateral output and input tariffs on sales, employment, and establishments across countries, industries, and stages of production. We find that third-country effects of the trade war are highly heterogeneous and largely offsetting, yielding moderately negative net effects overall. Most of the adjustment is driven by multinational enterprises reallocating activity across affiliate networks, while domestic firms respond much less.

Flooded with Uncertainty: Data and Design Pitfalls in Measuring the Granular Impact of Natural Disasters, with Thierry Mayer and Michele Fioretti (draft upon request)

Abstract:  Extreme weather events pose a growing risk to global production. The growing body of literature on the impacts of natural disasters, such as floods, on households, firms, and economic development largely relies on event study designs exploiting publicly available climate data. This paper documents the challenges and pitfalls associated with this approach using a commonly used database of floods. To do so, we study the responses and adaptations to flood exposure of firms in the automotive industry. We show that, in our setting, (i) floods are often missing or spatially / temporarily misallocated in the data; (ii) canonical Difference-in-Differences methods are sensitive to the design and lead to qualitatively different conclusions. We explore the mechanisms behind this dispersion and provide guidance on how to apply the methods to these datasets to deliver credible results.

Heterogeneous Regulations Harmonization in Global Production, with Monika Mrázová & Céline Carrère