Research

Drafts available upon request

• Build at your own risk. Integrating natural hazards into land use regulations
  [Solo-authored]
  Abstract

  Insurance coverage provides little incentive for households to adapt their dwellings to the natural risks to which they are exposed. This paper studies the housing market impacts of regulations that limit development in at-risk areas and mandate protective building norms. Using a novel national dataset and variation in the timing and spatial scope of policy implementation, I provide evidence that these regulations initially strongly limit new housing supply, but the long-term decrease in quantities is modest as the supply elasticity increases. I further show that with full insurance coverage, households do not value adaptation but still negatively value living in risky locations, which evidences risk aversion against natural hazards.

  
  

• Mayors strike back: Evidence from upzoning in France
  with Guillaume Chapelle and Camille Urvoy
  Abstract

  To address rising housing prices, national governments have increasingly sought to relax local land use rules through “upzoning” reforms. Yet the effectiveness of such top-down policies remains uncertain. This paper examines the impact of a reform in France that abolished floor area ratios (FAR) from unique zoning data in the Paris Urban area. We exploit within-municipality variation in pre-reform FAR coverage through a difference-in-differences strategy. We show that local governments responded by tightening other regulatory tools, such as maximum heights and building coverage ratios, and increasing permit refusals. These countervailing responses undermined the deregulatory intent of the reform, resulting in negligible effects on real estate markets. Our findings highlight the political and institutional limits of central government efforts to override local zoning decisions.

  
  

• What is France worth ? Residential land value estimates
  with Guillaume Chapelle, Arthur Lemoine, Alain Trannoy, and Etienne Wasmer
  Abstract

  This paper provides a new methodology to estimate the value of residential land at the local level. We assemble a unique, quasi-exhaustive dataset of land transactions across metropolitan France from 2010 to 2019, covering both single-family and multi-family developments. We document a striking divergence in land price dynamics: multi-family land prices grew five times faster than single-family land prices. We develop two families of predictive models, namely hedonic regressions and machine learning algorithms. Hedonic regressions show more credible aggregated and microeconomic results, despite worse predictive performance on multi-family markets.