This paper studies urban rental prices for half a millennium (1500–2020) and seven cities: Amsterdam, Antwerp, Bruges, Brussels, Ghent, London, and Paris. Based on a dataset of 436,000 rental cash flow observations, we build continuous annual indices of housing rents, which we employ to study the long-term developments in rental cash flows, as well as their predictability. We find that real rent growth has been limited, but with large differences across cities: average annual growth rates range between 0.12 percent for the Belgian cities to 0.30 percent for Paris. At the market level, we show that sluggish supply adjustment implies that past population growth negatively predicts current rental growth. At the individual asset level, we find that past excess rental growth rates are predictive of future rent revisions, and that increasing steepness of the term structure of contract rents is predictive for future rent levels.
A neural networks that accounts for spatial correlation and time dynamics?
New paper out: A closer look at urban land and structure values – this is important for national accounts and for analysis of real estate risk over time. With John Clapp and Jeff Cohen.
Immobilienanlagen: Angebliches Renditewunder entzaubert / Le « miracle » de l’investissement immobilier démystifié
Summary of our total return paper in German and French.
New working paper with John M. Clapp: An new valuation approach for urban land.
New research accepted for publication at the Review of Financial Studies (RFS) suggests that returns to real estate are solid but not exceptional: No sign of a housing risk premium puzzle.
My previous website went down in flames (or rather: is now hosted in a black cloud).
New research accepted for publication at the Journal of Real Estate Finance and Economics: This paper couples a traditional hedonic model with architectural style classifications from human experts and machine learning (ML) enabled classifiers to estimate sales price premia over architectural styles, both at the building and the neighborhood-level.
How do ML-models arrive at their predictions? Do they do what we hope they do — or are corners cut?
Aesthetic Preferences for Residential Architecture: Finding Ground Truth with Machine Learning Approaches
This paper first collects binary classifications of house pictures from a large group of participants and then trains personalized ML classifiers for each participant.