Working Papers

  • Machine Learning, Building Vintage and Property Values 

    (with Erik B. Johnson, Alabama) This paper makes three contributions: First, it introduces an algorithm that collects pictures of individual buildings from Google Street View. Second, it trains a deep convolutional neural network (CNN) to classify residential buildings into architectural styles, taking into account spatial dependencies of building vintages. Third, it investigates whether architectural styles influence house prices. For re-sales, the architectural style is a significant determinant of transaction prices while no such effect is found for new buildings. This indicates that any premia are for vintage-related quality characteristics and not for a home’s beauty.

  • 500 Years of Urban Rents, Housing Quality and Affordability 

    (with P. Eichholtz and M. Korevaar) We study urban housing rents and quality from 1500 to 2017 for Amsterdam, Antwerp, Bruges, Brussels, Ghent, London, and Paris. Based on a dataset of 436,000 rent observations, we relate rent developments to wages and consumer prices. Real rents have developed similarly in the long term, but reflect shorter-run differences in local economic and political conditions. Long-run growth in real rents has been limited. The ratio of wages to market rents was stationary before 1900, but grew considerably after that. Most of the increase in housing expenditure that did occur is attributable to increasing housing quality rather than rising rent

  • Market Concentration & Asset Pricing: Evidence from the Housing Market

    (with Joseph T. L. Ooi, NUS) This paper examines the impact of concentrated market share of suppliers on asset pricing based on a sample of 854 new private housing developments launched in Singapore between 2003 and 2014. The market structure for new housing is less than perfectly competitive due to spatial heterogeneity and land supply inelasticity. We find evidence of property developers taking advantage of their localised market power to sell new housing units at higher prices. The result highlights a significant consequence of industries becoming increasingly more concentrated in the hands of a few dominant players or monopolies in the “new” economy era.

  • That’s what we paid for it – The Spell of the Home Purchase Price through the Centuries 

    (with Piet Eichholtz, Maastricht University) This paper is the first to provide empirical insights into the long-term nature of the loss aversion bias. Using a database of Amsterdam housing transactions spanning 324 years, the paper studies the question whether loss aversion was present in centuries past, whether its effects were stable across these centuries, and whether the psychological effect of the purchase price on selling behavior eroded with time and through the occurrence of important events.

    The purchase price of the house is found to have been a psychological anchor, below which home owners were reluctant to sell their home. This result holds for 17th and 18th century Dutch home owners as well as for those who followed in their wake, but loss aversion appears to get stronger over the centuries. The anchoring power of the purchase price was strong: it survived the death of the original owner when the house passed on to the heirs. It was however diminished by loss realizations in housing transactions in the direct vicinity, and even more so by the occurrence of wars involving foreign occupation. The aversion to a loss relative to the purchase price was also gradually reduced by the time passed since the purchase.