Regulation and Frontier Housing Supply

TL;DR

The paper develops a frontier-based approach to quantify housing regulation by separating non-land production costs from price using only observed prices and building heights. It identifies frontier costs from the support of demand and supply shocks and accounts for random and systematic quality via stochastic frontier analysis and bounding arguments, respectively. Applying the method to Israeli urban data (1998–2017), the authors find a mean regulatory tax of about $RT/P \,\approx\,0.48$ with notable cross-locality variation; bounds in 2017 place the mean regulatory tax between $40\%$ and $53\%$ when allowing for systematic quality differences over space and time. The results show regulation correlates positively with centrality, density, and prices, and they document substantial within-locality heterogeneity, implying localized regulatory stringency. Overall, the study provides a rigorous, instrument-free measurement of regulation in multi-floor housing that has important implications for urban policy and housing supply dynamics.

Abstract

Regulation is a major driver of housing supply, yet often difficult to observe directly. This paper estimates frontier cost, the non-land cost of producing housing absent regulation, and regulatory tax, which quantifies regulation in money terms. Working within an urban environment of multi-floor, multi-family housing and using only apartment prices and building heights, we show that the frontier is identified from the support of supply and demand shocks without recourse to instrumental variables. In an application to new Israeli residential construction, and accounting for random housing quality, the estimated mean regulatory tax is 48% of housing prices, with substantial variation across locations. The regulatory tax is positively correlated with centrality, density, and prices. We construct a lower bound for the regulatory tax that allows quality to differ systematically over location and time, by assuming (weak) complementarity between quality and demand. At the end of our sample, when prices are highest and our bound is most informative, we bound the regulatory tax between 40% (using a 2km radius) and 53%.

Figures (16)

• Figure 1: Each point represents an equilibrium price and height. At heights with decreasing economies of scale, the red curve represents the firm's frontier inverse supply. At heights with increasing economies of scale, the blue curve represents the firm's frontier average cost. The regulatory tax is RT. The deviation from the frontier is $U$.
• Figure 2: Parcel-level inverse supply and demand curves.
• Figure 3: Frontier AC and MC curves.
• Figure 4: Frequency of building prices in NIS (rounded to nearest 100) by height.
• Figure 5: The red, blue, and purple points are estimated standard deviations based on \ref{['varv2']}–\ref{['varu2']}. The red and blue curves smooth the estimates with series estimators.