Nimbyism, Pigovian Equilibrium, Spatial Correlation or all three? Modelling the Distribution of Residential Land and its Impact in 27 EU Countries

We use a new dataset of around 1 / 4 million survey points that allows us both to derive estimates of residential land on a per capita basis for 27 EU countries, and to model its spatial distribution. There is a fairly strong negative correlation between residential land per capita and population density, despite the fact that residential shares are typically very low. There is also a striking lack of correlation between residential land and per capita consumption, but with no indication that this reflects any true economic scarcity value. We model the spatial distribution of reseidential land allowing both for spatial correlation and the impact of a consump- tion externality from nearby housing. We assume that planning policy restricts land supply to match its price to marginal social cost. Our econometric results provide qualitative support for the model; but it is very hard to match our reduced form results to plausible structural parameters unless we assume a social planner who gives a far greater weight to the impact of the externality than to the welfare gains from new housing. In this paper we have analysed a new dataset of around 1 / 4 million survey points, taken from the European Land Use and Cover Area-Frame Statistical Survey (LUCAS), cov- ering 27 EU countries. This allows us both to derive national and regional estimates of residential land on a per capita basis, and model its spatial distribution and economic determinants, in light of a theoretical model in which restrictions on land supply attempt to mimic a Pigovian optimum. Our econometric results show that supply of residential land per capita is affected rather weakly by higher consumption per capita, but somewhat more strongly (and neg- atively) by population density. While this is qualitatively in line with what would be predicted by a truly Pigovian land supply, we show that it is very hard to rationalise the magnitude of these effects with plausible structural parameters.