Land demand is high in cities. Cities have access to good jobs, people and specialized consumption services. High demand and the limited supply of urban land translate into high land prices. In turn, high prices promote high density land uses. If a parcel in the square mile costs hundreds of millions, expect to find a tall office building there rather than a bucolic bungalow. Despite all this, it is common to find vacant or severely underused land in cities. Old factories, derelict dwellings, abandoned car parks; all are common images from urban life. But, for an economist, this is quite surprising. Is it that some land is simply unresponsive to demand conditions?
Having a good answer to this question is important. Unused land makes our cities larger and our commutes longer. In stark distinction with parks, or undeveloped green-field land, abandoned previously-developed parcels are usually quite unpleasant (in econ-talk, they generate strong negative externalities). Planning authorities and local governments often have specific policies to promote land re-development, by offering tax breaks, grants or financing options to developers of infill land. Understanding how demand affects re-development helps us to anticipate whether these policies will be successful. Moreover, it also gives us a hint on whether or not market forces can themselves achieve the goal of re-developing these sites.
In a recent Centre for Economic Performance, Urban and Spatial Programme discussion paper, I address this question by using data on brownfields across England. As you may know, in the UK brownfield is the name usually given to land which was previously developed but is currently underused or vacant. Policies fostering redevelopment of brownfield land in priority over green-field (“brownfield-first”) have been a feature of the planning system at least since the 1990s and have seen some resurgence of interest in recent years. Whether these policies are actually effective in encouraging brownfield re-development in response to housing demand remains unclear. In this paper, I use data on brownfield sites to test whether the prevalence of empty land in cities is indeed sensitive to demand conditions. In essence, I want to test whether high demand (or high price) areas are more likely to be re-developed than otherwise identical areas where demand is lower. The resulting estimate measures the price sensitiveness of brownfield re-development. If you are versed in economics, this is like estimating a price elasticity of land supply. There are several challenges when trying to estimate these parameters properly. I briefly discuss them below in case you are interested.
My findings indicate that brownfield re-development is indeed affected by local demand conditions. In the long run, re-development follows demand. Quantitatively, a 1% increase in prices leads to a 0.07% decrease in the probability of having brownfield land in a given hectare. This number looks very small but the baseline probability of finding brownfield land in my sample is also low, standing at roughly 1.5%. Hence, I conclude that price differences can induce substantial variation in the presence of empty land in our cities. Importantly, this is the case despite the tight planning restrictions that characterize the UK. This is not to say that planning restrictions cannot influence re-development. In my paper, I show that re-development of parcels in areas with tougher planning (such as the South East) is less sensitive to demand conditions.
To sum up, my results show that market forces do affect brownfield development in the long run. According to my estimates, a large but plausible increase of prices of roughly 20% would lead to re-development of a large fraction of brownfields in English cities. It also suggests incentive based policies such as grants or tax breaks could, at least in principle, achieve substantial long-run re-development. Finally, the paper shows that equilibrium land supply is neither fixed nor determined exclusively by sprawl; it can also be shaped by development within the cities’ footprint.
My analysis is based on data for previously-developed land sites in England, as recorded in the National Land Use Database of Previously Developed Land (NLUD-PDL). I use the 2007 version of this dataset, but newer versions can be found online. Data on prices are obtained from the land registry and data for land use changes are obtained from the Land Use Change Survey.
There are a few technical hurdles we have to overcome before we can obtain credible estimates of the price sensitivity of re-development. On the first place, because brownfield sites tend to generate negative externalities on nearby residents, they themselves affect prices. A growing literature has made this case over the last decade. As a result, there is substantial scope for reverse causality, from vacant sites to prices. In addition, re-development costs may be correlated with amenities that affect land prices directly. For example, in a monocentric city, low amenity areas away from the city centre may be harder to re-develop because of reduced accessibility. Other confounders, such as differences in the type of sites (e.g. residential vs. industrial) in different areas could also bias naïve estimates.
In order to overcome this problem I need a demand shifter, a variable that affects demand without affecting supply conditions. In this study, I use school quality as my demand shifter. A long literature has documented price effects of school quality in residential markets and the UK is not an exception (see for example Gibbons, Machin and Silva, 2013). Combining data from school quality with detailed maps, I implement a two-stage boundary discontinuity design which uses school quality as an instrument in a regression of brownfield location variables on prices. The boundary discontinuity design uses school admission boundaries to ensure I am comparing locations which are very close to one another but have access to primary education of different quality. This method is what allows me to obtain credible estimates of the price sensitivity of re-development.
Note: See Figure description in the text.
An illustration of the result is provided in the figure above. The horizontal axis represents distance to a school-admission (county) boundary, with negative distances corresponding to the side with low average school quality and positive distances corresponding to the side with high average school quality. The vertical axis measures the probability that a given hectare contains brownfield land. Fourth degree polynomials fitted on the raw data represented in solid lines. The discontinuity observed in the middle of the figure shows that brownfield land is less common in high demand locations. More details can be found in the paper.
Gamper-Rabindran, S., & Timmins, C. (2013). Does cleanup of hazardous waste sites raise housing values? Evidence of spatially localized benefits. Journal of Environmental Economics and Management, 65(3), 345-360.
Gibbons, S., Machin, S., & Silva, O. (2013). Valuing school quality using boundary discontinuities. Journal of Urban Economics, 75, 15-28.
Greenstone, M., & Gallagher, J. (2008). Does hazardous waste matter? Evidence from the housing market and the superfund program. The Quarterly Journal of Economics, 123(3), 951-1003.
 My estimates however, are based on very long run changes. Anecdotal evidence shows that re-development can take a long time. It took 50 years to re-develop the areas around the London Royal Docks or Battersea Power station.
 See for example the contributions in Greenstone and Gallagher (2008) and Gamper-Rabindran and Timmins (2013).