Evidence-lite zone
The weak evidence behind the economic case against planning regulation
This post contains extracts of key sections from my new working paper with Peter Phibbs. Here’s the abstract.
Economists have taken a strong interest in identifying the costs of land use regulation. But how good is their evidence? We examine three economic approaches to this question—regulatory intensity, zoning tax, and housing as opportunity—and summarise the type of evidence used and its weaknesses. Despite strong claims about enormous economic costs from planning and zoning, the evidence used in these approaches often relies on questionable assumptions to interpret the data. In most cases, there are realistic alternative economic theories that fit the data but generate quite different conclusions. Planners are cautioned against taking these results at face value and adopting these methods without considering the hidden assumptions.
Here’s a key paragraph on the regulatory intensity approach, which looks at correlations between metrics of regulatory stringency and housing outcomes—housing rents, housing asset prices, or the rate of new housing supply.
Consider a correlation between the quantity of new dwellings in a period and regulatory metrics across areas. There is an embedded assumption that the overall rate of new supply across all areas is the sum of independently determined supply in each area. But since filtering of new supply works across locations this is unlikely to be valid. If you compare two areas, one zoned for housing (Area A), and one for non-housing uses (Area B), all new housing will occur in the area zoned for it (Area A). This shows that zoning regulates the location of new housing. But unless there is a plausible counterfactual of what total supply would be if housing was zoned in both areas, this is not evidence of aggregate supply effects. One counterfactual is that the same total number of new dwellings per period would be built but split across both areas.
The best evidence comes from studies that look at whether housing rents, rather than prices, correlate with estimates of the stringency planning regulations (such as in Quigley and Raphael (2005)). But even here, many causal stories can fit the data, such as:
Expensive areas may enact more planning regulations to resolve land use conflicts.
Zoning regulations increase the relative desirability and amenity of areas.
Regulations enacted in formerly low rent areas reduce supply and increase rents.
Without observing the mechanism by which these stories operate, causality cannot be established. For example, increases in amenity will lead to higher rents, but supply effects should exist in the absence of quality changes. This is hard to untangle and is complicated by the fact that increased housing rents due to improving local amenity are traditionally understood as an economic benefit.
Here’s a key section on the regulatory tax approach.
The approach estimates the difference between the marginal price of a unit of land, being the additional price for a parcel one incremental unit of area larger, and the average price of land per unit area, using sales data for detached (single family) dwellings. If the average price is above the marginal price, the difference between these two values is thought to be the result of a regulatory tax imposed by the planning system, since without zoning rules this difference would be traded away (i.e. the per unit land area price would be the same for all lots in an area regardless of lot size). However, whilst this argument might apply for some goods, in the case of housing, it is impossible to trade away the small marginal pieces of land because to construct another dwelling a piece of land at least the size of the dwelling in one location is needed, so very small parcels of land at different locations are of little value.
A decade prior to Glaeser and Gyourko (2003), real estate economists had accounted for this physical reality and shown that private land markets will sustain an equilibrium where average and marginal costs are not equal (Colwell & Sirmans, 1993). They even explained that overly small lots that cannot be recombined with other lots could be worth zero, which would predict regulatory tax estimates of close to 100% of land value in some cases. Such large regulatory taxes have been found, with estimates of 91% of a housing lot price in Los Angeles and Detroit, 88% in San Francisco (Glaeser & Gyourko, 2003, p.30), 74% in Sydney, Australia (Kendall & Tulip, 2018), and 89% in Auckland, New Zealand (Lees, 2018). The zoning tax approach ignored established economic evidence that explained differences between average and marginal costs and instead relied on a new implausible assumption about how property markets work.
Here’s the key argument about the shortcomings of the housing as opportunity approach.
The housing as opportunity approach suggests that if people are not moving to successful cities or suburbs, despite a socio-economic payoff from doing so, there must be a regulatory barrier. Typically, it is assumed that the barrier is planning regulations, which create a housing shortage and high housing costs. This results in more inter-regional inequality, segregated cities, and lower economic growth, with some estimates suggesting that the reduction in growth over four decades is as high as 36% (Hsieh and Moretti, 2019). The unifying assumption of this broad literature is that market mechanisms, via mobility, tend towards equality and diversity across economic, social and racial dimensions.
However, the market-pricing mechanism itself generates location segregation in private property markets. For example, if wealthy people prefer to live near each other, they can use their financial firepower to pay a premium for that location, creating high-priced clusters, as occurred even prior to the invention of zoning. Many models, such as Schelling’s (1969) checkerboard model of voluntary market location choices, explain this process. They provide an alternative assumption about the direction that market forces operate, in this case towards clustering and segregation by social and economic factors. Supporting this alternative assumption is the finding that rezoning activity in New York has been associated with a census tract becoming “whiter”, or less diverse (Davis, 2021).
It is well known that people sort themselves into clusters at different locations. We know that there are Chinatowns in major cities around the world. We know that there are exclusive suburbs where rich people want to live. We even cluster based on our politics. This is what the tribal human species does. Assuming that market forces will lead to the opposite outcome is weird.
Weird is how I would also describe the assumptions that have been smuggled into the regulatory tax and regulatory intensity approaches. The purpose of our new paper is to highlight how all three approaches to examining the link between planning and housing prices are reliant on such assumptions.