My forthcoming Build, Baby, Build: The Science and Ethics of Housing Regulation gives readers a Big Picture tour of research on housing regulation. Like Open Borders, the new non-fiction graphic novel strives to show the radical implications of mainstream research. Still, due diligence requires me to search for serious critiques of the mainstream position. Most notably, what are the best arguments against the mainstream view that housing regulation imposes a massive “zoning tax” that drastically inflates the price of housing?
The most common technique for estimating this zoning tax goes back to this 2003 paper by Edward Glaeser and Joseph Gyourko. The essence of their method:
Look at housing prices.
Estimate the price of vacant land.
Estimate construction costs.
Now remember: In an unregulated market, long-run housing prices would equal vacant land prices plus construction costs.
The estimated total zoning tax on a property, therefore, equals (1)-(2)-(3).
Alternate perspective: If land you can build on is worth the same as land you can’t build on, the zoning tax is zero. Otherwise, it’s positive.
Though the Glaeser-Gyourko method has been hugely influential, it has one vocal critic: Cameron Murray from the University of Sydney. Here’s the quick version of Murray’s critique; here’s the long version. (Murray, in turn, heavily credits this early critique).
Murray has multiple criticisms. Let’s start with the simplest: Glaeser and Gyourko estimate (2) by comparing the prices of otherwise similar homes with slightly different amounts of land. Since the market puts little value on marginal land, they blame housing regulation. Murray objects, however, that this could reflect mere indivisibilities:
1. Land is not a quantity. People are not indifferent between a parcel of land of 500 square meters and five hundred 1 square meter parcels of land in different locations.
2. Housing lots come in ‘lumps’ and hence subdivision represents a ‘suitcase problem’ of choosing a fixed number of lots and their sizes to fit in the plot being subdivided. Even a new subdivision will therefore not always be able to have housing lots of the exact size where marginal and average land prices are equal.
3. Even if new housing lots are created optimally when subdivided, they won’t stay that way for long as demand changes over time at each location. A continuous and complete rebuilding of the city is required to maintain optimally sized lots.
To instantly grasp the point, check out Figure 1 from Murray’s shorter paper:
Even under laissez-faire, these scraps of land will have little market value, right?
Murray concludes:
While a lower marginal compared to average land price for housing lots has broadly been thought to be an indicator of the economic cost of planning constraints or development taxes, in reality this measure contains no information about the effect of planning controls.
Indeed, the popularity of the G&G method is somewhat surprising, given it was initially published alongside a critique that came to this exact conclusion.
My reaction: Murray is clearly right that low value of marginal land might reflect mere “lumpiness.” But to conclude that low value of marginal land “contains no information” is still false. At least in the U.S., minimum lot sizes of an acre and more are now common. It would clearly be possible to subdivide such lots and build additional homes - and the value of the land would clearly be far greater if owners had such permission.
Furthermore, under laissez-faire, much of the “lumpiness” would probably never have happened in the first place. In Murray’s Figure 1, low marginal value of land would have tempted unregulated builders to build 12 or 15 houses. While this conceivably simply shows builders’ lack of foresight, the strictness of minimum lot size regulations should give us the opposite default.
The cautious reaction to Murray’s critique would be to look for better ways to measure the price of land you can’t build on. Gyourko and Krimmel’s recent “The Impact of Local Residential Land Use Restrictions on Land Values Across and Within Single Family Housing Markets” (Journal of Urban Economics 2021; ungated version here) does just that. Instead of looking at the prices of extra land on existing houses, they look at the prices of large vacant lots. How large?
In Atlanta, the average parcel size is about 1.1 million square feet, or nearly 25 acres; the size distribution is skewed by some very large parcels, but even the median vacant land parcel in this metropolitan area (within 30 miles of the area centroid) is 10 acres in size. There are some large residential land tracts traded in the Bay Area, too. In the San Francisco and San Jose CBSAs, the mean parcel sizes are about 14 and 27 acres, respectively. However, the medians are much smaller at about 3 and 7 acres, respectively.
This seems to almost totally avoid Murray’s challenge, yet Gyourko and Krimmel still find large zoning taxes. Indeed, this better data shows that Glaeser and Gyourko’s earlier work was much too optimistic about zoning. Previously, Glaeser and Gyourko found that the zoning tax was only severe in places like the Bay Area and the Northeast corridor, and negligible elsewhere. Gyourko and Krimmel find, however, that the zoning tax is moderate to high in almost every central city where they have data. The big difference is between regions where you can avoid high zoning taxes by moving thirty miles from downtown, and regions where zoning taxes remain high as far as the eye can see.
The main problem with Gyourko and Krimmel is that their measure understates the effect of zoning. Instead of looking at the prices of vacant lots approved for single family development, it would be better to look at the prices of vacant lots where development is strictly prohibited. Furthermore, their approach fails to capture the additional zoning taxes on multifamily housing, skyscrapers, and so on. I don’t mean this as a criticism; as far as I know, Gyourko and Krimmel’s estimates are the best available. My point is that their estimates are conservative.
Murray has many additional criticisms of the Glaeser-Gyourko approach; see Table A2 of his longer paper. For me, the most notable are:
Citing Sommerville (2005), Murray states “that G&G’s method requires that
town planning regulations do not generate ‘any benefits that manifest themselves in a higher land price…’”
The Glaeser-Gyourko method detects zoning taxes in colonial Australia and ancient Mesopotamia.
If zoning taxes are large, then developers should be pro-regulation: “The greatest calls to remove planning constraints usually come from housing developers themselves. If removing constraints did radically reduce land prices, then housing developers are undermining their own profitability by calling for more relaxed planning controls. They should instead be lobbying for tighter planning controls. This alone should raise questions about the mechanism by which price effects from planning regulations occur.”
If Glaeser and Gyourko are right, deregulation would lead to absurdly large falls in U.S. real estate prices: “Matching these estimates to estimates of the total value of residential land in these cities in 2010 from Albouy et al. (2017) leads to the conclusion that US$2.3 trillion of residential land value in these cities alone was created by town planning rules. Removing planning constraints in these cities could rapidly wipe US$2.3 trillion in land value from the balance sheets of homeowners in these cities – a shock to household balance sheets similar to the late 2000s financial crisis, where home prices (house and land) fell 31% nationally according to the Case–Schiller Index. This seems implausible.”
My response:
Glaeser and Gyourko’s method unambiguously detects the effect of regulation on the cost of individual homes. They are actually open to the possibility that the social benefits of regulation outweigh these cost effects, though they are fairly skeptical about the magnitudes. Which is very consistent with my reading of the evidence.
I would not expect colonial Australia or ancient Mesopotamia to have big zoning taxes, though perhaps if I studied these cases more I would change my mind. But in any case, this just confirms that zoning is not the sole possible cause of low marginal land values. It shows that the Glaeser-Gyourko method is imperfect, not worthless.
It would make sense for developers with most of their net worth already tied up in land to be pro-regulation. The standard business model for developers, however, is to buy land, develop it, then sell out. They make their money by building new stuff, not by sitting on real estate and waiting for it to appreciate. As a result, developers’ support for deregulation is exactly what the mainstream view predicts. Indeed, given the ubiquity of NIMBYism, the lobbying of developers is practically the only reason anything gets built at all.
If U.S. real estate values can fall 31% for reasons that remain somewhat mysterious, why couldn’t they fall a similar amount because of massive deregulation? If the point is that sudden housing price falls are dangerous for the financial system, take comfort in the fact that deregulation will come slowly, if it comes at all.
Closing thought: You might think that the punchline of Murray’s work would be, “Housing regulation almost certainly has a huge effect on housing prices, but the Glaeser-Gyourko method of measuring this effect is unconvincing.” Strangely, though, he seems to treat his critique as proof that zoning taxes are basically zero. Murray again:
Regardless of this academic history, housing policy-makers globally have been receptive to conclusions based on G&G’s method and many have identified planning policies as the major determinant of land prices. Such a belief is likely to create a bias toward policies that sacrifice the amenity benefits of planning controls in order to pursue illusory home price benefits. There is thus a need for a more robust critique that can shift the academic debate and reach policy-makers. (emphasis mine)
On the contrary, it is the amenity benefits that are largely illusory. While density has obvious drawbacks like congestion, these are easily solved with peakload pricing of roads, parking, and so on. Stopping housing to stop congestion is truly using a sword to kill a mosquito.
Furthermore, density is a major neglected cause of amenities. Regulation is able to boost housing prices a lot because people are eager to live next to lots of other people, so they can enjoy the resultant job, shopping, and social opportunities. This is a strong sign that the net effect of density is positive. What housing regulation does is bar the free market from making much-desired density nice and cheap. Contrary to most YIMBYs, the main barrier to deregulation isn’t NIMBY selfishness. It’s straight-up economic illiteracy.
Subscribe to Bet On It
Caplan and Candor
Next week: “Land-use restrictions in ancient Mesopotamia.”
Higher density doesn't result in more congestion. Holding population and lane-miles the same, higher density reduces congestion. The average miles traveled reduces which reduces total trip-miles.
Land use restrictions can defeat this by requiring businesses, retail, and housing far away. Also, it makes everyone commute in the same direction making poor use of half the lanes.