Discussion about this post

User's avatar
Tim Helm's avatar

Your first problem is trying to read Glaeser literally.

Take him seriously, not literally.

A literal reading will drive you mad because his housing analyis is not designed for coherence. It is designed to hide the economic role of land as distinct from capital (i.e. as a factor of production fixed in supply and earning economic rents), to misidentify the effects of land speculation as consequences of poor regulation, and to pursue deregulation, all in order to provide financial benefit to land owners.

As per my NZAE talk, I believe that neither Glaeser nor anyone else who claims that regulation makes housing expensive has any coherent theory of housing supply.

None of them can answer the simple question "why are houses built?" with anything approaching a coherent explanation.

Ironically, they have many answers (rather, the same answer, given many times) for why houses are NOT built. Which is a really neat trick, because it turns their scientific inadequacies into a strength. If they can't explain why prices aren't lower and quantities aren't higher, because they don't have a theory that works when it comes to explaining the facts, it means they can blame their chosen villain, which is regulation. In this way, incompetence is a great hole to fill with ideology.

For the benefit of your readers, here are several related ways to describe the problems with almost all US literature on housing supply, including Ed Glaeser's work. They're a slightly different take on your column:

1. This entire body of work implicitly assumes that houses are built if and when (and because) there exist arbitrage opportunities in present-day prices. This is the totality of their theory of supply. Their theory is that if the price of houses exceeds the price of construction, construction will go ahead. By implication, they think developers and landowners are myopic (i.e. not forward looking). The process of housing development in their theory of housing supply is one of 'myopic arbitrage', where no developer looks beyond present-day prices when deciding whether to build housing. In other words, if construction is profitable today, then unless encumbered by regulation, construction will go ahead, because developers and landowners in this theory are believed to be not smart enough to realise that it might be even more profitable to delay construction until tomorrow. There is no forward-looking behaviour in this theory. By assumption, therefore, there is never any vacant land already feasible to develop (since all land is developed as soon as first feasible), and there is certainly never any speculation or land banking (choose your term: either way it means not developing housing despite the change of use already being profitable), and therefore there also is no reason for these economists to go looking for, let alone try to understand, these empirical phenomena. According to the old management cliche, "you don't value what you don't measure", but for US housing economists, a better version is "you don't need to measure what you don't acknowledge".

2. They do not appreciate that "construction cost" is not the same thing as "supply cost". In the real world, construction is only one of the opportunity costs of developing a new home. To supply a new home in the real world means giving up the option to develop that land differently in future. Sacrificing that option is a true opportunity cost of supply. The true cost of supply therefore equals the cost of construction AND the value of land options foregone. What is that value? It is identical to the price of land, as set by developers themselves, competing to buy that land in the knowledge that they will be able to sell houses for so much more than construction cost because land is physically scarce and land rents can't be competed away ("they ain't making it any more"). Glaeser wonders why house prices don't match his measure of production cost (MPPC). One reason is that he doesn't include the actual price of land as a cost of production. As you point out, Glaeser's MPPC just includes an arbitrary scaling factor, in lieu of actual land prices. And as you point out, this is only suitable for a world where all locations are identical and location services just happen to be worth 20% of the value of structure services.

3. They use metrics that are blind to history, specifically, the existence of buildings already on the site that are not yet worth tearing down. One of the most bizarre things about comparing house prices to MPPC as a measure of a "regulatory wedge" that drives up house prices is that, even setting aside my previous point, the true opportunity cost of production on any infill site also includes the loss of capital extant on the site. When you must tear down valuable capital in order to supply new housing, that entails an opportunity cost, namely, the present value of the services the existing building could provide with no further investment. In the real world, this value acts as a hurdle to supplying new housing. But it doesn't exist in the MPPC concept, nor in the "marginal cost of land" concept used in the Glaeser + Gyourko "average cost vs marginal cost" method used to "identify" the price effect of planning restrictions. That makes the MPPC and the AC vs MC concepts only useful for vacant land, which is something people replicating those methods in the Australian and NZ contexts are clearly blind to. One of my favourite quotes from Kendall and Tulip (2018) is this: "some barrier has to stop people subdividing properties at low cost and then selling them at high market prices. We discuss several possibilities... and conclude that the most plausible explanation is that subdivision is illegal in large parts of our cities". Could that barrier possibly, just possibly, be the fact there are already buildings on almost all of the land in the cities Kendall and Tulip measure in their data?!?

Expand full comment
Benjamin Heller's avatar

Ugh, Glaeser. It was hilarious for me, as someone who spent a lot of his career trading options and modeling options to read the 2018 paper. My first take was, hey, building on some land is exercising an option to develop. I'm an options trader. I know that it is not optimal exercise behavior to exercise any option the moment it goes one penny in-the-money. But that's what his analytical framework is based on! That's actually how I discovered Cameron Murray's work -- ran across his paper on the importance of considering land banking... Don't sleep on Gyourko either. The Wharton Residential Land Use Restrictiveness Index, which now pops up everywhere in the literature, is a hot mess. In my particular branch of finance, we do a ton of Principal Components Analysis, and the use of it to create the WRLURI is more like abuse. Taking the first principal component and making that your metric? OK, let's replicate it with their data. Hey, wait a second, some of the variables have loadings with signs the opposite of what theory (and common sense -- after all, we are talking about land use rules, not particle physics) would suggest. Oh, and he mixes dichotomous and continous variables in the PCA. I mean, just a friggin' mess. And now the literature is riddled with regressions where WRLURI is in the set of independent variables. Their work is like the PFAS of housing economics. Dangerous forever chemicals that get everywhere...

Expand full comment
14 more comments...

No posts