Bad housing economics never dies, it only accumulates more ignored "limitations"

When a method takes on a life of its own it becomes hard to kill

Announcement: I am in Wellington, New Zealand, and want to invite any Fresh Economic Thinking readers who can make it to a meet-up dinner and drinks on Wednesday 6th December from 5.30 pm at Fork & Brewer. Look forward to meeting some readers there. No need to RSVP—there will be space for you. But feel free to reply to this email if you like.

My colleague Tim Helm and I have recently explained (twice over) why a popular method used to assess the effect on total new housing dwelling development in Auckland due to upzoning is unsuitable for the task.

It hasn’t stopped the method from being defended. Nor has it stopped those who want to believe the result from carrying on as if there are no problems with the method that undermine its main claim.

And I am confident nothing will change.

I have been through it all before. A paper comes out with implausibly large effects from planning and zoning based on a method that makes little sense, yet because of the desire to believe the result it gets widely accepted, replicated, and integrated into the lore of the discipline.

All criticism is ignored.

The last time was back in 2018 when I discovered Harvard economist Ed Glaeser’s zoning tax method for estimating the effect of zoning regulation on home prices.

Here’s how that went down.

1. Ross Kendall and Peter Tulip used the method in a Reserve Bank of Australia discussion paper in 2018. It attracted a great deal of media attention (and my attention), as you might expect with results like this:
   ”We estimate that zoning restrictions raised the average price of detached houses, relative to supply costs, by 69 per cent in Melbourne, 42 per cent in Brisbane and 54 per cent in Perth.”

2. Given the size of the results, the method was of great interest. I replicated the method on my blog using pre-planning colonial Queensland land sales data from the 1850s. If this method established the price effect from zoning, then how could it be true that in pre-zoning Queensland, with a population of 15,000, the method also finds a large price effect?

3. I was part of a video trying to bring the discussion back to reality.

4. I wrote a very detailed academic piece about the method, which is now published in the journal Environment and Planning A. Interestingly, one of the reviewers wrote the following to me in their comments:
   “The main point of this paper is both correct and important: the popular hedonic price method of calculating a "regulatory tax" initiated by Glaeser and Gyourko (2003) (henceforth G&G) has little or no scientific merit, and should not be used.

   ...So the G&G method is ripe for criticism. The early critiques by Somerville (2005) and O'Flaherty (2003) were massively ignored, and their authors probably didn't notice because they thought that the G&G method was too ditzy to go anywhere. But as Murray points out, the method has become popular and its results have become influential. So Murray's critique is timely.”

5. I wrote a short summary for journalists and interested laymen about how ridiculous the method is. In the process of writing this piece, I discovered that this whole debate already happened in the early 1990s in this 1991 paper and this 1993 paper.

6. I defended my arguments about why the method makes no sense on Bryan Caplan’s Bet On It substack. Peter Tulip quickly defended it.

7. I’ve repeatedly explained all of this on Twitter for five years.

Now, people just keep using it anyway, list some “Limitations on Methodology”, and carry on regardless, as the image below shows.

Source: https://www.cdhowe.org/sites/default/files/2023-05/For%20release%20E-Brief_341.pdf

I think I can try one more time to explain the method and why it makes no sense.

Imagine you are a car maker, producing a variety of car models, such as hatchbacks, sedans, vans, utes, and more. An economist looks at the stock of cars on the road and thinks something like the following.

“You know, all the extra steel, alloys, plastic, and rubber used to make bigger cars could have instead be used to make more smaller cars. If the marginal price of an extra kilogram of car material is lower than the average price per kilogram of a car, then car-makers are not optimising their use of materials to get the highest price-per-kilogram return. There must be a regulatory barrier. We can compare the average price per kilogram of a car to the marginal price per kilogram and the size of this gap will indicate the degree that regulations are inhibiting car supply.”

As a car maker, you would think these economists were insane. Why should cars be priced at the same dollars per kilogram regardless of size? However, if there were regulations that were costly for you, it might make sense to go along with them and use this research to help justify policy changes that would suit your bottom line.

But deep down, if people kept using this method of weighing cars and comparing their weights to the prices and simply added “limitations on methodology” to their analysis you’d still think they were mad.

Alas, this is where we are at with zoning and housing analysis. The desire to believe is strong. And there is now another case emerging.

Another case of bad housing research

An influential 2019 paper by Hsieh and Moretti entitled Housing Constraints and Spatial Misallocation, which has been cited over 800 times, is riddled with errors. The claim in that paper is that the rate of economic growth in the United States was 36% lower for nearly half a century because of land use regulations and that GDP would be 3.7% higher without these regulations.

In 2021, Bryan Caplan noted some anomalous numbers in key tables. Bryan wondered how such a large effect on the growth rate over such a long period could have such a small effect on the GDP level after fifty years or so. The authors conceded that there were errors in their tables and that the true effect should be that GDP would be 14% higher without land use regulation, not 3.7%. An error that makes the result even more shocking!

But now, a complete replication by Brian Greaney using the code and data from the authors has found numerous coding errors. Even assuming their data and method make sense (which is a very important but different question), correcting these errors took the result—which had jumped from 3.7% to 14%—back down to 0.2%. Not only that but simply working through their model assumptions correctly reverses the underlying basis for the claim that removing land-use regulation would increase output. Greaney summarises that “their model predicts that the land-use deregulation experiment they propose would decrease output.” Debate continues, and it’s not looking good for Hsieh and Moretti.

But will this stop the desire to believe? No. The game now is to pick different papers to cite instead to support the claim of huge growth effects from land use regulation!

Bad housing economics never dies.

---

Advertisement – Fat Tail Investment Research

Fresh Economic Thinking is reader-supported (thank you). I will also post advertisements from time to time.

Fat Tail Investment Research is an Aussie outfit that publishes research on a range of investment options. The people there have been readers of my work for many years.

Sign up for free using this link for their investment insights about stocks that historically pay a consistent income AND could deliver capital growth.

---

Thank you for reading Fresh Economic Thinking. Please subscribe and share this article.

Paid subscriptions are available, but for now, only get you good vibes. In 2024, better things are coming. My new book, The Great Housing Hijack, will be released and with your help, Fresh Economic Thinking will evolve into Australia's newest think tank.