Spot the fallacy: "banning people from living within 1km of the city center would not decrease total welfare, because people just pay less in rent to compensate for a longer commute".

Two very different ways for upzoning to *not* increase land values: A) maximize housing supply via high-density upzoning, which maximizes land values; then apply land value tax, so post-tax land lift =0. B) restrict density and add poison pills; here, pre-tax land lift =0. 1/

How to tell when zoning is *not* a binding constraint: when you upzone one more parcel, its land value stays the same (ie. 'land lift' =0). So if upzoning a parcel does increase its land value, then you know zoning is a constraint on housing supply. 1/

Saying zoning is not a constraint because some homes get permitted is like saying your personal budget is not a constraint because you bought some groceries.

Basic fallacy in this Groundwork report: it treats profitability and zoning as separate constraints on new housing. But zoning constrains supply *precisely by* making projects unprofitable! Strict zoning -> scarce apartment-zoned land -> expensive apartment-zoned land -> lower profits. 1/🧵

New housing supply enables filtering of *other* homes through vacancy chains: someone moves into the new home, which frees up their original old home for someone poorer to live in. The old home 'filters down' in the short run. The new home also filters down, but over decades.

As all economists know, removing zoning constraints (higher z̄) leads to lower prices (p*) because it increases housing supply:

New housing in the city allows a family living in the suburbs to sell their cars and move to the city. Since their transportation costs are much lower, they can now spend even more on housing. Has affordability improved?

In their study of Sao Paulo's upzoning, Anagol et al. assume that developer profits are 10% of new home prices. This is a strong assumption: - no entry of new developers to compete away profits - not accounting for losses from failed projects

If developers always build up to the maximum (density, height, etc.) that’s allowed in your city, that’s a sign that your zoning rules are substantially restricting housing supply.

Bike lane logic.

Building higher-density housing lets you save on land costs, by putting more homes on the same parcel. But for the same reason, people are willing to pay more. Which factor wins? Data from Auckland shows that upzoning reduced the land cost to floor area ratio (THA, blue). 1/

DT asked for some of my thoughts on multiplexes… A few clarifications and bits of additional context below. 🧵

This deserves more attention. Moretti regresses patents-per-coauthor (Y/X) on cluster size (S), controlling for number of coauthors (X). But there are mechanical correlations between Y/X and X (negative) and between S and X (positive). 1/

If inequality increases due to higher right-tail income, then the demand curve could kink like this, with higher willingness-to-pay at higher prices. Eg. rich people weren't willing to pay $3M for a house, but now they are. But this has no effect on prices! 1/

Pop quiz: how does rising inequality increase housing prices? Explain using a supply and demand model.

Can AI detect the ten errors in Moretti 2021? I did a test of GPT5.2 vs GPT5.4 vs refine. Takeaway: current reasoning LLMs are useful, with room for improvement. 1/

This is just as bad as log(y+0.00001). Moretti defines leads and lags of cluster size using _n+k, referring to next/previous *observations*. But since the panel is unbalanced, _n-1 can refer to 5 years ago (instead of 1 year ago).

I have a separate writeup explaining the mover event study. It seems M21: - ran an event study (interact T with event time) - got a result that you'd expect from a distributed lag model (fig 6a) - then proceeded as if it was a DL model (hence fig 6b) 1/

Replication is (mostly) thankless work, big kudos to @michaelwiebe.bsky.social for doing these. Great to in this case see the replication (with lots of corrections of issues in the original paper) getting published.

🚨Replication alert🚨 I'm pleased to announce that my replication of Moretti (2021) is now accepted as a comment at AER. I find ten issues in the paper. My comment focuses on two major problems; in the appendix, I document eight (relatively) minor problems. 1/

New blog, new post:

Supply skeptics at a picnic:

If you think of land value as a residual of housing prices, you're blinded to how increasing the supply of land can reduce home prices. But housing and land prices are jointly determined in general equilibrium. Intuitively, more inputs reduces the price of the output. 1/

'Nexus' studies claim that development brings in poor workers who need affordable housing, which is a negative impact to be mitigated. This is fundamentally flawed: the studies do not measure a worker's housing need in the No Build counterfactual, so cannot prove a causal effect. 1/

This is actually an interesting research question: when apartments are allowed, do they bring in richer or poorer residents?

On "growth should pay for growth": homeowners are all too happy to accept higher property values (created by growth in population, jobs, and public infrastructure), but refuse to pay for the costs that make growth possible. Is that fair?

Spot the fallacy: "Expanding the urban growth boundary won't reduce housing prices, because the upzoned agricultural land just becomes more valuable, wiping out any cost savings."

Pop quiz: why does upzoning shift the supply curve for housing? Eg. rezoning agricultural land to residential. The supply function tells us: for a given price, what quantity of homes will developers produce? Shifting the curve means this quantity is higher at every price.

Schrodinger's zoning regulation: it doesn't prevent housing development, but must never be relaxed.

Upzoning reallocates a parcel from the house-zoned market to the apartment-zoned market. This ⬇️ the supply of house-zoned land (⬆️P_H ), and ⬆️ the supply of apartment-zoned land (⬇️P_A). The upzoned parcel itself increases in price as it switches markets (from P1_H to P2_A).

A new condo building in Honolulu has 310 income-restricted homes. These don't merely benefit the residents in the new units; through vacancy chains, those residents free up their previous homes for other poor people to live in. This vacancy chain mechanism also works for the 202 market-rate homes.

Interesting new paper on the cost of permitting. This is hard to measure, and these costs have long been thought to be significant. This paper confirms this, estimating permitting costs at roughly 30% of the wedge between construction cost and the price of housing. (HT @michaelwiebe.bsky.social)

Left-NIMBYism in a nutshell.

When apartment-zoned land is scarce, it's expensive for developers to buy land to build apartments. But when apartment-zoned land is abundant, it is cheap. Upzoning works by unlocking denser housing, which increases the supply of homes and reduces prices.

Tapping the sign:

Another new paper on housebuilding and vacancy chains, this time with data on every Swiss resident & housing unit! An interesting context given Switzerland's high immigration, very large rented sector and strong tenancy rent controls... frederickluser.github.io/files/Moving...

Common misconception: thinking that 'filtering' is a *mechanism* that produces affordable housing. No. The filtering rate is an *outcome*: is the next resident in a home richer or poorer? The underlying mechanisms are depreciation and quality-adjusted price growth. 1/

Two ways to get more homes: 1) Status quo: developers wait for ⬆️rents before they build more housing. 2) Reforms: ⬇️development costs, shift the supply curve; developers build more and rents⬇️. Lesson: unless there was a reform, any new housing we see is from the status quo.

People who say that housing supply constraints don't matter are making a prediction: that removing those constraints won't lead to more housing. How about we test it out?

New literature review post! I discuss research showing what happens to housing prices, rents, and displacement when demand increases faster than supply, and what happens when supply is allowed to keep up. 1/

New literature review post! I discuss research showing what happens to housing prices, rents, and displacement when demand increases faster than supply, and what happens when supply is allowed to keep up. 1/

These homeowners sold to a developer for $2-5M, much higher than the typical price in the neighborhood. But please, tell me how restricting development maximizes property values (aka, the 'homevoter hypothesis').

What is the NIMBY response to this? - conclude that, if demand increases prices, we should just block demand - claim that supply-constrained cities have smaller *long-run* price increases, due to less induced demand - deny the evidence

A new luxury tower is bought up by foreign investors. What would happen in the counterfactual where the tower wasn't built? Would those investors instead buy up existing homes and hold them empty? Or look to another city for investments?

People think that up-filtering is some gotcha against YIMBYs, when in reality, a shortage of housing is what causes homes to filter up (ie. pass from poorer to richer occupants).

Question about NSS 1999 data: are schedule 1 (consumption) and schedule 10 (employment) supposed to merge on household ID? S1 has a str11 hhid, while S10 has a str13 key_hhold; the latter prepends subround and subsample. When I fix it manually, only 2% merge. #econsky

New high-end housing helps overall affordability when it frees up low-end housing, or absorbs a demand increase. But does this apply to mansion districts? If a new mansion in Vancouver just brings in another billionaire from SF, then there's no effect on the low-end market. 1/

Simple test for whether there is a housing supply shortage: if you remove all constraints on supply (zoning, height, FAR, etc), does the quantity of homes increase? If yes, then supply was constrained. If no, then it wasn't.

If dwellings per person is increasing over time, is that proof there's no supply shortage? Here there's ~0.5 dwellings/adult, or 2 people per household. But if the desired size is 1.5 per household, then we should have 0.66 dwellings/person—0.5 is suboptimally low.