Yet another study confirms YIMBYs are right about everything. By Matthew Yglesias

www.slowboring.com

Yet another study confirms YIMBYs are right about everything

Matthew Yglesias

11 - 14 minutes

Lifting heavy things is a good way to build muscle, but working out once won’t make you much stronger. Giving poor families cash is an effective means of reducing poverty, but giving each household $1 isn’t going to change the poverty rate very much. Having more police officers on the street reduces crime, but adding one more Chicago police officer isn’t going to have a large impact. Putting carbon dioxide into the atmosphere contributes to rising global temperatures, but burning a single gallon of gasoline has a minimal impact on climate change.

These are just different versions of something we all know: small changes generally have small impacts.

Many people claim that new construction raises housing prices, either because they are confused or because they believe the demand induced by new amenities swamps the impact of increased supply. This is false, empirically, and in fact, the opposite is true — new construction reduces nearby prices relative to baseline. But, as in the examples above, a small change in the housing supply has only a small effect. Similarly, while I would expect regulatory changes to influence construction volumes, I also expect the scale of the change in construction to be related to the scale of the change in policy — small changes have small impacts.

This all seems quite obvious, but an Urban Institute study was recently written up in Governing magazine with the unfortunate headline “Zoning Changes Have Small Impact on Housing Supply.” I suspect this is related to the fact that one of the study’s authors, Yonah Freemark, characterized it that way on Twitter:

If you read the paper, “Land Use Reforms and Housing Costs: Does Allowing for Increased Density Lead to Greater Affordability?” I don’t think it conveys any information whatsoever as to whether zoning changes have a large or small impact on housing supply or whether upzonings are “enough” (enough for what?). What it does provide is the best evidence yet in the direction of change debate, delivering a knockout blow to the idea that regulatory relief has some perverse negative impact on affordability. Beyond that, all it tells us is that most changes to land use regulation are relatively small. This is not that surprising — small policy changes are more common than large policy changes.

Freemark is one of the named authors of the paper, along with Christina Plerhoples Stacy, Christopher Davis, Lydia Lo, Graham MacDonald, Vivian Zheng, and Rolf Pendall. And their paper does something a little different from earlier empirical work on this subject that I’ve cited.

Studies done by Brian Asquith, Evan Mast, Xiado Li, and Kate Pennington have looked at the price impact of new construction, not necessarily construction induced by policy change. The Pennington paper, in particular, is my favorite of these precisely because she avoids policy change and looks at new construction induced by buildings burning down, a kind of natural experiment. The existing policy in San Francisco generally prohibits buying an old building, knocking it down, and replacing it with a larger one. But if one burns and becomes uninhabitable, then you can replace it with a new building. That gives us insight into the possible impact of a hypothetical policy change, and what she finds is a win-win: local demand does rise when a new building is built, but supply rises even more, so the neighborhood becomes a better place to live and the rate of displacement falls. Good news!

The new paper takes a national look at policy change, which is hard because there are approximately seven zillion zoning entities in the United States. Their solution is to “use machine-learning algorithms to search US newspaper articles between 2000 and 2019” to try to get a comprehensive picture of changes in land use policy. They then mash that up with USPS and Census Bureau data to try to understand the impact.

I think these are the key findings of the paper:

    “Reforms that loosen restrictions are associated with a statistically significant, 0.8% increase in housing supply within 3 to 9 years of reform passage.”

    “[W]e find no statistically significant evidence that additional lower-cost units became available or moderated in cost in the years following reforms. However, impacts are positive across the affordability spectrum and we cannot rule out that impacts are equivalent across different income segments.”

    “Conversely, reforms that increase land-use restrictions and lower allowed densities are associated with increased median rents and a reduction in units affordable to middle-income renters.”

People of a certain mindset are going to find it borderline absurd that the headline conclusion of an empirical research project involving seven co-authors and state-of-the-art artificial intelligence is that supply and demand curves work like intro textbooks say they do. But this is a contentious subject that people do argue about, and Stacy et al. have the clearest confirmation yet that this is, in fact, the way the world works.

What the paper doesn’t tell us is what policy changes were actually made. The virtue of their method is that it let them chunk through the huge quantity of zoning adjustments made across tons of different jurisdictions in the United States. But all it knows is the direction of the adjustment — was it supposed to facilitate more construction or less? — not the magnitude. But oftentimes policy changes are small. Arlington County in the D.C. suburbs just had a huge political fight over a “missing middle” housing bill that doesn’t actually let anyone build larger buildings. The bill just allows for the subdivision of the kind of structures that are already allowed. The experience of Minneapolis has been that this does not generate much additional housing. That’s because this is, deliberately, a small change. It takes the symbolically important step of ending exclusionary single-family zoning while trying to avoid visible alteration of the built environment.

So I think the paper contains a really important headline conclusion, namely that supply skeptics are still wrong, and also reflects the reality that most land use changes are small. There’s nothing in there that should complicate YIMBY narratives or dissuade people who’ve understood the supply and demand dynamics all along.

Meanwhile, even as academics continue to refine their event studies, I think it’s important for people interested in policy to not lose sight of the big-picture, cross-sectional facts.

    A house in the New York, Boston, Los Angeles, or San Francisco metro area is much more expensive than one in the Atlanta, Houston, Phoenix, or Nashville metro area.

    The population is growing much more rapidly in the Atlanta, Houston, Phoenix, and Nashville metro areas than in the New York, Boston, Los Angeles, and San Francisco metro areas. 

How do we explain these two facts? Well, one popular account of (2) is that mismanagement in the blue states is leading people to flee en masse from crime and high taxes. But that’s inconsistent with (1), which seems to indicate high demand for these big coastal metros, albeit somewhat less so in the Zoom Era. The only coherent explanation is that metro areas vary in the extent to which high demand manifests as an increase in quantities versus an increase in prices.

This is not entirely about zoning, of course. It’s not a coincidence that the most supply-constrained cities are bounded by the ocean. It’s also worth noting that the expensive cities are actually larger and denser on average than the fast-growing cities. It’s not like if Boston adopted Nashville’s zoning it would get cheaper; that would actually be a downzoning. But the cross-sectional facts show us something important about changes. And they should help us to avoid things like this Bloomberg article, which wonders why Austin is getting more expensive if it’s allowing all this new construction. The article itself contains a chart that very helpfully shows that the vast majority of the Austin-specific increase in rents just reflects a national increase in prices.

What else happened during this period? Well, Austin’s population grew 22% between 2010 and 2020, Travis County grew 26%, and the Austin metro area grew 33% (!) while the overall national population grew about 7%.

In other words, there was a huge national increase in the relative demand for the Austin area compared to the average location in the United States. This nonetheless led to a modest increase in the price of living in the Austin area compared to the average location in the United States. And that’s because the Austin area added a lot of housing during this period — primarily single-family sprawl in the suburbs, but also some shiny new apartment and condo towers downtown. Meanwhile, note that within Austin (to say nothing of the suburbs), the majority of the land is colored bright yellow for various forms of low-density single-family zoning.

What would happen if Austin rezoned that land to allow for more density? We know the empirical direction of the change would be toward more construction and more affordability. But how much more construction and how much more affordability?

I’m annoyed that this new paper, which should be settling the induced demand argument, is instead being used to cast doubt on YIMBYism because I think there genuinely are important outstanding research questions in this space.

The most important one, by far, is the question of which policy changes actually generate a lot of new construction and how we can try to estimate this in advance.

There’s nothing wrong with incrementalism and passing small, politically feasible changes. But it’s not at all clear that the level of political opposition to a given land use change is proportional to the change’s influence on housing supply. “Ending single-family zoning” throughout an entire county, but in a way that generates very little actual construction, might be a worse strategy than rezoning one particular neighborhood for much taller buildings. There are also lots of issues beyond zoning that relate to lot occupancy and parking, and there can be complicated interactions between them. There’s room for more analytic work here to help policymakers make better decisions.

The other really big one is about parking.

One of the biggest projects in the pipeline in Washington, D.C. is this development called Parkside near the Minnesota Avenue Metro Station. I was talking to someone who’s been to community meetings in the area, and she says the questions people ask about this aren’t expressing esoteric supply/demand doubts. Residents believe (correctly!) that even a transit-oriented development project will raise the demand for local parking. This is not a high-brow concern that grantmakers and nonprofit leaders care about. But it’s a very important issue for the people who live nearby, and I think it’s the number one real-world driver of hostility to market-rate infill development. I think this is probably a problem you can solve by giving incumbent residents a property right to street parking rather than a right to veto new construction, but the world would benefit from more analytic work.

Finally, to the extent that “upzonings aren’t enough,” I think the basic insight is that some people are poor and therefore in need of help beyond what a good market system can provide. But is subsidized housing a good use of money, dollar-for-dollar, compared to a more generous Child Tax Credit? My guess is no, but that’s really just a guess. I wish I had a better sense of the actual impact of housing subsidies vs. spending on affordable housing construction vs. general financial assistance to poor people. This seems like a big open research question to me. But “does regulatory reform improve affordability?” is a question we’ve answered.