A Not-So-Brief Thought on Zoning
I’m skeptical liberalization will boost density.
It is a truism among urbanists that if we remove onerous zoning and land use rules that density will increase; that is, buildings will get taller, parking less common, and more people will reside in more classically urban spaces. And on its face, that makes sense. The thought experiment here is simple: in a place where construction averages 2 stories, removing a parking space requirement allows only 1–3 stories of plausible construction on that former place. But in an urban area where construcion may run to 20, 30, or more stories, you can really run up the population score by making comparatively small changes in land use.
Where this idea has trouble is in the stylized fact that places with stricter land use policies actually tend to be denser than places with less strict land use rules. Huh, that’s weird, right?
For this post I’m mostly spit-balling a theory that remains to be proven. But the graph above, and the one below, is the image I have in mind for what I’m trying to puzzle through.
Faced with the graph showing the positive restriction/density correlation, zoning-reduces-density storytellers respond, reasonably enough, that the impact of zoning is quite recent; land use laws didn’t really get going until the 1970s.
The problem with this story is that it is not true. The New York City zoning code of 1916 initiated skycraper setbacks that observably reduced density, and also prevented some neighborhoods from filtering down. The infamous height rules in the District of Columbia date to 1899. Los Angeles prevented industrial sites being placed near residential areas (that is, initiated a use-based zoning code) in 1908. By 1910, Baltimore had a racial zoning ordinance, and by 1917 Louisville’s racial zoning ordinance was being struck down by SCOTUS. That is, zoning was already alive and well by the 1910s, sufficiently so that people were suing over it. For a good account of the racial history of zoning, read on here.
But look, the point here is simple: zoning isn’t new. If zoning is what caused the problem, then we should have felt the impact a lot earlier (and indeed it’s possible we did; we don’t have good housing cost data from 1930, but maybe NYC was higher then too!). Whenever somebody tells you that a major city, like, say, New York, grew up before modern zoning codes could constrain them, smack that person. At the time of the 1916 ordinance, New York City had 5.3 million people. It continued growing strongly until 1938, then WWII, crime, suburbanization, etc all set it flatlining, and you know the rest.
The classic response here is that zoning caused cities to expand: Daniel Kay Hertz reminded me on twitter that NYC had greenfield development into the 1920s. But in that case we have to acknowledge that population growth in NYC stalled in the 1930s. Pray tell: how is it that zoning began its bite in the 1970s or 1980s, when NYC population was declining? Does anybody think NYC’s population decline in the 1970s was a product of zoning? No? Okay. And its current population increases only recently pushed it appreciably over the 1970 population levels.
So let’s consider what’s happening here. Zoning came into existence around 1900 and became widespread by the 1920s. To the extent its impact was not felt, that’s because cities were able to grow expansively, that is, by new greenfield development within city limits.
But here we have a problem. If your theory is that zoning initiated in a much earlier period doesn’t become “felt” in prices until construction runs into physical barriers, then you’ve basically just conceded that what actually caused price increases was physical barriers, that is, lack of greenfields. Because we are human, it’s easy to look and see the anthropogenic factors here, like zoning, but the thing that actually changed to directly trigger higher prices was not zoning. It was the shortage of developable land. Even with loose zoning, you’d (1) eventually hit the same limit and (2) face higher prices before that limit because non-greenfield development is substantially more expensive.
Furthermore, this suggests that some of the key determinants of when development-driven population pressures will bite are not things we conventionally think of as land use restrictions. Conservation areas, parks, green spaces, public plazas, streetspace, these may indeed be the biggest drivers of actual costs.
Say we have an island of 23 square miles and it has 1.7 million people. Say 15 percent of this island is devoted to parks, of which half is land that would otherwise be readily developable. Now say that, oh, 35 percent of the land area is occupied by streets, leaving 50 percent for construction. This makes 148,000 people per developed square mile.
Now let’s do some math. Assume we have reached maximum population density under the current zoning code, and that, if upzoning occurs, population will rise to meet the maximum zoned potential anywhere that construction occurs.
Say we make changes that upzone by 25%; so we raise height limits or something like that. How long will it take for the whole city to be built under this new code?
Well, the answer is, forever. It will never occur. Upzoning will never densify the whole city like this. The cost of knocking down and rebuilding a major skyscraper is too great. Upzoning won’t replace skyscrapers; it can only replace comparatively smaller developments. Bad zoning of the past will stick around for a long time to come.
But let’s say that in the first 5 years, a whopping 10% of the built environment can be replaced by buildings that house 25% more people, and they all fill up. That gives us a final population of 1,742,500 vs initial population of 1,700,000. We added 43,000 people. But keep in mind, this required us to bulldoze 10% of the city.
But maybe we envision something more radical. Say we envision upzoning that allows 500% greater density. Well, that’s gonna be trickier and costlier to implement I’d wager, so let’s assume this far more radical upzoning replaces just 3% of the city’s built environment. This gives us 1.9 million vs. 1.7 million initially; a 12% population increase! That’s impressive!
Or, at least, it looks impressive. I still think bulldozing 3% of a city’s built environment in 5 years time seems extremely unlikely, but YMMV.
Let’s consider an alternative view. Let’s assume that, with no change in rules for development, we simply alter the amount of developable land. We said just 50% of the land area is actually open for development. What if we bumped it up to 55%? Well, that right there gives you 1.87 million. If you can just reduce the amount of non-developable space, you can get more people in. Plus, building on former parks and streets is easier than building on former buildings, and cheaper. This is something more akin to greenfield development.
But hold on: isn’t this just Kowloon Walled City? Am I suggesting some nightmare dystopia where there are no green spaces left?
Kind of, yes. But more simply than that, my suggestion is that (1) land use regulations are in fact quite old, (2) to the extent they have bitten recently, the proximate cause is actually physical land constraints, (3) many of those physical land constraints are in fact socially constructed. We can choose more or fewer parks, more or fewer streets, more or fewer plazas, and these choices are probably not that much more or less contentious than zoning codes themselves in most cities.
Under Uniform Liberalization, Suburbs Will Win
Why this long digression when I wanted to talk about why upzoning might not boost density? Because I want to interrogate the conventional wisdom about how zoning works, and broaden our mental model of what constitutes land use regulation.
With that broader view in mind, let’s cast our gaze to the suburbs. Why is it that less regulated places also seem to be less dense? Is it that they’re newer? Well, maybe, but why would newness reduce density? Many commentators will point to policy choices, including FHA rules about loans and lending, and I think those are plausible. We can also note the policy choice of subsidized roads and transit; that is, no American pays the full cost of their own commute, neither drivers nor transit riders. Given that we subsidize all forms of transportation, diffusion becomes more likely.
But there may be a regulation-related answer as well. Many of the most tightly regulated places are not urban downtowns, but suburbs. Hell hath no fury like a neighbor complaining about that carport you added onto your house. And when they find the extra unit you built in the back yard, ho boy, you’re never gonna hear the end of it.
It may be that urban cores are more regulated, but they also have less potential. That is, even without zoning restrictions, the existing non-depreciated investments in real estate there are large and positive and unlikely to be bulldozed for anything other than some huge gain. Greenfield development is rare. Construction is costly.
But in the suburbs, doubling density is easy. It is challenging to figure out how Manhattan could plausibly double its density in a 10 year window. It would require extraordinary economic changes and movements. It is not challenging to figure out how West Orange, NJ could densify. That it doesn’t is often a product of regulation.
Were we to unilaterally liberalize zoning, some builders would see new opportunities in Manhattan. But it seems far more likely gazillions of suburban folks would see the benefit to building a cheap extra unit in the yard and renting it.
In terms of raw potential, it seems quite likely there is more “zoning-prevented housing” in the suburbs or in fairly low-density areas than in already high-density ones. The result could easily be that uniform upzoning boosts metro-wide population, but also causes a shift of population out of the center, into the ‘burbs, where geography may prove less of a constraint. The fact that less-regulated places also seem to be less dense suggests that this outcome is at a minimum plausible. That is to say, if density is your goal, deregulation may be a very uncertain way to get there because, while there may well be demand for urban cores (maybe), land use rules are just one of many supply constraints. Geography, higher construction costs, large existing investments, and the dramatically lower costs to adding equivalent supply in the ‘burbs all combine to suggest blanket liberalization could cause the typical household to reside in a less dense neighborhood than they did under stricter regulation.
Consider our 1.7 million person island before. Say it has surrounding suburbs that amount to 250 square miles, of which 33% is developed private space and 67% is parks, green spaces, agriculture, rough land, etc. Say that the suburbs have 2 million people. So 2 million people live on 82.5 square miles at 24,242 people per square developed mile, or 8,000 people per total square mile, including undeveloped space. This is a density of 1/6 to 1/9 that of the core.
Assume, again, that this area is already built to capacity under current zoning, land use, and land allocation rules and policies. Now assume that we convert 3% of the currently-used land to density 5 times greater than the current density, the same extreme scenario we applied to the core. Well, that gives us 240,000 net new people, a greater number. The same math holds for land expansion. We can come up with a basic model where a regulatory liberalization across a whole MSA allows new construction everywhere, but that new construction occurs more in places with good access to jobs and amenities relative to the time and money cost of construction. That is, the core of course has better access to jobs, so will get a fair amount of construction under a more liberalized environment, but huge cost differences in terms of time and money mean the outer rings may well get the bulk of any population increase from uniform liberalization.
Formal Regs Are Just One Kind of Regulation
There are several factors at work here. To begin with, transportation policy! It is curious to me that many urbanism advocates also push for more public transit and more rail, because it seems obvious that improving transportation access reduces density. If you want density, then you want bad transit, or at least expensive transit. The benefits of dense living decline with the time and money cost of commuting. So if you make commuting cheap and convenient, you get lower density. Rather, transit expansions are likely to great denser hubs around stations in the ‘burbs, but, on net, transfer more population out of cores into ‘burbs. So transit raises density at every place it touches, but lowers the typical “experienced residential density.” For a fixed-area MSA, it will raise total raw population density given the amenity value of easier commuting.
Next up: attitudes! As a report from Trulia showed, the main driver of restrictive land use policies might not be zoning at all, but the time it takes to get building permits. That is, in cities where builders can easily deal with bureaucratic requirements and don’t have many regulatory papers to file or meetings to attend, they build more. This should be obvious. Zoning matters of course, but non-zoning policy choices are important too. If a city liberalizes its zoning code but maintains onerous or inefficient permitting requirements, it may not get much bang for its buck.
Permitting is in some sense a policy choice, but in another sense is an institutional choice. That is, changing it may sometimes require not that public workers do something different, but that public workers are fired and replaced with people with different attitudes. Getting public workers who are motivated to process permits quickly and efficiently isn’t always easy and isn’t always as simple as changing the law. It requires changing the culture. The same is true for neighborhood responses. If formal zoning is nonexistent but neighborhoods and interest groups get a lot of say, then, again, a liberalizing policy change is less likely to yield actual increased housing supply. This isn’t just about changing a few laws, it’s about changing a culture (which, in fairness, most ardent YIMBY types understand and work to promote).
Removing Land Use Regs Won’t Make NYC Gigantic
Recently, economists Chang-Tai Hsieh and Enrico Moretti published an update to a previous paper saying that lighter land use regulations would greatly boost the population in New York City, San Jose, and San Francisco. Population and housing stock would rise dramatically to unprecedented levels, as would employment. Nominal wages would fall 25%, but because housing prices would fall too, real wages would end up pretty similar to what they started at.
First of all, their model is fascinating because they essentially argue that tight land use laws serve as housing-destroying wage-boosters. By restricting competition for jobs and boosting local cost of doing business, their model suggests that less housing yields higher wages. The result is that they argue that NYC/SF/SJ GDP share overstates “real” welfare in those cities by an enormous amount. I broadly agree with this view in their paper, that nominal GDP misallocates the spatiality of economic activity.
But where I part ways is their view that lighter zoning rules could feasibly cause such massive population shifts. Their illustration assumes that NYC had the 2008 median zoning restrictiveness back in the 1960s, after multiple rounds of restricting zoning codes had been promulgated. In other words, they assume an improper recency of zoning restrictions, as I showed earlier. The illustrative gains they attribute to NYC bear no relationship to reality whatsoever. Furthermore, they assume an outsized share of housing elasticity is due to formal regulation: they model a change in elasticity, then call it a change in policy. Well, much of elasticity is about physical geography, local attitudes, transportation networks, and all the other factors I’ve described. And even the policy components aren’t as simple as one might think. Land use rules aren’t even close to the only thing holding back NYC’s housing supply and population growth.
I’m also suspicious of their method, given that it is only aggregate data, not microdata. More scrupulous recent research has shown that migration is falling because of successful convergence, and that such migration as does occur likely reflects people moving from, for example, NYC to, for example, Charlotte, for equal or higher nominal wages, and much higher real wages. That is, it turns out that when you control for occupation and industry, the whole “moving away from productivity” story doesn’t make nearly as much sense. It turns out that computer programmers in Des Moines are not that much less compensated than computer programmers in San Jose. The location-specific effect of productivity is actually not extremely large. And, again, this leads into the standard critique that spatially locating productivity is sort of a crapshoot anyway. If a firm has mission-critical functions in 3 locations and each location is highly specialized in different areas A, B, and C, with measured productivity equal to, respectively, 1, 2, and 3, what is actual productivity? Our stats will say A-1, B-2, C-3. But we specified that all 3 are mission-criticial, i.e. you can’t produce widgets without all 3. So in practice, because this is a causal chain, there’s a good argument for assigning all of them the same productivity. But then that gets weird because wages don’t match productivity at all. All of that to say: figuring out the “correct” way to spatially represent true productivity is a task that still hasn’t been fully completed.
Conclusion
This post has run on long enough. It’s just my musings on how land use liberalization would actually impact density. I think it’s less clear cut than enthusiasts believe. I of course favor liberalization because land use rules are villainous communism and the right of private property is sacrosanct. That view is tenable under any model specification. But if your view is that liberalization is desirable because of some preference for density, then I am afraid you are likely to be disappointed. Formal policies are only one of many reasons why elasticities in dense urban cores are likely to be lower than in areas with more greenfields or interests at stake.
Check out my Podcast about the history of American migration.
If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!
Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!
I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.
My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.
