Here is a housing study that looked at housing supply -- in the US from 2000 to 2020 -- relative to median housing values. And here is the key takeaway:
What this chart is saying is that new housing is rarely added in cities with the lowest-value homes. The bar on the left represents municipalities whose median housing values are less than 50% of the metropolitan average. And this makes sense. If values are low there is likely little to no incentive to build. The math just doesn't work.
However, as home values increase, the incentive to build and the ability to finance new projects also increases, and that is what we see in the above chart. This also makes sense.
But something interesting happens in the highest-value cities -- housing supply once again starts to fall off. And it turns out that there is a bit of a sweet spot. Municipalities whose relative housing values are 110 to 130% of the metropolitan average actually produce the most overall housing. Any higher than that and things start to decline.
Why is that? The answer likely has to do with restrictive land-use regulations. The highest-value cities (and wealthiest suburbs) often have a lot of large single-family lots, as well as policies to ensure that this kind of built form doesn't change. This has the effect of both limiting supply and enshrining values.
So when it comes to housing supply, what you don't want are low-cost areas. But you also don't want the highest-value areas. What you want are areas that are doing well, but no so well that they start really restricting new entrants. This is what our industry often refers to as exclusionary zoning.
Now, one of the most common ways to respond to this problem is to develop an opposing policy, namely inclusionary zoning. But usually what this policy doesn't do is direct more supply to these high-value and low-density areas. Instead what it typically does is force the segment that is producing the most housing -- let's call it the 110 to 130% band -- to deliver more affordable housing.
It's a neat trick that sounds pretty cool, but it is not at no cost.
This is a fascinating little experiment:
From Oct. 12, 2020 to Jan. 3, 2021, Redfin ran an experiment on 17.5 million of its users across the US. As prospective homebuyers entered the site, Redfin assigned them randomly to either a group that was shown flood-risk information on each property or a group that was not.
The flood-risk scores came from First Street Foundation, a climate and technology nonprofit that works to make climate hazards more transparent to the public. In June 2020, First Street published the first public maps that revealed flood risk for every home and property in the contiguous US.
First Street scores properties on a scale of 1 to 10 based on the likelihood that they will flood in the next 30 years (which is assumed to be a typical mortgage term). A score of 1 means the property has "minimal" risk and a score between 9-10 is considered "extreme" risk.
So what happens once you start showing people flood-risk information? They, not surprisingly, start systematically looking for safer properties. After one week of users being exposed to this new information, prospective buyers who were previously looking at "extreme" homes started looking at homes that were about 7% safer.
After 9 weeks, these same "extreme" home buyers were looking at properties that were about 25% less risky. And for some buyers, in particular those working with a Redfin agent or partner, their flood-risk tolerance dropped by over 50%. (Embedded in this data might be a sales pitch for working with a knowledgeable Redfin agent or partner).
Also interesting is the fact that below "severe" flood risk (a score between 7-8), there was very little change in behavior. "Major" flood risk, it would seem, isn't all that concerning to most buyers. It needs to be "severe". Nevertheless, it is noteworthy that people will in fact make behavioral changes when presented with clear climate-risk data.
The relationship between car ownership and urban density is a fairly intuitive one. Below are two charts from a study by Francis Ostermeijer, Hans Koster, Jos van Ommeren, and Victor Nielsen, showing how urban density is inversely correlated with car ownership. In other words, the more people with cars, the less dense that a particular place is likely to be.
But there's an interesting chicken-and-egg question here. Does Atlanta, which is near the bottom right in the above chart, have a lot of cars because it wasn't dense enough to support other modes of transport, or did the prevalence of cars in Atlanta cause the city to spread out and become less dense? And that is exactly what the above researchers set out to determine.
To do this, they started by looking at the presence of commercial car manufacturers in the above geographies in the 1920s. One of the things they found was that having a car manufacturer in your city at this time appears to have had no effect on population density. But over the long run, rising car ownership seems to have had a sizeable effect on reducing population densities in those places.
The conclusion they draw from this is the title of this post: cars have made cities less compact, rather than low population densities causing people to go out and buy more cars. This makes some sense to me because cities were doing just fine before we invented cars. But like all transportation innovations that allow us to move faster over longer distances, the car encouraged decentralization.
There are, of course, all sorts of possible implications for a finding like this. But the authors specifically mention developing countries where car ownership may still be relatively low. This is something to be mindful of because if you put most people into cars, history strongly suggests that it will impact the kind of city that you end up building.