Does Density Increase Housing Prices?
Even if it does, allow more housing!
You are reading Economic Forces, a free weekly newsletter on economics, especially price theory, without the politics. Economic Forces arrives weekly in the inboxes of over 8,000 subscribers. You can support our newsletter by sharing this free post or becoming a paid subscriber:
A common debate within housing policy circles—and the too-online crowd, more generally—is whether building houses will decrease housing prices. Places with lots of housing have high prices! What gives?
The most recent discussion on the topic came from
at , who concluded density likely increases local prices. He was arguing most directly against a series of newsletters fromat. Since I started writing this newsletter, Scott wrote a follow-up post as well.In today’s newsletter, I will do my best to convert their different arguments into standard economics reasoning. After thinking through the implicit, positive models of housing prices, my current conclusion is that it is unlikely (but not impossible) for density to increase housing prices. Since it’s possible, we should take seriously that density causes higher housing prices.
“Look, building housing doesn’t even make housing cheaper!” may be one response. However, I will then argue that “density causes higher prices” is not a reason for a victory lap for the NIMBYs who oppose building more housing. If density increases housing prices, that actually strengthens the normative case for more housing!
But before we can get around to the policy debate, we need to be careful about what we mean by “density increases housing prices.” As always, we need to go back to the basics of price theory.
Basics of Causation
Economics—whether theoretical or empirical—helps us establish causal claims about the world. If X changes, then Y changes. If you prefer the framing of counterfactuals: If X had been different, then Y would have been different.
One form of this type of reasoning is basic comparative statics. As I recently explained,
With comparative statics, you imagine something outside of the system you are studying (in jargon, an exogenous parameter) changing: the tax rate rises, oil prices spike, or an outsider develops a new technology. You then trace the effects of the parameter on some outcome variable (an endogenous variable).
Causation is always hard, but it is especially hard in economics. The two variables we talk about the most (price and quantity) don’t cause the other. In most cases, talking like quantity causes price or price causes quantity will confuse you. Both are endogenous to most markets.
This is what Scott seems to be doing when he says
So if Oakland became bigger, it would become a more appealing destination for these people at some rate (making it more expensive) and get more supply at some rate (making it less expensive).
But how did Oakland get bigger? Did people fall from the sky? Did Oakland allow more building permits? Did Oakland subsidize people to move there? Did new employment opportunities appear? (The latter is also endogenous, but we can sometimes meaningfully separate the housing market from the employment market.)
You cannot just plot housing prices and quantities and expect to learn anything about the causal relationship. Correlation is not causation, as we all know.
That is why economists have the apparatus of supply and demand. It forces us to be explicit about what is the original mover in this chain of causation. We can then talk about something changing on the supply side or the demand side of the economy and then trace through how price and quantity change.
In the follow-up post, Scott is clearer that he is thinking of a supply shift.
My claim is that marginal changes - like Oakland building an extra 10,000 units, but everyone else staying the same - will most likely increase Oakland prices.
In the words of supply and demand, Scott is arguing that this type of supply increase will cause house prices to rise, although “50 million units, that would soak up the entire excess demand and probably lower prices.” Let’s focus on the “10,000 units will increase prices” claim since that is the surprising one.
This flies in the face of basic economics. The usual mechanism of supply and demand is what I’ve called elsewhere a “dampening mechanism.” An increase in the supply because the local government allows more houses drives down prices, but that decrease in price raises the quantity that demanders demand. The price decreases less than if demanders didn’t change their behavior, which dampens the overall change in prices.
Notice I’m saying “allowing more housing to be built,” which is a way of saying that the government lowers the cost of new housing, which we draw as a shift to the right of the supply curve. If I just said, “Oakland built more housing,” that could be a movement along the supply or demand curve. That’s a statement about an equilibrium quantity, which is exactly what we want to avoid.
Housing Demand Slopes Up!
People are quick to admit the above logic for apples. As people consume apples, they want the next apple less and less. Diminishing marginal utility and all that. That’s why we draw the demand curve sloping down. With downward-sloping demand and upward-sloping supply, the above argument goes through.
But housing is different, we are told. How?
As an intermediate thought experiment, maybe the demand curve for housing in the Oakland is flat. This would be the case if there were many people who want to move to Oakland whenever their wage increase in the city, minus the increased cost of housing, means the move makes financial sense. There is just a bunch of people willing to pay P_E. If the city of Oakland allows more housing, more people will move there, but they will bid up prices to exactly where they were before. Housing prices won’t fall.
In this case, any increase in housing built does not lower the price of housing in the city, although it may lower the price outside of the city, as people move out of other areas. Scott grants that.
But Scott’s argument about housing goes even further. In his words, “if Oakland became bigger, it would become a more appealing destination.” It could be because big cities are more fun or there are agglomeration effects to work, so the pay is better.
In this framing, Oakland gets more appealing as people purchase more housing. That is a shift along the demand curve. In other words, the demand curve for housing slopes upward. Nick Rowe provides some explicit micro foundations for upward-sloping demand at the market level. I’m just going to assume it is that direction.
Relative to Oakland with a flat demand curve, with an upward-sloping demand curve, the price increases, and the quantity demanded increases even more. Sometimes people call this “induced demand,” but I don’t find that language helpful.
If agglomeration or escalating effects dominate, prices will rise with more supply. However, even now, it’s not quite the perpetual motion machine that Matthew Yglesias criticized since any individual increase in housing permits generates a finite move along the demand curve. But you easily could imagine a further complication of the story where more and more permits are issued, and so prices rose more and more. And once permits aren’t the constraint, policy subsidizes construction to further shift the curve to the right.
But are perpetual motion machines in economics so ridiculous?
Before you start saying upward-sloping demand is ridiculous, it is the exact same logic that applies when we have downward-sloping supply curves, which makes more intuitive sense to me. As I wrote in my newsletter on AI and dampening vs. escalating mechanisms,
If supply is downward sloping, you no longer have a dampening mechanism but an escalating mechanism. In that case, increased demand drives down marginal cost, which increases the quantity supplied, which further increases the quantities demanded. If that’s the full model, this process goes on forever. This is how endogenous growth models work, like those discussed above. Technological progress begets progress.
Over long time frames, it does seem like perpetual motion machines are possible in economics. Technological progress begets progress.
At a micro level, we can think of manufacturing plants that learn by doing and become better and better. If Apple made one smartphone, it would be incredibly costly. As they increase production, marginal costs drop.
The perpetual motion machine here would be that as the manufacturer ramps up production, output increases, which pushes down the price, which further increases the quantity demanded for the smartphone. Instead of a dampening effect, the escalating effect keeps driving down the cost of a smartphone.
How plausible is this story? It’s hard to tell. We cannot directly compare the price of the original iPhone in 2007 vs. what that same phone would be today. I assume cost would continue to drop for Moore’s Law reasons, but Apple chooses to respond to those cost savings from learning over time by increasing quality. We have a much more powerful iPhone today than in 2007. But they COULD make the original iPhone today for cheaper than in 2007.
But how general is this story? It is important to consider the scale and scope of these things. If Apple could just continually drive down the cost of their phones, why not just ramp the production up to a billion phones right away in 2007?
The reason is that there are the usual forces causing the supply curve to slope up. Eventually, you hit other capacity constraints. You may be smarter at making phones, but can you get the raw materials? What about the workers?
In the same vein, there are forces causing the demand curve for housing to slope down. What if a new building removes the parks? What about traffic congestion?
Overall, we need always to ask, which effect dominates when and where?
Think of a labor supply curve. Higher wages may cause people to work less. It makes intuitive sense that labor supply curves can bend backwards, which is one form of downward-sloping supply. But that clearly doesn’t hold everywhere. For most people, labor supply curves slope up, as basic supply and demand predicts. Across many domains, there is reason to believe that the curves slant the way we expect them to, at least locally, and the dampening effect dominates.
My reading of the housing literature, like Matt’s, is that empirically, increasing supply lowers the price, implying standard-sloping curves. As a recent paper by Asquith, Mast, and Reed in the Review of Economics and Statistics puts it, “If buildings improve nearby amenities, the effect is not large enough to increase rents.” Yes, amenities could drive up prices, but that’s not what serious people find in the data.
Should We Build More Housing?
Everything so far is about sorting out the positive, causal effects of allowing more housing construction. What happens when the supply curve shifts? While we usually think of increasing supply causing a decrease in prices, it is reasonable in some cases to imagine that it may raise prices. Before my YIMBY friends get too upset at me for saying that this is possible, note this is purely a positive claim about what is possible. It is possible for an increase in housing to cause housing prices to rise.
The NIMBYs will point to the inability of new housing to lower the price of housing and say, “See, building housing won’t help any, so our NIMBYism is okay.”
That is bass-ackwards. If the housing market is characterized by upward-sloping demand curves, we want housing construction to ramp up! If you can move up the demand curve, housing is more and more valuable. Keep it moving. Take advantage of the beneficial escalating effect. This is what Matt Yglesias means by the perpetual motion machine. But for slow thinkers like me, this implication of the “density raises prices” argument wasn’t entirely clear until I progressively worked through shifting the lines. That’s one benefit of attempting to boil down arguments to simple models.
Again, the parallel to downward sloping supply is helpful. If Apple really could move down its downward-sloping supply curve, it would be able to make phones cheaper and cheaper. That’s great. That perpetual moving down the supply curve is what we call technological progress.
The same is true for housing. If the demand curve is upward-sloping, housing is a perpetual motion machine that is creating more and more value for people. In that case, turn it up to 11!
If I'm understanding Alexander's argument, he believes that a denser population itself makes a city even more desirable to live in. To me, that sounds like a positive externality, which strengthens the argument for policies which encourage more housing in dense areas. Is that a sensible way of thinking about agglomeration, or is that already "baked in" to the upwards sloping demand curve you drew?
I think this argument is unnecessarily complicated. A denser Oakland with the better amenities Scott assumes--greater consumption variety, more walkable layout, agglomeration effects on productivity--is not the same good as it was prior to the density. Improving the quality of the good mid-discussion is not ceteris paribus, which is a crucial assumption for the partial equilibrium analysis most are doing when they discuss more supply leads to lower prices. Of course there could be general equilibrium affects that *appear* to raise the price of the same good, but that is caused by secondary effects that feedback into the initial analysis, some of which can even change the good in question (as with neighborhoods). A denser Oakland is a different Oakland (if amenities increase as Scott assumes), just like an iphone 14 is not the original iphone. No upward sloping demand curves are needed to explain any of this.