How Much Do Supply & Demand Move Home Prices?

I estimate the post-2008 national shortage of homes to be roughly 15 to 20 million units. It’s a true shortage in the pedantic sense that economists use the word. Lending regulators and local land use regulators have capped effective production of new homes in every American city so that now every home is bundled with a plot of land that sells at a premium because the land is the permission slip to have a home. There is a shortage of housing, and the high price of buying or renting one is “the bribe you must pay the land to buy the house”.

Figure 1

Before 2008 there was a regional shortage, so prices and rents were elevated regionally, but there wasn’t a national shortage of homes. Los Angeles had a shortage, so L.A. residents had to keep upping the “bribes they paid to the land” until the next family chose to move. When they moved, there was an affordable home in Phoenix for them. So, before 2008, the average price of homes across the country was elevated, but the total number of homes constructed wasn’t a national “shortage” in the pedantic sense of the word.

If the shortage is approximately 15 million units, then that has been associated with about 20% excess rent inflation and about 70% price inflation. (The price inflation part is a bit complicated.)

There are about 150 million homes in the US, so a 15 million unit shortage is about a 10% shortage. It would take 10% growth in the number of homes nationwide to eliminate the nationwide bribe we must pay for homes on urban land.

Roughly speaking, let’s say a 1% increase in homes per capita would reduce rents by 2% and prices by 3%. (Inflated land prices are associated with higher price/rent ratios, so changes in rents under shortage conditions lead to larger changes in prices.)

That’s a pretty realistic demand elasticity, I think, reflected in the literature.

Figure 2 compares the annualized number of completed new homes to the number that would hypothetically be required to keep rent inflation equal to core non-shelter inflation.

Figure 2

Figure 3 shows the same measures as a percentage of the existing stock of homes and also includes the 1 year growth rate in employment, which tracks pretty well as a leading estimate of the number of homes required to maintain neutral rent.

Figure 3

A couple things to keep in mind on Figure 3:

First, it doesn’t look like employment and rent-neutral home completions tracked well before 2008, but, remember, then, rent inflation was regional. It was high in the housing poor coastal metros of the Northeast and West that were losing workers and low in the South and Midwest cities that those workers were moving to. Where homes were being adequately built, employment growth was positive and rent inflation was moderate.

The mortgage crackdown in 2008 spread rent inflation nationwide after the initial foreclosure crisis played out, so the national trends are more aligned after 2008.

Second, that spike in 2021 might make it look like more supply couldn’t have helped keep rents and prices in check. But, keep in mind, of the 15 or 20 million homes missing, about 5 million of them would be vacant. By 2021, we were scraping the bottom of the barrel of vacancies that might be turned quickly into occupied units. If we hadn’t been underbuilding for years, 2021 would have been characterized by a declining vacancy rate rather than rent inflation.

We’ve sort of been putting together a natural experiment. In Trump’s first term, first immigration declined, then the Covid pandemic briefly almost took population growth to zero. Then, there was a recovery in population growth and a pendulum swing back to strong net immigration during the post-Covid Biden boom. But, by 2024, either from natural mean reversion or from Biden administration attempts to slow immigration, population growth started to slow down.

Figure 4

These population trends might be more important in Trump’s second term than cyclicality will be.

AEI estimates that net immigration could even turn negative after having spiked to nearly 3 million net entries in 2024. That could whipsaw population growth from about 1% to 0%, or even slightly negative.

That would be enough to flatten rents at current levels if non-shelter inflation remains on a 2% target. If new construction can return to increasing by about 100,000 new units more each year than the previous year, that would slowly bend the rent inflation curve down by about 0.1% each year. Stringing several years of that construction growth together could knock rent inflation down another 1% by the 2030s.

Rent is about 1/3 of the consumer price index, so if the Fed aimed to keep the CPI at 2%, then rent, under those conditions, would run at 0% while everything else averaged 3%. Housing makes up a smaller portion of the PCE price index, so if that tracked 2%, rent trends associated with a 2% inflation target would be a little lower.

My point here is that supply is a slow-moving beast. We aren’t going to crash the housing market or reduce the demand for new homes with either slower population growth or much more construction. Someday, years or decades from now, sustainable annual construction activity might eventually be near or below where it is today. That’s a long time from now in any scenario.

In the meantime, slower population growth will have minor effects on land and home prices. It will speed up the pace of valuation normalization a small bit. But, it will have to be paired with a relentless growth of new housing construction for many years, and there is little we can do, on the margin, to stop that as the capacity returns to do it. Doing something big like outlawing large investors in housing could stop it from growing. That’s my biggest fear.

A couple other questions to ask

Conventional estimates of the housing shortage tend to be a few million units. What do all of the economists making those estimates think demand elasticity is for housing? Do they think a couple million homes is going to lower rents 20% and prices 70%? If not, what do they think is making rents and prices so high in a way that is unrelated to supply conditions?

And, how about the period before 2008? What quantity of excess supply did they think necessitated a subsequent cumulative decline of millions of new homes? What supply conditions could lower prices across the country, on average by more than 20% (and more than 50% in many neighborhoods) while rents only declined briefly by a few percentage points? And where do they think they see that supply in, say, Figure 2?

It kind of seems like conventional wisdom about 2008 and the period preceding was so malformed that speaking in terms of supply and demand elasticities gave results that were nowhere near the scale of price and quantity changes that occurred. So, everyone kept the supply and demand story: too many reckless loans pushed prices unsustainably high, creating a glut, which inevitably normalized. But, nobody did the math. The math was nonsense because the story was nonsense. There was no supply-and-demand justification for any construction downturn in most cities, and certainly not the permanent erasure of 3/4 of local single-family production in city after city. But they kept the story and just ditched the math.

So, there are 2 sides of a coin here. A glut of supply does not and cannot explain the housing bust in 2008. It’s not even a matter of the scale being exaggerated. There is nothing there. It didn’t happen.

That means that today, going forward, the best and the least we can hope for and expect is a slow, drawn out normalization of housing supply and housing costs. Using a basic demand elasticity estimate actually gives you a pretty good key on the last 30 years of American housing activity and price trends. A boring 2% rent impulse related to each 1% of missing supply. You just have to rip 2007-2010 out of your model. That was a different thing. It had nothing to do with a national supply and demand mismatch. It created a supply and demand mismatch, but it wasn’t caused by one.

Take that out, and housing costs are a boring line moving relentlessly up and to the right. (Australia and Canada, who didn’t have a moral panic about mortgages in 2008, have housing markets following that line.) It’s just going to keep moving along that boring line. I think 2 decades has been long enough to let recency bias create an overfocus on 2008. Accept the boring. You’ll be happier for it. And wiser.