The Good, the Bad,...

I recently wrote a post with some generalized comments about recent supply-skeptic chatter. The chatter was about a paper titled, “Supply Constraints do not Explain House Price and Quantity Growth Across U.S. Cities”. There, I discussed a generalized issue with the supply-skeptic literature. The housing supply shortage is so bad that it actually induces unprecedented types of trade-offs within a city’s housing market. Rents rise regressively as families with low incomes must endure excessive housing costs to resist regional displacement. The average price of homes in a supply-deprived city is really just the middle point of the regressive rent inflation that tends to remain moderate at the top end of the market and increases sharply at the bottom. The slope of that regressiveness is determined by the unwillingness of families to be displaced much more than it is by the aspirational demand of newcomers that are moving to a city.

This is the classic downward sloped price/income line that lies under much of my quantitative work here. The resistance, but eventual surrender, family by family, to displacement, increases both local home prices, as they pay to resist, and local incomes, as they surrender and move away. Research that uses local income trends as a proxy for demand without accounting for this bias ends up mislabeling the consequences of obstructed supply as higher demand. It looks like rents are just rising because the local economy is so strong that wages are rising.

Two cities with equivalently strong economic conditions, but with different supply conditions, will look like 2 cities with different economic conditions.

It turns out that that isn’t such a big problem with this paper. But, before I get to that, I want to touch on a couple of other papers. One good and one bad.

The Good

“Mortgage Supply and Housing Rents” by Pedro Gete and Michael Reher speaks the truth! It is dated 2017, but I don’t think I’ve seen it before. The abstract:

We show that a contraction of mortgage supply after the Great Recession has increased housing rents. Our empirical strategy exploits heterogeneity in MSAs’ exposure to regulatory shocks experienced by lenders over the 2010-2014 period. Tighter lending standards have increased demand for rental housing, leading to higher rents, depressed homeownership rates and an increase in rental supply. Absent the credit supply contraction, annual rent growth would have been 2.1 percentage points lower over 2010-2014 in MSAs in which lending standards rose from their 2008 levels.

Preach it, brothers! They use different methodology than I do. I mainly use local incomes, FHA market share, and denial rates as proxies for credit dependency. They use lender sensitivity to post-2008 capital stress tests as a proxy for local credit access. It’s good to know the truth is buried somewhere in an academic journal out there, visible enough that others are citing it. This, along with various papers about economic migration patterns, builds a story. The academy largely hasn’t recognized the depths of that story, but at least the puzzle pieces are starting to accumulate.

The Bad

Another paper cited is “Why is the Rent So Darn High?” by Greg Howard and Jack Liebersohn. It’s a supply skeptic paper.

Because of changing location demand. In a spatial equilibrium framework, we show that three-quarters of the CPI rent increase in the United States from 2000 to 2018 is due to increased demand to live in ex-ante housing-supply-inelastic cities. The quantitative importance of location demand is greater if people are mobile in response to rent changes. Empirically, we show that people have high long-run mobility by estimating that income changes have similar effects on rents across cities regardless of housing supply elasticity. The cross-sectional pattern of location demand implied by our model matches patterns of labor-market and amenity changes.

Please correct me in the comments if I am incorrect on any of my points. I am not always able to fully deconstruct the PhD level statistics. So, I am willing to be corrected if I have missed something. This is the sort of paper that makes me sad, because the academy largely ignores the importance of the “good” paper quoted above, and I think it has poisoned years of research with a set of errors by commission and omission that will be hard to untangle.

This paper has several of the common issues that I frown on. They use rising incomes as a proxy for demand because of local economic growth. They measure rents at the city level, so they miss all the intra-metro trading among families resisting displacement. So, they perceive the average metro area rent as a reflection purely of aspirational newcomer demand. And, since they don’t account for the deep effect that the mortgage crackdown had on construction, they have a distorted estimate of supply conditions.

They use the standard estimates for local supply elasticity. But, before 2008, over half a million new homes were sold every year at price points below $300,000, and after 2008, that fell to practically nothing.

That is because the mortgage crackdown drove prices down below replacement value for entry level new homes everywhere. Supply is inelastic if the prices of existing homes are lower than the prices of equivalent new homes. So, every city had similarly inelastic supply after 2008, and the standard coefficients based on local housing politics and geography just don’t apply to this period at all.

Of course, using that data, it looks like supply conditions don’t matter. They might as well have used a random number generator as their estimate for each city’s supply conditions. And since supply conditions were similar everywhere, even if they did have a good estimate of supply conditions, they didn’t differ enough to matter much.

So, yeah, from 2000 to 2018, locational demand has been the prime factor driving rent inflation. And it especially looks like the prime factor when outmigration of low-income families causes local incomes to rise even more when demand isn’t met with new supply.

They conclude, “national expansions of housing supply can have effects, but it does not matter for rents where the housing is built. Again, because the population elasticity is high, an additional house in a rural or urban area will have the same consequences for rent.”

Basically, they are saying that if supply is inelastic, then when someone new moves into a city, someone else just chooses to move away, with no effect on rents. I think that’s what they are saying. Effectively, they are claiming that, on the margin, families accept displacement with little resistance.

Again, please correct me if you know this paper and I’m wrong.

A recent post helps think about these problems. I think they’ve got it a bit backwards. Figure 1 and Figure 2 help to think through it.

Aas a housing shortage really becomes binding, rents and prices become regressively inflated. Basically, families self-select according to who is willing to pay to avoid displacement. As you move down the income distribution, more families are forced to migrate away and the ones that are left are the families who are most willing to accept high costs to remain.

In the earlier post, I argued that this condition didn’t lead to a national shortage of homes, because when LA lacked housing, affordable homes were built in Phoenix. But, it did lead to rent inflation because the families in LA were willing to pay higher rents to avoid moving. The regional shortage didn’t affect the total national count of housing units much but it did cause rent inflation.

Figure 1

I think what is happening is that these authors are taking the metro area average. In LA, the average rent was higher. And, as low-income families give up and move away, that causes average local LA incomes to rise, and also raises wages for the remainers where labor is in short supply in low wage jobs. They measure that as a demand factor. That’s the classic story. Families want to move to cities where incomes are higher. And, rents rise along with that higher income. So, they conclude that rising incomes in LA are causing rents to rise, and since rising incomes seem to explain rising rents, it seems as though some family on the margin decided to move away to make room, with little friction involved.

Figure 2

After 2008, homebuilding cratered everywhere, so every city, like Phoenix, started to look more like LA. And, so they conclude that supply conditions don’t matter because rents went up just as much in supply friendly cities like Phoenix as they did in supply obstructed cities like LA.

Since they don’t think families mind moving, they conclude that you could build an extra house in Kalamazoo rather than Phoenix or LA, and some family would decide to move to Kalamazoo with little fuss.

They conclude that the expensive cities are in high demand, so that obstructed supply might limit economic growth by preventing families from moving there, but on the whole, if LA doesn’t build more homes, it won’t affect rents.

Of course, EHT readers know that the mortgage crackdown means that Kalamazoo hasn’t been building many homes either. There is no rural place that has elastic supply conditions, because of the mortgage crackdown.

I think the key difference in their conclusions and mine comes from using metro area averages versus recognizing the asymmetrical effects on rent and prices that the supply shortage creates. And it leads us to amazingly diametrically opposed conclusions. They conclude that families move easily and, because of that, rents just reflect local income opportunities, so supply doesn’t matter.

I conclude (which I will present in more detail in a paper that is currently with editors) that nearly all of the increase in home prices and rents over the past few decades is caused by the resistance of families to displacement, so the lack of supply is devastating.

I think this comes down to whether you base your model of the housing market on an “LA price/income” estimate of 10x or whether you base it on a model of “LA price/income” that is 14x in the poorest ZIP codes where families are being displaced and 5x in the richest ZIP codes where they are not.

I feel very comfortable with my position here, but, again, I will readily retract any of this if I am unfairly interpreting their explanations.

Actually, this has gotten long enough. I will save my comments on the other paper for a part 2.