A "Shortage" or "Crisis" versus Regular Scarcity: Part 1

Aziz Sunderji at Home Economics has an interesting new post about housing affordability.

He makes a valid point. One of the commonly cited measures of housing affordability concerns is the percentage of income going to rent, and HUD defines a household as housing “burdened” if they spend more than 30% of their income on rent.

But, the percentage of the typical family’s income that goes to rent is almost completely a function of income. Figure 1, from his post, makes his point quite clearly. Families in the 3rd decile of incomes spend 45%-50%, give or take a few percentage points, of their incomes on rent because they are in the 3rd decile of incomes, and no change in housing supply conditions is going to lower the rent expenses of the average American family in the 3rd income decile down to the level of families in the 5th or 6th income decile.

Figure 1

As Sunderji notes, families at a given position in the distribution of incomes have a very strong tendency to spend a similar portion of their incomes on housing over time. But, at any given time, families across incomes systematically spend much different portions of their incomes on housing.

There is some value in measuring the percentage of households that spend more than 30% of their incomes on rent. In a way, I think this strengthens the case for the 30% threshold as a warning sign, because if half of households are normally above that, and currently 55% or 60% are above it, that is a strong sign, not just of marginal constraints in housing supply, but of constraints so overwhelming that they move families away from these very stable long-term patterns.

As I’ve probably noted before, if someone from Ohio gets a job offer in New York City, they would never cut their calorie intake by 70% in order to make their budget work in New York, but they very well might move to a 70% smaller home. Housing is a bundle of some essential necessities and a lot of luxuries. We are very willing to trade off those luxuries to keep our spending in line. But, there are also endowments as part of that bundle (living in a hometown, being near family and friends, knowing ones neighbors, knowing local support services, etc.). We dislike trading those away nearly as much as we dislike trading away necessities like a roof and a kitchen. So, when the newcomer moves into the 70% smaller home in New York City, New York City housing politics require that someone else must trade away their necessities and endowments to move out of New York City. Resistance to those moves is why New York City is expensive, more than demand from the newcomer is. The newcomer is bending. The marginal leavers in a city decades into persistent displacement are not keen to bend on the housing necessities they are holding on to. And, when one of the small portion of displaced New Yorkers who value the endowments of their life in New York City even more than the necessities of shelter and privacy becomes the next marginal displaced resident, they remain in New York City and give up their shelter.

(Don’t take this to mean that New York City’s problem is too many newcomers. They have a normal number of newcomers, but their housing policy makes those newcomers disruptive.)

That is why housing expenses have this pattern and other categories of consumption don’t. We bend a lot to keep spending within the normal range. Though, it happens in chunks. We don’t all trade around the housing stock every month to keep our budgets intact. And, we don’t easily trade-off elements of housing when housing expenses are so out of line that they require us to trade down, especially if that means giving up endowments or necessities.

That explains the downward slope of the rent/income pattern. Families with lower incomes are paying, mostly, for necessities and endowments. However, I think there are some tricky issues with housing now that make the spending levels within each decile in the chart look more stable than they actually are. And there are some subtle things to consider about housing consumption that I think this chart helps to inform. And, I have one or two complaints about Sunderji’s post.

First, A Complaint

I have a commitment to pointing this out when people do it. Sunderji writes, “More building might get you more square footage for your 30% (the median new home grew from 1,500 sq ft in 1970 to 2,300 sq ft by 2020). But it won’t change that 30%, which seems to be the equilibrium share that humans allocate to shelter when they have any choice at all.”

Sunderji tends to be a skeptic of the supply crisis narrative. Supply skeptics have to benchmark to 1970 and whistle past the the most recent 3 decades on this point because households haven’t been getting significantly smaller, and new homes haven’t been getting larger for some time. New homes have been getting smaller, on average, since 2007. The average new home today (inclusive of both single-family and multi-family) is roughly the same size as it was in the mid-1990s, even though the average real income is up 70%. That’s a huge differential.

Figure 2

This actually understates the problem a bit. Land values under existing homes are inflated. Builders compensate for the worsening affordability by shifting to smaller homes. But, they also compensate by building new homes in ways to minimize land cost - in Phoenix instead of Los Angeles, or on a smaller lot in Phoenix, or in a worse location in Phoenix. So, new home prices haven’t been as inflated as existing homes. The Case-Shiller index, which tracks the value of unchanging existing homes, has inflated much more than measures of new home prices for this reason. New homes aren’t just getting smaller. They are reflecting compromises on several margins.

Sometimes, you’ll see claims, using new home data, that homes aren’t getting that much more expensive, or that land costs aren’t that elevated. Well, if every home in Los Angeles has a half million dollar scarcity premium on it, that will show up in the Case-Shiller index, but it doesn’t show up in new home prices because Los Angeles is associated with an insignificant portion of new home activity. It has a large scarcity premium because it has an insignificant portion of new home activity.

Anyway, homes got larger and families got smaller from 1970 to 1995. There wasn’t a housing crisis in 1970 or 1995. A small subset of think tankers have managed to seed the discourse with this talking point that obscures the most recent 30 years of an important social and economic trend. If you are a common consumer of housing discussions, this won’t be the last time you see that 1970 comparison but no comparison to 2010 or 2000.

In the first half of that time period, when homes were getting larger at the same rate as real incomes, rent inflation was moderate. In the second half, homes stopped getting larger in line with incomes, and it was during that period that high and regressive rent inflation started to dominate housing economics. This isn’t about things like households valuing location over size. Supply curves have been moving to the left faster than families can backtrack, compromise, or displace themselves.

Now, if it was just a matter of low, or even slightly declining productivity in the construction sector, I think the main point of Sunderji’s post would be correct. People would continue to live in homes with rental values anywhere from 15% to 60% of their incomes, mostly depending on what their income was. The typical family’s home would be smaller than they otherwise would be, but probably larger or nicer than their parents’ and grandparents’ homes had been. Everything would be, more or less, fine. There would be less housing than there could or should be. But, it wouldn’t be a “shortage” or a “crisis”.

But real housing that is losing pace with real incomes by a couple percentage points a year, on average (and in an asymmetrically regressive way) is a shortage and a crisis. And one sign that it is a crisis is that the quantity of housing is diverging from real income growth because families are making all sorts of compromises around housing consumption, and in many places their spending on housing, after making all those compromises, is still rising enough to force them into decisions under duress, like accepting regional displacement from a region where they have strong personal ties.

Sunderji notes, “Easterlin’s paradox was resolved by understanding that happiness correlates with income rank, not income level. Sunderji’s paradox resolves the same way: what matters for your rent burden isn’t how much you earn, but how much you earn relative to everyone else. We are all competing for the same locations, and location is in fixed supply.”

All of that is true until we get to the last sentence. The last sentence isn’t exactly false. But it is overemphasized. And most of the limits on the utilization of location are political and arbitrary. It’s an example of the damage that spatial equilibrium models and agglomeration economy models have done to this discourse. Locational value is real. But, it doesn’t explain the recent elevation in housing costs. And, while housing costs were inflated in some undersupplied cities before 2008, it didn’t equate to a national shortage of housing units because those units were built in other places. They just weren’t built in a few cities where a lot of people already lived and would have preferred to stay. The painful migration required by that problem led to so many homes being needed in other places before 2008 that it created the infamous housing bubble in Arizona and Florida. That was a regional shortage, and it was a crisis for a lot of families in Los Angeles and New York City, but it wasn’t associated with a national shortage of units or a national housing crisis.

This creates a sort of political horseshoe in housing. There are some economists who are very focused on the shortage, but they are convinced that it’s a location thing. They think homes got very expensive because certain prime locations became very valuable, and those specific locations need to loosen their local land use rules. That sort of turns the housing shortage into a demand story. And, it makes it harder to counter poor supply-skeptical arguments like the one recently being raised at the San Francisco Fed.

I must admit that I was still sort of thinking of it that way when I wrote “Shut Out”. For instance, I wrote, “We had a housing supply bust—first in the places where people want to live, in places where there is more economic opportunity.” If I was rewriting the book today, I would de-emphasize the importance of demand. I don’t think demand from newcomers is necessary nor sufficient to elevate housing costs in the expensive cities. And it doesn’t deserve an important presumptive role in a story about the 5 major metropolitan areas that had the lowest rate of new home construction, and the highest rate of net outmigration, becoming expensive before 2008. Especially, with another 2 decades of hindsight where every other city started growing more slowly and also started becoming more expensive in the same way.

In any case, the overemphasis on locational value sort of hands all the trump cards to supply skeptics. It supposes that high costs in places like New York City are paid aspirationally. If an increase in the number of families with housing costs that are more than 30% of income reflect an abundance of newcomers, then the case for understanding and acting on the national supply crisis is undermined.

Across the US, the families spending the most on housing are families trying desperately to stay in place in cities that are growing more slowly than they used to. And the most expensive cities are the cities where the most families are giving up on staying in place. That’s why it’s a crisis. Today, there aren’t many cities to move to anymore where your arrival will increase the number of new homes rather than increasing local land rents. It would take 15 to 20 million new homes to eliminate land rents and allow for natural household formation. That’s why it’s a shortage. (And, there is no correlation between the increase in land rents in cities across the country and locational value within each market. One way I would put it, in the language economists would use to identify, pedantically, a shortage, is that inflated land values are the bribe you pay to the land for the right to build a house in a market where that right is limited.)

So we have supply skeptics, like Sunderji, who believe there isn’t a supply crisis because square feet per capita increased 40 years ago. We have supply-focused economists who undermine their own case because they note that the problem was limited to only a few prime locations 25 years ago. And for 20 years, rents have been eating up the incomes of the most vulnerable American families from Albuquerque to Ypsilanti because their neighbors couldn’t qualify for the mortgages they would have used to buy new homes on the edge of town. (And, Albuquerque and Ypsilanti are as allergic to dense infill construction as cities that were already expensive before 2008 are.)

What are you buying when you buy a house?

Why does spending on housing look like this? For better or worse, and for both natural and political reasons, families segregate by income across an urban market. Much of what we pay for a home is the price of living in an area with a given socio-economic character. The causation sort of goes both ways. One reason some neighborhoods are valuable is because they are too expensive for poorer families to live in. The homevoter hypothesis says that homeowners support zoning so that their home values remain high. I think homeowners are actually trying to maintain the socio-economic position of their neighborhoods without increasing home values, which is one reason the mortgage crackdown was so popular. And it is why neighborhoods rarely lobby for targeted upzoning that would give all the homeowners in the neighborhood massive windfall gains.

This is also why that Ohioan may be willing to reduce their housing footprint in New York City by 70%. Because, due to higher construction costs and decades of displacement caused by zoning, there are 70% smaller homes in New York City that are located in neighborhoods with an acceptable socio-economic character for them. They would be less likely to move to a 70% smaller home in Ohio because that home would be in a neighborhood with a lower socio-economic character.

The income deciles in Figure 1 are for renters. Homeowners tend to have higher incomes than renters. The median homeowner has an income that would probably be in the ninth decile of renters. So, imagine the top 9 deciles of homeowners roughly spanning the income range from about the 5th decile of renters to an income a bit higher than the top decile of renters.

And, the pattern of home prices for homeowners is different than rent is for renters, as shown in Figure 3, from a recent post. In the left panel, Phoenix and Cincinnati in 2002 were examples of cities without a supply crisis. Among homeowners, housing costs are relatively similar across incomes. In cities with enough housing, a price/income ratio of about 3x is common, and usually price/income ratios in the neighborhoods with the lowest incomes are about the same as in the richest neighborhoods - maybe a 1/2 point more in the typical city.

Figure 3

At typical mortgage rates, for a new buyer with a small down payment, that puts housing costs, including the mortgage payment, some owner expenses, taxes, and insurance, at around 35% of income, on average. First-time buyers frequently have starting debt-to-income ratios above 40%. Just as price/income ratios are relatively uniform across amply supplied markets, there isn’t a particularly strong correlation between debt-to-income ratios and borrower income.

Housing costs for owners are more complicated, because the cash flows don’t match the consumption like they do for renters. But, using the typical numbers for first-time homebuyers, and accounting for the fact that homeownership is increasingly likely in higher income deciles, Figure 1, accounting for both owners and renters, is much more uniform, starting at 60% and then settling at around 35% or so for most income levels, rather than declining to 15%. Mostly, what families with above average incomes are purchasing when they buy a home instead of renting is control.

This means that price/rent ratios are systematically higher in neighborhoods with higher incomes. And, wherever families have the means to be owners, the cost of ownership is always much higher than the cost of renting. Keep that in mind when you see claims that the housing market is overpriced because it is cheaper to rent the average home than it is to buy it with a mortgage. That isn’t really a margin around which housing markets revert. The 2010s was a really odd period because it was defined by the aftermath of the mortgage crackdown. A lot of families were cut out of the buyers market, and those who were still able to buy got bargains the likes of which they will never have again. It is really weird for the typical home to be more affordable to rent than to buy, and it is a strong piece of evidence that homes were too cheap after 2008.

On the other hand, in poorer neighborhoods, owning is usually cheaper than renting, which is why poor single-family neighborhoods are dominated by small scale investor landlords. It’s a good investment. It’s a good investment that the FHFA and CFPB have been tasked with blocking working class tenants from making.

Now, as the right panel of Figure 3 makes clear, owning has become more expensive in poor neighborhoods. That doesn’t necessarily mean that homeownership is a worse investment in those neighborhoods than it used to be. Ownership costs more because rents are higher, and the income ownership buys you (either in cash rents as an investor or in rental value as a homeowner) is higher.

Why then does Figure 1 show rent/income ratios across incomes that are about the same as they were in 2005? I think there are several reasons why the housing crisis doesn’t show up clearly in Figure 1.

But, this post is getting too long, so I will save that for Part 2.