New paper!

Home Price Trends Point to a Worsening Lack of Supply

Paper #4 of my recent series of papers on prices and rents at Mercatus is out.

Paper 1: Rising Home Prices Are Mostly from Rising Rents

Paper 2: Price Is the Medium Through Which Housing Filters Up and Down: A Proposal for Price/Income as an Indicator of Housing Supply Elasticity

Paper 3: Reassessing the Role of Supply and Demand on Housing Bubble Prices

Paper 4: Home Price Trends Point to a Worsening Lack of Supply

Paper #2 is the conceptual lynchpin here. My thesis is that regionally constrained supply creates a predictable systematic pattern in housing costs. Costs rise the most in neighborhoods where families have the lowest incomes. The relative costs are driven by intra-metro migration patterns that a lack of adequate supply triggers.

Figure 1, from the new paper, should look familiar to long-time readers. There is a reliably linear relationship between local incomes and local price/income levels on housing. That relationship is flat in an amply supplied market, and it tilts down in an undersupplied market.

Figure 1

Atlanta, and many other cities, used to be even more flat than Atlanta in 2021 is in Figure 1. That is because housing filters down in a healthy market. Here is how I would put it:

Where housing supply is ample, demand is largely driven by incomes. Families, on average, will spend a predictable portion of their incomes on housing. Demand for housing is non-homothetic, meaning that spending tends to decline as a proportion of income as incomes rise. I think this is mainly due to the fact that what constitutes housing is a bundle of services that combine strict necessities with luxuries. A rich family might own a home that includes basic tools for eating, sleeping, and cleanly defecating, but also entertaining, playing, socializing, and showing off. The basic parts of that bundle are so essential that a family may be willing to spend every last dollar and more to retain them. That same family will give up some of the luxuries at the drop of a hat if they require a dollar more than their income can comfortably afford.

This means that, where incomes are low and/or the cost of the necessities of shelter are high, families systematically and predictably spend more of their incomes on housing. Where incomes are higher, the total portion of a family’s income spent on the bundle of services we call “housing” declines, and families predictable add luxuries to the bundle in scale to that proportion. The less families spend on housing, the more luxuries they will include in the bundle.

In a supply constrained market, like LA in Figure 1, almost every household is in a financial state of deprivation where they spend more on the housing bundle than is comfortable. In Atlanta (and especially in cities like Atlanta in the past when housing was more plentiful), almost every household is in a financial state of satisfaction in terms of the necessary part of the housing bundle, and so the difference between families in Atlanta isn’t the proportion of their income that they spend on housing. It’s mostly the amount of luxuries they choose to include in their housing bundle.

That is basically a description of Figure 1 and why the slope of the trends in Figure 1 in each metropolitan area at any point in time reflect the scale of housing deprivation they are experiencing. Twenty years ago or more, most cities were relatively flat, and housing deprivation wasn’t a particularly important issue. Today most cities are somewhere between Atlanta and Los Angeles and housing deprivation is important everywhere, and very important in a few places.

What this means is that in places with flatter price/income patterns than 2021 Atlanta (which used to be most cities), the cost of homes is largely a function of incomes and the nominal value of home is largely a product of changes in residential investment and maintenance. Also, different types households move between neighborhoods over long periods of time. All those other moving parts - homes changing slowly over time and the residents moving around the region slowly over time - play out to maintain the relatively stable nominal housing costs that keep prices that are about 3-4x the income of their residents regardless of their incomes.

In terms of a model, “affordability” in that context is relatively fixed, and everything else changes to keep it so. Homes cost 3-4x income, and they can be 800 square feet or 4,000 square feet, full of electronics and upgraded interiors or white walls with a couple of mattresses and old chairs, etc. There is a huge range of relative value that can adjust to stay at that 3-4x income range. You really have to sustain a ridiculous local political regime over a long period of time to look like LA in Figure 1.

Figure 2 shows the price/income levels of homes across Atlanta and Los Angeles in 1996 and 2005. The y-axis scale is set equal to Figure 1 to highlight the change over time. The local effect of supply deprivation took hold in the 2000s in Los Angeles, and the migration statistics corroborate the process that was heating up - LA residents flooding away from deprived, expensive housing markets. But, before the turn of the century, LA was much more like Atlanta than it was like 2021 LA.

Figure 2

The localized effect of inadequate supply that increased home prices in the 2000s in LA was interpreted as the result of a credit bubble. That’s the topic of paper #3. The way the 2000s was reported in much of the academy was that a credit bubble inflated prices, and it especially inflated prices where supply was inelastic, like in LA. I argue in paper #3 that there was a very large effect from supply constraints that pushed up prices in LA in the way that inadequate supply does (shown in Figures 1 and 2). It also indirectly pushed up prices in places like Phoenix because of the wave of housing refugees that flooded out to those cities, which created extra demand in cities like Phoenix and raised prices in a way that was likely unsustainable and worthy in some respects of being called a bubble. (But, still, it was triggered by the lack of housing in cities like LA.) And, while all that was happening, there was a credit boom that was quite a bit less important of a factor driving prices higher. Where it did drive prices higher, it did so at least as much in cities like Atlanta as it did in cities like LA.

The regional differences between LA and Atlanta which are obvious in Figure 2, and which were the primary trigger for the moral panic about mortgage lending, were really entirely a supply issue, which is why those differences remain today, even though mortgage lending to sub-760 credit score borrowers has declined by more than 70% compared to 2005 and 1996 norms.

The moral panic led to temporarily and unsustainably low home prices in cities like Atlanta, which were the cities that didn’t have the price spikes which had triggered the moral panic. That is something I go into in the new paper.

I will spend a few posts reviewing the new paper, which has three main points:

1. One pattern that is associated with greatly obstructed regional housing supply is a negative correlation between population growth and housing production. This peculiar pattern may provide another tool for identifying supply-deprived markets and quantifying the scale of the shortfall.

2. Price, rent, migration, and housing production trends after 2008 were much different than trends had been before 2008. This points to a universal shift to inelastic supply. Every city became more like LA.

3. The most likely culprit for this shift is overly regulated mortgage access, which has undercut supply the most where incomes are lower and new housing production is more sensitive to access to mortgage funding.

I will use some posts to outline those points in more detail. Then, I will publish a final post that will link to all of my summary posts of the 4 papers.

I am currently editing one additional paper that continues to use the framework from paper #2 to quantify the effect of inadequate supply on aggregate real estate values. That will come out later this year.

These four papers form the basic framework that aligns with the model I use here at the newsletter.