Subscribe now

Read more

Subscribe now

Before 2008, a handful of cities on the blue path had an especially large cycle - the Contagion cities. That cycle was largely caused by the outflow of households from the Closed Access cities which spiked migration rates into the Contagion cities, creating the need for a housing boom.

Famously, the combination of those trends that were caused by a lack of housing was entirely blamed on cyclical excess. In fact, I think it is unavoidable that secular trends will be initially mistaken as cyclical, because cycles happen more quickly, we always have them, and cyclical changes are more salient to us.

So, the problem I think you can see playing out in Figure 1 is that as a secular supply problem puts real estate valuations on an upward slope, the conventional and relatively universal reaction is to say, “Oh, no! The cycles keep getting worse! We have a bubbles problem!” You can see this reaction all over the place. It’s probably the modal belief among lay investors, and maybe even professional money managers.

Then, when the cycle normalizes, it normalizes at a much higher price level. Again, you can see these reactions all over the place. Many high profile economists and pundits from, say, 2009 to 2011 were warning that home prices needed to decline even further to get back to normal.

Of course, by 2011, the American home value, represented by, say, the Case-Shiller index, was an average of low-tier homes across the country’s interior that were well below any reasonable equilibrium price, and homes in the Closed Access cities that were still elevated, but that were well below the prices that would be justified by the high rents created by their regional shortages.

Then, the next cycle that played out after 2020 was actually a very mild cycle, nationally, as Erdmann Housing Tracker subscribers know. Yet, the man on the street reaction is again, “This is an even larger cycle than the last one.” We are just further up the orange line now. And, the orange line moved up even more sharply because the engineered collapse of the new home construction market after 2007 created a shortage across the country, and now every region has experienced a decade of regressive rent inflation similar to the Closed Access cities.

That condition will be temporary unless we keep working hard to make it permanent, by, say, making new rental housing illegal while we continue to exclude families from mortgage access and homeownership. The mortgage regulators are just doing the Lord’s work, protecting families from themselves, but the big investors that have replaced those families in the market are buying homes out away from families, don’t ya know?1 Did I mention those intuitions?

Anyway, I have realized that the chart plays out a bit differently on the way down, though, again, it is the salience of the cycle that drives poor intuitions. The Closed Access cities are still riding the orange line up. Actually, the Closed Access cities have generally reached a ceiling on the real estate value / GDP metric. They have reached a functional maximum. When cities initially lift off the blue line, most families react by paying elevated rents to avoid local or regional displacement. But, after that process plays out perennially for years or decades, more families reach their limit, and react by accepting displacement. So, the aggregate metro area real estate / income ratio asymptotes, and more rent inflation requires an equivalent rise in average local incomes which is achieved when poor families choose to accept their fate and move away, raising the average income of the remainers because of leavers’ absence. And, then some clever academic writes about how there isn’t really a housing shortage because home prices just rise in line with incomes. The intuitions. Did I mention them?

The cities that had been on the blue line before 2008 have been riding up their own orange line since then, but temporarily. Cities like Austin that are leaders in multi-family re-legalization and other cities that just still have plenty of greenfield space in the suburbs and exurbs cannot permanently maintain elevated real estate values. And, as the national capacity to build slowly returns, those cities will slowly revert back down to the blue line.

On the way up the orange line, the intuition-impaired ignored the huge cost differences and housing-induced migration patterns that resulted from local shortages, and confused local conditions for national conditions. So they looked at the Case-Shiller index and decided that prices were cyclically too high in 2005 when prices in most cities were pretty normal. And they decided that prices were still cyclically elevated in 2011 when prices in half the country had been beaten so disastrously low that the construction industry was permanently handicapped. Of course, it is common to see the claim that the homebuilders were consolidating and expressing market power at that time - purposefully throttling construction activity to keep margins high. There is even at least one academic paper about it.

On the way back down to the blue line, the cycle is still more salient than the long-term trend. So, even if a city is attracting enough capital into their local housing market to moderate long term rents and prices and return to the blue line, that moderation will play out cyclically. Prices and rents fall faster during a local cyclical downturn, and under current conditions, they might still rise during a cyclical expansion, even if a city is doing better than average.

This is actually really weird. In the 20th century, rents and prices didn’t rise and fall sharply over the business cycle, because urban land didn’t have a scarcity premium. The price of homes largely reflected locational amenity value and the cost of construction. When cycles moved up or down, the number of homes being constructed changed, a lot, but rents and prices didn’t typically fluctuate more than a few percentage points relative to incomes over the course of a cycle.

Some of the intuition-impaired will report that builders won’t build so much that rents decline, because they foresee that too much building will cause rents to decline, and they pull back on construction activity. And, that was very true of the 20th century. But, it hardly took any change in rent trends at all to trigger very large cyclical shifts in construction activity. It doesn’t take long for builders to notice that units aren’t selling or renting at a pace that can sustain a given trend in construction activity. Builders were very sensitive to changing demand because rents weren’t elevated.

But, rents are elevated today - specifically land rents. New construction will lower land rents, and land values under these conditions. Builders might pull back in a given market in a way that is correlated with rent trends and land price trends, because under the conditions of a national capacity constraint, inputs will be attracted to markets where the shortage is associated with the highest land scarcity premiums. This will make those misapplied 20th century intuitions seem to be vindicated. But, the places with local regulatory and geographical conditions most amenable to more housing construction will bend their curves the most over the full cycle. Nashville, Raleigh, and Austin are good examples of cities that have pushed further back to a more affordable trend even while dealing with unusually high growth rates.

Of course, Austin is the standout. The way to pull scarce resources into your market while the industry is at capacity is to remove unnecessary local constraints to the most fitting forms of urban housing. On a per-capita basis, both Austin and Las Vegas still permit fewer single-family homes than they did at the 2005 peak because the mortgage crackdown specifically derailed single-family construction. But, in Austin, multi-family permits reached double their 2005 level and are still roughly similar to that level, even after a cyclical pullback. Las Vegas multi-family permitting is down as much as single-family permitting is, compared to 2005. (And, in case some readers are new here, one result of those housing intuitions was a mass hallucination that there had been a building boom in Las Vegas in 2005.)

A partial view of the cycle might fuel the intuition that Austin’s recent good track record on rent trends was just the result of a cyclical downturn. But since 2015, rent trends in Austin have fallen more than 20% compared to Las Vegas while Austin has grown at twice the rate.

I think, under current conditions, reforming multi-family land use regulations was probably the only way that Austin could have produced such a winning outcome. Austin still had an up and down cycle. Construction in Austin has moved up and down with that cycle. Your monkey mind might tell you that it’s all about the cycle and that your long experience in 20th century housing markets taught you that builders won’t produce into rent moderation.

In the 20th century builders did nothing but build into rent moderation. Sometimes that meant building a lot. Sometimes it meant not building much. The vast expanse of American cities lost moderate rents after 2007, and where growth pressures are the highest, they also lost moderate prices. Those high prices won’t last. Cities that are supply-oriented will get back to moderate housing markets more quickly. And, after they do, you’ll be able to say that developers won’t build into more rent moderation. When the scarcity premium is gone, that will be true. They will build until rents are moderate, and then construction will fluctuate to maintain moderate rents.

Until then, those pesky intuitions will provide a lot of counterparties for those who see the coming building boom, and find wise ways to invest into it.

Subscribe now

Subscribe now

I saw this interesting, short Minneapolis Fed post, and figured I owed inflation an update. Figure 1 compares core CPI inflation to core PCE inflation. CPI has run hot because housing is a larger component of the CPI. Fortunately, the Fed technically targets the PCE price index, so maybe the bias has been lower than it could have been, but CPI tends to drive the public conversation.

The one period where CPI didn’t run persistently higher than PCE was in the peak years of the housing boom when the rest of the country was finally building enough to fully counter the rent inflation of the Closed Access cities. Rent inflation briefly converged down toward general inflation at the peak of the boom.

We would be living in a very different economic and political environment if that boom had been allowed to continue, but it wasn’t. I choose my words carefully, there. That sentence probably sounds like cheap conspiratorial stuff. But, in this case, clearly, obviously, there was a concerted public policy effort to slow down housing construction, and it was very popular. And, so, rent inflation moved back up, and the CPI trend ran hot again in the 2010s.

Recently, the CPI has dipped below PCE. One reason is that rent inflation hasn’t recently been elevated. Neil Mehrotra and Michael Waugh at the Minneapolis Fed disaggregate the PCE price index into core goods, core services excluding housing, and housing (rent). You can see that housing isn’t inflating the index currently.

In Figure 1, you can see that CPI tended to run pretty close to the Fed’s 2% target in the 2010s after the financial crisis settled down. PCE averaged about 1.6%. In Figure 2, from Mehrotra and Waugh, you can see that without housing, it would have been closer to 1.2%. I would prefer to see non-housing core inflation at 2%, and that’s basically what JPow! (the once and forever Fed king) has managed since July 2022. That was a combination of services near 4% and goods near 0%. Stated core PCE inflation was still close to 3% because of housing.

In Figure 3, I have pulled housing out of the CPI, and compare it to an adjusted estimate of Zillow’s US rent measure. You can see that inflated rent after 2022 was entirely due to lags in the methodology of measuring rent inflation. Before 2022, rent was inflating the index because rent inflation was excessive. Since 2022, it has largely been a methodology quirk. Over the past 12 months or so, Zillow’s measure suggests that rent has been especially moderate. It has recently recovered. I don’t think rent inflation will return permanently to 3%+, but I do think it will temporarily be above 2% before slowly declining back to and then under 2% when new home completions rise - if we let them. To the extent that increased completions will need to be rentals because of the policy detritus from the last time we aimed to slow down housing production, we might collectively act to keep ourselves underhoused again, in which case rent inflation will again remain elevated.

Anyway, back to the Figure 2, they find that core goods inflation is raising core PCE inflation by about 0.7% - roughly half related to the AI boom and half related to tariffs. The AI boom appears to be competing with housing for some inputs, so some of the recent inflection back to higher rent inflation showing up in the Zillow measure might be somewhat related to the AI boom.

For 4 years, the Fed has hit the target I like the most. The risk of a monetary dislocation seems higher now. Tariffs, immigration, etc. make it harder to know what trends to target, and the in the moments where new Fed chair Kevin Warsh was in the room during important inflection points in monetary policy decision making, he has had very wrong positions - such as being hawkish in the summer of 2008.

Core services excluding housing may be a key item to watch going foward. I like where it has been for the past couple of years - just under 4%. I think that equates to 2% non-housing inflation. If that measure takes a sharp turn, it might be a sign of coming nominal economic trend shifts and dislocation.

With the AI buildup and the defensive position of residential construction under current conditions, it seems like any coming cyclical shifts will parallel the 1997-2002 period.

Subscribe now

More on that below, along with this month’s Erdmann Housing Tracker update.

Subscribe now

Read more

If single-family homes sold continues on such a flat trajectory much longer, we’ll have to start wondering if it has a de facto national cap like multi-family does.

I’m kidding… I think.

Subscribe now

Read more

I’ll try to figure it out below.

Subscribe now

Read more

Subscribe now

Here are various observations. Staveski writes:

A 2019 research paper published by the Federal Housing Finance Agency (FHFA) used a combination of nonpublic and proprietary data to understand how mortgage lending standards changed from 1994 to 2019. Using a statistic called the stressed default rate, the authors estimated the share of mortgage originations each year that would default in the event of a financial crisis similar to the Great Recession.¹² By that metric, the mortgage market had far less default risk in 2019 than in the mid-1990s; neither period was as risky as the early and mid-2000s. The authors estimated that 13.5% of borrowers would have defaulted in a crisis in 2019, compared with 18.4% in 1997. (See Figure 2.) A 2022 update to this FHFA report found that credit risk decreased further in 2020 and 2021, reaching its lowest level in modern history—a projected crisis-induced default rate of 10.4%.

Figure 2 is from the Pew post, and the next figure is from the Urban Institute. In the Erdmann Housing Tracker, I adjust the credit component manually. It is calibrated to changes in home prices measured after 2007, and I generally set it to track the New York Fed’s estimate of the median credit score on new mortgages. The tracker has been attributing a significant drag on home prices in the last few years to even more tightened lending, and both of these indicators corroborate that trend.

One pedantic point I would make about this is that I can understand why modelers naturally use the 2008 crisis as the benchmark test for default risk, but it really isn’t appropriate. Most of the defaults after 2007 were caused by the severe tightening of lending standards that collapsed home equity in credit-sensitive neighborhoods. This wasn’t some uncontrollable business cycle event. We could choose not to do it again. In fact, we couldn’t do it again. There aren’t that many marginally qualified borrowers left to exclude.

I have a couple of pedantic comments on the piece. Staveski notes that down payments are lower than they were in the 1990s, though one of the measures he cites shows down payments growing in recent years. One measure shows the average loan-to-value at origination rising from about 84% in the mid-1990s to about 86% in 2020.

He also notes that debt-to-income ratios (the percentage of income going to the first mortgage payment) have increased, too, from about 33% in the 1990s to 39% in recent years. He equates these shifts with a loosening of standards, and in some technical way, it is. But, in both cases, this is more about home prices changing than it is about borrowers changing.

Consider the down payment. The average home price is up about 42% since the 1990s relative to incomes (high tier prices are up less and low tier prices are up more). So, if the 20th century norm was to spend 3 years’ of income on a house, now families have to spend 4.25 years’ of income for the same house. That means that in the 1990s, a 16% down payment took about 48% of their annual income. A 14% down payment today takes about 60% of their annual income. So, they are actually making larger down payments than they used to. If underwriters had been operating with stable standards, down payments would have declined as a percentage of the home value. In fact, only moving from 16% to 14% is likely to represent tightening, even on this margin, after accounting for high home prices.

Similarly, the debt to income ratio would have increased from 33% to 47% purely due to rising home prices. So, again, the fact that it rose to 39% likely represents relative tightening in lending standards, given the rising cost of homes.

Debt-to-income is actually an interesting measure under current conditions. In the before times, when housing supply was adequate, price/income ratios were relatively stable across a market. So, debt-to-income ratios weren’t really particularly highly correlated with borrower income or quality. But, since the shortage has a highly regressive effect on home price inflation, the shortage now does cause debt-to-income ratios to be highly correlated with borrower income or quality.

This has a couple of implications. First, when credit access was tightened, it didn’t really cause the average price-to-income ratio to decline. Pew shows it declining temporarily from 39% in 2007 to 35% in 2012. But, of course, home prices declined by 35% relative to incomes over those years, and mortgage rates declined. Here, I’m not claiming that the relative stability of debt-to-income means that credit standards weren’t tightened. Clearly they were. Debt-to-income just isn’t a great signal of credit standards. Or, it least it shouldn’t be. So, debt-to-income can be cited to claim that credit standards have changed or not changed in ways that I don’t think are really accurate.

And, now, if families with lower incomes were allowed back into mortgage markets, the condition of the housing market would tend to force them to take on higher debt-to-income levels and make smaller down payments as a percentage of their home values. The shortage is the reason these measures are elevated.

A good rule of thumb is that where you see a focus on debt-to-income and down payments, you’re reading the output of a credit hawk. They will get everything wrong. I’m not saying that is what Staveski is doing here. He is reporting on these facets of borrowing in a way that he needs to, and is appropriately putting his focus elsewhere.

Ironically, tightened lending standards created the supply crisis that created the lending stresses that the credit hawks use to claim that lending standards can’t be safely loosened. They are basically NIMBYs working through the credit channel instead of the zoning channel. There is no feasible standard that can overturn their point of view. It starts with a presumption that the main cause of rising home prices is mortgage access itself, and then these confused causations create a vicious cycle where the answer is always more exclusion.

And, of course lending standards can have some effects on housing markets. The Erdmann Housing Tracker has a Credit component, after all. But, the effects of loose lending are generally greatly exaggerated and blamed for high prices that were caused by other sources of supply and demand imbalances. And, looser lending access general works through rising price/rent ratios. Where access has been cut off, in the long run, rising rents have more than made up for the declining price/rent ratio. Tight lending made housing less affordable, especially for families that can’t get mortgages.

A margin where I wonder if there was some empirical reason for Staveski’s choices is the focus on sub-700 credit scores. Maybe it’s just a matter of data sources. Staveski reports that the average credit score on approved mortgages used to be in the high 600s and now is near 740. The New York Fed puts it at more like 720 rising to 770. So, I think this is mostly a product of citing different credit score sources.

But, in Staveski’s Figure 7, he shows originations to scores under 660 to be down about 75% since the 1990s, and scores that he bins as 661 to 850 are down about 25%. I’m sure he had good reasons to focus on those score ranges. But, I would expect that, even on the scale that he’s referencing, if he split that into a 661-720 bin and a 721-850 bin, the lower bin would also be down more than 50% and the higher bin would show originations as high or higher than they were in the 1990s. The tightening has gone well into prime borrower territory.

He notes:

When policymakers tighten credit standards, they disproportionately affect the very people whom public policy often aims to help: young adults just entering the housing market, lower-income families, rural communities, and Black and Hispanic households. Although some of these potential borrowers might not be financially prepared to take out a mortgage, others are excluded because of a thin or nontraditional credit history, or because the federal government’s credit standards are historically high. While tighter standards have made the mortgage market safer, they have also made it harder for some qualified individuals to achieve homeownership.

Hear, hear.

The federal agencies had actually made great strides in the late 1990s expanding credit to underserved markets through improved underwriting methods. The expansion of homeownership that resulted from it had nothing to do with the housing bubble or the subsequent default crisis. But, we lost all of that with the 2008 tightening.

Staveski also points out that during the Covid shock, the FHFA made great strides in creating programs that help delinquent borrowers avoid default and foreclosure. The program was a great success with little downside. Mark Calabria, the former head of the FHFA deserves kudos for managing that process.

Credit hawks tend to want to make borrowing as expensive and potentially harmful as possible. It’s sort of in the same category as monetary hawks who always think we need to be disciplined by recessions. Even the 30 year fixed rate mortgage, itself, is a product designed to front-load cash outlays to make it a more difficult substitute for paying rent. And, efforts to reduce the standard amortization period aim to make the substitution even harder. So, we have a lot of room for making mortgage markets more functional and accessible in ways that aren’t financially damaging to borrowers. And, innovations over time that we aren’t fully utilizing can produce even better outcomes than what was normal in the 20th century.

Bringing back access to mortgages for underserved populations and pairing it with what we have learned about default mitigation could be a real step to better housing and household financial conditions. I salute Pew for standing up on this topic. They will be attacked in some circles for saying plainly true things. It’s long past time to reverse the mistakes of 2008.

Subscribe now

The housing shortage creates a distinct pattern in prices in a given market. You can describe it in at least two ways. Figure 1 shows the pattern. You can describe it as a fixed premium on each lot that contains the right to a single unit of housing. That’s how I model it in my Metro Area Analysis Packages. One way to put it is that when there is a shortage, you have to pay a bribe to the land to get access to a house. There is a price for that right that is relatively uniform across a metropolitan area.

Notice in Figure 1, our eye doesn’t want to interpret the change as a uniform increase, but note that all the upward pointing arrows are the same length.

You could also look at Figure 1 proportionately. The lots that used to be very cheap are now much more expensive than they used to be while the premium adds less to the value of lots that were already worth a lot. That’s the model I use in the Erdmann Housing Tracker. How far, in percentage points, is any given home’s price from the cyclically neutral price that doesn’t include the scarcity premium? That’s more akin to what our eyes pick up from the figure, I think.

You can see how the pattern in Figure 1 would support various intuitions about home prices. It is very common for people to intuit that homes are more expensive because buyers want amenities. In the shortage market in Figure 1, it still costs more to buy a nicer home in a better location. So, observation will seem to confirm their intuition even though the effect of a shortage on home prices has exactly the opposite effect, in proportionate terms. Homes with fewer amenities have the highest proportional price appreciation.

You can also see how the effect of the shortage isn’t directly observable to developers. I use a hypothetical lot on the edge of the suburbs as an example of a lot that would normally have only marginal value under normal conditions. But, there isn’t a line in the sand that denotes where that lot is. The red line continues out into the land farther from the city center or with a less desirable location.

Subscribe now

Read more

Subscribe now

Read more

Subscribe now

After a decade of rolling around in this data, I have come to believe that, in fact, endowments (including a sort of lifestyle benchmarking) are the only part of housing that we are willing to pay so much for. And, the reason housing is in such a mess is that wherever families are spending twice what they should have to spend on housing, almost everyone inflates other explanations to bridge the huge explanatory gap between what they should have to pay and what they do have to pay.

Agglomeration economies. Credit bubbles. Speculation. Interest rates or monetary policy. The business cycle. Federal subsidies. All of those factors can, here and there, explain a 10% or 20% aggregate variance in housing expenditures as a portion of our incomes, and sometimes larger temporary variances or variances in how we consume housing. Housing discourse over the past 30 years - academic papers, the Financial Crisis Inquiry Commission, countless books and memoirs - has been almost entirely a debate about which of these real aggravating issues the debaters favor inflating to explain the high cost of endowments.

Before 2008, New York City, Los Angeles, San Francisco, and Boston became exceedingly (and regressively) expensive because the number of existing residents that needed to be displaced each year in order to shrink the population to fit in their stagnant stock of homes kept rising. The way housing markets clear in those cities is that rents keep rising until enough families choose displacement over failure and leave. Proportionately, rent inflation rises the most in the poorest neighborhoods under those conditions, and the leavers tend to be the residents of those neighborhoods.

Those Closed Access leavers became Contagion city arrivers, and by the mid-2000s, enough of them were arriving in places like Florida and Arizona to overwhelm those local markets and create housing bubbles there. In order to put a stop to those bubbles, all the proponents of the barely relevant factors like credit bubbles, speculation, interest rates, monetary policy, federal subsidies, and the business cycle found common cause in supporting policies that would create a deep recession.

And, the policy that was so popular, and came so naturally, that both the public and the academy barely know that we did it, was permanently blocking something like 15 million families out of mortgage access over the course of 2008 - amounting to a slowdown of about $1 trillion annually in liquidity from mortgage originations, and slamming the brakes on aggregate spending (and income, and employment, etc.)

Eventually, we lost our taste for recession and financial ruin, but we never regained our taste for 20th century mortgage access. The 20th century path to recovery had always come through recovery in new home construction. But, even the champion of that process opposed aiming for its recovery this time, because he was dedicated to some number of the wrong explanations for a housing-glut-that-never-was, like interest rates and monetary policy.

With a lack of mortgage access, there was a lack of liquid money and a lack of construction employment. Home prices had been pushed so far below the equilibrium price level that could have even induced construction growth that marginal stimulus couldn’t even help. And, in the face of all those challenges, after we had all jumped off the economic cliff together, hand in hand, the remedy that was left to us to try to stabilize the macro-economy was massive fiscal stimulus.

Figure 1 shows the growth of federal debt held by the public since 1970. Fully 40% of the debt outstanding was a direct result of the mortgage moral panic. Nothing else. If we hadn’t tightened mortgage access at the scale that we did in 2008, total federal debt would still be at a very manageable 60% of GDP.

The mortgage crackdown created the crisis. It created the liquidity collapse. It prevented the normal recovery of new home construction from ending the recession. And, it created a massive wealth shock that sucked $5 trillion dollars out of American household net worth.

The literature on the financial crisis isn’t really at odds with my assertion that those outcomes caused the call for fiscal stimulus and the sharp rise in public debt. Most of what economists like Atif Mian and Amir Sufi write about in the conventional literature about the period is the problem of the wealth shock and its effect on employment and private and public spending. They just never think to question the inevitability of the wealth shock, because they relied on all the wrong explanations for why housing had become expensive.

So, my assertion here isn’t particularly broad. I’m just adding the last little bit of connective tissue, that the mortgage crackdown was the proximate cause of the wealth shock. Figure 2, from an earlier post, shows the results of a simple regression between 3 month changes in home prices in Detroit and the beginning FHA market share in each ZIP code. Something similar to this happened in every city across the country. If your neighborhood had depended on New Deal mortgage programs for the past 80 years, you were screwed. If it hadn’t, you were fine.

Figure 3 gives a sense of scale on these changing measures of wealth and debt. By 2006, mortgages outstanding were equal to about 70% of GDP, but equity in those homes had risen to more than 100% of GDP. The common belief at the time was that the rising mortgage borrowing had caused the rising home values and equity. The mortgage boom was, at most, an aggravating factor amounting to a few percentage points difference in total residential real estate value. We collectively ran a national experiment after 2008 to test the thesis of whether high real estate values had depended on increased mortgage borrowing, and few seem to have bothered to pay attention to the results of the experiment.

I have added dashed lines to point to what likely would have happened if prime mortgage access had been maintained, if the Fed had aimed for recovery in new home construction by the end of 2007, and if the highly regressive wealth shock that followed in the absence of those choices had not happened.

There had been some cyclical froth, but by the end of 2007, aggregate home valuations were cyclically neutral. The decline in home equity after 2007 was the result of the mortgage crackdown. Total home equity would be worth less today than it is because, in the aggregate, home values are inflated today by the shortage. If we had built more homes, the total value of all homes would be lower than it is. It is the rent on the land under those homes that is inflated, and that land is bundled with the homes.

Mortgages outstanding were a bit elevated. There had been a lending boom; it just wasn’t responsible for elevated home prices outside of being a minor aggravating factor in some ways. So, mortgages outstanding would have likely declined a bit over time under normal conditions after the subprime boom ended.

And, public debt would have probably still been near about 40% of GDP when Covid arrived.

Finally, just for kicks, in Figure 4, I compare mortgage debt outstanding and public debt to total owner-occupied real estate value. Here again, I have added rough approximations of where the trends would have gone without the mortgage crackdown.

The increase in mortgage leverage after late 2007 came from collapsing home values related to the mortgage crackdown. And, the end of the subprime boom would have led to a moderate decrease in mortgages outstanding over time.

Federal debt would have risen moderately, but would still probably be less than 30% of the value of owner-occupied real estate. The reason it would have risen a bit is because aggregate real estate value would be a bit lower if we hadn’t cratered the construction market.

The Irony

Of course, the irony is that in terms of either ideology or personality type, the Venn Diagram of people who had a moral panic about mortgage borrowing and people who are concerned about the federal debt have a lot of overlap. We have a much larger federal debt because we made it much more difficult for families to qualify for mortgages. The federal debt panic is, effectively, completely the result of the mortgage panic.

Let me be clear. This isn’t an either/or thing. I’m not saying that the federal debt somehow replaced mortgage debt on the national balance sheet. Not at all. Today, mortgage debt plus federal debt combined are worth about 140% of GDP. If we hadn’t cracked down on mortgage access, total mortgage and public debt would be closer to 100% of GDP.

The attempt at safety backfired because it was based on false pretenses. We made ourselves more collectively leveraged by blocking mortgage funding.

And, it’s sort of funny, because as I write all of this down, it all seems quite obvious to me. And, I suspect that for readers who have been “Erdmann-pilled”, the intuition here is pretty easy. But for those who haven’t been Erdmann-pilled, this is probably a non-starter. If this was the first thing you showed someone about my work, they might be likely to decide I was a sophist and would vow never to read another thing I ever write.

It’s hard to know the difference between science and witchcraft when you don’t know the science. I’ve been at this long enough to know that there are a lot of empirical blind spots in the canon about what supposedly has happened to our housing. I might still be engaged in witchcraft. Maybe I have my own blind spots. But, I do know that there is plenty of witchcraft in the canon. As long as that’s the case, there will be true things to write that probably hurt my ability to engage in public discussion more than they help. I don’t know what to do about that except for plowing ahead and writing what I think. This is what I think.

Subscribe now

Subscribe now

I would like to add some emphasis to their findings.

Their cross-sectional analysis really highlights 2 of the empirical patterns I discuss a lot here at EHT. The housing shortage is a supply problem, not a demand problem. The housing shortage hits the poorest households the worst. And, so the worst hit households are the poorest households in the cities with the worst supply conditions.

The problem has gotten worse on all those margins over the last 20 years. In 2005, about 16%-18% of young adults earning less than $20,000 lived with their parents in all types of cities. By 2024, in the most expensive cities, it was 35%. And, by 2024, even 25% of young adults making $20,000-$65,000 were living with their parents in the most expensive cities.

The Urban Institute has a reasonable list of solutions. Relaxed zoning and building code regulations in the housing-poor cities. They also note that policies that subsidize rent aren’t very helpful because: “Used in isolation, demand-side tools will be partially absorbed by higher housing costs where supply cannot respond.” I would go farther than that.

The numbers here point to about 4 million unformed households, just from this 25-34 year-old age group. We are another 5 million vacancies short of traditionally functional housing markets. So, just accounting for unformed households in this one age-group and missing vacancies, we can account for nearly 10 million homes, which is most of the way to the 15 million unit lower end of the range of my estimate of the shortage.

Why does the shortage show up so starkly among this age group? As with all things housing, I think it is the product of very different supply and demand elasticities. This shows up all over the place. I would argue that the demand of an aspirational newcomer is much more elastic than the demand of an existing resident in a given city. That doesn’t matter where supply is elastic, but the existing resident becomes the price setter in a city where the shortage is so severe that rents have to keep rising until some families choose displacement. Wide variance in supply and demand elasticities create vastly different outcomes.

On this margin, I think the demand for new entry level units, such as small apartments, by young adults is much more elastic than the demand for those same units from working class families that are also in the market for homes that don’t include a lot of luxuries. The trade-off the young adults have to make is to simply stay where they are. The trade-off the working class families have to make is quite the opposite - they would have to choose to leave to keep their housing expenses low - specifically leave to go to a neighborhood or city with worse amenities, maybe even negative amenities, like crime. So, working class families are the price setters and young adults choose not to have housing expenses.

Supply is what it is. Under the conditions where these young adults aren’t forming households, the supply curve is effectively vertical. New supply hasn’t responded to higher rents. So, any rent subsidies or down payment assistant that helps these young adults form new households will require displacing more of the working class families who are currently setting the price. In a context where the housing market equilibrium is the balance of the rate of household formation, the rate of displacement, and rent inflation, rent subsidies to encourage household formation will lead to displacement. The rent inflation the Urban Institute points to is just the mechanism through which that displacement is determined.

That’s a permanent state in the Closed Access cities, but it is temporarily the state of the market across all major cities under current market conditions. Under the Malthusian limit - even if it’s an arbitrary one that we have chosen to be at - every positive will create a matching negative.

One final point I would like to highlight comes from one of the figures in the Urban Institute post. The madness of the 2008 moral panic is hidden in plain sight everywhere. This chart is a good example.

Look where all the increase in young adults at home happened - 2006-2015. That’s really weird! That is precisely the period where household formation went haywire, compared to any previous modern period. Homeowners flatlined from 2006 to 2015, and there was a surge of renters. Most of the young adults from 25 to 34 that move out of their parents’ homes first form households as renters.

So, if they weren’t forming households, who were all the new renters? They were families that were losing their homes through foreclosures and short sales or who were freshly locked out of mortgage access.

But, weren’t we supposed to be crawling out from a glut of homes? In 2011, Fed chair Ben Bernanke told Congress that economic recovery was still slow because we were still dealing with the oversupply of homes that was keeping housing starts and construction employment from recovering. And, this was considered a completely reasonable thing to say. I don’t know if anyone ever even pushed against him on that comment. Renowned housing economist Ed Leamer wrote that there were a million too many construction employees in 2015.

Four million additional young adults were living at home by then.

What a cultural change that must have been, if we had a glut of homes! So many that we couldn’t even find anything useful for former construction workers to do. And even with this sea of unwanted homes drowning us, American young adults suddenly said, “No. I’m just fine here.”?

It may be true that, even though the conventional view of leading economists are indefensible in hindsight, there is little Bernanke could have done to truly address the problem. More monetary stimulus would have mostly been inflationary because the CFPB and FHFA weren’t going to let enough households get mortgages to kickstart the single-family market. In fact, the only persistent source of inflation in the 2010s was rent inflation. Rents were already going up without monetary stimulus because the sudden loss of millions of homeowners meant that rent inflation had to accumulate to get home prices back up to the level that could lead to more construction again.

Apartments could have provided those young adults with new housing, but cities across the country were playing “Red Rover, Red Rover” on apartment approvals. While the single-family construction market is still, to this day, 20 years past the 2006 peak, at recessionary production levels, starts on new apartments had bumped back into the locked arms of America’s municipalities by 2013. The apartments those young adults would have moved into weren’t going to come.

Whenever you hear a room full of conventionals talking about the glut of homes after 2006, add the Urban Institute chart above to the endless list of charts that are screaming into the void.

Subscribe now

I have some reactions to this offer and some back of the envelope chicken scratches to offer about it, for subscribers - mostly as they relate to my expectations about the homebuilding industry and investment potential of some specific builders.

Subscribe now

Read more

Subscribe now

It’s depressing that this is such a cherished assertion on the right, because the solutions to the problem of housing affordability generally involve deregulating supply. These are pundits and policymakers that would naturally be on the correct side of that argument. And, instead, they engage in these ridiculous mis-readings of the data, they entertain fallacious narratives about how the Federal Reserve affects mortgage rates and home prices, and they treat every originated mortgage to a borrower with less than an 820 credit score as some sort of government subsidy. So, on the most important long-term problem that free markets can fix, its defenders are fighting for it like Everett Ulysses McGill.

Of course, at the other end of the political spectrum, there is a common ailment to be congenitally opposed to markets. The lack of markets in housing has very regressive economic effects - it makes poor families poorer. But some argue that the housing affordability problem is caused by poverty, being unable to see that the poverty is caused by the lack of markets.

The growing YIMBY movement and Abundance movement have to build from the middle outward since the ends of the political horseshoe are so incapable of understanding and addressing this issue.

One can understand how right-wing pundits have this reaction against the left-wing claims, because the left-wing claims are off-base. The problem isn’t poverty making housing less affordable. The problem is that housing is causing poverty. So, I can see how there is a motivation to push back and to say that, no, actually, we don’t have a poverty problem. That reaction is correct. But, until they stop being confused about housing, they are going to twist that correct reaction into an unhelpful intransigence.

I don’t think these outdated assertions about the state of housing affordability are going to go away, but, maybe, just maybe, the starkness of the reality, which I have tried to lay out in the charts below, can help remove the scales from the eyes some members of the audience of the pundits with the outdated assertions - to help move them to the middle where YIMBY and Abundance reforms eschew the errors of the ideologues.

There are a few intermingling claims that are repeatedly asserted - households have been getting smaller, homes have been getting larger, and so if households are having trouble affording housing, they could just live in homes like the ones their parents and grandparents lived in and stop their bellyaching.

Invariably these claims will cite housing conditions in 1970, or thereabouts. The reason they cite dates that old is because the story they are telling hasn’t been true for 30 years.

On to the pictures. But, first, one caveat. I am using the average square footage data for all types of homes from the Census Bureau’s Quarterly Completions by Purpose and Design. This is only for new homes. Sometimes when analysts use this data, I complain about that because in any given year new homes are only a very small portion of the stock of housing. And, the shortage specifically causes the cost of older homes to increase the most, so, using new home price data is sometimes mis-used to claim that there isn’t a housing cost problem. While the unchanging existing homes inflate in place, homebuilders make a lot of decisions about where to build, what to build, etc., to keep prices of new homes lower. One obvious issue is that the cities that have become the most expensive - NYC, Los Angeles, etc. - are expensive specifically because home builders can’t build there, so the most expensive cities basically don’t show up in the new home price data.

But, that’s the best data we have on home size, so it’s what I have to use here, and I don’t think it causes any biases favorable to my point.

On all these figures, the x-axis will be the aggregate value of all residential real estate as a multiple of aggregate US personal income. Household size and home price/income data is available from 1947 to the present. Square footage data is available from 1974 to the present.

The y-axis in Figure 1 is the average members per household. In the 20th century, the size of the average household declined sharply. That required building a lot more homes per capita. And while we were doing that, the aggregate price of homes remained flat - between 1.4x and 1.8x aggregate personal income.

In the 21st century, household size stabilized and home prices jumped from 1.6x to 2.5x aggregate personal income.

In Figure 2, the y-axis is average square footage of new homes. From 1974 to 2000, that jumped sharply from 1,400 square feet to 2,000 square feet while home prices remained between 1.4x and 1.8x personal income. Today, the average size of new homes is less than 2,000 square feet and price/income is above 2.2x.

Finally, the y-axis in Figure 3 is square feet per household member. From 1974 to 2000, that increased sharply from 467 square feet to 772 square feet. Aggregate home values ranged from 1.4x to 1.8x aggregate personal income. Today, the average square feet of new homes per household member is 780, roughly the same as in 2000.

During the 20th century, new homes got much larger with no strong effect on price/income. In the 21st century, homes have become much more expensive in a way that has absolutely nothing to do with how large they are.

In some way, surely, building smaller homes should make them more affordable. These false assertions have such staying power because it seems like that statement is a truism. But, homes are necessarily paired with land, and the reason we have a housing crisis is because obstructions to building and funding housing have been so binding, that the premium families must pay for the land under the homes has been outpacing our ability to shrink them through changing the composition of new homes. What seems like a truism is actually very much not true. It is unrelated to our problem. And the assertions based on it are so wrong that it would be true to call them lies.

Subscribe now

Subscribe now

The problem with those charts is that the decision to seek meaning from those charts reflects a lack of understanding about our housing crisis. It reflects the problem of not even knowing what question to ask.

There are at least 4 margins through which “Mortgage Affordability” changes:

Interest Rates

Obviously, mortgage rates are important in a practical sense for leveraged buyers and especially for leveraged buyers that can’t qualify for a mortgage large enough to purchase the home they would choose. This can sometimes be the case because American mortgage regulations and programs have long been influenced by a sort of conservatism where front-loading cash outflows (which the 30 year fixed rate mortgage does) to make mortgaged purchases more financially stressful in the early years is meant to keep buyers from overconsuming to protect them from themselves.

That is especially the case in high-inflation scenarios, so that “mortgage affordability” measures look really bad in the 1970s and early 1980s. There are a lot of problems with using the measure this way. A 30-year mortgage has 360 payments, so using a measure of only the first payment when that payment is going to decline sharply in real terms over those 30 years isn’t very informative.

And, it also happens to be the case that mortgage rates just don’t have much important effect on home prices in the long-term. If they did, “mortgage affordability” charts would be a relatively flat line instead of ranging from 20% to 50% depending on what mortgage rates are.

My figures here don’t really address that issue. I will take “mortgage affordability” at face value to make some other points.

Geography

My figures here will also not address this problem. There wasn’t much geographical difference in home prices until the late 20th century. There were some dense locations where, say, unit size might be traded off against location, or where families in regions with amenities like reliable mass transit might spend more for housing because the home included access to amenities that kept other expenses lower.

But, by 2008, the Closed Access cities included neighborhoods that commonly contained homes worth more than 10 times their residents’ incomes (versus a traditional 3x multiple). They are still in that range today, but they aren’t such outliers any more, because many other cities now have neighborhoods where home prices have risen to double the 20th century 3x norm.

Under shortage conditions, high housing costs are highly regressive. In Figure 1, I compare the typical mortgage affordability over time for families with 3 different income levels (based on 2026 incomes). In the 20th century, this chart was basically just a chart of mortgage rates. By the 2000s, the national number was not really informative. There were many cities that basically looked like the red line for all households. The blue line (mortgage affordability for poorer households) was mostly being driven higher by the extremely high costs of homes in the Closed Access cities where few families with lower incomes could newly purchase homes any more.

Today, there is still some geographic variance, but now many places are expensive, so the elevated mortgage expenses are reflective of a national problem.

Incomes

In Figure 1, I compare typical mortgage affordability at 3 different income levels. In the 20th century, this margin didn’t matter much. Homes in all neighborhoods across all incomes sold for similar multiples to their residents’ incomes.

This is the housing crisis. The problem is that mortgage affordability for a family making $50,000 is nearly twice what it is for a family making $250,000. This is a new thing. The scale of it is massive. And it is increasingly the case everywhere. And there is only one conclusion that the average “mortgage affordability” measure represented by the black line points to: the person that made the chart doesn’t have a clue what the problem is.

For families with the highest incomes, housing for the past 20 years has been more affordable than ever before.

Rent

The final margin I will discuss here is rent versus owning. The main source of our affordability problem is that the extreme tightening of mortgage access after 2007 turned us into a nation of housing haves and have-nots. The lack of affordability is a side effect of that problem. It is why the income-correlated costs have risen so much. And the high costs are from inflated rents. The haves, who can get mortgage funding, have been able to buy homes at a discount because excluding the have-nots pushed prices down. That’s why rents increased: for years, homes were too cheap, so builders couldn’t profitably build more.

In Figure 2, the dashed line is what mortgage affordability would be for the average family with $100,000 income if homes didn’t have a discount from the mortgage crackdown.

Think of the dashed line this way: it’s the price that renters households would be willing to pay to own their homes, if they had access to mortgage capital, based on price multiples that were common in the 20th century.

In Figure 3, I compare mortgage affordability across incomes, and I include the undiscounted affordability measure for families with $50,000 incomes - what the typical renter with $50,000 income should be willing to pay in mortgage payments to own their home.

Families with $250,000 incomes can generally qualify for mortgages, so there is no discount in their neighborhoods. Mortgage affordability in those neighborhoods was extremely easy when mortgage rates were low. Now, mortgage affordability in those neighborhoods is right at the level it had been at for years before 2008. It’s easy to see how pundits who live in those neighborhoods might not have a clue what the problem is.

In those neighborhoods, the typical starting payment for a new home takes about 30% of the buyer’s income. In the 20th century, you could have said the same thing about the typical family with a $50,000 income.

But, today, the typical family with a $50,000 income has to rent because we won’t let them buy a home. And the mortgage payment that would equate to their rent expenses would take 60% of their income. Twice what it should take.

And the reason that number is so high is that the supply of new homes has been so lacking that housing costs have been rising on many families faster than their own incomes have been rising. And that puts them in the position of making hard decisions. Move out of the entire region? Move their kids into a worse school or a neighborhood with higher crime? And faced with those sorts of decisions, they choose financial distress.

The distribution of costs is the crisis. The average expense of families who are allowed to take out mortgages isn’t even feeling a part of the proverbial elephant.

Subscribe now

Subscribe now

But, also, I think there are some subtle justifications behind Marohn’s assertions. I think this is one of those 20th century vs. 21st century issues where a century of reasonable and accurate expectations about how the housing market operates just don’t work any more. Marohn pushes back against treating Austin canonically (in his words) like a supply success story, but his caveats amount to obfuscations. Yet, that is only true because we are in this odd, temporary, supply-shortage context where everything is turned upside-down. In the 20th century, his caveats would have been canonical - and rightfully so.

And, in order to get to those points, I should lay out the Erdmann Housing Tracker best guesses about what is driving housing trends today and how we got here.

• Before 2008, regressive rent inflation in housing-poor cities like Los Angeles led to regressive price inflation and outmigration.

• Before 2008, the scale of that outmigration became large enough to create housing bubbles in a number of markets like Florida, Arizona, and Nevada, and raised home prices temporarily and across the board in those markets.

• After 2008 a permanent and deep crackdown on mortgage access led to the collapse of home prices in neighborhoods with lower incomes. This reversed some of the rent-related price inflation in cities like Los Angeles and created price deflation in neighborhoods with lower incomes in all other cities across the country that was novel - not a reversal of any previous inflation.

• Even in 2014, new home sales were still about 1/3 of their pre-2008 peak. By this time, the effects of the mortgage shock had largely played out, but construction had been so deeply wounded that capacity to build new homes had been permanently lost, so that the recovery of demand for new homes after 2014 has increased faster than the construction industry’s capacity to meet it, and prices have started to inflate across the country in the same way that they inflated in cities like Los Angeles before 2008 - regressive rent inflation leading to regressive price inflation. Now, however, there are no major cities for families to move to where new housing can be built without leading to rent inflation.

• Since mortgage access was the main new obstruction to new home sales, the collapse in capacity was largely driven by the collapse in single-family construction. Multi-family construction recovered relatively soon after 2008 but hit a cap at around 400,000 units nationwide annually that was the result of universal zoning obstructions.

• The post-Covid years have been facially volatile, with temporary regional migration spikes, new supply interruptions, etc., but none of these temporary noises in investment and consumption patterns have the permanent importance of the previous bullet points.

• We are now in an odd context where building a home in one location must, generally, come at the expense of a home in a different location, and the construction market clears through the relative inflation of land in different locations and construction inputs. A city can increase its own capacity for growth with zoning reform, but total national capacity is still growing at a limited pace.

Marohn’s Comments on Austin

Marohn summarizes his observations from the blog post about Austin in this tweet.

In the post, Marohn writes:

Austin did build a lot of housing. Rents did decline. What struck me was not what the report said. It was what the report didn’t seem interested in asking.

Because the Austin story is strange, especially if you back up far enough to see all of it…

If I begin the Austin story in late 2021, I can tell a story about rents falling. If I begin in 2015, I can tell a story about rents rising dramatically before they fell, despite the construction of an “enormous number of homes.”

This is so wrong it’s weird. He’s using volatility associated with Covid to complicate the long-term story. Figure 1 shows Zillow’s typical rent estimate for the US, Austin, Raleigh, and Nashville.

There was a little bump in rents in Austin in 2021 when it took on a spike in population growth. But, over the entire period since 2015, it is an extreme outlier on rents. It more than made up for that bump.

Before Covid, rents were rising similarly across the country. This relates to my points above. Housing everywhere was growing too slowly to meet neutral demand for household formation. I think rent inflation was relatively uniform across the country then because where demand was higher, it led to more land inflation locally and more input inflation generally until construction rates settled at similar rent trends in all markets.

But, you don’t need to go down my rabbit hole of explanations to think that Marohn is tossing dust in the air here. Even before Covid, rent inflation was relatively lower in Austin, and it was the fastest growing metro area in the country when that happened. By. A. Long. Shot.

Does that bit of up and down noise after Covid, followed by a downtrend in rents that itself makes Austin an outlier (By. A. Long. Shot.) somehow cloud the supply story here? How?

But, back to my rabbit hole. The reason construction trends and rents were relatively uniform across the country in the 2010s was because the mortgage crackdown lowered the prices of existing homes and suppressed demand from single-family homebuyers similarly across the country.

In this weird period, supply across cities was proportional to demand, but was short. In the 20th century, supply was proportional to demand, and supply was added until rents moderated and the price of land with marginal locational value was reduced to marginal value. Marohn’s intuition about declining rents leading to reduced supply applied in that context. It was true that supply was self-moderating. But, that was only because rents were moderate, and the price of marginal land was never elevated.

Today, there are cities where a lot with no particular amenity value, but simply with access to utilities and permission to build a home, has a value in the six figures before a shovel touches the ground. In this context, when supply is available, more supply will continue even as rents fall.

Now, as I mentioned, I think that the national supply conditions are slowing down that process. If a city’s rents decline enough that builders are unwilling to bid up the price of construction inputs, those inputs will be drawn to other cities. Rents should generally, slowly decline together across cities for the same reason that they became inflated together after 2008.

But, this is why, under current conditions, the results in Austin are especially striking. Rents are going down much faster than anywhere else, and the city is growing faster than anywhere else. It is attracting capital to its housing market.

How?

Because it is also an outlier in terms of zoning reforms. Figure 2 shows the rate of single-family permits per capita (x 1,000) across metro areas since 2015 on the x-axis and the total rate of permits per capita (x 1,000) on the y-axis. For almost every other city, the difference in quantity of housing has been largely the difference in single-family homes. It has been the equilibrium between inflated land, input prices, and mortgage access. Austin’s built different. Austin built more because they approved much more multi-family. Multi-family demand isn’t suppressed by the mortgage crackdown. So, where multi-family units are legal, they attract capital from other markets and bring down rents.

Under current conditions, they have been claiming supply from other regions in the only way that a city can - by allowing multi-family construction that is legally bound up everywhere else. And the results are clear. There’s Austin and there’s everyone else. It is canonically a supply success story. What are we even doing here?

Finally, Figure 3 compares multi-family permits per capita. Again I am using Nashville and Raleigh as comparisons. They are also outliers. Raleigh is the second-fastest growing major metro area behind Austin since 2014, but it’s not particularly close. And Nashville is also among the top growers.

All three have grown more with below average rent inflation because all three have been winners on multi-family zoning reform. But, even among the outliers, Austin is an outlier.

Later in the tweet thread, Marohn wrote “As rents soften, permits collapse. Builders slow down building.”:

He’s building on that 20th century intuition. When rents and land values aren’t elevated, builders won’t keep building as rents fall. That was true in the 20th century. It only seems true, in general, today because of national supply constraints.

As you can see in Figure 3, Austin building slowed all the way down to where it is a bit less of an outlier among the outliers. In 2025, Austin still permitted more new homes than any other major metro area except for Raleigh. It permitted about 11 new units per 1,000 residents in 2025 compared to the US average of about 4. After cumulative double digit relative rent deflation over several years. You might not be able to discern that from Marohn’s chart in the tweet.

What exactly is the complication here? Marohn is trying to get you to drive through the fog with your high-beams on. Low visibility is a choice.

Subscribe now

Subscribe now

FRED only has housing inventory data since 2000 because the Census data has a break there, but the Census publishes that estimate back to at least 1965. So, I downloaded this data and smoothed out the break to see how this ratio looks over the longer-term. (I also expanded the count of adults to include all Americans 20 years old or older.)

Here is what the ratio looks like since 1965. Marriage peaked in 1970, so there is a little bit of a natural downward trend in this ratio over time, because under any context some unmarried adults have roommates or live with their parents. So, there will be less than 1 home per unmarried adult. There are more unmarried adults today than there were in 1970, so this ratio naturally will be farther below 1 than it was in 1970. But, clearly, there was a big drop after 2008.

In a regression of the stock of homes against the number of married and unmarried adults (with the y-intercept forced to zero), from 1965 to 2007, there were approximately 96 homes per 100 married couples and 83 homes per unmarried adult. The r-squared of that regression is ~1.0.

Figure 3 shows the number of housing units annually since 1965 and the number of housing units we should expect to have, based on the number of married and unmarried adults.

Figure 4 shows the difference between the actual number of homes and the number of homes we should expect to have based on the pre-2008 consumption of housing by married and unmarried adults.

Based on this measure, there are about 9 million fewer homes than there would have been in the typical year before 2008, given the current adult population.

I ran a similar regression with adult couples and unmarried adults from 1965 to 2007 to estimate the expected number of rented homes, owned homes, and vacant homes.

Figure 5 shows the number of rented homes versus the number of rented homes we should expect based on pre-2008 consumption patterns.

Figure 6 shows the number of owned homes versus the number we should expect based on pre-2008 consumption patterns.

There are about 3 million more rented homes than expected and 6 million fewer owned homes than expected. But, keep in mind, the shock that led to the national shortage was the mortgage crackdown in 2008. That caused the number of homeowners to flatline for a decade. In Figure 6, that shock amounts to about 10 million households who were owners in the 20th century and are renters today.

Accounting for that shock, there are about 7 million fewer rentals than expected and 4 million more owned homes than expected. How should we interpret that? I think one way to think about it is that some of the decline in homeownership after 2008 was from foreclosures and short sales on homes deep underwater whose owners could still qualify for mortgages, but were temporarily pushed out of the market. That would mean that the decline in homeownership was temporarily larger than the permanent shock to mortgage access would create.

In the other direction, there are likely some number of homeowners who purchased homes before 2008 and managed to hold onto them, who would not qualify for mortgages to purchase their first home under current standards. That would mean that the total decline in homeowners is still smaller than the permanent shock to mortgage access would create.

I’m not sure exactly what the scale of those factors is, but the net shortage of occupied homes is certainly in rentals. In other words, accounting for the effects of the mortgage shock would move the black line in Figure 5 above the red line and the black line in Figure 6 below the red line.

Figure 7 shows the expected number of vacancies. Based on the number of adult couples and unmarried adults, we should expect about 6 million more vacant homes than there currently are.

Again, that points to more rented homes. Rental homes are much more likely to be vacant than owned homes are.

This estimate is lower than my favored estimates. I normally point to a shortage of 15 to 20 million units. Keep in mind, this is based on stable housing consumption patterns since 1965. As American families get richer over time, we demand more housing. For instance, here in Arizona, there are thousands of retired middle class couples that own a home in the Midwest, the Pacific Northwest, or Canada, and who winter here in modest mobile homes. These aren’t billionaires. They are regular families engaged in additional consumption of housing that was less likely a few generations ago. So, the fixed estimate of homes per adult likely biases the expected number lower over time.

Subscribe now

My higher estimates for the shortage are benchmarked to demand that rises over time as it did in the past. I know that the shortage makes everyone cranky, so that every pied à terre, AirBnB, resident who hasn’t filed the proper paperwork, or empty bedroom is supposed to be a moral stain on our souls. The reason you feel that way is because we have a shortage. Benchmarking to demand for units that doesn’t grow with our incomes is benchmarking to shortage conditions. I’m not going to do that even if the housing reactionaries can’t understand what the shortage has done to their souls.

You might be tempted to argue that it is inappropriate to refer to any deviation from a comfortable quantity of units as a shortage. That’s normally a reasonable point. But, the housing shortage doesn’t play out like shortages in other goods and services might. It plays out through distribution. It plays out through regressive rent and price inflation. So, when total consumption is below the expected trend, it is systematically associated with both a decline in real consumption among many households for whom that decline is relatively painless (e.g. a young adult staying with her parents for a couple extra years) and a rise in nominal expenses where it really hurts (e.g. the young working class family paying rents much higher than they should have to pay to stuff themselves into an outdated and small apartment because regional displacement is the alternative). This is all a package. It all plays out as part of one single process. Filtering. Vacancy chains. Etc. It might seem like we can point to certain homes and say, “We could do without those homes.” But, it doesn’t work that way. Every home is part of a housing ecosystem. The existence or lack of any given home affects every family. The demand that would normally be expressed as seasonally vacant homes, etc. grows as we get richer. And, when it doesn’t get met with supply, the rents for that working class family get pushed higher. To benchmark to anything else would be a fib. Call it a shortage. You have my permission.

Details below the paywall.

Subscribe now

Read more

Recent temporary trends and noise in inflation measures have made very short term inferences difficult. I’d say that’s still where we are, with continued general sideways movement of home prices and construction trends, in general.

Figure 1 compares the CPI estimate of core inflation excluding shelter (inflation without food, energy, or rent) to a monthly rent inflation estimate using Zillow’s ZORI rent estimate.

June CPI estimates took a big dip, so relative to CPI estimates, rents have taken a tick up this month. This is a correction I have been speculatively expecting. It sort of happened all at once this month, which isn’t uncommon in CPI data, as Figure 1 makes clear. We might see more of this, more gradually, going forward.

More below.

Subscribe now

Read more

Subscribe now

The Fed authors note, “Measuring homeownership by the person instead of by the home is particularly important for comparing characteristics of homeowners: by age, by geography, and across time.”

The authors write:

Why would the difference between the two homeownership measures be so closely tied to housing costs? The most obvious explanation is that in places where housing is cheaper, more adults are buying their own homes. Of course, monthly expenses are only one factor people consider when they are deciding how—and with whom—they’d like to live. Housing situations might change when people get married. Some adults might want to take care of their aging parents or give their adult children a more affordable place to live. Some homeowners might prefer living with other people, leading them to rent out a spare room instead of using it as a guest room or an office.

Personal preferences influence some of these decisions. But economic conditions, such as the price of housing, are another influence. For example, the more expensive it is to rent or own a home, the more appealing it is to split those costs with others. As higher housing costs motivate more adults to live together, the difference between the owner-occupancy rate and the HPOP grows.

And, there is a surprisingly strong relationship between high housing costs and the gap between the traditional homeownership rate and the HPOP. Here is their chart.

They consider both spouses in married couples to be homeowners, which is a reasonable way to compute the ratio, but it does mean that, all else equal, if marriage rates are declining over time, HPOP will also have a tendency to decline even under stable housing conditions, as unmarried adults are less likely to own homes than married adults are.

One thing to keep in mind, also, is that recent distortions in the American housing market have affected housing supply and housing costs much more than homeownership. Figure 3 from their paper is an important reminder that homeownership is not the place to focus to find the most consequential outcomes of the housing crisis.

It is still true, and will likely always be true, that under any circumstances, being 70 years old versus 30 or 40 years old will be a stronger indicator of whether you are a homeowner than housing supply conditions are. But, this doesn’t mean that there isn’t a crisis in housing supply. It mostly means that homeownership isn’t where the crisis creates the most brightly blaring red flags.

Imagine a set of 100 homes and 100 residents (to keep it simple), with 65 owners and 35 renters. If you take away 10 homes, and there are now 62 owners, 28 renters, and 10 adult roommates of on kind or another, the homeownership rate would be a bit higher, the homeowner-to-population ratio would be a bit lower. That is basically the scale of our current condition. The change in those ratios would be a symptom of the lost supply of homes. But, the more important consequences would be the 10 missing households, and the elevated rents that fall most heavily on households that cannot easily merge or delay formation. The regressive accumulation of rising costs, in proportion to each household’s lack of choices, is the more important factor to understand regarding the scale of the problem.

That being said, this new data should be a helpful addition to discussions on this topic.

Subscribe now

Subscribe now

Read more