Housing supply isn't constrained in Phoenix

Scott Sumner has a post over at econlog today about the mystery of low housing starts.

But what if supply is also constrained in Phoenix and Las Vegas?  I don’t have any good explanation for what that might be so, but the data strongly suggests that there is some sort of supply problem.  The high prices are back, but construction remains severely depressed......I have no idea why supply in these markets is so constrained.  I’ve read articles that make vague references to the cost of land and labor, but no real explanation of why things are so different from 2003.

As Scott frequently points out, bubbles are not nearly as numerous as they are made out to be.  It is important to remember that what happened in cities like Phoenix in 2005 was an extreme anomaly.  A combination of population flows and capital flows that briefly pushed both the demand for real shelter and the funding available for it high enough that the local short run supply curve became disruptively inelastic.

In most cities at most times, where demand hasn't pressed quantity demanded so far up the supply curve that it becomes inelastic, changing demand only has minor effects on price.  When markets are relatively normal and supply isn't extremely inelastic, changing demand mostly affects quantity.  Even in 2006 and 2007, housing starts underwent extreme fluctuations before prices followed them down.  So, the signal in Phoenix is a reasonable reflection of changing demand.  (There was another anomaly in housing, after the crisis, where severe limits to lending pushed prices down in many cities.  That, again, though, was an extreme anomaly.  As Scott shows in his post, slowly this anomaly has reversed, also.  So, Phoenix was subjected to two anomalous housing events.  An extreme upswing in prices - what you might call a bubble - followed by an extreme downswing in prices that was also far from a level that long term fundamentals would justify.)

Source

You can see this in population and migration data.  One effect of extremely tight lending that has prevented aspirational middle class homeownership since the crisis is that population and migration trends that had been steady since WW II into places like Phoenix have been sharply curtailed.  Migration briefly declined to nothing in Phoenix for a few years after the crisis, and has since recovered to a level where the difference between Phoenix population growth and US population growth is only about half what it was pre-crisis.

Phoenix builders could build thousands more homes each year without much rise in cost.  Actually, prices in the low tier existing housing stock in Phoenix probably need to rise a little bit more to make building profitable.  This is a sign that lending regulations are the key variable moderating demand.  Building rates are highly correlated with income and with home prices, both within and between metro areas.  Low tier demand is in retreat because it has been pressed into a landlord's market.  This shows up both as a decline in inter-metro migration and in a retreat of real housing consumption combined with high rent inflation in low tier and rental markets.  The FHFA and CFPB's have-nots are either stuck in place or in retreat.  This also shows up in American Housing Survey data that suggest household size has continued to slowly decline among homeowners but has reversed and started to climb for renters, since the crisis.

What about the issue of costs?  I suspect there is something to that.  Low interest rates do make land more expensive. (Back yards are much more rare in the new neighborhoods in Phoenix than they used to be.)  Regulations, etc. have all probably risen somewhat since the crisis.  So, low tier prices might need to rise even more than they would have needed to without those issues.  But these are marginal changes that can't be responsible for such universal and extreme shifts in building rates around the country.

Source

On the topic of labor costs, however, I think there is something interesting here.  Here is a graph of construction employment in Phoenix as a percentage of total employment.  This did briefly rise during the boom, and then collapsed to very low levels after the crisis.  This is especially striking if you believe, as I do, that Phoenix never had an oversupply of homes, only a sharp negative demand shock, which is clear in the population chart above.  (I think the 2010 dip is probably due to a data revision, which is likely due to population growth that was lower in 2008 and 2009 than shown.)

Economists such as Peter Boettke and Arnold Kling talk about the economy as a coordination problem.  I think this is a valuable and useful way to think about the economy.  The problem here is that this extreme disemployment in construction has been universally accepted as something that was necessary.  Economists have treated the collapse in construction employment as the necessary correction that had to take place, and blamed the slow recovery on the scale of that correction.  But, what if that wasn't a correction at all?  What if that was the disequilibrium.  Consider the scale of the damage we have imposed on our own economy through monetary and credit strangulation that we were able to permanently disassociate 3-4% of the Phoenix labor force from a reasonable and useful local industry.  This damage has been so diligently imposed that the dislocation remains in place a decade later and these workers have disappeared.  Builders complain of a labor shortage.  What happened to them?  Did they give up? Did they move away?  Did they have to go through a difficult (and unnecessary) transition to other work?

Hysteresis has been an idea sometimes used to explain the slow recovery.  Here's your hysteresis.  Even today, I commonly see people react to slow housing starts by asking, "Aren't we still working off the oversupply of the bubble?"  That is absurd.  It was absurd when Bernanke asserted it in 2011.  It was absurd in 2005, frankly, though one can certainly understand how conventional wisdom got it wrong at the time.  The fact that this idea is still floating around the zeitgeist in 2019 is a signal of how misguided conventional wisdom has been about housing.  The persistent unemployment of those workers was a policy goal shared by both populists and technocrats.

A nice review of "Shut Out" at CATO

David Henderson, one of the frequent posters over at econlog, who I have always enjoyed following, has a very nice review of "Shut Out" at Cato.  It begins:

In his recent book Shut Out, Kevin Erdmann, a finance expert and visiting fellow at the Mercatus Center at George Mason University, has two main messages. The first, which is not controversial among economists, is that restrictions on residential construction in coastal California and the urban Northeast have constrained supply so much that housing in those areas is virtually unaffordable for people in the lower- and middle-income classes. His other message is more controversial: the financial crisis last decade was not due to a housing bubble but, rather, to bad policy decisions based on the idea that there had been a bubble. Whereas I was already convinced of his first point, I, like the majority of economists, was skeptical of his second. But because of all the data and reasoning he brings to the issue, I now find myself at least 90% convinced.

Please click on the link (pdf) for the rest.

I need to get that darn second book finished to address the other 10%.

The Divergence in Incomes and in Resource Usage

Recently, I was listening to Russ Roberts at EconTalk interview Andrew McAfee.  The topic was the surprising change in trends in resource use.  It appears that as economies grow, at first resource use increases, but eventually economic growth comes from more efficient use of resources instead of through the brute force of added resources.  Surprisingly, the use of many resources has been declining for some time in the developed world.  Not just in per capita terms, but in total.  Now, getting richer seems to mean using less.

They mentioned that the divergence seemed to happen around 1970.  Here is a graph of real GDP growth, iron and steel, and cement use, all indexed to 1970, using data from McAfee's website.

Although I don't think they mentioned the parallel in the program, I immediately thought of this graph that is frequently cited in the income inequality debate.  The source of this graph has made it quite clear what they think caused the divergence.

It seems likely to me that these issues are linked.  As economic growth became decoupled from the Malthusian quest for more resources, it became associated with rising services and status competition.  There could be a number of things going on here.  First, if it is easier to meet basic physical needs, there may be less motivation to increase income above a certain threshold.  Also, the real economic value of services and status items may be more difficult to track because it isn't based on the blunt measure of a physical quantity of inputs.  Variable inflation rates may be more difficult to track.  Think of the difference in rent between San Francisco and Little Rock, or groceries at Whole Foods vs. Wal-Mart.  Or, the price of a last-minute business class airplane ticket vs. an economy ticket.  Or, the vast number of services created by the internet that are commonly provided for free.  Think of the cost of Bloomberg financial services vs. the huge amount of data sites like Zillow make available for free.  The value of things versus the price of things has become highly variable.

In any event, these developments seem certainly to be related, and the transition away from a resource based economy seems like a much more relevant trigger than President Reagan.  I suspect there is a combination of mismeasured well-being and variance in well-being that is largely played out in status seeking services.  Thus, measured inequality seems high even though most households can purchase basic goods at real costs that are far below what they were in 1970.

I wonder if those who give Reagan such an important role in relative measured income growth after 1980 would feel such a strong intuition about the first graph, and hail Reagan as the president who curtailed resource usage.

An interview with ALEC

I really appreciated being allowed to share my work with ALEC at their recent conference in Phoenix.  I saw a lot of great nuts and bolts activity going on there in the service of creating an equitable and economically vibrant nation.

Here is a short interview from the conference.

November 2019 CPI inflation

Inflation remains in a holding pattern - about 2.3% core inflation, which consists of 3.3% shelter inflation and 1.6% non-shelter core inflation.

The yield curve is, similarly, in a holding pattern.  How this plays out will depend on unanticipated real shocks over time and the future bias of the Federal Reserve.  My inclination is to continue to believe the short term outcome will mostly accrue to declining Treasury yields and the depth of that decline will greatly depend on the willingness of the Federal Reserve to support short term NGDP growth.  The slow moving train wreck (or not) continues.  As long as non-shelter core inflation remains below target and the yield curve remains effectively inverted, my expectations will be somewhat bearish.

Momentum in equities when reputational risks are high

This blog originally was supposed to mostly be about investing tactics, but I got sidetracked when I discovered the housing issue that has ended up taking over.

Here is a post on the topic I used to dwell on - tactical investing for high returns by being insensitive to reputational risk. A recent example of this issue is Hovnanian Enterprises, a major homebuilder that is still so down on its luck as a result of the housing bust that as recently as July, it was being threatened with delisting from the NYSE.

All along, they have mostly just needed the market in new housing to recover.  They need revenue to fully regrow back into the financial and organizational framework that had developed under the Hovnanian name in 2005.  They have so much organizational and financial leverage that small increases in revenue will translate into large increases in market capitalization.  This will further be enhanced by knock-on effects of recapturing the value of tax assets and written down developments, and paying or refinancing debt at more favorable terms.

After the July delisting notice, Hovnanian released results of two quarters which have provided strong evidence that revenues will be growing and these positive developments will be coming.  Normally, efficient markets would internalize these developments immediately, and a firm's share price would immediately jump to a level reflecting the new expectations.  Financial research has shown a momentum effect.  In other words, a trend in share prices doesn't happen 100% at once.  It mostly happens at once, but there does appear to be some predictable serial correlation.  A recent trend shift or a recent positive or negative shock to a share price will tend to continue in the short term, to a certain extent.

But, in unusual positions where reputational risk has become acute, this momentum effect can become very large.  I am sure there were some institutional holders who were forced by their own rules to unload shares when Hovnanian received the delisting notice.  At some point, the reputational danger of having owned a stock or of recommending a stock, can overwhelm the objective value of it.  In these cases, the market for that equity becomes very tepid.  It's sort of like very thirsty wildebeests coming upon an oasis.  They very understandably approach it carefully at first, not sure if it is safe.  But, almost inevitably, the whole herd will be lined up at the shore, sucking up water vigorously.  To someone who happens to have been at that oasis when the herd showed up and recognizes already that there are no crocodiles, this process can seem excruciatingly slow.

Here is the Hovnanian stock chart from the past 6 months.  The positive shocks are noticeable after each quarterly announcement, but even those shocks took place over several days.  Following the initial quarterly shock, there was further upward drift for some time.  The more recent positive shock also took several days to play out.  It will be interesting to see if a positive drift follows this shock also.  If revenues do climb from here, the share value is likely many times the current market value.  Getting from here to there will reflect a combination of objective results, expectations, and changing reputational risks.  It is the implicit position on reputational risks that can provide very high returns over time, but it comes with the occasional risk of losses, and those losses will necessarily be devastating and embarrassing.

Source

Great Review of Shut Out in the Economic Record

Declan Trott, an economist with Australia's Department of the Treasury, has written a very nice review of Shut Out for Economic Record, a journal of the Economic Society of Australia.  It offers a concise and well-written overview of the book's thesis.

Unfortunately, there is a pretty hefty paywall.

A couple of brief excerpts:

But what if this fall in prices were not the inevitable bursting of a bubble, but an unnecessary and self-inflicted panic? This is Kevin Erdmann's contention. . . This is a provocative thesis, not to be accepted lightly.  Yet Erdmann has assembled a formidable battery of data and argument to support it.

. . .

It is the detailed documentation of the (housing bust), and the treatment of the entire episode as a panic rather than a bubble, that is Shut Out's key contribution relative to the academic literature. 

. . .
And, I was flattered by his closing comment.

As a member of the PhD tribe, I occasionally found myself wishing for more equations and regression coefficients, and simpler charts.  But still, somebody should give him an honorary degree. 

That sounds just like my editors and internal reviewers.  Why does everyone hate complicated charts so much? Anyway, this review was a pleasant Thanksgiving surprise.

A postscript on the review of the crisis

Upon re-reading my summary of the housing bubble and financial crisis, I suspect some skeptical readers might find my summary lacking because it may seem as if I am writing off what was clearly a boom in residential investment and home building.  It may be worth a clarification.

There certainly was an increase in building from the mid-1990s to the mid-2000s.  But, in terms of either the number of homes per capita or the rate at which they were being built, nothing was outside of historical norms.  Residential investment seemed high, but part of what is accounted for as residential investment is brokers commissions, which don't really add to the housing stock.  Brokers commissions were high, however, because the shortage of urban housing made existing homes too expensive.  Subtracting commissions out of residential investment reveals a long term decline that was briefly interrupted with rates of residential investment similar to the 1970s.

There wasn't really a national building boom.  There was a moderate rise in building.  The reason it seemed so disruptive is that the Closed Access cities can't allow a sustainable amount of building.

That means that any time Americans try to increase our real consumption of housing at the same pace that our incomes are rising, a disruptive migration event must occur, because if the consumption of housing expands and some cities cannot expand their local stock of housing to accommodate it, those cities must depopulate.  That's what happened before the financial crisis.

Policymakers since then, whether they understand it or not, have been trying to avoid this disruption by either keeping incomes low (through tight monetary policy) or by reducing demand for housing (through tight lending standards).  That has reduced the migration out of the Closed Access cities, but it has come at the expense of living standards for Americans everywhere.

A review of the crisis narrative

Over at econlog, a commenter has asked me for a comprehensive review of the standard narrative and my objections to it.  His summary of the standard narrative is clear and concise, but thorough, and I thought it might make a nice template for posting a summary of my new narrative.

His description of the standard narrative is indented, and my responses are not.

A variety of factors (securitization introducing a principal-agent problem, organizational changes in banks/GSEs, regulatory encouragement, etc.) led to much looser standards for lending. This included:
No-documentation loans.
A growth in subprime lending.
Shrinking requirements on down payments.

These factors were all definitely at work.

This led both to an increase in for-occupancy home purchases by people who used to be renters, and in speculative home purchases (which were now easier to finance, and looked profitable as home prices were rising).

The private securitization boom, which is associated with all of these developments, lasted from roughly the end of 2003 to mid 2007.  Homeownership rates had been increasing since the mid-1990s, but they peaked near the beginning of that period, and then declined.  The relatively high level of homeownership was generally due to age demographics.  Homeownership rates for all working-age groups were about the same they had been in the early 1980s, at the high end of their recent ranges, but not unprecedented, and by the end of the subprime boom they were back in the middle of the long term ranges.  American Housing Survey data suggests that this was because, both, the rate of first time homebuyer activity declined, and an increasing number of existing owners sold out.

Of course, someone has to own every home, so this means that investor ownership increased.  In some volatile markets, some of that activity was speculative and ill-considered.  It probably hastened the early defaults in those markets because investors are more likely to default when equity becomes slightly negative than owner-occupiers are.  But, the investor activity was more of an effect of volatile markets than a cause of them.  Prices were nearly topped out by the end of 2005, and most speculative activity happened in 2006 and 2007.

In short, it is implausible to blame speculating investors for the rise in prices from 1997 to 2005 and it is implausible to blame rising homeownership from 1997 to 2004 on the loosening standards of the subprime boom.

So, what did cause rising prices?  In at least 2/3 of the country, prices weren't outside of historical norms relative to rental values.  They were slightly high, which can be explained with low long term real interest rates, but not unusually high.  In 5 primary cities [NYC, LA, Boston, San Francisco (+ San Jose), and San Diego] prices were high because rents were very high and were rising.  Fundamentals fully account for high prices in those cities, and this is more obvious with every passing year.  Their rents are high because they allow an astoundingly low quantity of building.  I call them the Closed Access cities.  Loose lending may have added demand to the buyer market in those cities beyond what was previously possible, but it was generally allowing borrowers with high incomes to buy in cities where rental expenses are also outside historical norms.  Households with lower incomes were flooding out of those cities at the time by the hundreds of thousands each year. 

A smaller set of regions had something more akin to a true bubble - prices that were likely to retract at some point in the natural course of things: Arizona, inland California, Florida, and Nevada.  I call them the Contagion cities.  They were the primary landing ground for the Closed Access outmigrants, and the primary cause of their brief positive spike in home prices was that they were generally overcome by in-migration.  The demand was for actual shelter.  Families were moving to these places, in droves, specifically to drastically lower their housing expenses.  Recently, I have been working on preliminary evidence that during the periods where prices were rising in the Contagion cities, there was no unusual rise in borrowing.  That happened after prices and rates of new building in these regions had peaked.  Borrowing at the state level tends to lag both the building booms and the price spikes.

It would be very difficult for these cities to overbuild because they natural have heavy in-migration, and at the time it was higher than normal.  In fact, at the metropolitan area level, in the 2003-2006 period, rates of building were especially correlated with population growth.  Population growth in Contagion cities suddenly collapsed when the migration event out of the Closed Access cities collapsed.  This was happening by the end of 2006.  By then, the Fed should have been trying to stabilize housing markets, not slow them down. Yet, even in late 2008, the main criticism they faced was that they weren't destabilizing housing and financial markets enough.

The shortage of homes in these cities is the fundamental cause of the housing bubble and ill-informed policy reactions to it caused the financial crisis.

This rise in home prices was not sustainable (80% increase in 6 years, much faster than inflation), and eventually slowed/ended, this happened concurrently with raises in the interest rate (and thus in the rates of adjustable mortgages)
This would not have created a crisis by itself (housing markets have had downturns in the past) except for the fact that many homeowners either:
Couldn’t afford their mortgages and could no longer refinance them using new equity from price appreciation.
Had “negative home equity” and lived in no-recourse states, making it cheaper to default than to keep paying their mortgages.

The Fed had inverted the yield curve by the beginning of 2006.  To the extent that monetary policy is communicated through interest rates, the peak of the housing boom coincides with them.  Adjustable rates have little to do with the default crisis.  Defaults were highly sensitive to cohort (how soon after you borrowed did prices begin to collapse).  2007 was the worst, followed by 2006.  The yield curve was inverted and the short term Fed Funds rate was at or near the 5.25% high point throughout the period when those mortgages were taken out.  Rising rates on adjustable mortgages have nothing to do with the default crisis.

Falling prices (negative equity) were by far the largest factor leading to defaults.  Lending standards were tightened sharply during 2007, so it is true that it was harder for borrowers to refinance.  The drop in homeownership in 2007-2008 was mostly among homeowners with high incomes in the "bubble" areas where prices were collapsing the most sharply.  Declining middle income and lower-middle income homeownership rates were a very lagging event, really not happening until after 2008, after lending standards had been sharply and permanently tightened, which caused a largely unacknowledged second housing collapse that was focused mainly on low tier neighborhoods, and which affected nearly every city in the country.  The bottom of prices around 2012 was not a return to normalcy, it was a self-inflicted collapse in credit constrained markets that were now locked out of mortgage access.

This led to a snowballing increase in delinquency rates which started prior to the crisis and lasted through the recession. It was also unique in that it happened in a correlated fashion across the country, unlike prior downturns which tended to be local.
This then impacted the financial sectors as many instruments built on securitized mortgages were discovered to be worthless, and entire companies went bankrupt.

The fact that it was correlated across the country is a solid signal of how wrong the standard narrative of its causes is.  Cities have huge differences in prices, rents, rates of building, vacancy, etc.  It is implausible that overbuilding or unsustainable prices could have done this. It was the result of national policy choices aimed at doing it, first by tight monetary policy that began to limit liquidity and change sentiment (leading to collapsing new building rates beginning in 2006) and continued to push markets into further disequilibrium as it remained too tight until the end of 2008.  By the end of 2008, lending standards had been tightened (average FICO scores of approved borrowers moved from around 710 to 750 over the course of 2008, a huge shift, which largely remains today). So, from the end of 2008 onward, high tier home prices stabilized but low tier prices had their worst declines after that.

A postscript.

Comments on the Quarterly Report on Household Debt and Credit (2019 Q3)

Here are a few updates on the data.

 First, mortgage originations by FICO score.  This continues to remain near levels it has been since 2009.  In fact, the average FICO score of borrowers started moving up in the second quarter of 2007, just before home prices started to collapse.  They basically hit the new plateau in the second quarter of 2009.  As I have shown, much of the devastating loss of equity in entry level homes happened after 2008.

This is the actual cause of the housing bust (and the financial crisis). The general collapse in home prices came after credit tightening, and the continued additional collapse focused on low tier housing came well after credit tightened, after it settled permanently at the new normal.  To this day, the consensus response to that claim is that it had to happen in order to bring credit standards back to normal.  But, borrower standards were normal.  The typical FICO score of borrowers in 2006 was the same as it had been in 1999.  The squeeze continues.

Total mortgages outstanding seems to be settled at about 3-5% annual growth.  And total number of mortgage accounts outstanding fell from about 98 million in 2008 to 81 million in 2013, where it remains.  That would be a bit laggardly in a fully recovered market, but it is very laggardly in a market with a severe shortage of housing and a rent expense problem.

Second chart shows the balance of debt of different types.  Good job America!  We have managed to push all that borrowing out of HELOCs and into credit cards, because the lesson we all learned from the financial crisis was that unsecured debt is preferable to secured debt. I read the terms on my credit cards and I can't help but shout "Stability! Prudence!"  We're so much wiser now.  Kudos everyone.

Remember, if you sell your house short because you're 30% underwater, that's really bad.  But, if you have to sell your house because you hit a rough patch and nobody will lend on more than 80% LTV and/or perfectly documentable income, that's just being reasonable.  If that happens to you, try being a little gracious about it.  It was for your own good, silly.

Third, debt outstanding by age (adjusted to per capita). Some analysis of the crisis sets it up as rich (savers) vs. poor (borrowers).  But, really, the only reason it looks like that is because the crisis was more a matter of old (savers) vs. young (borrowers).  Borrowing was moving up as much for the old as it was for the young, but older borrowers tend to be less leveraged. The older groups have increased their borrowing since the crisis.  That is because they didn't tend to own homes with high leverage during the boom, so they escaped the housing collapse with less damage.  And, that has allowed them to continue borrowing after the boom, because borrowing scales with wealth and income, to a certain extent.  The younger borrowers took a hit in the foreclosure crisis and are now catching up.

Unless there is a return to looser lending, though, it seems like there is a limit to how much catch up can happen.