Monday, December 28, 2015

Housing, A Series: Part 94 - The Housing ATM did not prolong the boom.

When I began my study of the housing market, I did not realize the central role of localized supply constraints.  So, as I revisit some of my early analysis, I realize that it can suffer as much from aggregation as the sorts of analysis that I am pushing back against does.

One aspect of the boom that I have given some credence is the idea that FICO scores and general borrower characteristics were overstated during the boom because marginal borrowers could tap their "Housing ATM's" to cover up the fact that they were not able to sustainably pay off their mortgages.  I don't think the evidence justifies this as a major effect, but it was at least plausible.

But, if we look at the housing market, as we should, as two markets - the supply constrained market and the open market - even this idea loses credibility.  Why?  Because home price increases were concentrated in a few cities.  For home buyers in most of the country, there wasn't an unusual level of home equity to draw on.  So, if the crisis was precipitated by the unsustainability of subprime loans, in 75% of the country, those borrowers should have been defaulting well before 2007.

Here is a Fred graph of annual home value changes.  This is the S&P/Case Shiller National Index.  I have also added the Federal Reserve's measure of aggregate real estate value.  This value includes both the increase of existing properties and the addition of new properties, so it tends to run slightly higher than the Case-Shiller measure.  But, looking at both, we can see that there is a very long term tendency for homes to gain about 5% per year, over the long run.  I have also included a measure of housing leverage (100 minus equity proportion), on the right hand scale.

The housing market in the early 1990s, in nominal terms, was the worst market since WW II.  In the late 1990s, prices increased at slightly above trend, though this was mostly mean reversion, and, in the aggregate, households appear to have deleveraged as a result.  From 2000 to 2005, there was significant price appreciation that wasn't associated with deleveraging.  This is the period associated with the "Housing ATM".  This graph probably understates the issue a bit, because in the 1970s and 1980s, when home prices had previously risen strongly, they had risen along with broader inflation.  In the 2000s, since price increases were due to housing supply constraints, and not monetary expansion, inflation outside of housing was low - generally below 2% - and the gains to homeowners were real capital gains.

If we imagine a typical household, with income of, say $40,000, and a $150,000 home, the typical year might provide them with a potential $7,500 capital gain.  A period of time with 10% housing price appreciation would provide them with a potential extra $7,500 gain.  This is certainly enough to provide noticeable cover for a household budget that is in the red.

 But, if we look at this by city (below I include the 20 largest MSAs), what we see is that the Closed Access cities had very high rates of price appreciation - generally running above 10% after the year 2000, and sometimes much higher.  By 2004 and 2005, 7 or 8 out of these 12 cities saw 20%+ annual price appreciation.  These cities really did provide significant gains to owners. *

Outside of these cities, though, the story was much different.  The Closed Access (and "Closing Access") cities represent just under a quarter of the US population.  Almost all of the remaining three quarters of the country had a different experience.  Here is a graph of the Open Cities and the Old Economy cities.  Together these represent about 1/6 of US population.  But, in this graph, I also included the results for the US, nationally.  I have not adjusted this to remove the Closed Access cities, so the US outside of those cities would have seen annual price increases somewhat below the aggregate average shown here.  These areas tended to run along the long term 5% trend.  Phoenix is the extreme outlier in this group.  But, even in Phoenix, the extreme price movements were very late in the boom, so that subprime buyers in Phoenix would not have been capturing very unusual capital gains in the 2000 to 2004 period with which to inflate the appearances of their credit-worthiness.

When we look closely at the disaggregated national data on home prices, clearly most subprime mortgages were made in areas without much of a "Housing ATM" - areas representing about 75% of the population and 85% of housing starts.  The creditworthiness of most subprime borrowers would not have been inflated by unusual capital gains.  And, if the crisis was caused by the inevitable default of millions of mortgages to unworthy borrowers, there is no reason why those defaults would have failed to show up until late 2007 and after.  In these cities, we generally see fairly normal price behavior that suddenly collapses in 2006 and 2007.

All of these cities saw home prices begin to fall in nominal terms between late 2005 and early 2008.  There is no reason why the many cities that never experienced unusually high price appreciation should have suddenly experienced price depreciation in 2007 when the rest of the economy was still strong.  Home prices and mortgage affordability in most of the country were not high in 2005, yet by mid 2006, mortgage affordability in the Open Access cities was falling, in concert.

If bad credit risks were the reason for the housing bust, then in the 3/4 of the country where there never was an unusual amount of home price appreciation, we should have been seeing higher defaults all along.  But, we didn't.  Defaults were very strongly correlated with the time of purchase.  Mortgages originated in 2006 and 2007 had much higher defaults than mortgages originated in other years, and the higher defaults happened across mortgage types.

Before I was in the habit of thinking in terms of the different cities, it seemed plausible that the falling default levels from 2000 to 2005 were a product of rising prices.  But, when we think in terms of cities, this is a strong counter-argument against the credit-driven theory of the bust.  Subprime delinquencies were falling throughout the expansion.  At most, we could say that the rising prices of the Closed Access areas were pulling delinquencies down somewhat.  But, even after making that adjustment, we are left with delinquencies that were unremarkable until the collapse began in 2006 and after.  In 2004 and 2005, there was a sharp transfer in mortgage originations from conventional mortgages to subprime mortgages.  Those mortgages performed well even though most of them never experienced significant capital gains.  But, in 2006 and 2007, suddenly new credit dried up, prices collapsed, and many recent home buyers defaulted together.

There is little reason to believe that, on the national scale, there was a way for rising home prices to completely hide a growing problem of unworthy home buyers.  Look back at that first Fred graph.  When we remove the demand-side explanations from our story of the housing boom, that extreme and sudden drop in home prices should cause our own jaws to drop with it.  We shouldn't have expected it.  We shouldn't have stood for it.  And, we shouldn't have demanded it.  But, we did.  When we believe the demand-side narrative, that drop seems inevitable - even reassuring.  But, if the demand-side narrative is wrong, then our monetary authorities absolutely had a duty to avoid it.  That is what they are there to do.  I don't see that as an indictment of them so much as an indictment of our insistence on centralized monetary control.  They did what we demanded of them, and it will probably be at least a generation before most observers really question what happened.

*  Keep in mind, even in cities where prices were rising sharply, they were rising because of sharply increasing rents and rising expected future rents.  Renters in those cities, who don't have a mortgage, are facing rising costs every year.  So, a marginal household that uses a subprime mortgage to buy a home may still be taking on the safer position compared to renting, even if they have to tap equity to make it sustainable.  If the renter expects to see rent inflation of 5% per year, then a position where they take on a mortgage that either manually or automatically contains a negative amortization of 2% or 3% per year is the safer position, compared to renting.  They have become speculators in the local housing market, which requires them to bet that rents will continue to rise (because that expectation is embedded in the price of their home), but this wasn't by choice.  This was foisted on them by the reality of their closed access local housing market.


  1. You may want to consider this chart, Kevin. The private lending that lead to the crash went bonkers in mid to late 2003. Massive lending took place compared to prior years. Lending DOUBLED in the crazy years:

  2. And also, Kevin, according to BAC, mispriced risk caused the bubbles in stocks. And the Fed did those. Since risk was mispriced in the housing bubble, it was also caused by the Fed. The Fed caused the bubble and the crash.

  3. Great post. The US needs a few more "hip" cities. Places people want to live. Let's move to Milwaukee and start talking about its architecture, cuisine, history and so forth. Erect a statue of Warren Spahn. Bring back the Braves. Re-shore manufacturing from expensive Japan, bring in tech. Yahoo!