Friday, June 30, 2017

Housing: Part 238 - Home Price Changes over time

I've been meaning to try this for some time, and finally got around to it.  This is a motion scatterplot of over 10,000 zip codes from 2000 to 2017, showing how home prices changed through the various stages of the boom and bust.  The measure on the x-axis is median zip code home price.  The measure on the y-axis is the 6 month change in zip code median home price (not annualized, continuously compounded).  I have highlighted LA, New York City, and Phoenix.  (Data from Zillow.)

Closed Access Price Appreciation:

Until late 2003, prices within each MSA tended to rise as a group, and prices in the more expensive MSAs were rising more.  This is the fundamental Closed Access problem.  Expensive cities were becoming more expensive.

Subprime Boom:

Around the end of 2003, we start to see the transition away from the GSEs to the private mortgage securitizations (subprime and Alt. A).  This shift was stronger on the west coast, and we can see here how there is a separation between LA and New York.  Note that initially, LA prices increased as a group.  There still was little difference in top and bottom tier housing markets.  Also, there was little impact in Phoenix at this time.  And, we can see that the effect of the shift to private securitizations was to increase prices in the high cost cities.  The trendline for the aggregate national plot becomes very steep during this phase.  This is because the looser terms of the private securitizations were facilitating home purchases in Closed Access cities by households with high incomes.

Rate Hikes & Subprime Boom:

Fed rate hikes began soon after the private securitization boom began, in June 2004.  As the Fed Funds rate increased, the main effect was a decline in top tier price appreciation in LA.  So, during this period, from mid 2004 to late 2005, there was a big difference between low tier price appreciation and high tier price appreciation in some cities, but in LA, this is because all home prices were rising sharply in 2004, and then high tier price appreciation retreated after the Fed began to raise rates.  It was after the retreat of high tier price appreciation in LA that prices in Phoenix finally shot up.

Now, maybe this is a just-so story.  But, I think this is a case where thinking of Fed policy in terms of interest rates is problematic.  I think we are seeing three separate effects here.  First, as rates continued to rise, migration out of Closed Access cities was very high - for both renters and owners.  NGDP growth was starting to moderate.  Homeownership had peaked and was starting to decline.  And, sentiment was starting to turn south in the housing market, partly because of these factors.  This was leading to tactical selling and outmigration of homeowners.  At the peak, which was during this time, about 2% of Closed Access homeowners were moving out of the Closed Access cities annually, net of in-migration.  That is a significant amount of selling pressure, and I think this is a major factor in the downshift of price appreciation during this time.  And, a lot of that pressure was from households with significant equity positions who had owned their homes for a long time.  It appears that there were a lot of families whose home values were far out of scale with their general income and wealth, and they captured the capital gains as the boom peaked.  These appear to have generally been households that aren't particularly leveraged.  So these sales weren't particularly interest rate sensitive.  They were more sensitive to sentiment and expectations.

Second, in the low tier markets in LA, the private securitization market was still the dominant factor.  It was still funding starter homes in Closed Access cities, although by 2005, the flow of first time homebuyers was starting to decline.  But, it appears that aspirational buyers were still entering the housing market at this time in LA.  Again, this was probably not activity that was particularly rate sensitive.  These loans tend to have higher rates than conventional loans.  And, the Fed Funds rate doesn't necessarily have that strong of an influence on long term mortgage rates.  It definitely didn't during this period.  This lending channel remained strong because it wasn't that sensitive to the Fed Funds rate.

Third, the combination of tactical sellers and priced-out renters both led to a massive outflow of population into cities like Phoenix, and that is the main factor behind Phoenix's late price surge.  Certainly this price surge was also facilitated by the loose terms of the private securitization boom, both for first time buyers and for investors.  But, I think migration explains the timing of these events - why the Phoenix surge was so late in the boom and why it occurred after LA price appreciation had already started to wane.  This is also not related directly to interest rates, which is made clear by the fact that the entire boom in Phoenix happened after the Fed began to raise rates.

So, I don't think interest rates, per se, have much to do with these trends.  Money supply and expectations seem more important.

Inverted Yield Curve & Subprime Boom:

When the yield curve inverted in late 2005, which is an important signal of financial dislocation and coming economic contraction, we see an immediate and sharp reaction in all markets.  Really, this should have been the extent of the contraction.  By the time the CDO panic, there had been a significant amount of monetary tightening.  The yield curve had been inverted for nearly two years with predictable results - declining NGDP growth, declining housing starts and investments, moderating or falling prices, the initial drop in employment growth.

By the end of this period, home prices were beginning to decline, but this decline was led by the top end, both locally and nationally.  Within MSAs, it was high tier markets that tended to fall into declining territory first.  And, nationally, we can see the trendline start to fall below 0%, and it has a negative slope when it does.

CDO Panic

By the time of the CDO panic around August 2007, the private securitization market was dead, and the other mortgage conduits didn't expand to take up the slack.  Since private securitizations had been facilitating entry into Closed Access housing markets, at a national level, the drop in demand was most felt in the expensive cities, so at a national scale, it was expensive markets that dropped the most.  The national trendline really goes negative.  But, within those cities, it was the low tier entry markets that had large numbers of recent new buyers with large mortgages who were vulnerable to default and who were now locked out of mortgage access.  Even though they were in entry markets, this was largely young families with high incomes.  We can infer this because the initial drop in homeownership during this time was among young families with high incomes.

GSE Conservatorship & Financial Crisis

Here we can see how these low tier markets continued to drop for months or years after the GSEs were taken over, while top tier markets stabilized.  We can see this in both axes.  First, we see the long tails down to the left that represent low tier markets that, even in 2009 were declining by 10% or 20% or more, every six months.  And, this is such an extreme drop, we can see those dots moving left over this time, as the median prices in those zip codes were decimated.

Buyers in these markets were locked out of mortgage markets by the tight lending standards of the GSEs.  This period and after was when the vast majority of defaults happened.  Really, by the CDO panic of late 2007, nominal home prices at the national level were in unusual negative territory.  The declines between then and September 2008 were gut wrenching and far outside of any modern experience.  And, after that happened housing markets at the low end continued to experience losses for years, that, by themselves, would have registered as generation defining events.  Any relative valuation gains during the boom had been reversed by now.  In most cities, like Phoenix, there had never been any unusual gains in the low tier compared to the high tier, but the low tier losses after 2008 are massive.  (edit: You can really see this by just watching the Phoenix zip codes over the entire period.  High tier markets tend to lead low tier markets, slightly, in both boom and bust, but there is never much difference between high and low tier markets during the boom.  There is an idea that the bust was just the rewinding of the boom.  But, in Phoenix, a true unwinding would have effected all zip codes equally.  This wasn't an unwinding.  This was a massive dislocation targeted at low tier markets.)


That wasn't enough for us, though, and in July 2010, we passed Dodd-Frank.  And, with its passage in this time lapse graph, we can see the market that was just finally starting to stabilize take another pause.  The declines by this time were not as sharp as they had been in late 2008 and 2009, but we can see low tier home prices stall with continued price declines for another couple of years after that.  There were significant valuation discounts in those markets by then.  There was absolutely no reason to restrict lending and demand in those markets.  The rate of first time home buyers had been very low for years by then.  If anything, there had been a deficit of lending at the margin.  And this is not a subtle point.  The lack of reasonable lending in low tier and entry markets had been extreme.  Finally, in 2012, prices ceased their decline. (edit: You can see this by focusing on the national trendline.  It was just starting to move back up to zero when Dodd-Frank was passed, and just as Dodd-Frank passed, the left end of the trendline moved back down, and didn't recover back to zero until 2012.)


  1. Way cool graphs. Waaay cool.

    Well, a single anecdote is more compelling than reams of data, and so I can confirm that when I lived in L.A. I knew two different people who sold properties in L.A. to reinvest in Phoenix, in the time frames under discussion.

    That means the extensive work Kevin Erdmann has done has been validated.

    Our Fed does not need data; they cite anecdotal labor shortages as a reason to raise rates. The data is not there to suggest wages are rising. The website shows restaurants hiring hostesses for $9 an hour in Indiana, where there is a 3.2% unemployment rate.

    So anecdotes are very important, especially if well told.

    If there was a well-crafted news story on some guy flipping houses, or one poor family that bought more house than they should have, then that is the story of the 2008 house-price decline.

    This raises a possibility, easier said than done: You find someone who bought a house who fits your profile of the homebuyer who bought in the boom. Build a telling anecdote around them to start your book, and then say. "And this was the typical homebuyer of 2006-7, in a closed access city."

    People remember and believe anecdotes.

    1. Thanks Ben. I have found a much better way to embed the graph now. Much cleaner.

      If you can get me in touch with your friends, please send me an e-mail. Maybe I can include those anecdotes in the book. You are absolutely correct.

  2. Ben talking--

    Well, those people left for Phoenix way back when, and then I left for Thailand. I will see if I can hunt them down.

    I suppose you could run ads on Craigslist for people who anecdotally fill the bill. I will mull this over. It should not be hard to find present-day buyers in closed access or other cities. Going back eight years is a trick.

    You live in Phoenix. I guess you could do a sort on Trulia etc for all people who bought a house in 2007-9. But that hardly gets you close.

    The graph actually looks like it is rotating.