Monday, January 11, 2016

Housing, A Series: Part 101 - The Effects of Supply vs. The Effects of Demand

Since there are two distinct housing markets in the US - Closed Access and Open Access - there were two distinct housing booms - the Price Boom and the Starts Boom.  Consumption can change because of changes in supply or demand.  Economic intuition should give us two distinct sets of expectations about a housing market with a negative supply constraint vs. a housing market with a positive demand shock, or expectations we might have about a market with both.

In terms of homeownership vs. renting, the trends across the country were similar.  Here is a graph of homeownership rates in selected states.  So, to the extent that a shift in demand was related to expanded lending to marginal homebuyers, it was a national phenomenon.  (The rise in homeownership rates may have been mostly simply a demographic phenomenon, unrelated to housing supply and demand - simply a transfer of tenants to owner-occupiers as part of normal life-cycle trends.  My preliminary estimate suggests that about half of the growth in homeownership was related to age-demographics.)

Among the largest cities, Phoenix, Dallas, Houston, and Atlanta are the best examples of housing markets with minimal supply frictions.  From 1995 to 2005, Phoenix and Atlanta issued about 16 housing permits per resident and Dallas and Houston issued about 10, compared to a national rate of less than 7.  So, if we want to check our intuition about the effects of a demand shift, these cities should be a good example.

I would expect economic intuition to tell us that more generous credit and low interest rates might lead to a positive shift in demand.  This could lead to higher home prices, but in an efficient housing market, a small price increase would lead to more homebuilding.  More building would increase supply, and rents of existing housing stock would decline as a result of the new supply.  Households would consume more housing at relatively lower rents.  Consumption would increase, but prices and rents would be moderate.

In fact, I don't think my expectation here is particularly controversial.  Sophisticated observers who described the housing boom as an unsustainable bubble did so because they thought supply and prices were lagging due to market frictions, but they expected new supply to eventually bring down rents and prices.  Where I would part ways with the bubble narrative is that I think it is implausible to expect home buying demand to push home prices nearly as far away from intrinsic value as it has been presumed to have done.

But, you don't have to take my word for it.  We have markets which are known to have flexible housing supply.  And, those who have been following this series know what these cities tell us.  During the boom period with strong housing starts, Price/Rent ratios in these cities were level.  (I will address the late price surge in Phoenix later.)

And, likewise, Price/Income was fairly stable, with maybe a slight rise over the period.

So, these areas demonstrate the classic response to demand shifts when markets are efficient and supply can respond.  When demand shifted up, rent inflation declined, housing starts increased,   This conveyed economic benefits to the broad range of households.  The signs of a demand shift are not rising prices.  The signs of a demand shift in homeownership are rising starts, rising real housing expenditures, and, because homeownership creates supply for housing consumption, relative falling rents.

After 2005, housing starts collapsed and prices collapsed, even in the Open Cities.  This can't be the result of an upward shift in supply, because starts were collapsing.  This has to be the result of a downward shift in demand.  This downward shift generally remains in place.

But, in the Open Access cities, the main signal of demand was high housing starts, not high prices.  The source of demand before 2006, in these cities, wasn't so much generous credit as it was migration from the Closed Access cities.  It also so happens that some of the secondary cities that had the most pronounced late-cycle price bubbles, like Phoenix, were cities most exposed to waves of migration from the Closed Access cities.  Phoenix wasn't different from Atlanta or Dallas because it had broader access to private MBS funding.  It was different because it was closer to California.

 Housing starts are dismal in the Closed Access cities, but during the heart of the boom, according to BEA table CA1, population growth was especially stagnant.  At the end of the boom, population flows into the Open Access cities moved higher, and we can see here that the flows into Phoenix, compared to Dallas and Houston, were earlier and stronger.  This doesn't seem like it should have been enough to cause a price spike in Phoenix.  But, for reasons I haven't pinned down, housing permits were not keeping up with demand.  At the new neighborhoods, there were lotteries at the time to apportion new homes to buyers.  Maybe there is some bureaucratic annual maximum of about 0.15% permits/population.  But, it actually looks like, in the midst of this spike, permits began to decline at the end of 2004 in Phoenix, while population, rents, and prices were all spiking.

I would gladly ascribe higher prices to expectations of persistent rent inflation, but, given Phoenix's historical posture of expansion, persistent rent inflation seems unlikely.  Yet, the fact remains that even in this city that seems to be a prime example of speculative disorder, experienced high population inflows, high rent inflation, housing starts stagnating or falling from their normally high levels, and clear real-time constraints on supply.

The conclusion, in any case, we can come to most confidently is that there has been a negative demand shock since 2005 in the Open Access cities.  Before then, there was either a positive demand shift that was moderated by efficient and elastic supply, or there simply wasn't anything we would describe as a significant positive demand shift.  Housing starts were typically strong in these cities, but they don't look unusual, compared to the 1990s.

In the Closed Access cities, we have clear supply constraints.  So, the question here is, was the boom mostly a reflection of supply constraints, or was it a combination of supply constraints and unsustainable demand pressures?  We know that prices increased to much higher levels in the Closed Access cities than it did in the Open Access cities.  One way we might decide how much of this was influenced by generous lending would be to compare these cities after the bust to the Open Access cities after the bust.  If credit was the fuel, then the removal of that credit should show up as lower prices.  Certainly, during the bust, prices and starts collapsed together, just as in the Open Access cities.  But, in spite of extreme constraints in mortgage credit markets since 2007, where even today total nominal mortgages outstanding has fallen back to 2006 levels, prices have begun to rise.  Rents are rising, too, because supply has especially been constrained since 2005.  So, Price/Income in some of the Closed Access cities are now reaching levels approaching the peak of the boom.

Some might argue that this is due to loose money and QE, but why would QE lead only to nominal expansion in housing and not in broader consumption?  And why only housing in just a few cities?  If the answer to that is that some QE purchases were of MBS, which supported the mortgage market, wouldn't we expect some of that to operate through an expansion of mortgages outstanding?

There was a negative supply constraint, which, beginning as early as 2006, was combined with a negative demand shock.

We should ask ourselves another question.  If this was a credit-fueled bubble, why did housing starts collapse in the cities that didn't have a price bubble?  Why would the subprime buyers in Dallas and Houston, paying 2x - 3x their incomes for homes be the primary source of the collapse instead of the subprime buyers in California paying 10x their incomes?  Why did home prices collapse before rents did and well before defaults did?  (In fact, rents climbed as housing starts collapsed.)


  1. Maybe housing starts in Phoenix were rising, but they just weren't keeping up with the strongly rising population. There was some rent inflation in Phoenix during the housing boom, but not much.

    "But, it actually looks like, in the midst of this spike, permits began to decline at the end of 2004 in Phoenix, while population, rents, and prices were all spiking."

    -Permits per capita, not permits.

    And a rapid expansion of demand from Californian housing investors seems to have been highly important to the Phoenix price boom. The contemporary accounts all mention this.

    And did you see my Miami, San Francisco, Boston post on Against Jebel al-Lawz? What's your explanation for the Boston price and rent hikes of 2000-2004 other than, as I suggest, speculators?

    1. I note that from 2004 to 2005 the construction of 1-unit structures in Phoenix-Mesa-Glendale declined, while that of multi-unit structures rose.

    2. Yeah. I might have to do some sleuthing on that. Something was going on in Phoenix regarding single family home permits. It's strange.

      I didn't see that post. Boston looks like a standard closed access city to me. Very low starts, high incomes, high rent. It doesn't seem like a mystery to me.

    3. Was its housing stock diminishing between 2000 and 2004?! Its population was in steep decline.

  2. Most high end housing in LA and SF are cash transactions. The banks gave huge credit lines and also rich people paid with cash. Just watch Selling LA and you can see that high end properties are in cash, and it is commonplace. I have a relative who buys on credit in a close to high end place, near Palos Verdes, and the banks are requiring that he sell investment property. The banks are making slaves of successful people. Cash is king.

    1. Yes. The idea that those houses doubled in value because of some sketchy mortgage programs is implausible. LA has a supply problem.

    2. I would not argue against you, there Kevin. But LA investors went by bus to Las Vegas and elsewhere to push prices up the last decade. Rents did not reflect the values they were pushed to. Even the little bubble in Las Vegas the last few years is dying out except in preferred areas, where rents and prices are crazy.

      Question Kevin, isn't NGDP targeting used in Japan where the BOJ is buying assets like Reits? And isn't this speculative rather than productive? Could you have a rise in NGDP that is not productive to society?

    3. The stability of the trend is more important than the level. NGDP is just a meaningless number. But, in an economy filled with nominal contracts, unstable NGDP leads to dislocations which are largely unnecessary. If the BOJ had a long-standing NGDPLT policy, they would probably have a much smaller balance sheet.

    4. So, then, Kevin, you would define what happened in the last decade in housing as instability rather than a bubble, is that correct? Also, how many years would a commitment to NGDP targeting be to make it most effective and would a government say, we are committed to this many years? And can you further explain how this works in this Bill Woolsey statement: "Market monetarists not only recognize this, we celebrate it. The whole point is to get rid of changes in real output caused by nominal shocks."

    5. I have this blog called idiosyncratic whisk. I've written a lot about this. You should read it sometime. Scott Sumner has an interesting blog about the details of NGDP targeting.

    6. I was going to go to sleep, but you kept an old man up. I searched around and found this which is easier for me than other ways of understanding the subject. It does not explain, however, how Woolsey would get rid of changes in real output. Is the "how" the buying of assets by the central banks. Is that the main source of "how" to get it done? That was my question. Don't be so grumpy. :)

  3. Great post.

    You know, some stories are simple. There is excess capacity in global auto production market, about 20 million units annually. We have had no inflation in new US auto prices in 20 years. US housing?

    Supply and demand, baby, supply and demand.

    1. You want inflation in autos, Ben? Really? I know, you are a used car salesman. :)Just kidding.

    2. Ben, AS/AD is not quite the same thing as S/D. In any case, your conclusions of what is going on do not follow from your premises.

  4. Yeah. I might have to do some sleuthing on that. Something was going on in Phoenix regarding single family home permits. It's strange.
    Cleaning houses