Monday, April 25, 2016

Housing: Part 139 - "Demand" side vs. "Supply" side and the NGDP measurement problem.

One aspect of my story, as it relates to monetary policy, is that, contrary to popular opinion, monetary policy was not particularly loose at any time during the 2000s, and that eventually, really starting at least as far back as the sharp pullbacks in GSE originations in 2003 and 2004 after the accounting restatements, a series of demand shocks eventually built up until they caused a recession.  Or, I have been calling them demand shocks.  Maybe I am using the wrong terminology.  Give me advice if anyone has any ideas.

The problem is that we normally think of bank credit as "demand" because we would expect it to fund additional spending and lead to an expansion of currency.  But, I contend that the higher mortgages of the housing boom were not associated with demand or expanded spending.  They were a reflection of transfers to previous Closed Access real estate owners.  To the extent that Closed Access homes were sold at profits, the old owner received a windfall and the new owner took on new debt that was a reflection of future high rent payments.  I don't see any reason to think that those mortgages were funding net new spending.

But, since credit is usually associated with demand, I have been referring to the series of negative shocks to the mortgage credit market as negative demand shocks.  I think I need to find better terminology.  Because, those mortgages fund housing starts where housing starts can be funded, so really, negative shocks to mortgage credit were negative supply shocks.

And, this is the strange way that the recession was related to a sort of perverse NGDP behavior.  Since housing demand is inelastic where supply is constricted, the shocks in the mortgage market were causing housing starts to collapse, which led to the high rent inflation of 2006 and 2007.  This meant that real GDP growth was falling.  Nominal GDP growth held up somewhat because of the high rent inflation.  I say that both measures were in recession territory by the end of 2006.

But, we didn't call it a recession, because employment was still strong.  Employment was strong because the Fed had inadvertently created a supply-shock recession based on supply shocks that they didn't recognize.  And, since housing demand is inelastic, the more they pulled back on the credit levers, the higher rent inflation and NGDP went.  This meant that there was no sticky wage problem, there was just a ratcheting up of costs.

We ended up with monetary offset - against itself!  First, the Fed and Treasury created a supply shock by ratcheting down the mortgage market, then they reacted to that by adding a demand shock by cutting off currency growth.

I have been attributing this completely to rent inflation.  But, now, I think maybe import inflation plays a role here.  I will try to address that in an upcoming post.

I have been arguing that NGDP growth has been overstated since 1995, and was especially overstated in 2006 and 2007 because higher rents in Closed Access cities are measured as inflation, when they are more akin to a tax (of Closed Access wage earners) and transfer (to landlords).  If I just adjust GDP with the different inflation rates (with or without shelter inflation), there is a difference between the two.  But comparing total GDP to GDP (ex. housing) gives very little difference.  I assumed it was because of some trick in GDP growth calculations that I didn't understand.  Today, it hit me.

Migration has become a core part of the story, and I realized this forces all the shelter inflation to be mitigated with substitution.  There is no way for it not to be mitigated through substitution because the core act in the story is that when a high income household moves into a Closed Access city, a low income household has to move to Phoenix.  So, there is some rent inflation in San Francisco and a new home in Phoenix of much lower value than the home in San Francisco.  Shelter has been about 18% of nominal consumption for decades.  A longterm flat trend.  After that migration is triggered, spending on housing in both cities remains the same as a portion of incomes.  The cost comes from the family that must move out of San Francisco, lowering both their income and their housing expenses.  The counterfactual doesn't come from the difference in non housing GDP and total GDP.  The difference comes from total GDP and an unmeasured potential GDP.  We can estimate this by applying different inflation rates, where the substitution isn't accounted for.  But substitution has to make the difference disappear in GDP figures because the substitution is the story.


  1. Fascinating post.

    Obviously, the Fed and other macroeconomic policy makers need to consider US housing markets.

  2. Wow! That's a yolk! The displaced don't really need to move to Phoenix, they just have to move far enough away where their cost and income go down. They pay another "tax" by commuting to where the jobs remain. The sub-divisions far from Chicago are always attractively priced, but the 2 hour commute is not free of cost, actual or lost wages due sitting in traffic.

    1. Yeah. You're right, there is a gradation of trade offs. The trade off ends up creating a substitution between monetary expenses and non-pecuniary expenses. A lot of wasted resources. That's where Phoenix comes in. Eventually, the cost problem reached far enough into the suburbs, that living in Riverside in a relatively expensive house and commuting into LA for the extra $20,000 in wages becomes the less favorable option than taking the pay cut. And, if you're going to take the pay cut, you might as well move to Phoenix where the houses are cheaper.