Monday, February 24, 2020

Housing Part 363 - Did increasing debt cause rising home prices?

I've been playing around with some data on home prices, debt, and construction employment, by state.  First, here is a graph covering 4 distinctive periods of time, comparing changes in home prices to changes in construction employment. (The construction employment measure I am using is the proportion of state employment that is in construction. So if at the start of the period, 5% of the state employee base is in construction, and at the end of the period it is 6%, that registers here as a 20% increase in construction employment.)

Note that there is a surprisingly stable relationship here, throughout the different phases of the boom and bust.  This includes states like California and states like Texas. (Here, I am using the 11 states for which the New York Fed publishes quarterly per capita debt statistics.) There is truly a supply response to rising prices that appears to be generally universal across geography and across time.  The problem, of course, is that in the Closed Access areas, the base level of construction employment is very low and prices are very high, so these relative changes unfortunately are heavy on price changes and light on construction changes.

Source
This is all well enough as it is.  What I would like to reconsider today is the role of debt in this relationship.  Generally, this relationship is taken to be obvious.  Here is a graph comparing home prices and mortgage levels.  Before the crisis the relationship seems unassailable.  Before moving on, I suppose I should point to the obvious divergence after 2011.  Should that give us pause regarding this relationship?

If I was to, say, suggest that, rather than having had a housing bubble, we had a moral panic about lending, which created a one-time 30% or so drop in home values because the new lending standard added a sort of liquidity premium to home equity investments, so there was a one-time price shock then prices continued upward reflecting fundamental value.  Wouldn't a graph of that event look exactly like this?  I have added Canadian data here for a counterexample.

One problem here is that there is no controversy about the potential for a lack of liquidity to push prices lower.  Home prices would go even lower if we made mortgage lending completely illegal. But that doesn't generalize to prices above a reasonable, liquid equilibrium.  The fact that more generous lending today would cause prices to rise (reducing the liquidity premium on the yield earned by real estate owners) doesn't mean that more generous lending would lead to an irrational increase in prices.

Given current interest rates, US home prices are clearly very cheap compared to rents in most places.  Here is a graph of construction employment, mortgage affordability, and rent affordability in Atlanta.  In 2008, a bunch of construction workers were laid off, and homes went on a 30%-off sale.  At the same time, the FICO score of the average borrower shot through the roof.  Lending tightened dramatically.  These market shifts are so extreme, the only reason that the shift toward affordable ownership vs. renting isn't the most talked about issue of our day is because when the body of canonized wisdom is incorrect, it literally blinds us to reality.  You can't see things that you can't look at.  An example I use is that we don't question gravity after watching a magician levitate.  Gravity is canonical. That's fine. Gravity appears to be a true concept. Making it canonical saves us loads of time and effort. But, if we believed that some people had special powers to call on angels to lift us into the air - if that was canonical - we would leave the magic show with a deeply confused and dangerous confidence about how the world works.  Why bother looking to see if there were ropes or hidden platforms? Obviously the guy called on some angels.  You could be like, "But, mom, I saw a cord attached to a harness.  That's how he did it." And your mom would get angry. "What an insulting thing to say about a man who has power over angels."

But, let's leave that all aside.  Let's look at the bubble period.  This graph compares rising debt levels and rising prices between states.  Similar to the first graph above, but the y-axis now is the change in debt rather than the change in construction employment.

Before the crisis, there was only one 2-year period (from the end of 2003 to the end of 2005) where there was any relationship at the state level between debt and home prices.  From 1999 to 2003, debt rose at about the same rate in all eleven states.

One intuition we might have about that correlation is to say, well, sure, we should expect that.  There was a mortgage bubble, and in places with inelastic supply, prices went up, and in places with elastic supply, they built too many homes.  But, remember the first graph.  The change in construction was positively correlated with rising prices, not negatively.  Debt in states like Ohio and Michigan was not related to a significant rise in construction or prices.  The only state that is an outlier from 1999-2003 was Texas, which saw debt rise at a lower rate than most other states, even though Texas had a healthy building market.

One problem is that the canonized narrative is a hodge-podge of different stories.  They all make sense as individual parts of a broader narrative that properly puts supply constraints and rising rents at the center of the story. But, they really don't fit together well within the canonized narrative.  The idea that households were both desperately cashing out housing ATMs and also engaging in a bidding war on entry level housing is a tough pair of assertions to pair up.  I think there is some truth to both stories, but the true version makes more sense if we remove the presumption that the "big story" here is debt leading to an unsustainable price bubble.  The price bubble was largely an equity bubble.  Where mortgage debt increased, it was generally where there was a combination of available home equity and declining rates of local economic growth that caused demand for liquid assets.  So, until the end of 2003, prices were unrelated to levels of debt.

From the late 1990s until 2008, mortgage debt increased from about 43% of GDP to 73%.  From the end of 2003 to the end of 2005, that figure increased by about 8%.  So, 8% out of 30% of the rise was associated with rising home prices, at the state level.  Even there, that doesn't mean that the entire increase in home prices during that time was caused by expanding mortgage issuance, but at least it's plausible that some of it could be.  Now, it could be that the 6% increase in mortgages/GDP after 2005 was recklessly underwritten and ultimately destabilizing, but it had nothing to do with rising home prices.  And, much of the 16% increase that happened before 2004 happened in places with neither unusually rising prices nor rising rates of construction.

It is likely that much of the correlation even in 2004 and 2005 between debt and price growth is a lagged reaction to the price growth of the previous few years.  Buyers requiring more debt to buy more expensive homes and homeowners having more access to home equity.  That makes debt a lagging factor.

That clearly is the case during the period from the end of 2005 to the end of 2008, which was characterized by declining prices while debt was still increasing.  During that period, the relationship between changes in prices and changes in debt became negative.  That is because both declining prices and increasing debts were largely the product of prices having been driven higher before.  Prices had more room to drop and homeowners had more equity to draw on during the early recession period.  At least, until prices collapsed so much and lenders pulled back, so that they didn't have access to equity any longer.

Then, after 2008, there really was a highly positive correlation between rising prices and rising debts, or, more to the point, declining prices and declining debts.

The subsequent events and the policy postures that they called for take on a much different hue if causation largely goes from rising prices to rising debts than if causation largely goes from rising debts to rising prices.  Unfortunately, we went all in on the latter when the case for it was not necessarily strong.

1 comment:

  1. By splash painting your uPVC windows and doors, you can rapidly have a shining new look to the outside of your home, at a less expense and chaos than supplanting! Likewise, a pro paint showering organization can likewise splash paint your carport doors to coordinate. Zenit ramen en deuren

    ReplyDelete