Wednesday, April 2, 2014

Economic growth was not fueled by unsustainable consumer debt.

There appears to be widespread belief in the story that for the past 30 years, economic growth was unsustainable because it was fueled by unsustainable debt accumulated by a squeezed middle class.  Here is a graph of the type that usually accompanies this story:

That story is wrong.  The real story is exactly the opposite, in fact.  This isn't a story of the American middle class mortgaging their future in order to consume today.  This is the story of the American middle class attempting to pay for tomorrow's consumption today.

I believe that this is another case of a sort of money illusion that arises from changing real interest rates and the typical methods though which Americans buy homes.  Even before getting into the details, the commonly believed story is suspect, because, (1) marginalized or economically stagnating populations tend to deleverage and (2) it would be very strange if most of the households in an economy were experiencing stagnation and the most pronounced result of that stagnation was a sharp and sustained bidding war for housing.  In fact, these developments point to a strong and aspirational middle class.

I think that digging into the details makes it clear that this is all about housing.  Here is a graph of mortgage debt, other consumer debt, and home equity, all as a percentage of GDP.  Non-mortgage debt is insignificant compared to mortgage debt, and follows a fairly linear, slightly increasing, trend that goes back at least 60 years.  The issue comes down to housing.  Here, we can see that mortgage debt grew slowly until the 1990's, then leveled off for a decade, then skyrocketed for a decade along with home prices and equity, then collapsed.  Below, these series are shown in nominal dollars, along with GDP.  In both graphs, it is clear that mortgages and home market values were rising together.  It was only with the collapse in home prices that the level of mortgages became out of line with the level of home equity.
 Here is a graph of debt service levels.  Note that while home equity and mortgage levels started rising in 1998, mortgage debt service didn't start rising until 2005.  2005 was when real interest rates bumped up slightly after bottoming out in 2002-2004.  Until 2005, regardless of the nominal price of homes, homebuyers were not committing to any more cash outflows, relative to rent, than they had before.  In 2005 and 2006, rents were increasing at more than the general rate of inflation (in fact throughout the 2000's rents were generally increasing) and home prices had begun to drop slightly, so even during this period, cash outflows of homebuyers relative to renters were not excessive.

So, what's my point, exactly?

My point is that, in terms of lifetime household consumption, the equilibrating variables will be the cash flows of owning versus renting.  Rent levels aren't going to react strongly to interest rate levels.  They will roughly move along with income levels.  So, the monthly cost of buying a home is also going to roughly move along with income and rent levels.

Think of home ownership as basically a security whose coupon is the rent payment (net of estimated repairs, maintenance, taxes, etc.) on the home, with a maturity value equal to the future sale price of the home.  That future sale price will itself be a function of subsequent future rent payments.  So, assuming that home values and rents will increase at the general rate of inflation (which is to say, assuming no speculative motive on the part of the buyer), the maximum current value of a home is the net rental payments over the functional life of the home, discounted by the real interest rate of the given duration.

Here is a rough estimate of real 30 year treasury rates over the past 35 years.  It's a rough approximation, but it tracks pretty well with TIPS rates during the time when TIPS have been available.  The actual duration of a 30 year bond is usually less than 20 years, because of the coupon payments, and long term interest rates tend to be more stable than short term interest rates.  On the other hand, the change in real rates on 30 year treasuries understates the change of rates at a given duration, because as rates decline, the duration of a bond increases.  So, the rates in 1990 reflect a bond with a duration of about 12 years, while the rates in the 2000's reflect a bond with a duration of about 16 years.

In any case, zero coupon rates beyond 10 years tend to co-move.  It is not hard to see that real rates of very long duration securities like a home have fallen significantly.

If we estimate a home value based on a 50 year functional life (with no residual value, for simplicity)and a net rental value of $1,000 per month in current dollars, at a real discount rate of 5%, the home is worth $220,000.  At a real discount rate of 3%, the home is worth $310,000 - a 41% increase in nominal value.  As we push the functional life of the home into the future, the nominal current value of the home increases more with each incremental decline in the discount rate.  One can quibble about the functional life and the proper interest rate to use at that duration, but the end result is clear - very long duration real interest rates plunged in the last 35 years, and home values should have been very sensitive to this change.

My point is that, in the analysis of aggregate households in the face of changing interest rates, this nominal home value is meaningless.  What matters to households and to aggregate measures of cost of living, etc. is that net monthly rental cost of $1,000.  The nominal price of the home is a transfer between two households.  And as long as the housing market is liquid and the real net rental expense remains $1,000, the market value of the home and the size of the mortgage are a mirage.  They could amount to 10% or 1,000% of GDP, and the meaningful amount of debt held by the public would still be a reflection of their ability to live in homes that cost $1,000 per month - and this value did not change significantly over this time period.

Further, in the face of falling real and nominal interest rates, there were ONLY two feasible outcomes - (1) continued unremarkable behavior in the implied rent expense of homes together with significant nominal increases in home values and mortgage levels, or (2) a breakdown in the housing market which would block most potential home buyers from the market and present remaining buyers with very high returns on investment.

We had #1 until the Fed engineered #2.

I say, forget about all the other theories about the unsustainably indebted middle class.  To have expected Americans to keep their aggregate real estate values below 70% of GDP, you might have well asked them to stop breathing.  In the aggregate, it couldn't have been done - the outcome we saw was the only functional possible outcome, and it had nothing to do with debt-fueled consumption.

Side Note

If I was a homebuilder, and I believed that the period up to 2006 was reasonable, and that it was 2007-2008 where the housing market was broken, and if I had options on developable land, I would not be in any hurry to develop those plots.  If the Fed can keep itself from torpedoing the money supply again (a big if in today's hawkish environment), those lots should see a 30-40% appreciation over the next couple of years as long as the banks rediscover real estate loans coming out of QE3.