Friday, August 28, 2015

Housing Tax Policy, A Series: Part 54 - Two Stories of Housing Supply and Demand

A review:

Excessively accommodative monetary policy led to over-expansion of credit and mal-investment, leading to a massive over-consumption of housing.  Where that overconsumption couldn't be met with supply, prices skyrocketed.  As predatory lenders approved households for more and more housing, households bid up the price of homes, pumping up the price of owner-occupied properties as households moved out of their rented apartments into their shiny new McMansions which were more house than they should have ever been allowed to own.

As regular readers know, I have previously pushed back against this narrative.  Well, now that I have reviewed the data, I must confess that this narrative does fit...


...It fits the 1970's, that is.  Which, it so happens, was a period that actually had excessive monetary accommodation.  Well, the narrative fits if you remove the rapacious bankers and replace them with reasonable households.

I might need to explain the graph.
The green line is core inflation - a pretty good proxy for monetary policy, given that our stated policy is an inflation target.
The purple line is shelter inflation, relative to core inflation.  Owner-equivalent rent inflation only goes back to the early 1980s, but since it accounts for the bulk of the Shelter component, shelter is a very good proxy that goes back much farther.  I have not adjusted this measure to remove non-owner rent inflation, but this tends to have a small effect, which would mostly just amplify the patterns we see here.
The blue line is rent inflation for non-owners, relative to core inflation.

As we enter the 1960s, inflation is low and shelter inflation is unremarkable, but beginning in 1966, inflation kicks up and remains high until after the Volcker adjustment.  Notice the interesting pattern throughout that period.  Whenever inflation shoots up, relative rent inflation moves down and owner-equivalent rent inflation moves up.  Notice, too, how the late 1960's and late 1970's episodes, which coincide with the sharpest owner-equivalent rent inflation also see run-ups in homeownership rates and in real estate values.  That is what a demand-side housing boom looks like.

What caused me to look at this more closely was a very interesting speech by Philip Lowe, Deputy Governor of the Reserve Bank of Australia.(HT: Matthew Klein via Noah Smith)  (I will have more to say about that speech in my next housing post.)  One factor Mr. Lowe mentions regarding recent home price behavior is:
In the 1970s and 1980s, regulation of the financial system and high inflation served to hold down land prices artificially. They did this by limiting the amount that people could borrow. When the financial system was liberalised and low inflation became the norm, people’s borrowing capacity increased. Many Australians took advantage of this and borrowed more in an effort to buy a better property than they previously could have done. But, of course, collectively we can’t all move to better properties. And so the main effect of increased borrowing capacity was to push up housing prices, and that means land prices.

I'm not sure about much of this anymore.  First, I'm not sure why we can't all move to better properties.  There is a lot of unused space 100 to 1000 feet above the major cities, in Australia and the US, and building costs aren't as high as one might think.

I originally thought that unprecedented low mortgage payment levels in the 2000s reduced barriers to credit, and that this was a cause of low cash yields (high price/rent) in the 2000s.  Mr. Lowe's assertion seems intuitively reasonable.  But, this assertion is, surprisingly, not supported by the data, which shows strong single family housing starts, rising owner-equivalent rent, rising home values, and rising homeownership rates during both the 1970s and 2000s.

And, even more surprising, even while all of these indicators of growth were strong during both periods, growth in real housing expenditures stepped down to permanently lower levels coincidental with both periods.  That's a pretty strange thing to see during a building boom.

Looking at the 1970s, the demand for homeownership was probably related to the high inflation of the time.  Not only did homes provide a hedge against inflation, they provided a tax-sheltered hedge against inflation.

These various trends are what we would expect to see in an efficient market.  The incentive toward ownership would lead to reduced renter demand and increased owner demand.  Owner-occupied expenditures would be inflationary, because the tax benefits of ownership (imputed rent paid to oneself is tax exempt, as well as much of those inflation-fueled capital gains) would shift the demand curve of owner-occupied housing to the right.  (Another way to think about this is that the increased tax benefits created by high inflation pushed pre-tax owner-equivalent rent up, but after-tax O-E rent may not have changed substantially.)

Higher rents and low real interest rates may have pushed the nominal price of homes up, and higher demand for homeownership may have pushed up single family housing starts.  But, the trend in real housing expenditures suggests that marginal households were lowering their real housing consumption in order to transition into homeownership.

This explanation (easier access to credit because of low interest rates) was part of my original explanation for the housing boom of the 2000s, but upon closer inspection, I don't think the evidence points to a very strong effect.  There are many confounding factors that create false impressions that overstate this effect.  In addition to all of the moving parts described above, when real long term interest rates fall, they cause the intrinsic value of homes to rise at the same time that they would appear to increase demand for homeownership due to easier access to credit.  But, the lack of an increase in real housing expenditures during these periods of low real rates suggests that it is the rise in intrinsic value that is operable.  Households weren't moving up into more valuable homes (in terms of rental value) during either period.

Another confounding clue is the change in tax laws in 1986 that created new value for home mortgages because of the mortgage interest tax deduction.  Note that O-E rent did begin to fall along with home ownership after inflation began to subside in the early 1980s, but permanent reductions in inflation happened slowly, and the added tax advantage of the mortgage deduction moderated the market reaction.  I think this probably helped homeownership remain at about 64% into the 1990s and produced some moderate O-E rent inflation in the late 1980s, relative to renters inflation.

Beginning in the mid-1990s, rent inflation for both renters and owners began to run persistently higher than core inflation - with renter inflation tending to run slightly higher than owner inflation.  As I have pointed out before, home prices didn't begin rising until around 1998.  Half of the recent (temporary) increase in homeownership happened before any substantial rise in home prices.  This rise in homeownership doesn't appear to have any significant effect on prices or housing consumption.  This could be related to Community Reinvestment Act policies, but is unlikely that it is related to mortgage rates.  Rates in the 1990s were lower than the late 1970s, but they were still relatively high.  And, there were some increases in non-traditional lending, but nothing like the scale of the 2000s.

This period doesn't have the rent inflation signature of the 1970s.  And, along with the persistently high level of renters inflation, there is no cyclical supply response in multi-unit housing starts during this period.  The demand curve for home ownership hasn't shifted to the right; the supply curve for rented real estate has been stuck to the left.  The first time credit access really became a factor in housing markets was around 2006.  Since then, homeownership rates have collapsed, and for the first time, there has been a sharp divergence between the expected rate of return on homes and the rate of return we would expect in a market with broad access.

Here is a graph I have posted before, I think.  Before 2008, the inflation premium inferred by the difference between real returns to homeowners and effective nominal mortgage rates was a pretty good approximation of expected inflation.  In other words, housing markets were efficient.  In fact, considering the persistently high rate of rent inflation since the mid 1990s, the implied return on homes in the 2000s looks fairly conservative by this measure.  Even if effective mortgage rates at the time were reduced by the use of adjustable rate products, and a duration adjusted inflation premium at the time was more like 3% instead of 2%, this would not have been unreasonable.  But, after 2007, this long-standing arbitrage relationship broke down.  Real returns on home ownership have been higher than nominal mortgage rates, even in the face of the return of high rents.

I think there is a lot of confusion about supply and demand in housing that comes from conflating home ownership with housing consumption.  The presumptions we make about housing are the equivalent of seeing a sharp rise in bond purchases, and concluding that people must be spending and consuming more.  Home ownership is a claim on future cash flows, and it appears to me that the price of those claims, in the aggregate, is as systematic and regular as we would expect from any security with variable, but relatively stable, cash flows.

Home prices are efficient.  Nominal housing consumption is pretty stable over time.  Nominal housing expenditures seem to have reached a stasis around 1960.  Notice how the slight dip in relative nominal housing expenditures is during the inflationary period where home prices and demand for home ownership is rising and real housing consumption is stable. (The bumps in real housing expenditures around 1975, 1980, and 2008 are mostly related to drops in real incomes associated with deep recessions.)  The secondary effects of inflation increased demand for ownership, not for consumption.  This seems to have had more effect on the housing market than limited access to credit did.  Remarkably, prices and net returns on investment appear to reflect reasonable estimates of long term real required returns.  Real returns on homes were declining in the late 1970s, as did real long term bond rates.  To the extent that high mortgage rates created a market reaction, it appears that nominal expenditures - rents (paid and imputed) - adjusted while home prices remained efficient, leaving no unusual profit opportunities.

Supply is the irregular variable here - the variable that isn't either stable or predictable.  As real housing supply declined from the mid 1990s to the crisis, relative to other expenditures, nominal consumption remained stable, and prices were an efficient reflection of constricted supply.  Now, the systematic nature of prices has broken down, and will remain broken until either owner-occupier credit becomes vital again or investor ownership continues to climb long enough to fill the void.  Until then, rising rent will be the story of housing.

The obsession with "bubbles" and confusion about housing means that rising rents will be widely perceived as a "bubble", regardless of whether home prices fully recover or not.  As long as supply remains low, high rents will make just about any price seem high, even if it's 30% below intrinsic value.  The resulting destruction of demand imposed to fix the "bubble" will do nothing to solve the supply problem.  But, I suppose our lower real incomes will make us miss the homes we couldn't have built anyway a little less.  We won't have to deal with any of those gauche rich people trying to buy condos in San Francisco.  So, we'll have that going for us.

No comments:

Post a Comment