Tuesday, November 17, 2015

Housing, A Series: Part 85 - Housing Prices were sustainable because of migration

I think the migration patterns of the closed access cities are an important and underestimated factor in the characteristics of the housing market.

Here is one of many strange patterns that we see in housing markets.  There is a positive correlation between incomes and Price/Income of local real estate.  As incomes rise in these cities, households don't just increase their housing budget proportionately; they spend more of their budget on housing.  I believe that this explosion of incomes in just a few cities is due to the constraint on housing.  The constraint on housing explains the higher incomes and the peculiar pattern of spending more of those incomes on housing.

Here we see that these outlier closed access cities have unusually high rents.  The high prices are based on rents and expected future rents.  (Washington, D.C. is a special case where high incomes are not caused by a housing constraint.)

Next we see that home prices are even higher in these cities than we might expect from the higher rents.  This implies that the cities with high rents also have high expected rent inflation.

The result of these relationships between price, rent, and incomes means that in these cities home prices are very high as a proportion of incomes.

It seems like there might be a sort of disconnect here.  If rents are already high in these cities, how can we expect rent inflation to continue?  I have even walked through a model of housing expenditures myself that says households reach an upper limit when rent affordability reaches a level 10% or 15% higher than other cities.  But, this is where migration creates a distortion in our analysis.

If rent as a proportion of income tends to run about 50% higher in the closed access cities than it does in the open access parts of the country, then we might expect home prices to be 5x or 6x income there instead of the 3x or 4x levels we see in the rest of the country.  So, housing selling for 9x or 10x incomes seems like it must be speculative and unsustainable.

But, the tricky part here is that the value of that home is based on the present value of its future rental income.  Under current policies, out-migration of financially stressed households and in-migration of high income households is inevitable.  It should be a part of home buyer expectations, and thus home prices.  This means that the income of the household who currently owns that house is not the binding constraint on its price.  The price of the house, and its future rents, should reflect the income of the household that will live in it 5, 10, or 20 years from now.  In the open access market that we tend to assume is operable, there is no persistent migration, so we can almost always assume that the income of the tenant household is stable.  But, because the tight constraint on housing in these cities will create a transition of higher incomes into the city, the Price/Income of the city's current inhabitants will be high, because they are living in a home that is likely to be sold to a higher income household when they are finally forced to migrate out of the city.  The housing market is so efficient that it has already priced in that household's future economic stress.

Median rent is now about 45% of an $80,000 median income in San Francisco.  What will happen is that migration will support an increase of 30% in rents.  Median rent will remain 45% of the median income, but the median income will now be $104,000, partly as a result of migration of low income households out of the city.  And, today's price is much closer to 6x that new household's income than it is to the current tenant's income.

It is common to point to these cities and claim that these high home prices are clear signs of credit-fueled speculation.  But, given the supply constraints in these cities and the ongoing population flows that result, what do these observers expect home markets to do?  Given the severe supply constraint, home prices that are unsustainable for their existing owners are a sign of efficiency.  The city itself is unsustainable for its current residents.  They have been and will be leaving.

Of course, in these cities homebuyers have to use much higher levels of credit and creative financing.  They have to buy a home at the price that reflects the future value it will have to in-migrating buyers.  They don't have a choice.  Given the severe supply constraint, high mortgage levels are a sign of efficiency.

And, of course, the real estate in these cities will be more volatile and more susceptible to a downturn.  They have the characteristics of a growth stock.  Their value is highly dependent on that future rent inflation.  Given the severe supply constraint, collapsing prices in response to a negative liquidity shock and income shock are a sign of efficiency.

A new economic letter from Glick, Lansing, and Molitor at the San Francisco Federal Reserve Bank (HT: CR) is almost a paragraph-by-paragraph example of how the assumption that these things are the product of lending and excess demand seems so irrefutable, yet is so wrong.  We were so convinced that the demand explanation was right that nobody bothered to realize that all of these facts are just as emblematic of a supply constraint.  We have been so sure of ourselves, that nobody has noticed how weird it is that the excess lending only applied to a few peculiar geographic areas.  We just thought that there must have been something in those areas that was especially enticing to those frothing speculators.

The letter references a Gelain, Lansing, and Natvik (2015) article that claims to reverse engineer the housing boom in a way that suggests loose lending standards were to blame for unsustainable home prices.  They claim that rational expectations models fail because they would require a rise in rental expense along with a rise in home values, and aggregate rent expenditures did not rise during the housing boom.  They find that a model dominated by persistent and backward looking buyer expectations and related changes in lending standards explains this behavior better.  They point out that several studies have found that home prices rose more sharply in areas where subprime loans were more prevalent.  But, it seems to me that they are assuming away the supply factor, and this causes them to reverse the causation on these factors.  Persistent buyer expectations of price appreciation, more lenient lending standards, and higher consumption of households in appreciating areas are all unsurprising outcomes from a housing supply constraint in a city with high demand for labor.

And, what the studies that saw these connections between rising mortgages and home prices should have also noticed was that those were areas with high rents and rent inflation.  What the authors failed to notice was that the stable rent/income ratio was achieved through geographically selective growth and migration - high income households bidding up rents in the closed access cities and middle income households building new homes in open access cities.  More than 80% of new homes built during the boom were built in areas with little or no home price appreciation.  (I touched on this in the last post. I hope to follow up on it some more in an upcoming post.)  Why didn't it strike anyone as strange that in all of those cities with sharply rising mortgages and home prices, where the demand excesses were supposedly centered, there were relatively few new homes being built?

Disaggregating by city, we can see quite clearly that home prices and rents were highly correlated.  Rational expectations certainly describes the pattern of prices and rents between cities.  Here is a comparison of the top 20 MSAs by home price and rent in 2005, at the top of the market.

Looking at rent 10 years later, in 2015, compared to 2005 prices, we see that even after subsequent rent appreciation, there is a systematic relationship between price and rents.  Rents have risen proportionately to the peak prices, even though (1) the patterns of rent inflation and population migrations were interrupted by the recession, reducing increases in rents in the closed access cities, and (2) continued repression of the mortgage markets has caused rents to rise unnaturally even in the open access cities.

But, even with those factors at play, obviously there were strong connections between price and rent appreciation during the boom.  The model that the article develops specifically favors the lending standards explanation for the boom because it results in a disconnection between home prices and rents and rejects a rational expectations explanation because that results in a connection between prices and rents.  That does appear to be what happened in the aggregate data, but when we disaggregate by city, their model doesn't describe price and rent behavior anywhere.

Back to the FRBSF letter:
An accommodative interest rate environment combined with lax lending standards, ineffective mortgage regulation, and unchecked growth of loan securitization all helped fuel an overexpansion of consumer borrowing. An influx of new homebuyers with access to easy mortgage credit helped bid up house prices to unprecedented levels relative to rents or disposable income. The run-up, in turn, encouraged lenders to ease credit further on the assumption that house prices would continue to rise. Similarly optimistic homebuilders responded to the price signals and embarked on a record-setting building spree...
Disaggregating by city, we can see that this is incorrect. The influx of new homes was almost completely within the parts of the country that saw moderate price increases.  The price run up and the construction boom happened in different places.  Homebuilders did not respond to price signals.  They built in the lowest cost cities.  There was no record-breaking building spree where prices were high.
But when the various rosy projections failed to materialize, the housing bubble burst, setting off a chain of defaults and financial institution failures that led to a full-blown economic crisis.
Here is a chart comparing the prices of the median home in each MSA in 2005 near the peak of the boom with the subsequent change in price.  Not only have rents continued to climb, but as we have escaped the liquidity crash, home prices have recovered at least as strongly for high priced cities as they have for low priced cities.  Except for the projections of a stable money supply, what projections have failed to materialize for home buyers in the boom?  Vacancy rates have tended to remain low in the closed access cities, and rent inflation rose sharply in 2006 and early 2007 after housing starts collapsed.
In contrast, the recovery in new housing starts has been more sluggish; the series remains roughly 50% below its prior peak—suggesting that homebuilders are exercising caution in light of the substantial overbuilding that occurred during the mid-2000s. The pattern in Figure 1 also suggests that there may be further upside growth potential for the housing market; continued high house prices should contribute to more building activity and more construction jobs.
In the decade from 1995 to 2005, in New York City and coastal California, where Home Price/Income reached an average of 9x at the peak, there were 2.8 housing permits issued for every 100 residents.  In the rest of the country, where the average Price/Income never reached 3x, 7.2 permits were issued per resident.  Homebuilders aren't exercising caution.  Lenders are.  There was no overbuilding related to high prices.

I really want to try to be polite here, because really we have all missed the boat on this issue, but I just have to point out that this is the San Francisco Federal Reserve Bank.  Everyone involved in preparing this analysis about "substantial overbuilding" triggered by high prices that were unmoored from rents, at the end of the day, turned off their computers, walked outside, and went home to the most expensive, constrained, undersupplied housing market with the highest rents that we have ever managed to conjure.

Depending on whether their analysis is correct, their closing sentence: "Nevertheless, given that housing booms and busts can have significant and long-lasting effects on employment and other parts of the economy, policymakers and regulators must remain vigilant to prevent a replay of the mid-2000s experience." is either the most comforting or the most frightening economic statement you will read today.


  1. Excellent blogging. Egads, of course housing supply is constrained---there is extensive zoning up and down West Coast. This is obvious!

    1. You can say the entire series, all 100000 parts of it, are obvious. Location, location, location, and yes first tier properties retained their value better after the 2006 Bubble was burst...'nuff said.