Thursday, January 14, 2016

Housing, A Series: Part 103 - How supply constraints lead to instability

One of the factors in the housing bubble that is difficult to appreciate and difficult to convincingly document is the volatile influence of rent expectations.  Part of the bubble story is the expectation that there cannot be persistent excessive rent inflation because higher rents will induce new supply, eventually bringing down rents and prices.  Since rational homebuyers should anticipate this, home prices that rise above the level justified by current rents, incomes, etc. are a signal of irrationality - a bubble.

Of course, there can't be a sufficient market supply response in places like Silicon Valley, San Francisco, New York City, and Boston.  So, instead of creating a supply response, rising rents and prices in those cities create persistent rent inflation expectations.  This leads to volatility, because the effect of changing expectations on a long-lived real asset are strong.  The difference in the value of rental cash flow 40 years from now with no unusual cash flow growth vs. excess rent inflation of even 1% is large.  In fact, permanent rent inflation of 4% to 5% would basically drive intrinsic home prices to infinity.  Homeowners in the Closed Access cities have to deal with a sort of St. Petersburg paradox:

Question:  How much is a house worth in San Francisco?
Question in Reply: How accommodative will local housing policy in the year 2040 be?

The possibility of sharp regime shifts in policy and the strong influence those future policies have on the values of properties, today, makes the values of those properties very volatile, and it also makes it easy to yell "irrational exuberance" in a crowded buyers' market, because the only way to falsify that claim is to quantify the rational expectation of San Francisco housing policy in the year 2040.

The problem is, you have to have an expectation.  It is unavoidably important to the value of homes in San Francisco.  This is why the scoffing dismissals of San Francisco home buyers who were expecting 10% capital gains every year are slightly off the mark.  The value comes from rent.  And, very small changes in rental growth create huge shifts in property values.  So, what if we reframed this as a question of rent instead of property values?  Would it have been crazy for San Francisco home buyers to expect persistent real rent increases of 2%?  Because that is about all they needed, even at the top of the market, to justify their purchase prices.

You know where buyers don't have to worry about local housing policy in 2040? Dallas, Houston, Atlanta.

So, what is more likely?  That there just happen to be rapacious, short-sighted bankers and gullible borrowers in all the places with constraints on local housing expansion?  Or that constraints on local housing make rational valuations very difficult, very important, and very volatile?  Certainly, pre-paying those future rents by buying the property usually requires the use of credit.  Certainly, some households vulnerable to the uncertainty of rising rents will be more agreeable to some uncertainty in their credit terms.  So, certainly, the extent to which a generous credit market is available will effect the ability of prices to reflect expectations.  The question, then, isn't whether generous credit is responsible for rising real estate prices.  The question is, "Were the expectations implied by those prices reasonable?".

Has anyone even attempted to answer that question?  This is one of the illusions at the heart of the housing market "bubble" narrative.  Prices can rise because of positive demand pressures or negative supply pressures.  Rising mortgage levels are not a sign of rising demand pressures.  They are only a sign of rising prices.  But, it seems that rising mortgage levels have been taken as the sign of demand pressues.  Nobody has asked the important question: "What was the correct price for homes?"  It sure seems like it should be around 3x incomes and mortgage payments should be 28% or less of incomes.  Those rules of thumb have worked most of the time, for a long time.  But, that goes out the window when median renters in places like San Francisco are spending 40% to 50% of incomes on rent.

So, one of the problems I will have as I formalize my findings is that if I have a model that seems to show that expectations were reasonable, skeptics will say, "Who says your model is right?"  And they will have a point.  But show me the model that the consensus is using now!  It looks to me like we've drawn a straight trend line, and when prices moved away from it, we all said, "Ah, ha! It's those rapacious bankers and irrational home buyers."  I ask, what is your model?  A model that can attribute anything it can't explain to irrationality and unsustainable greed, and treat that as a confirmation of the model, explains everything.  It's the ultimate exercise in question begging.

Aaaaannyway.... back to the issue that motivated this post.  The fact that expectations are so powerful creates a real problem for the potential to solve the supply problem.

Here is a table of home valuations, based on a very simple standard financial model of discounted future rent cash flows.  Here I have based it on the present value of 100 years of rent.  These valuations happen to fairly closely match valuations we see in US housing markets.

The left column has "Open Access" assumptions.  These are roughly the median US input values and home price values we see today.  In the mid-1990s, net rent, in nominal dollars, was about $4,000, and median home values were about $100,000.  So, adjusting rent, this also roughly describes the US housing market in the mid-1990s, before the boom. (Net rent is an approximation of rent after expenses, taxes, and depreciation.)

The next column represents base US median "Open Access" home values similar to market conditions in 2005, after real interest rates had declined from the levels of the 1990s.*

The next column represents the effect on price from a sharp rise in rents.

The next column (the middle column) reflects a city with high rents and an expectation that they will continue to rise.  Adjusted for rent inflation since 2005, this column roughly approximates rent, inflation, rates of return, and home prices, at the peak of the boom in 2005, in San Francisco and San Jose.

The next column reflects a city with high rents, high expected rent inflation, and an increase back to higher real long term interest rates.  This reflects the current rent, inflation expectations, and real estate rates of return, and median home price for the San Francisco-San Jose consolidated metro area.

The sharp contraction in mortgage markets has led to a divergence between returns available from home ownership and returns from real treasury bonds.  If those (implied) yields converged again, I think real estate values, given current conditions, in San Francisco, would rise toward the level shown in the middle column.  The middle column reflects a city with the severe building constraints of San Francisco, but with a healthy mortgage market that funds a reasonable level of housing starts, nationwide and pushes prices to non-arbitrage ranges.

OK, So what's the point of my fancy-schmancy table?

Here's the take-away.  Let's start at the column where interest rates have already risen to 4% and the median home in a San Francisco-like city is worth $726,812.  This is similar to today's context.

Let's say that San Francisco solves their housing problem enough to reduce rent expectations, but not enough to reduce actual rents.  This is actually not a natural resting point.  Stable real rents are a reasonable expectation in an open access environment, where the cost of unconstrained new building creates a stable floor under expected future home prices and potential new supply creates a ceiling.  But, there is no equilibrium that would remove volatile future rent expectations while keeping rents in San Francisco at the artificially high levels of today.  But, taking this as a conceptual first step, this alone would reduce home values by about 13%.

But, if San Francisco realistically solved their housing supply problem, we would expect rents to eventually revert back to "Open Access" levels.  This is represented in the last (right hand) column.  I have tried to use relatively conservative assumptions here.  In the high-rent-inflation scenario, I limited real rents in the long-term to double their current levels.  In the mean reversion scenario, I have limited real rents on the downside to a level that is still 50% higher than the US median.  This is about where San Francisco rents were before the 2000s.  So, even the mean reversion scenario leaves San Francisco a very expensive city.  And, even in this scenario, as modeled here, it would take more than 35 years for rents to revert back to that level.

So, if San Francisco credibly solved their housing problem at the abstract policy level, even before tenants actually began to see relative rents declining, home values might decline by 50%.  This is the power of expectations.  I don't see any natural policy resting place between those two extremes of expectations.  And, I don't see how a mitigating policy can counter that.  In fact, that expected 50% decline is based on today's price levels, which I would argue have already cut home values there by about 25%.  So, by undercutting the national mortgage market so sharply that it hasn't expanded in nearly a decade, we have managed to avoid less than half of the potential decline in San Francisco real estate.

The sorts of policy shifts that we have to see in our leading cities in order to take the cap off of our aggregate national growth potential, to reduce income inequality, and to stop the painful segregation of low income households migrating out of those cities and high income households migrating into those cities, will be unavoidably painful.  If, in the meantime, we manage to fix our national credit and monetary policies so that we stop harming potential homeowners in Texas and Georgia, then home prices will generally rise, and eventual solutions in the Closed Access cities will be even more painful.  I don't think those real estate owners generally deserve to be harmed.  They played the hands they were dealt, which included the expectations created by Closed Access policies.  But, I don't see how we avoid it.  And, they certainly won't be getting a lot of sympathy from the broader population.  It's just so easy to claim, in hindsight, that the banks and home buyers were unhinged speculators taking irresponsible bets.  It will be a game of "The bankers did this to us: Round 2.", complete with grainy video footage and Stacy Keach voiceovers.  And even the solution will cause the airwaves to call out, "The system is rigged!"  Volatility will always be blamed on the capitalists.  And, it is true that the price of San Francisco real estate is more volatile than the official price of Venezuelan toilet paper.  But, at least the sound and fury would be coming from necessary solutions and not unnecessary deprivation.  In a complicated world, it's probably the best we can hope for.

* Actually, some of the price levels of the mid-2000s, even at the aggregate national level, reflected some of the rent inflation expectations caused by the Closed Access cities.  Rent inflation and Real Rates of Return have a similar but inverse effect.  So, a 0.3% rent inflation expectation and a 3.3% required Real Rate of Return would lead to a similar valuation.  In either case, a net discount - growth rate of 3%, together with median nominal rent levels at the time of around $6,000, lead to a valuation of just under $200,000, which are all in the ballpark of the measures at the time.


  1. Excellent blogging.

    As I said before, a generalized inflation rate of 3% would probably cushion long-term homeowners, and besides that more building in closed cities would probably encourage more net in-migration slowing the decrease in housing costs.

    Print more money and build more housing and don't stop for the next 10 years!

  2. You are generally right about closed cities, Kevin. But bankers in open cities sucked. The bubbles happened in outlying areas, Modesto, Sacramento, Stockton, Fresno, Tulare, Bakersfield, San Bernardino, Las Vegas, Reno, and in the states of Florida and Arizona. There are millions of people who live in those cities. Even Atlanta crashed.

    If you make the argument that the closed cities are not bubble cities, you may be onto something. However, in Manhattan Beach, as I recall, prices were barely sub 1 million and the credit crisis dropped them to 750k. Then they bounced back up in relatively short order and are above 1 million. So even closed cities are subject to banker fears and any bigger crisis could undermine confidence even in those cities. But generally, you are correct about the closed cities, but those were not where the big money was lost.

    Defending bankers in these narrow cities may be legitimate, but liar loans, or loans toxic in any way, or as you say, uncertain credit terms as a weak euphemism, can be very hurtful. And if you have to sell at the wrong time, with all this volatility, you can be screwed. Don't buy with this volatility.