Tuesday, November 24, 2015

Housing, A Series: Part 88 - Supply, Demand, and Economic Migration

I want to continue on the topic of part 87, but before I do, I think it's worth a reminder, before we get too far down the road of talking about subprime loans and increasing housing demand, of several background issues on this topic.  Thinking through supply and demand dynamics is useful, but I don't want to give the impression that, on an aggregate scale, there is much evidence of demand-side excess of home building.  (Keep in mind, higher quantities demanded by home buyers means higher quantities supplied for home occupiers.  Even if these buyers and occupiers are frequently the same, this issue should not be confused.)

1) Real housing expenditures have been falling since the 1980's as a proportion of total spending, and only briefly moved back up slightly as a result of falling incomes after the crisis.  There was no trend of increased aggregate real housing consumption during the boom.

2) Between 1995 and 2008, year-over-year rent inflation was higher than core inflation every single month (minus one outlier).  Oversupply generally leads to falling prices relative to other goods.


BEA data, Survey of Consumer Finance, and author's calculations
3) Securitizations, in general, including public and private, were stable as a proportion of total mortgages throughout the period.  Until 2004, the decline in Ginnie Mae securitizations, which frequently have low down payments and higher interest rates, more than countered the gain in private subprime loans.  After that, there was a sharp rise in privately securitized mortgages, which was offset by a decline in GSE securitizations.  The prevalence of non-conventional mortgages has little relation to rising prices and none to homeownership rates, which topped out in 2004.

4) The average new homeowner during the boom had household income at least as high as existing homeowners.  There is no correlation between rising homeownership and lower homeowner incomes.

5) The rental values of the homes of average owner-occupiers were stable or declining during the boom, while housing costs for renters were rising over the same period.  This is not what we would expect to see if households were using easy credit to move upmarket.

In any event, demand-side factors were not at the center of the housing market for most of the boom, so what I am talking about here is how this secondary factor may have interacted with the supply issue.

I have been describing the country as having open access housing markets with a few closed access cities.  Here, I want to divide that with a little finer detail.

based on 20% dp, 30 year fixed rate, median income and home
First, we have the Old Economy cities.  These cities generally have low or negative population growth because they have lost their economic draw.  They have low rents and home prices.  There were many subprime loans made in some of these cities, but there is little competition for existing housing stock.  It is probably a bit unfair to put Chicago and Philadelphia in this category.  They both actually seemed to have housing market behavior that we would expect in an unconstrained market with low long term real interest rates - moderately rising home prices and real housing consumption (in terms of rent) without excess rent inflation.  Some of that growth may have been accommodated by relaxed lending.  Mortgage affordability as a proportion of income rose during the boom, but was still in the range of the early 1990s.

As I have pointed out in other posts, the sharp decline in mortgage expense in cities like these, along with sharp declines in housing starts in mid-2006 were already a signal of disequilibrium.

Next are the Open Access cities.  This is where the building happened.  Most of the other 62% of the country that lies outside these 20 largest metro areas has a profile similar to this - low cost, higher than average housing permits, and low price/income.  Phoenix prices rose sharply toward the end of the boom, which I think was related to California out-migration, and this may be attributable in some ways to generous credit markets.  Phoenix still belongs in this group because it has low cost and high growth potential, but actual growth temporarily outpaced potential growth.  Phoenix, along with the inland California cities, Las Vegas, and Florida is a candidate for home prices in the late boom that would be difficult to justify with expected cash flows.  In these limited cases, there is some evidence of positive demand from credit availability leading to prices above obvious fundamental values.

Next are what I am calling here the "Closing Access" cities.  The Florida cities and Riverside are a sort of wild card.  They tend to be characterized by very high population growth and low median household incomes.  They, like Phoenix, could have some price appreciation that was inflated somewhat by generous credit markets.  But the rest of the cities in this group are closed access cities in their infancy.  They have limited housing policies, which has pushed rents above the comfort zone, but for the most part, households have been able to counter that by reducing their real housing consumption - paying the same rent as they would in open access cities, but settling for a smaller or less convenient unit.

This combination of high rent inflation and household housing budget adjustments means that rent expense tends to be low in these cities, but home prices have been climbing.  I would identify this as a supply issue, because those high home prices are justified by expected future rent inflation.  And, low real long term interest rates would have an especially strong effect on home prices in these cities because the expected persistence of rent inflation means that the value of homes in these cities is more dependent on far-future growth rates in rents.  It also means they are more susceptible to sharp downturns because of changing growth rates, whether in rent inflation or in the expected income of future tenants.

Finally we have the Closed Access Cities.  These cities have housing supply problems that are so bad that future rent inflation will be facilitated by migration out of the city.  Despite having 15% of US population, and generally high incomes, these cities only accounted for 6% of housing unit permits from 1995 to 2005.

Before the bust, in Open Access and Old Economy cities, rent claimed around 25% of the typical household's income, and this translated into Price/Income ratios of around 3x-4x, which is roughly the high end of the long-term range of home prices where there is no expected unusual rent inflation.  Since the slight rise in home prices was a product of lower long term discount rates, not supply constraints, both rent affordability and mortgage affordability were low (with Phoenix as a brief exception).

In Closing Access cities, rent claimed around 25% of the typical household's income, and this translated into Price/Income ratios of around 5x-6x during the boom.  This is because future rent inflation of a given housing unit will claim a higher proportion of a given household's future income in a city with restricted housing expansion.  This is reflected in today's price, but not in today's rent.  Rent affordability was low, but since home prices were affected by both low discount rates and expected rent inflation, mortgage affordability was not low.

In Closed Access cities, rent claimed around 30-40% of income and Price/Income ratios were up to 11x.  Both rent affordability and mortgage affordability were high.  In these cities, rent affordability could not be mitigated by future reductions in real housing expenditures nor by rising rent as a portion of income.  So, in Closing Access cities, home prices reflect future rent inflation that will be imposed on the present occupants.  In the Closed Access cities, home prices reflect higher rents imposed on future occupants with higher incomes than the current occupants.  Without that migration, these price levels would be unsustainable.  Median mortgage payments on a conventional loan reached 55% to 70% of median incomes in the coastal California cities.  In 2005, aggregate national mortgage affordability was at about the same level that it had been in the late 1980s.  But this was divided between 3/4 of the country where mortgages were as affordable or more affordable than they had been in the late 1980s and 1/4 of the country where mortgage affordability was driven by these supply issues to be far outside any previous range.

I have shown how even though incomes in these cities have grown abnormally, most if not all of that growth has been claimed by rent.  The only way to avoid the ratcheting of higher rents without moving is to try to become an owner-occupier.  Imagine the cash flow demands of that position.  Of course these households were looking for creative financing.  The problem is that the unsustainability of their residency in these closed cities was already embedded in the price of homes, so the only choice was to either move away or use creative mortgage terms to pull down the cost of buying.  These prices were efficient.  A lot has been made of survey information where buyers expected home price appreciation of 10% or more.  This is just one of many oddities that came out of our mania regarding the housing boom.  Since when do we use surveys as a measure of price efficiency?  Do analysts issue a sell opinion on Apple Computer because they interview retail buyers and find that many of them buy Apple based on naïve brand affinity?  No.  They do a discounted cash flow analysis and compare their valuation to the market price.  If we do that with aggregate home values, across these cities we find that home values were justifiable with moderate rent inflation of about 1-2% above core inflation.  Can anybody claim that this is an unreasonable expectation for coastal California real estate?  Ten years past the peak, I'd say it looks too conservative.

If we look at this supply and demand chart for housing, we can see what sort of behavior we would predict in each market from a rightward shift in the demand curve, either due to falling real long term interest rates or due to more lenient lending terms.  These factors would allow a household to bid up the price of homes without increasing their cash outflows.  We should expect low real long term rates to be fully accounted for in the marginal price, because they would change the intrinsic value of the home broadly across the market.  Lenient lending terms would have a smaller and less ubiquitous affect on home prices because these create more risk, so the individual risk profiles of each household would create idiosyncratic demand reactions.

Previously, I have used this chart to think about housing consumption (rent).  Here, I have tweaked it to think about home buying.  On the x-axis, the proxy for quantity is the real rental value of the home.  The y-axis is the price of the home.  One thing to keep in mind when thinking through these scenarios is that rising housing consumption in real terms means that households are building new housing stock.  This does not affect existing home prices.  Only inflationary increases in rent affect home prices, both by increasing existing cash flows from rent and by increasing expected trends in future cash flows.

In an Open Access city, a small increase in home prices will trigger more supply, so in open access cities, a rightward shift in demand would lead to rising, non-inflationary rent/income, level mortgage affordability, and slightly rising prices.  This is what we see in those cities.  Note, elasticity of demand doesn't make much difference here.  It is the elasticity of supply that creates these general responses.

In "Closing Access" cities, real rents would rise more moderately, triggering some rent inflation, and home prices would rise more sharply, partly because of the increase in real rents shown on the graph, but additionally because of rent inflation.  Again, this is what we see.

In Closed Access cities, among existing tenants, rent would already be maximized, so a rightward shift in demand would not increase housing consumption, in terms of real rent.  In fact it couldn't, because supply is inelastic.  But, lower rates and lenient terms would feed the bidding war for existing housing.  So, in these cities, rents would rise moderately.  This would be inflationary, not real.  And prices would rise more acutely because households in these cities would be more willing to use lenient lending terms to remain in the city.  So, prices would rise due to both the effect of persistent rent inflation and due to the effect of lower mortgage payments increasing the potential purchase price of buyers.  Since supply is inelastic, in these cities, the demand shift will flow strongly into home prices.

This is exacerbated by the resulting migration.  High income workers with skills that can be leveraged in these cities have less elastic demand for housing in these cities than existing occupants, so as demand shifts rightward, existing occupants are out-bid for housing as these high income in-migrants become more able to afford homes at imputed rents above levels that the existing tenants can afford.  So, incomes in the city rise.  This would not cause rent/income levels to rise, but it would cause high rent inflation, leading to high home prices, leading to high mortgage costs for the existing occupants.

And, again, this is what we see in those cities: moderate increases in rent/income (no higher, really, than in any of the other types of cities during the boom years), but high rent inflation, and very high prices and mortgage/income ratios.  Here, the elasticity of demand also has little effect on real housing consumption.  The real housing stock is stagnant.  Demand can't raise real rents, so it causes rent inflation.*

So, yes, both low real long term interest rates (which are a market phenomenon, not a Federal Reserve phenomenon) and lenient lending terms can increase demand for home ownership.  But, the effect of that demand is completely dependent on supply.  Where there are no supply constraints, we see classic econ 101 - rents on the existing housing stock decline and real housing consumption increases.  Where there are supply constraints, rent inflation and prices rise.  If California and New York City had Texas housing policies, this period would have been characterized by broad-based economic improvement instead of economic stress.

Source
Here we can see how rent inflation behaved in Open and Closed Access cities from 1995 to 2005.  (San Francisco has some idiosyncratic movement because of its exposure to the internet boom and bust, so that it has high average rent inflation for the entire period, but low inflation during the 2000s.)  This is the signature of strong demand in different supply contexts.  After 2005, demand has been limited by access to credit, not by equilibrium cash flows.  So, tenants are bidding up rents, which has led to similar rent inflation in all types of cities, since the hobbled mortgage market can't provide funding for a price response that would trigger real housing expansion.

The Open Access cities should have been our measure of demand.  There was no increase in mortgage costs in those cities (excepting Phoenix, briefly), but there has been a sharp decline in mortgage costs relative to income since 2006.  A demand bubble never led to unjustifiably high home prices (outside of possibly a few cities, briefly, that amount to a small portion of the aggregate), but a demand bust has led to unjustifiably low home prices since 2006.  We cannot solve the supply problem by creating liquidity crises.  The only way we can pull down home prices in Closed Access cities is to pull down real incomes.  That has been our implicit national policy since 2006.  I would like to try a new policy.




*Rent inflation is understated in these cities, because what little housing stock expansion there is gets valued as if its market value is its real value.  But, really, its market value is an inflated value.  If supply were allowed in these cities, we would expect to see sharp deflation in the rents on those new properties.  If supply were allowed in Dallas, we wouldn't expect sharp rent deflation because, well, supply is already allowed, and the rents there represent unconstrained value already.  This would be hard to measure for many reasons.  One reason is that actual costs in the Closed Access cities are inflated because where builders can manage to build, there are excess profits available, relative to raw building costs.  The gap is filled with costs created by city bureaucracies and by generous labor agreements, as we see in New York City where new building policies are bundled with demands from local craft unions.  This is one of many ways where the economic rents created by limiting the entry of new labor into the high wage competitive sectors in these cities gets transferred to workers in other sectors.  These are the types of rent extraction that are nominally sticky downward and tend to hasten a city's descent when competition reduces the profits of the industries that made the economic rents possible.  When Google starts shifting the core of its value-added operations to Austin, or when Computer Science grads from US colleges start moving to Seoul to begin their careers, shorting Case-Shiller San Francisco futures will probably be quite lucrative, because there is a long way to fall, and no clear ways to avoid the ensuing dislocations.

6 comments:

  1. How does land value play into your thinking in closed cities? I live in NYC and it would seem that even if some building constraints were relaxed overall project costs wouldn't fall that much because land values are anchored in owners' minds.

    ReplyDelete
    Replies
    1. Well, really, all of the changes in market value come through land. We just lived through a period where home values dropped by more than 25% in a lot of cities in just a couple of years. Owners' mental benchmarking didn't prevent that. I think the impotence of owner benchmarking has been pretty much settled by that recent experience.

      Delete
  2. I'm quietly wondering if the fastest way to destroy your city isn't to let in tech people.

    Because tech people use 100 sq. ft. of office space each. So what you'll end up with is Amazon adding 40,000 people to 5 skyscrapers right off 1 freeway exit.

    At which point, you either need to build shoeboxes in the sky (best option), or watch as $10 Billion/year of tech workers working in the same 5 skyscrapers eats up every piece of vaguely tolerable housing for miles in every direction.

    And of course, when you're looking for "High-wage workers that can tolerate the wage ratchet", tech workers just about take the cake. Netflix replaced 30,000 Blockbuster employees scattered around the nation with 2,000 people in Los Altos*. So:

    A) Each Netflix employee is worth $150/hour.
    B) You have a choice of being in Los Altos making some number less than $150/hour or making $0. (Hyperbole, but).

    That sounds like a recipe for disaster if there's any NIMBYism at all ever.

    *This is an old number from a couple years back.

    ReplyDelete
    Replies
    1. Just build.

      This problem appears in major cities across the Anglosphere. San Francisco and Silicon Valley are especially bad because of the network value of tech firms, but if Texas had California's housing policies, you might just as easily be leaving comments saying, "The fastest way to destroy your city is to find oil." Housing is the causal factor here, and cities in the US and around the world stand as examples that it doesn't have to be that way.

      Delete
  3. Print more money and outlaw property zoning. Result: Full Tilt Boogie Boom Times in Fat City. And what would be wrong with that?

    Side note: Although, in a city such as Los Angeles, regular increases in gasoline taxes and increasing subsidies for mass transit might be appropriate.

    ReplyDelete
  4. Read your blog its really informative and helpful keep updating with newer post on emi calculator housing loan

    ReplyDelete