Monday, November 9, 2015

Housing, A Series: Part 80 - Measuring the Real Estate Access Premium

I have been playing around with the awesome data sets at Zillow, and the picture of the housing constrained economy comes out through this data clearly.  Zillow has some data that goes back to 1979, and I would divide the time frame into three broad periods. (1) 1979 to 1995, when housing constraints weren't binding and incomes were rising broadly across cities, both before and after housing costs. (2) 1995 to 2005 when unstoppable force of the rising value of location and skill in certain cities ran into the immoveable object of housing markets in the modern urban developed world.  This period is associated with a migration of high incomes into those cities, low incomes into the rest of the country, and the capture of much of the additional income those cities create by real estate owners. (3) 2005 to 2015 when we crippled the housing market nationwide so that the same pressures remain in the high income cities, but now real estate owners across the country are capturing a higher proportion of incomes because access to ownership is now widely constrained.

In housing post #76, I included a number of graphs showing that migration patterns, housing affordability, and incomes for most of the country reflected what we might expect with classical assumptions and an open society.  But, the cities with housing constraint problems (mostly California and New York City) portray a limited access society, where limited housing fuels high incomes and high rent inflation, so that we see a strong out-migration from the cities with the highest incomes and costs.

Here are 3 graphs that show how this relationship has changed over time.  Rent affordability had not changed much between 1979 and 1995.  In these graphs, we can see the normal stable relationship in the rest of the country, where there may be a slight rise in housing costs in cities with migration inflows.  But, housing markets react fairly efficiently, new units are constructed for the new households without much change in costs.

In 1995, the problem cities where populations are being dislocated by housing constraints had slightly elevated costs, but were still relatively normal.  Then, through 2005 and 2015, we can see the ratcheting up of housing costs.  In these cities, the higher costs are leading to out-migration.

The national housing crisis has caused rent to rise nationally, but rents are rising across the board in the other cities.

I have posted this next graph before, but I want to walk through it again, to describe what I think we see in the data.  Most areas have relatively efficient housing markets, so that if population or incomes rise, the housing stock expands to accommodate the new demand for more valuable housing (in terms of rent value).  In cities characterized by relatively open economic and housing policies, there are two mediating factors here.  One is that housing stock expands to pull rents down to the comfortable range of 20-25% of income.  The second is that economic opportunity tends to be bid away as workers move into a city to take advantage of opportunities.  So households and builders react to rising income opportunities by increasing the supply of labor and housing, which moderates both income and rent.

Washington, D.C. happens to be a special case here, because it looks like it is squarely in the right hand section of this graph.  Rent is low as a proportion of income in Washington, D.C.  But, that second factor isn't in effect in Washington.  Incomes are very high.  In this case, the cause of high incomes is not housing.  It is (I presume) economic rents coming from the federal government.  The source of the high incomes is fairly immune to competitive pressures.

There are constraints to housing in Washington.  In addition to other typical zoning limitations, there are sharp height restrictions.  However, the suburbs have been growing robustly.  So, prime locations are scarce, but there is some housing available.  In these two graphs, we can see that the Washington MSA has decent population growth.  So the value of existing real estate rises as the city grows.  New households buy homes on the comfortable margin while existing households in prime locations see rising rents.  So, Washington has a unique signature - rent/income is in the comfortable zone, but rent inflation has been as high as anywhere.  So households in Washington can generally remain in the comfortable zone of housing expenses, and may choose to substitute location for size as their incomes rise, which, in the aggregate data, shows up as inflation.

In cities where housing constraints become more binding and where there aren't open-access suburbs capable to absorbing new population, households eventually reach a lower limit in real housing consumption, where they are unwilling to downsize any more in order to retain a comfortable level of housing expenses.  In these cities, rent/income ratios start to rise.  Some households will still be shifting to less valuable units (in terms of rent), so we see rent inflation in these cities that is mitigated partially by lower real housing consumption as a proportion of income.  Faced with this problem, households who prefer not to leave the city will accept higher housing costs to a point.  Since there is a limit to new population growth, residents capture the economic rents from the geographical advantages that have led to higher incomes.  The gains will tend to eventually accrue to real estate owners, as households bid up rents in an attempt to capture those higher incomes.  So, incomes after rent expenses will stagnate even as gross incomes rise.

Eventually, the constraint becomes so severe, as in Manhattan and Silicon Valley, that households hit the upper level of sustainable housing expenses (the left section of the supply/demand graph above).  At this point, marginal changes in the city's income and housing consumption will become a product of migrations.  Higher income households move into the city, bidding up rents so that some lower income household is forced to leave.  So, as incomes rise further, rent/income remains the same, but the new tenants have higher incomes and pay higher rents.  In these cities, median households may even see incomes after rent expenses falling even as gross incomes rise because the migration patterns will be pushing them further down the distribution of the city's household incomes in a ratcheting up of costs until they become the next family that leaves.  In these cities, real estate owners may be capturing all of the median household's new income plus more.

Of course, in many cases the household is the real estate owner, so the complete picture frequently includes families scraping by on a cash flow basis in California until they sell their home and move to a less expensive location where they are wealthy relative to the locals.

Here is a scatterplot comparing rent affordability and relative household incomes.  There are two series shown - one for metro areas in 1979 and one for metro areas in 2015.  In 1979, there is a big blob with no correlation.  Differences in rent affordability at that time reflect various local factors.

But we see two distinct changes when we look at 2015.  First, the entire blob moves up.  In other words, across the country, rent has taken a higher proportion of incomes.  That is because we have created a national shortage of houses, and households are spending above their comfort zone for housing just about everywhere.  Secondly, we see a few metro areas bursting far outside the norm in both income and rental affordability.  Where high incomes lead to high rents, Washington DC is the sort of outcome we might see.  High income households are extracting rents from the rest of the country, which is pushing local incomes higher, but those rents aren't related to limited access to housing.  They are related to limited access to politicians.

But, in most of those outlier cities, there is a relationship here that should be shocking.  As household income rises, households are spending a higher proportion of their income on housing.  My contention is that it is the constriction of new housing supply in these cities that is creating the higher incomes.  These cities need to have some local value with geographical barriers to competition.  And, I think the interesting foundation to this problem is that that value is the dense presence of highly skilled labor in industries where a connected and innovative labor network is the source of value.  All of this income should be conveying to the rest of the country and the world as consumer surplus.  But, artificial barriers to population density allow producers to capture some of that consumer surplus, much of which must be shared with local real estate owners.

That consumer surplus we were supposed to get from the internet?  Here it is.  The growing amount of the national income going to the financial industry, in spite of revolutionary reductions in trading costs and costs of diversification?  Here it is.  It's all in the pockets of urban real estate tycoons and suburban homeowners, thanks to renter advocacy groups, NIMBY protesters, and zoning boards.

This is why spending growth seems like a mirage coming out of the housing ATM.  All that consumer surplus is flowing into real estate.  The solution isn't to shut down the ATM.  The solution is to build homes.

More graphs coming in a follow up post.  I think I will show that all those rising incomes, and more, are going to rent.  The median households in those cities are worse off than the households in the open access cities, even with those higher incomes.  Much of the increase in income inequality we have been measuring is a false positive due to aggregating incomes in closed access, high cost areas with incomes in open access, low cost areas.


  1. So one thing your point leaves out is taxes.

    1) If you assume marginal tax rates at about a third, then a $30,000 increase in income should get you at most a $20,000 increase in rent when you're playing with nominal numbers.

    2) Partly because most of the bad regions are in CA/NY, and partly because progressive taxes, as income rises, takehome percentage should fall. So not only is the median renter in SF paying 40-50% of their income in rents, but the median household is probably paying a good 35-40% of their income in all various forms of taxes.

    Now 20% of SF is pretty good money, especially once you leave SF (aka that place where your beer has to pay the bartender's $4000/month rent), but it's still a point worth thinking about.

    /Of course, there's a counter point where people are getting away with paying 45% of their income in rent because 20% of SF is pretty good money in a way that 20% of Cleveland isn't, but.