Tuesday, November 10, 2015

Housing, A Series: Part 81 - The Cost of Access

Following up on yesterday's post, I want to look at the shape of incomes over time in various metro areas some more.  I am using data from the top 97 metro areas that have data for this time frame in the Zillow database.

One of the interesting aspects of recent increases in income inequality is that the source of changing distribution of household incomes seems to be a similar change in the distribution of firm incomes.  Also, as I have found, and as Matt Rognlie found with much more detail and sophistication, the coincidental shift downward in labor incomes was largely a shift to real estate income, not corporate income.  Recently, researchers who have been focused on income inequality have also been coming to the conclusion that it is mostly a shift within labor income, not so much an increase in the share of income going to employers.

I find that income by city shares this same trend of increased variance.  A few cities at the top of the distribution have been exhibiting income growth unrelated to the rest of the country.  I believe that all of these findings tie together to point to one source of the problem - housing constraints in our most productive cities.  This leads to a high level of income for firms and workers in these cities and a transfer of income from laborers to real estate owners.  The localized housing constraint is a cause that can explain all of these patterns.

Here is a graph comparing 97 metro areas, ordered by median household income as a proportion of the national average, in 1979 and 1995.  The next graph shows the same cities, in the same order, but in that graph, the y-axis is the median household income after rent expenses.  There was not a housing constraint problem during this period.  There was a slight rise in the variance of median household incomes between cities.  So, the highest income cities saw incomes rise slightly more than other cities.  And, as the next graph shows, incomes after rent followed the same pattern.  Since these cities are ordered by gross income, the second graph is a little messy.  I have fitted 3rd degree polynomial lines to the distributions to get a cleaner picture of the general trend, and we can see from those lines that the shift in incomes after rent mimics the shift in incomes before rent.  Households in more productive cities were retaining those income gains, but the income gains were slight.

Next, we move to the same two graphs, except this is for the period 1995-2015.  Here we see a sharper divergence of the top cities.  Income inequality between cities increased more during this period than it had before, but that increase is concentrated in the few highest income cities.  However, this is reversed when we look at incomes after rent expenses.  The top two cities fared well (San Jose and especially Washington, DC), but for the other 95 metro areas, the change in income distribution after rent is the reverse of the change before rent.  The poorest cities actually gained on the average, and high income cities lost.  In other words, workers in the high income cities were paying all of their income gains, and more, to their landlords (which in many cases are themselves).

Households in San Jose retained some of their additional income because there was still some positive relative population growth there in the 1990s, allowing some relief in the real estate bidding war.  The new housing stock would have tended to supply very high income households who were moving into the area.  I expect San Jose to revert to the pattern of the other productive cities as the city matures and succumbs to political obstacles to growth in its housing stock.  The economic rents captured by the median household in the Washington, D.C. area are retained by the household because the economic rents there are not caused by the housing constraint; they are created by other factors, presumably largesse related to the federal bureaucracy.  I'm sure, according to the theory of efficiency wages, this will lead employees of Washington's main industries to work even more diligently to keep supplying the rest of us with the blessed fruits of their labors.


Here is a graph that combines the graphs above for 1995 to 2015.  I have labeled the 2015 Income after Rent points of the problem cities.  Except for Washington and San Jose, what we see is that during this period, household median income among most cities didn't change much, either before or after rent.  The strange behavior in Income after Rent comes from these few problem cities.  In New York City and the California cities, housing costs were high in 1995, and they increased so sharply from 1995 to 2015 that Income after Rent in these cities dropped by 7% to 14% of the national average over those twenty years. (We can see several cities that had unusually low incomes after rent in 1995 which recovered by 2015 - the dips in the light blue line.  These cities were characterized by high housing costs that declined over this time.  They tend to be secondary cities in California and New England.  I am not sure how that peculiar signature relates to the rest of the story.)

Residents of those cities really are experiencing a different economy than the rest of us.  They really do live in cities where middle class households have experienced two decades of decline.  This decline now dominates our national political conversation, but it has little to do with national policy.

Here are a few more graphs that may help to highlight the pattern.  The first is a graph (in current dollars) of the annualized growth in median income, rental expense, and income after rent for the US and the closed access cities.  We can see here again how high gross income growth did not translate into high income growth after rental expense.  The US median annualized income growth during the period was about 3% before and 2% after rent.  New York and San Francisco saw about 5% growth before and about 1 1/2% growth after rent.  These are nominal dollars, so the median families in those cities have seen falling real incomes, after rent, for 20 years!  And, it is worse than that.  Since this problem creates a steady migration flow of low income households out of these cities and high income households into these cities, a household over time that remains in the city will be moving down the distribution over time.  The household with income higher than 60% of the other households in New York or San Francisco in 1995 was probably below the median by 2015.  In other words, the income growth of the median household may overstate the experiences of the typical households.  (This may be mitigated by other factors like lifecycle income patterns.)

Here are some other graphs.  I was hoping that I could include real-vs-inflationary effects in these graphs, but there is just too much of a difference between aggregate national expenditures and median expenditures to do that easily.  I think it would be a useful exercise, but it would require a lot of work to estimate inflation specific to the median household.  (Here is an interesting recent paper <HT: John Wake> on this topic that deals with the compositional effects of rent inflation, since new housing tends to be higher value housing.  I think the composition of households within cities also has an effect on measured rent inflation.  But, to be honest, I haven't been able to wrap my head around this issue.)

The first graph is in nominal terms, for the US from 1995 to 2005 and 2005 to 2015.  The x-axis is the growth in the nominal median income.  The y-axis is the growth in nominal median income after rent expense.  The idea here is that, the square indicator represents the potential growth in income after expense if households do not increase rent expense at all.  In a static context where households maintain a constant proportion of housing expenditures as incomes grow, we would expect rent expense to grow along with incomes so that income after rent falls near the 45 degree line.

We can see that this pattern held for the aggregate national data from 1995-2005.  Rent inflation was high during this period, so contrary to conventional wisdom, households were not increasing their real consumption of housing.  They were not moving into more valuable houses (by rent).  They were living in homes that, on average, were less valuable (by rent) in real terms, but rents were rising faster than the pace of inflation.  But, in nominal terms, housing expenditures were rising in step with incomes.

From 2005-2015, first, we can see that income growth has been very low because of the Great Recession.  But, in addition to that problem, the collapse of the US housing market has led to rising rents nationwide, so that income growth after rent is not even keeping pace with the low level of gross income growth.

The next graph looks at the closed access cities in terms of growth relative to the US median household for the entire 1995-2015 period.  In other words, this is how the median household is faring in these cities compared to the median US household that already has income after rent that is lagging gross income.  The measures in this graph are the change in median income relative to the US.  For instance, San Francisco's median income grew from 137% of the US median to 153% from 1995 to 2015 - excess growth of 11.2% (153/137).

As with the previous graph, the top square is the rate of growth in income after rent these median households would have received if they did not increase their rent expenditures, relative to the US average, over this time.  As above, in a world with stable proportional spending, we would expect households to increase their housing expenses, on average, at a rate similar to income growth.  So, in an unconstrained world, the circles would fall near the 45 degree line.  In other words, we would expect incomes after rent expenses to grow at a pace similar to gross incomes.

Looking at Washington, DC, as I did in the previous post, we have an exogenous influence on incomes that has created tremendous income growth there relative to the rest of the country.  And, here, Washington looks like an open access city, with income before and after rent growing at similar rates.  But, in Washington, this is practically all inflationary.  Households are bidding up housing stock in prime locations with their new higher incomes.  Washington is a special case where, because the income growth is unrelated to housing, the higher incomes are leading to higher rents in the closed access parts of the metropolitan area.


Boston and San Jose are closed access cities that haven't quite reached the tipping point where all income gains go to real estate owners.  The median households in those cities retained a small amount of their income gains.  In San Jose, incomes rose 23% faster than in the rest of the US, but the median household only retained 6% extra income growth after rent expense.  The migration patterns are especially strong in San Francisco and San Jose, so much of the income growth there is probably compositional - high income households moving in and low income households moving out.  So, San Jose is realistically probably in the category with San Francisco, LA, and New York City as experienced by households living there.

Those cities are in a category where housing is so constrained that real estate owners are capturing all of the relative income gains plus additional income, leaving households poorer after housing expenses, even if they have higher gross incomes.  In these cities, I think this evidence points to a reversal of causation compared to Washington, DC.  The lack of housing is pushing up costs.  As long as local industries are capable of retaining some economic rents as a result of the local closed access policies and firms in other locations are unable to compete away those rents, then much of those excess profits must flow to labor to entice workers to the closed access city, and those extra wages then flow on to the landlord.

This is why income inequality among workers appears to be highly correlated with income inequality at the firms they work for.  They are both capturing excess profits from the same phenomenon.  The source of monopoly profits are the workers themselves, who are protected from competition because there is no housing for additional workers.  Firms who can get exposure to these closed access cities share these excess profits along with the workers.  However, since there aren't nearly as many political constraints to commercial real estate as there are to residential real estate, and since shelter expenses are a smaller portion of firm expenses, the firms are able to retain much of their excess profits.

Among the workers, it looks to me like there is actually more income inequality than aggregate measures tend to show, because low income households at the threshold of sustainable budgets will send all of their marginal income to the landlord.  Higher income workers may be able to retain some of their new income if their income is high enough that they can stand to reduce their real housing expenditures to a more comfortable portion of their budget.

This is one way to view the migration patterns - when a poor household moves out of a closed access city and a high income household moves in, the rent of the unit becomes a smaller proportion of the new household's budget.  The closed access policies won't allow rents to decline relative to incomes, but something has to give, and the migration pattern accomplishes this by pushing up local incomes.  This is why it looks to locals like high income in-migration is the cause of the crisis.  And, the migration pattern is the reason that relative incomes after rent are falling, because the inflow of high-productivity workers to the profitable local industries allows those firms to increase their productive activity and capture more economic profits from the rest of the world, which then funnel through the new workers to the local real estate owners.

I think this migration pattern has much to do with the extreme income growth at the top of the income distribution, too.  These closed access cities are much more valuable to high income workers, and the migration pattern we see is a reflection of that value.  Highly productive workers who can benefit from these valuable professional networks in tech. and finance can especially increase their incomes as a result of the closed access, and they spend a much smaller portion of those incomes on housing.

The irony is that the best policy we could institute for reducing all these forms of economic inequality between workers and capital is to build hundreds of thousands of high end housing units in the core areas of San Francisco, San Jose, and New York City.  In fact, this is the only policy that can accomplish that task.

5 comments:

  1. Great one for you: in 1979 you could rent an office in downtown Los Angeles for under $3 a square foot per month. You still can.

    Another one I just got from Fred St Louis: there has been no inflation in automobile prices since 1997.

    I think you hit it: in many regards, the economy has bifurcated, into housing and everything else

    ReplyDelete
    Replies
    1. Interesting.

      Yep. Well, education, health care, and finance have their problems, but they are more diffused so they are harder to measure. The thing about housing is that the problem is localized, so that the higher costs imposed by regulatory constraints can be measured and compared. It's too bad the other problems are so broadly applied. They are probably doing at least as much damage, but it's not as easy to measure.

      Delete
  2. if health care costs fell under an NGDP targeting regime while everything else continued on trend, would that entail looser monetary policy?

    some states seem to b piloting allowance of Nurse Practitioners as more complete substitutes for doctors, and there are some new active measures for demand repression by the feds and states. I look at this in a fairly Soviet "how many bushels??" kind of way, I'm afraid - and so receptive to competing views - but I don't think most medical care does much for people, so I'm cautiously enthusiastic about efforts to shrink its incidence. a certain portion of medical action - vaccines, antibiotics, antiretrovirals, some other drugs - gets a lot of bang for the buck, further gains get progressively more painful and expensive

    ReplyDelete
    Replies
    1. I'm with you on your general POV on health care, if I understand what you're saying. On monetary policy, I'd say there are two ways to think about it. One way is to use the Fed's stated target. Above target is loose below is tight. In a more general sense, I think there is a general range of monetary expansion that is functional, and mp is loose or tight if we are outside that range. A lot of people think we had a housing bubble that was due to monetary expansion, so they refer to 2000s mp as loose. I would call 1970s policy loose because it seems clear that high inflation was causing economic dislocation. And I call mp in 2006-present too tight, regardless of inflation or ngdp because the bond markets and real estate markets are prevented from reaching functional equilibrium prices in ways that would be cured with higher inflation.

      Delete
  3. early 2000s monetary policy was "too loose" because the Fed said so. the Fed chair at the time of our tightening, and during the Great Recession, was a leading scholar on the question of printing's efficacy, and on tightening's dangers. the lady who says we should probably tighten soon is a Berkeley liberal whose political party faces reelection in the coming year. with doves like these..

    I wonder if structural reductions in cost growth in component sectors of the economy would/will reduce the likelihood of false positives. if Sumner, Krugman and Romer can't get FOMC seats, maybe structural adjustment gives the central banks more breathing room for monetary offset, just like fiscal cuts

    ReplyDelete