The Crisis in Deeply Underwater Mortgages, Unemployment Edition

Make sure to check out the Negative Equity Breakdown by Calculated Risk Blog.

I got a hold of the negative equity data by state and decided to plot the percent of owner-occupied households that were more than 50% underwater, so the incredibly deep underwater mortgages, against unemployment. Here is percent of mortgages underwater 50%+ underwater versus June 2010 U3 unemployment by state (click through for larger graph):

Now here’s the interesting part. I took the 1 year difference of U3 unemployment by state, so June 2010 minus June 2009, and plotted. A positive number here means unemployment is increasing over the past year:

We ran some back-of-the-envelope R modeling here and got an estimate of a homeowner making between 7 to 11 years worth of payments before he or she gets above water:

(Thank god FHA is going to penalize so-called strategic defaulters with a 7 year lockout on getting a loan! A real credible threat when they are going to be paying 9 years of worthless debt off by staying in the property at a 150 LTV.)

These estimates are in line with some professional analyst work I’ve seen on this question of how long communities will be underwater. Mind you, this is conditional on housing stabilizing and the Fed starting to hit some inflation targets, both dubious propositions these days.

So unemployment increased more in places with a lot of underwater mortgages in the past year. Question: As the worst mortgage debt in the highest unemployment states continues to be foreclosed on, and no mortgage relief, cramdowns, or inflation is inbound, how much will this become a spiraling problem? Foreclosures depress housing values making unemployment worse increasing foreclosures? The link between how a further local depression in housing values could increase unemployment needs some more causation analysis, but I think there’s a real, and worrisome, problem here.

How many more states will end up in a Nevada spiral before this is done?

UPDATE: Forgot to link to this Annie Lowrey piece: “If You Cannot Sell Your House, You Cannot Move.” The Post notes: “With many people locked in homes by underwater mortgages, only 1.6 percent of Americans moved between states in a one-year period that ended in March 2009 — a labor stagnation not seen in half a century.” How much is this a driver of unemployment?

UPDATE II, August 4th: The second graph, of year in changes, is mostly driven by the upper-right outlier of Nevada. Without it, the results are positive but not significant. I don’t see a relationship between deep underwaterness and change in unemployment when the outlier of Nevada is removed. The same holds for percent of all mortgages underwater with Nevada out.

This entry was posted in Uncategorized. Bookmark the permalink.

18 Responses to The Crisis in Deeply Underwater Mortgages, Unemployment Edition

  1. You know your evidence is consistent with a structural interpretation of the crisis, i.e. the “its not aggregate demand (so the Fed/Treasury can’t do anything)” view?

  2. Mike says:

    Forgot to link to the Lowrey/Post pieces, updated added at end.

    Will,

    Definitely, though the Fed and the Treasury can do tons about bad mortgage debt, from inflation to eminent domain to running a PR campaign to get cramdown to all kinds of things I don’t even know about yet. In some ways, it might be easier.

    And it’s obviously not an all or nothing issue. My question is how much is this the driver of persistently high unemployment?

  3. Oh I see, inflate away the bad debts. This would be great for those with underwater mortgages but I’m guessing it might not be efficient overall (or fair to the owners of bank assets).

  4. Mike says:

    I wouldn’t say inflate the debt away, but deflation is going to go be a double crunch if people are in a debtor’s prison called “the home they live in.”

  5. “if people are in a debtor’s prison”

    *Some* people; the rest of us, not. The point is that your set of policies have nothing to do with the Fed’s mandate and little to do with the AD debate. If you want the rest of us to help out these people with underwater mortgages, make an argument for why we should help them (and, given there’s trade-offs, not others), i.e. make a fairness argument.

    Personally, my “help people that are suffering” budget would be better spent on the poor directly via lump sum transfers. As an extra special AD bonus (and as a tie-in to your discussion of liquidity constraints), these lump sums would vary with the business cycle.

  6. chris says:

    How much is this a driver of unemployment?

    ISTM that this could be measured fairly straightforwardly: how many job openings are going unfilled for lack of qualified applicants? Of those, how many are jobs where a qualified applicant exists, but is located somewhere else and can’t relocate?

    Since ATM the answer to the first question is “damn near zero”, I don’t think immobility is driving unemployment to any great extent. While it might benefit an individual to be able to move somewhere the local economy is less crappy, that would just make them the 50th applicant for a job in State X instead of the 250th in State Y — even if they actually get the job and therefore personally benefit, it’s only by displacing someone else and the macroeconomic picture doesn’t change.

    Immobility can’t be the cause of the number of job openings being much less than the number of people looking for jobs. Immobility-driven unemployment would necessarily be characterized by excess job-seekers in one place *and* excess unfilled jobs in another, and the latter doesn’t presently exist.

  7. Mike says:

    Chris, that’s a good point. We are at like 5+ job seekers per job opening. I’ll see if I can find that data cross-sectionally. This spiral effect will impact some specific states with other conditions (notably Nevada), but it isn’t driving the lack of aggregate demand; I’m surprised it isn’t spurring the Fed to try and hit their inflation targets moreso. As Yglesias suggested today, to not do what they are supposed to doubles the problem.

    • John Thacker says:

      “This spiral effect will impact some specific states with other conditions (notably Nevada), but it isn’t driving the lack of aggregate demand;”

      If you’re positing Nevada having specific conditions that distinguish it from the great mass of states, then your second graph, far from being “the interesting part,” is practically meaningless. Remove the outlier Nevada that drives nearly the entire least squares model (as any single point with independent variable very different from the rest of the points will do in a least squares model), and the slope nearly disappears.

  8. Michael Turner says:

    I think the structural/non-structural debate is a bit sterile. It’s more a geographic thing, and I believe your correlations would only tighten sharply if you could get data at the level of county (or multi-county metro area) rather than state.

    After all, the housing bubble itself was fairly geographically structured. OK, there might have been some spillovers into manufacturing-intensive regions. Perhaps a company making granite counter-tops would tend to be located closer to a quarry than to any booming destination market. But I bet if you look at how manufacturing and business services have declined (sharply, almost as much as in construction), you’ll see strong regional correlations as well between deep-underwater (can we call them Deep Water?) mortgages and unemployment rates.

    Wouldn’t it make perfect sense anyway? Realtors running around town will want more FedexKinko outlets; construction workers running around will want more fast-food places for quick bites; local lumberyards will staff up, even expand, and so on, rippling through the local economy. None of these businesses are construction or finance per se, but they would be sources of new jobs created on the back of local speculative demand for housing.

    IIRC, during the bubble, there were towns in Florida where almost every new job was directly related to either the local expansion of housing or the local increase in house prices. How refreshing it must have seemed to be making real things happen on real estate. Not like that dot-com B.S., eh?

  9. Pingback: Links: 2010-08-04 | Credit Writedowns

  10. Joel says:

    The correlation on your second chart is very weak. In stats we were always told to be wary of a single data point with an overly strong influence on a regression. If you take out the outlier (Michigan?) and re-do the regression, I expect it would be close to a horizontal line and not at all supportive of the relationship you suggest.

  11. John Thacker says:

    What’s the model and correlation without that one big outlier on the upper right (Nevada)? Looks to me like the second graph would have a near zero slope and poor R^2 without Nevada.

  12. Mike says:

    Joel & John,

    Yeah I’m backing away from the second graph – it’s mostly driven by Nevada (the outlier), I thought I had checked it with the outlier removed but hadn’t (had actually checked the first graph twice with the outlier removed, which is very significant). Still positive slope, but not very significant.

    So to update, I don’t see a relationship between deep underwaterness and change in unemployment when the outlier of Nevada is removed. The same holds for percent of mortgages underwater with Nevada out.

    • John Thacker says:

      Thanks for the update! I appreciate your quick response.

      I’m a professional mathematician (a probabilist by training), and the second graph threw up some caution flags. It looks like a fairly textbook example of when to be cautious. One data point with a very different value on the independent variable exerts a lot of leverage in your basic least squares regression.

      Of course, it’s not to say that when a state gets in a true “death spiral” like Nevada that the two factors don’t influence each other, just that we’d like some more evidence with other points at the bad end of underwater (well, in reality we hope that other states don’t get in that pattern!) Though in that case perhaps the effect would be non-linear, since we don’t see too much of an effect in the 49 states with more normal values.

  13. Anonymous coward says:

    You know, my undergrad physics lab course instructors used to call such graphs “starry sky”. Such experimental data was never accepted, you had to redo the measurements. Kill the single outlier point in each graph, and even the pretense of a relationship vanishes. Even so, declaring a correlation at an R value of 0.4 is some serious sticking out of the neck ^^

  14. Pingback: IMF: Large Majority of “Structural Unemployment” is Foreclosure Crisis, Underwater Mortgages « Rortybomb

  15. Pingback: FT Alphaville » Underwater mortgages and US housing

  16. Pingback: FT Alphaville » Negative equity still declining, for now

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s