More on Erik Hurst and FT Alphaville on Structural Unemployment

Sorry for my regular readers, but I’m going to keep hammering on structural unemployment issues at this blog.  I’m hearing rumors that Davos is going to be entirely about “The New Normal” and I want to be ready if that’s the case.  Structural unemployment is also doing a lot of the work of saying that the economy needs to bleed out the badness and be punished for the good times, a long-standing argument.   There’s definitely a problem with housing and foreclosures, but they only speak to a part of the cyclical crisis we are going through.

In response to an editorial we discussed, where Raghu Rajan cited the work of Erik Hurst saying that “structural unemployment may account for up to three percentage points of total unemployment. In other words, were it not for construction, the US unemployment rate would be 6.5% – a far healthier situation than today,” Greg Mankiw runs a clarification from Hurst:

I saw that you linked to Raghu’s writings which described some of my work in progress assessing labor market mismatch and its potential effects on current unemployment rates (joint with Kerwin Charles and Laura Pilossoph). There was, however, an error in Raghu’s assessment of our work. I am emailing Raghu as well. Raghu reported that we are finding that upwards of 3 percentage points of total U.S. unemployment can be explained by structural forces. That is not what we have found. Preliminary back of the envelop calculations suggest that upwards of 3 percentage points of the unemployment rate in high unemployment rate states like Nevada or Arizona may be due to structural forces – not 3 percentage points of total U.S. unemployment. The amount of total U.S. unemployment explained by structural forces will almost certainly be much less.

Excellent that Hurst responded so quickly. I like Rajan’s work, but this is really a bad thing to do, throw out a number that bold without checking in a major editorial. It was self-evidently wrong; A quick data glance finds 1.75 million unemployed construction workers unemployed in December 2010; a quick back-of-the-envelope tells us that if all of them found a job tomorrow that would only bring down unemployment 1.1%.

(And remember our “natural rates” of unemployment – If I’m pulling data right, the annual average unemployment rate of construction workers in 2005 was 726,000. So if we got the construction workers “structurally unemployed” instead of naturally searching it would be a lot less than that.)

In some ways Hurst’s number even feels low for Nevada.  3% unemployment for Nevada related to the housing bubble is very consistent with a report the IMF put out – The Great Recession and Structural Unemployment – which found find that structural unemployment is 1%-1.75% nationwide, with skills being 0.5%. That means housing hurdles run from 0.5% to 1.25% of unemployment.

Which was looking at data through end of 2009;  Nevada stands at 14% unemployment right now, a huge outlier when it comes to states.

And that paper don’t have a full grasp yet.  I think these issues are more complicated than straightforward matching of workers to jobs. Housing booms in states are correlated with underwater homeowners afterwards. Underwater homeowners are correlated with foreclosures. Foreclosures are correlated with state budget gaps, which were altered assuming increases in property tax revenue.

Those are four different stories:  Too many construction workers, people underwater lacking mobility, mass foreclosures with spillover effects and increased uncertainty about the housing sector, and anti-stimulus of state cutting their budgets.  The last is a demand phenomenon, not a matching one, and the third has to do with legal and investment mechanisms.

For some reason we can’t let this go as a narrative. Here’s John McDermott of FT Alphaville:

Back in the summer FT Alphaville’s Cardiff Garcia identified that the following reasons were being cited, by one economist or another, for the errant Beveridge curve:

1) A mismatch between the available jobs and the skills of the jobless
2) Reduced labour mobility because of homeowners stuck with underwater mortgages
3) The perceptions of employers towards the long-term unemployed
4) Jobless benefits that function as disincentives to working
5) The reluctance of employers to hire in the midst of a fragile economic recovery
6) Other stuff we haven’t thought of

It’s worth briefly noting that arguments for 2) have strengthened, while doubt is increasingly shown in regard to 4) — the disincentive affect of benefits. In a self-admittedly back of the envelope bit of research on Wednesday, Goldman Sachs analysts estimated that the unemployment rate is only 0.5 percentage points higher than it would’ve been without extended UI….At least another two reasons exist for why we’re reluctant to completely throw our lot in with cyclical proponents just yet.

Our economy has had a major demand shock and is going through the largest Recession since the Great Depression.  Cyclical drivers should be the default option here, not structural ones.

Mobility: We need to clarify that point 2. The link refers to a Washington Post article from Summer 2010 that states: “The labor migration rate is down sharply since the start of the economic downturn in 2007 and is just half the rate of a decade earlier, according to William H. Frey, a Brookings Institution demographer who has analyzed Internal Revenue Service and census data.”

In November 2010 Greg Kaplan and Sam Schulhofer-Wohl of the Minnesota Fed found that these results were the result of a bug introduced into the Census data in 2006 (“We show that the significant drop in the annual interstate migration rate between the 2005 and 2006 Current Population Surveys is a statistical artifact. The Census Bureau’s imputation procedure for dealing with missing data before the 2006 survey year inflated the estimated interstate migration rate…The 2007–2009 recession is not associated with any additional decrease in interstate migration relative to trend…”).  We cover that paper here.  So that shock to mobility didn’t happen, just a gradual decline.

It’s also not insane to think that people will strategically default to move large distances to go get a better job (or a job period). We ran a letter from a strategic defaulter here who left their house behind to build their career. It happens. It doesn’t happen in a huge wave because, remember, there’s not really places to move to. Every state has a higher rate of unemployment than it did before the Great Recession, roughly double across the board.

There are a lot of places you’d like to leave, but not many places you’d like to go.

Unemployment Insurance: Best evidence tells us that a little over 0.4% of unemployment is related to UI (616,000), though through multiplers it’s increasing employment (793,000 new jobs) more than it is increasing unemployment, so net job creation.  The duration of New Entrants, or people not getting UI, is very high, telling us that it’s not simply a UI story.

At this point there should be quantitative triggers if structural unemployment was tripping up our economy, and we don’t see them.  What we see is a patient who was shock paddled from almost dying into a coma, and needs another shock paddling to get them up and going again.