Ed Dolan's Econ Blog

As Jobs Surge, Term Structure of Unemployment Remains Distorted

The US economy added 321,000 payroll jobs in November, the best in almost three years. Strong upward revisions to September and October numbers boosted the 12-month gain to 2,756,000, a new high for the recovery.

The official unemployment rate was unchanged at 5.8 percent. The broad unemployment rate, U-6, which takes into account discouraged workers and involuntary part-time workers, fell to a new low of 11.4 percent.

Although these data indicate a return to normal in many respects, distortions remain. One of the most conspicuous is an elevated rate of long-term unemployment. As the following chart shows, the term structure of unemployment remains substantially different from the prerecession pattern:

On the one hand, we see that the short-term unemployment rate, made up of people who are out of work for four weeks or less, has not only returned to its pre-recession level, but has actually dropped below it. In 2007, short-term unemployment averaged 1.66 percent of the labor force; now it is 1.61 percent. Most unemployment in this category is voluntary, representing a minimum time needed for job search, interviews, moving, and perhaps a quick vacation between jobs. It includes people who find work soon after entering or reentering the labor force or find a new job quickly after leaving a former one.

At the other end of the spectrum, we find people who are out of work for 27 weeks or longer at a stretch, some much longer than that. Most of the painful stories of labor market distress come from this category. It includes young college graduates who look for months without finding work in their field of study. It also includes people who have been laid off after years of steady work and can find nothing acceptable to replace it. The chart shows that long-term unemployment has fallen significantly from its peak, but at 1.8 percent of the labor force, it remains far above its 2007 average of 0.8 percent.

Some observers have attributed the high rates of long-term unemployment to the availability of extended unemployment benefits, that is, benefits beyond the 26-week norm that prevailed in most states before the recession. A recent study from the Boston Fed concluded that there was some truth to that hypothesis. However, the study did not support the popular idea of deadbeats who only pretend to look while their benefits lasted, and then, when benefits finally run out, quickly take a job that they could have taken at any time. Instead, the study found that the most common pattern for the long-term unemployed was to draw benefits as long as they could, and then withdraw from the labor force entirely without ever finding work.

Evidently, though, other factors are at work as well, since extended unemployment benefits have been substantially phased out since 2012 without bringing the long-term rate back to its pre-recession level. Other suggested influences include technological change, globalization, and demographic factors. It is also worth noting that the rise in long-term unemployment is not entirely new to the Great Recession that began in 2008. We can see from the chart that even after the short, shallow recession of March to November 2001, the long-term rate failed to return to the level of less than 0.5 percent that it reached in 2000.

Clearly, there are many unanswered questions here. I will take a closer look at some of them in future posts.

15 Responses to “As Jobs Surge, Term Structure of Unemployment Remains Distorted”

BryanWillmanDecember 6th, 2014 at 2:27 pm

I have now seen several (possibly overlapping) reports on various web discussion forums of a phenomenon where some UI program ran out, and a pool of people suddenly appeared looking for work. And where found to be utterly unemployable by anybody within sight.

In other words, some fraction of the long term unemployed may have been "no longer competitive as suppliers of market labor" the day they were laid off, but kept searching least enough to maintain benefits as long those lasted, and when they ended would quite likely have tried a variety of things (that didn't work) before looking for a sustainable exit.

For that fraction the issue isn't the length of UI (or other) benefits, but rather, they were already unemployable, finally got shaken out in the recession, and virtually certainly will not find sustained market employment again. Whether it was 26 weeks or 99 weeks or 3 weeks, they were and are done.

One wonders why so many people surfaced all at once with the recession…

Ed Dolan EdDolanDecember 6th, 2014 at 3:08 pm

One wonders why so many people surfaced all at once with the recession…

Because it had been a long time since the last severe recession?
Because of aging baby boomers, lots of people within a few years of retirement?
Because of accelerating technological change or globalization?

Take your pick or mix and match

BenL8December 6th, 2014 at 10:40 pm

The employment to population, E/P, ratio for age 25 to 54 dropped from 80% to 75% in 2 years, 2008 to 2010, and from almost 82% in 2001, the peak of this ratio, a 7 point drop. Today it's at 77%, and I call it a 40% recovery after 5 and a half years. Or is it a 28% recovery after 14 years of a jobless recovery/major recession?
The total population E/P dropped from 63.0% in 2007 to 58.2% in 2011, a fall of 4.8%, and today that would represent 11.9 million lost jobs. Total population E/P is up from 58.2% to now 59.2%, a decrease 3.8%. After falling 4.8% we have recovered 1%. We have a 21% recovery after 5 and a half years. But many claim that the demographics caused so many workers to drop out between 2007 and 2011. I have seen reports from the CBO and OECD saying that half of the drop in LFPR and E/P was due to demographics, both aging baby boomers and too few 25 year olds aging into the 25 to 54 year old group. I've seen two reports from the Fed. saying 1/4 of the drop is from demographics, and the takes the middle at 1/3rd. In any of the cases, we haven't recovered, just 40% recovery of the age 25 to 54 group. And a 21% recovery for the total population, unless you factor in demographics, which is probably a major fraction of the cause.
Let's look at private sector employment — from Jan 2000 to Nov 2014 up 7.7 million, a growth rate of
43,300 a month, or a 7% increase over 14 plus years, nearly 15 years, about half a percent per year. 36 million people joined the working age population, and only 7.7 million found work. This says that 21% found work, when the normal rate is that about 60% would find work. Go figure. I wrote a longer more tortured, if possible, version at my blog,

__Tom__December 7th, 2014 at 1:59 am

I think the Boston Fed paper's conclusions were no surprise.

If you've lost your job and you either don't want to return to work or can't find a job that would be worth returning to work for, it makes perfect sense that you would exhaust your benefits before leaving the labor force.

But not many people are really going to sit and collect benefits right up till they're exhausted and then quickly get a job. That strategy only makes sense if you:
– Can't find a high-paying job (people who can don't wait till they've exhausted benefits)
– Prefer collecting unemployment benefits to working a low-paying job
– Prefer working a low-paying job to leaving the labor force
– Can easily find and are confident of their ability to find a low-paying job even after a long spell of unemployment

Granted people who meet all four of those criteria exist, but they're not typical.

Ed Dolan EdDolanDecember 7th, 2014 at 10:38 am

Certainly, you can focus on the weak spots in the jobs picture, or in the strong spots. Both are there. I would like to add a couple of specific comments:

(1) You are right that economists differ considerably in their estimates of how the changes various indicators are due to demographics. In my view, that is because for each demographic group, economic factors are all at work. For example, other things being equal, growth of the >65 group tends to lower the employment population ratio, yet the same time, because of better health and because low interest rates make it harder to finance retirement, the participation rate for that demographic is going up. For another example, a big part of the drop in participation rates for the <25 group is that more young people are staying in school, but what is driving that? Is it the realization that more education improves lifetime earnings, or is it the feeling that with no jobs available, you might as well hang out on campus?

(2) When you look at the very slow pace of the recovery up to the most recent period (43k per month as you point out), you have to place a big part of the blame on the climate of procyclical fiscal austerity that has prevailed since the expiration of the early Bush and Obama stimulus packages. I have argued again and again that with common-sense fiscal policy that is balanced over the cycle but countercyclical during booms and recessions, we would have long since been done with this recovery and would not even be using a term like "Great Recession" to describe it.

__Tom__December 8th, 2014 at 5:09 pm

One thing worth noticing is a comparison of November 2014 and 2013

Nov. 2013: 522,000 net new jobs before seasonal adjustment, 284,000 after
Nov. 2014: 497,000 net new jobs before seasonal adjustment, 321,000 after

I'm not fussed about whatever the reasons were for this year's seasonal adjustment being so much smaller (I noticed there was one more weekend day this November). And I don't mean to take anything away from what has been a quite strong year for job growth. My point is that, on the loosey goosey scale of monthly jobs numbers, this November was actually about the same as last year's November, and not really the indicator of possible trend-breaking that it's being made out to be.

Ed Dolan EdDolanDecember 9th, 2014 at 8:21 am

One month's numbers never tell you much except what happened that month, and that is not even accurate for the unrevised first version of payroll numbers, which are subject to substantial revisions in additional to the seasonal adjustments you point to. The more impressive figure in the recent report is the 12-month job gain that is on a steady upward trend.

__Tom__December 9th, 2014 at 1:21 pm

But actually, it was only the seasonally adjusted 12-month job gain that rose. The not seasonally adjusted 12-month job gain fell by 25,000 jobs (522 – 497, see above). Given that year-on-year numbers shouldn’t have any seasonal fluctuation component, the big jump in the seasonally adjusted number is odd. I’m thus even more inclined than usual with such jumps to assume it is noise.

Again I don’t mean to take anything away from the pace of ~2.8m net new jobs per 12 months, which is quite good. I’m just skeptical of the November number as an indication of acceleration upward from that pace.

Ed Dolan EdDolanDecember 10th, 2014 at 6:57 am

No, we are not talking about the same thing. You need to look at Table B-1 of the the BLS press release ( ). When you do so, you can compare Nov 2014 to Nov 2013 for both seasonally adjusted and seasonally not adjusted data. Both increased. It is true that the seasonally adjusted number increased more, but both went up.

I believe that the reason the 12-month figures do not totally wash out the seasonal adjustment is that the industry mix of jobs changes over time, and different sectors get different seasonal adjustments. For example, notice that NSA construction jobs went down from Oct to Nov, as in normal in the season, but SA construction jobs went up, because the decrease in NSA was less than usual. On the other hand, NSA retail jobs went up by more than 400,000, but SA retail jobs only by 50,000, because part of the increase in retail jobs is purely seasonal.

__Tom__December 11th, 2014 at 5:12 am

Hi Ed, well we're hashing out a very minor technical point, but you've confused me so now I'm curious what you mean. When you say the Nov 14 vs Nov 13 NSA net job gain "went up," relative to what?

I'm not going to guess what exactly the reason is for the smaller 2014 seasonal adjustment versus 2013. I understand that different sectors get different seasonal adjustments but I don't know why any would get a bigger adjustment in 13 than 14. The one suspected factor I mentioned already is that Nov 14 had an extra weekend day vs Nov 13. My other vague guess is that the BLS are using some kind of algorithm that adjusts the size of the seasonal adjustment from year to year, but I don't know why that would increase this year's adjustment. As I said I'm not fussed about it. Seasonal adjustment is an extremely tricky if not impossible thing to do consistently well. I generally prefer NSA year-over-year, which is why I looked at those.

Ed Dolan EdDolanDecember 11th, 2014 at 10:49 am

You ask: " "went up," relative to what? "

The 12-month job gain means the difference between the Nov. 2014 total jobs and 2013 total jobs. For SA, table B-1 shows this as 140,045-137,311=2714. Is that the question?

__Tom__December 11th, 2014 at 9:10 pm

Ah, so you merely meant the Nov 14 SA and NSA levels were higher than the Nov 13 SA and NSA levels? Ok, no argument there. The other thing I don't question is that the revisions of months before November improved the Y-o-Y count.