David R. Henderson  


Quickly Admitting You're Wrong... So you want my opinion?...


As a former senior economist at the Council of Economic Advisers, I get in the mail near the start of every month the Council's report "Economic Indicators," prepared for the Joint Economic Committee. I like old-fashioned hard copy, but it's available to the public on line as a pdf.

It's worth perusing once in a while, if only to check the data against your impressions. In my perusal of the latest report, published on April 6, what I found most interesting were the data on unemployment.

Some highlights for the March data:

Unemployment rate for blacks or African Americans: 6.9 percent.
Unemployment rate for whites: 3.6 percent.
Unemployment rate for Asians: 3.1 percent.
Unemployment rate for Hispanics or Latino ethnicity: 5.1 percent.
Unemployment rate for men 20 years and over: 3.7 percent.
Unemployment rate for women 20 years and over: 3.7 percent.
Unemployment rate for both sexes, 16-19 years: 13.5 percent.
Unemployment rate for married men, spouse present: 2.1 percent.
Unemployment rate for women who maintain families: 5.6 percent.

These are all strikingly low for the 21st century, even the teenage unemployment rate that, while high, is way lower than it has been in almost all months in the last 10 years. I put in italics the one I found most striking.

Also interesting was the duration of unemployment spells.

The average duration, while having come down from its 10-year peak of 39.4 weeks in 2012, is still high, at 24.1 weeks. Part of this decline is due to the growing economy, but I suspect that most of the decline since then is due to the expiration of the long-term unemployment insurance benefits, early in Obama's second term, which gave people in some high-unemployment states benefits for up to 99 weeks, up from the normal 26 weeks. One indicator that the extension of unemployment benefits is one of the main causes is the difference between the number of people receiving unemployment benefits from state programs and the number receiving unemployment benefits from all government programs, including the federal extended benefit program.

The number of people receiving insurance benefits from state programs fell from 3.3 million in 2012 (this is the weekly average for the year) to 1.9 million in 2017. (It was up to 2.6 million in January 2018, but my guess is that much of that increase is due to layoffs of people who were hired for the Christmas season. I will be surprised if the average for 2018 turns out to be much higher than 2.0 million.) But take a look at the number of people receiving unemployment insurance benefits from all programs, including the federal extended benefit program. It fell from 6.0 million in 2012 (again, a weekly average) to 4.6 million in 2013, and 2.7 million in 2014. (The extended benefit program ended at the end of 2013, but in some states, depending on their state unemployment rate, the extended benefits were not relevant before then.) That suggests a strong role of extended benefits in causing a high average duration of unemployment spells. By 2017 and January 2018, it was virtually identical to the state number.

Other confirming evidence of the role of extended unemployment benefits is from the percent of duration spells of 27 weeks or more. It was a whopping 41.1 percent in 2012, but fell to 33.5 percent by 2014 and 28.1 percent in 2015.

The median duration is down to its pretty normal 9.1 weeks.

Note that there are two ways an unemployment spell can end: (1) the person gets a job or (2) the person drops out of the labor force.

I posted back in 2011 about the effect of extended benefits on unemployment.

Comments and Sharing

CATEGORIES: Labor Market

COMMENTS (13 to date)
Kevin Erdmann writes:

There has been a longstanding relationship between unemployment durations under 26 weeks and durations over 26 weeks. In other words, you could predict how quickly long term unemployed workers would return to work if you knew how quickly their cohort had been returning to work in the first 26 weeks of unemployment.

That relationship broke down after 2008. Here is a graph showing the actual duration of unemployment for long term unemployed compared to the predicted duration over time.

As you note, this is probably a combination of slower employment re-entry and lower rates of dropping out of the labor force.

Philo writes:

I see no italics.

John Fembup writes:

Philo: unemployment rate for married men, spouse present 2.1%

David R Henderson writes:

@John Fembup,
@Kevin Erdmann,
Here is a graph showing the actual duration of unemployment for long term unemployed compared to the predicted duration over time.
Thanks. What is the prediction based on?

john hare writes:

To me the official unemployment rate is a political construct. I think participation rate is more honest as suggested in your option 2.

Kevin Erdmann writes:

By using the counts of unemployment for the time groups under 26 weeks duration, I estimated the average duration of unemployment for those unemployed less than 26 weeks.

Then, I used that figure and the published figure of average duration for all unemployed workers to estimate the average duration of those unemployed for more than 26 weeks.

Until 2008, a simple regression of the average duration under 26 weeks (lagged 2 years) gave a pretty good estimate of the average duration of workers over 26 weeks. But, after 2008, the actual duration of workers unemployed for more than 26 weeks was much higher than the average duration of those under 26 weeks would have predicted. This suggests that in 2008 rates of exit shifted down at the 26 week mark compared to earlier times.

Philo writes:

@ John Fembup:

Thanks, my eyes weren't that sharp. Maybe it should have been put in bold.

RPLong writes:

Kevin, has the new trend persisted? To David's point, did the expiration of the extended benefits program set the trend back to its pre-2008 relationship with short-term unemployment?

Kevin Erdmann writes:


I haven't looked at it for a while. After the end of EUI, it seemed as though there were two groups of long term unemployed workers: (1) those who were then returning at fairly normal rates, and (2) a contingent of very long term unemployed who were at durations of 100 weeks or more, but were still identifying as unemployed rather than out of the work force. They presumably used up their EUI long ago, but something caused this to persist. I don't think anything like that has happened before. That group slowly declined over time. I think the over 26 weeks group may still be slightly oversized, compared to past patterns, maybe by 0.1% or 0.2% of the labor force.

Vic Volpe writes:


Note the late '60s.

James E. Forrest writes:

Love your analyses, quick question: as I recall, the actual employment rate is around 3 points lower (62% vs 65%?) than historical levels. My question is this: what are the people who are not participating in the economy doing, and how are they supporting themselves? How do they live?

Colin writes:

It would be interesting to see the "Married men, spouse present" underemployment rate over time. It looks like the unemployment rate is generally around 2-3 percentage points lower than the "All civilian workers" rate; in 2017, they were 2.4 and 4.4 percent while in 2009, they were 9.3 and 6.6 percent. Knowing the underemployment rate among both groups at these times would be useful to parse out the relative magnitudes of two prominent effects that come to mind: it is easier for a married man to seek (and thus find) a new job if he has a spouse's support for him to invest the time into this and that married men who are the primary breadwinners for the family may feel more pressure to take a job so they can provide for their family, which could mean taking a job they are overqualified for. Anybody know where this information could be found or have other reasons why married men's unemployment rates would be so low?

Anisha Meka writes:

It is interesting to see that the unemployment of teens in the ages of 16-19 is at 13.5 %. I wonder if this rate decreases during the months of June- August. A decrease in the rate would occur because a vast majority of teens look for summer jobs during this time. At other times of the year, it would make sense for the unemployment rate to be higher, because most teens are at school.


Return to top