December 08, 2003

Note: The Strange Case of the Unemployment Rate That Did Little in the Early 2000s

Look at what has happened to the U.S. employment-to-population ratio--estimated from the BLS household survey--over the past half century:

The employment-to-population ratio falls in each recessionary period.* Back in the old days, the rule of thumb was that the rise in the unemployment rate (in percentage points) was about five-thirds as large as the fall in the employment-to-population ratio (in percentage points). Thus the 1973-1945 recession saw the unemployment rate rise by 4.4% while the employment-to-population ratio fell by 2.4%. The 1979-1983 recessionary period saw the unemployment rate rise by 5.2% while the employment-to-population ratio fell by 3.1%.

But in the most recent 2000-2003 recessionary period, the employment-to-population ratio has fallen by 2.7% while the unemployment rate has only risen by 2.1%. The old pattern would have led us to expect such a fall in the employment-to-population ratio to have been accompanied by a rise in the unemployment rate of not 2.1% but 4.5%. More than half of the additional people who would have reported themselves as unemployed in a previous big recessionary period... aren't. They're reporting themselves as out of the labor force instead.

Why? What's happened to change the relationship between changes in employment and changes in the labor force? And what does it mean? (It's not the self-employed: this is from the household survey.)

It might be the sheer length of the downturn: a longer downturn may induce more people to give up looking, and produce more discouraged workers out of the labor force. But the 1979-1983 period was also prolonged, and although I remember Larry Summers and Olivier Blanchard worrying about how prolonged employment declines might discourage workers and produce a version of the European structural employment disease, it didn't.

It is a mystery to me.

*The employment-to-population ratio also rises over time as discrimination against women is severely reduced, and women find jobs outside the home in amazing numbers in the 1970s, the 1980s, and the 1990s.

Posted by DeLong at December 8, 2003 02:18 PM | TrackBack


I remember a few news articles - and have some personal anecdotes - about "marginal workers" that became regular employees for the first time, well, ever, in the late 1990s boom. Maybe they're mvoing back out.

Posted by: Jason McCullough on December 8, 2003 02:51 PM


I think Jason is right. Can you say "workfare?"

Posted by: Ann Nonymus on December 8, 2003 03:00 PM


A few weeks ago, Austan Goolsbee, an economist from U. Chicago, had an op-ed in The New York Times about unemployment rates that might shed some light.

He argued that the unemployment rate is artificially low, because some people who used to be classified as unemployed are now classified as disabled, and therefore not in the labor force.

The article is now taken down off the Times website, but I found it here:

Two quotes:
"the unemployment rate has been low only because government programs, especially Social Security disability, have effectively been buying people off the unemployment rolls and reclassifying them as 'not in the labor force.'..."

"By the end of the 1990's boom, this invisible unemployment seemed to have stabilized. With the arrival of this recession, it has exploded. From 1999 to 2003, applications for disability payments rose more than 50 percent and the number of people enrolled has grown by one million. Therefore, if you correctly accounted for all of these people, the peak unemployment rate in this recession would have probably pushed 8 percent."

Perhaps if you added the increase in disability enrollment since 1999 to the official unemployment count, you'd find that the old 5/3 rule still applies.

Posted by: Clark Williams-Derry on December 8, 2003 03:29 PM


In the computer industry, at least, I think it's happening because many unemployed people have been completely unable to find new jobs (without emigrating to the third world at least) and are changing careers instead.

They're highly proficient in a field requiring massive amounts of specialized knowledge and training. A field that's in huge demand but that many people are no longer willing to pay a fair price for any more, and that some people now even call "low-skill." In other words, they're knowledge workers who are being told they're a commodity.

What would you do?

Posted by: Chris Hanson on December 8, 2003 03:39 PM


Jason McCullough: "'marginal workers' that became regular employees for the first time, well, ever, in the late 1990s boom. Maybe they're mvoing back out."

I heard similar stories about all these students leaving college early to cash in on the dotcom boom, no doubt going back once the bubble burst. I'm sure some people came out of retirement for the same reason. Dunno if they both should get lumped into "marginal workers" or not.

Posted by: fling93 on December 8, 2003 03:48 PM


BTW, Prof. DeLong, you might want to consider scaling that enormous graph down a bit. Images that big wreak havoc with some browsers.

Posted by: Ann Nonymus on December 8, 2003 03:56 PM


>They're highly proficient in a field requiring
>massive amounts of specialized knowledge and
>training. A field that's in huge demand but that
>many people are no longer willing to pay a fair
>price for any more, and that some people now even
>call "low-skill." In other words, they're knowledge
>workers who are being told they're a commodity.
>What would you do?

Mainly, we're declaring bankruptcy, losing our houses, getting divorced, and ending up homeless on the street begging for spare change.

This is especially hard for those of us who ran up the student loan machine trying hard to get degrees, only to find that said degrees were worth less than the paper they're printed on, and that the liability on the loan is not absolved via bankruptcy proceedings - I.E., even if you're making $7.00/hour at McD's, Sally Mae will automatically deduct $300/month from your paycheck.

Me, bitter? Naaah.

An interesting sidenote of the technological revolution - remember the promise of being able to work from home? Well, bosses want bodies in the chair, so good luck getting permission to work from home. However, the exact same technology that you aren't allowed to use to save 3 hours a day in commute time is used to send your job overseas for 20-30% of your going rate.

Not that people in India, China, or Russia get to work from home either.

Posted by: Thane Walkup on December 8, 2003 04:16 PM


How does this ratio compare to the civilian participation rate as published by FRED of the St. Louis Federal Reserve? Is it safe to say that this ratio is the product of the participation rate and the ratio of employment to total labor supply?

Posted by: Hal McClure on December 8, 2003 04:44 PM


The increase in "disability" explains perhaps a point of the difference. My post

expands a bit on the Austan Goolsbee oped mentioned above. (Both based on the Autor and Duggan piece linked in my post.)

Posted by: Alex Tabarrok on December 8, 2003 06:10 PM


Doesn't "the unemployment rate that failed to rise 2000-2003" seem rather like the flip side of "the unemployment rate that failed to fall, 1997-2000"?

We will remember that back in 1996 Krugman was pretty harsh on those who lobbied to target 4% growth, saying that five years of 4% average growth implied an unemployment rate of only 1.5%, which obviously was impossible, so there! After which the economy promptly experienced five years of average 4% growth with the unemployment rate never going below 4%.

What happened was that the bubble economy created a labor market so hot that it brought people who normally wouldn't work into it in unprecedented numbers. Retirees went back to work, people were recruited with "bring your pets to work", etc. The work force expanded so the unemployment rate didn't fall as the 'simple model' predicted.

Well, that being the case, when the bubble economy ended and the extraordinary employment lures ceased, why wouldn't the people who were recently lured into the work force go home without joining the ranks of the unemployed -- after all, they weren't unemployed to begin with.

IOW, if we keep in mind that the unemployment rate first fell by 2.5% less than expected when employment was rising, why would we be surprised that it subsequently rose by 2.4% less than expected when employment later fell off?

Posted by: Jim Glass on December 8, 2003 09:20 PM


Prisoners. That's another component of this. We've added a huge number of people to the prison population.

Posted by: Dave Johnson on December 8, 2003 09:46 PM


It's hard to tell what is going on unless things get broken down more, say by sex and age classes, and by occupation categories.

There are other historical things that I was wondering about too, like why the employment figures in the 50's look so much better than the 60's, which were supposed to be the boom times.

Regarding the increase(?) of disabled? What is the disabled % of the working population and has that increased, not the actual number or percent increase over what it was in the 80's would be a more relevant figure.

Also, regarding the disabled, I think there has been an increase in repetitive stress injuries related to computer work and high volume processing plants. Another issue is mental illness- what percent are mentally ill? It's not really fair to start from the assumption that low income people are trying to get the easy life on SSI and SS disability. I have seen people trying to get off SSI and get a job and it can be very difficult.

Posted by: northernLights on December 8, 2003 10:18 PM


So we have the following:
- Marginal workers opting out
- Workers classified disabled
- Knowledge workers career shifting (or just getting by under the table)
- Army Reserve/Nat'l Guard callups (assuming this is what Mr. McClure refs with civilian participation rate)
- Incarceration (is this rate up in last 3 yrs?)

I'm certain that lumping all together could account for this mystery.

Prof. DeLong suggests the upward trend since the 1960's is largely due to women who "find jobs outside the home in amazing numbers in the 1970s, the 1980s, and the 1990s."

If so, are we at a point where the marginal contribution of a second parent working outside the household is nil? My brother and his wife reached this point well over a year ago. After factoring in extra car, child care, prepared meals, etc. (not to mention other potential harm to family harmony) perhaps many families are making due with one income.

A number of my former colleagues (I'm a devalued unemployed "knowledge" worker cleaning houses under the table -- harsh work but very glamorous) have reached this point and are full time Moms and Dads.

Interesting to note the ratio is reverting to the levels held in the late '80s when many conservatives trumpeted the decline of the family due to two income households. Is this an undeclared goal of the current administration?

Posted by: Merry Maid on December 9, 2003 12:04 AM


Most of you have already touched on this point, but I would like to lend it some extra weight. Looking at the BLS series data on US employment since 1940, the civilian labor force as a percentage of the total civilian non-institutional population steadily increased from 1942-1951 (from 57.2% to 59.2%), held more or less steady from 1952-1968, then increased from 1969-1997 (from 60.1% to 67.1%). It held steady at the historical high of 67.1% from 1997-2000, then, as you all have implied, it began to decline. Clearly, this number cannot grow indefinitely. It has a ceiling: 100% is the clear mathematical limit, but some other, unknown number may be the limit under ordinary (non-wartime) economic conditions. It is notable that the workforce percentage did not increase during the best years of the late 90's boom, which suggests that, at least for the time being, 67.1% may be the current ceiling, above an equilibrium level, to which the work force may now be descending in a normal corrective process.

Posted by: AAH35 on December 9, 2003 12:45 AM


Hmmm. Lots of interesting theories. Particularly the disabillity arguement, but one would assume that a surge in SS disabillity claims would have occurred in previous recessionary periods, thus not explaining the phenomenon(unless SS admin judgements have become more compassionate this time around ;o ).
I was also wondering about the BLS sampling technique. I don't know exactly how they do this, if it was by phone or by door to door survey, but I have noticed since losing my land phone line and using cellular for everything i never get surveyed(or telemarketed for that matter) and was wondering if that would skew the demographics of the sampling pool. I.E. my parents are rather technophobe and get phone surveyed and polled all the time, and are apprehensive to go cellular or caller ID for that matter.
Also (because those BLS statisticians run such a tight ship) I assume nothing has changed on the survey that would make a respondant more inclined to report that they are not looking for work.
My other thought was due to the tremendous growth of the temporary classification of employees. "temp" labor supply seems to move more fluidly in response to demand. So perhaps alot of employees previously working as temps are classifying themselves as not looking because a temp job usually carries no benefits and little incentive to search for a job that will last shorter than the search. Chances are that a BLS survey will hit them when they are actively looking for a job as opposed to between jobs-not looking and employed as a temp is probably less than it would a economically traditional labor pool participant. Caveat in that many of these working temps are also looking for permanent full time jobs, thus competing for jobs that "traditional" unemployed people are also seeking.
An interesting macroeconmic metric would be %of population actively applying for jobs, regardless of current employment status.

Posted by: Clayton Ullerich on December 9, 2003 04:23 AM


From :

"And in fact, what we found suggests that for men in the workforce, today's number actually rivals the 10.8 percent record of 1982, because, it turns out, there are four factors suppressing today's official number, at least for men: Millions more discouraged workers than there were in 1982; millions more on disability; nearly 1.5 million more incarcerated men; and finally, there's a demographic factor. Today's is an older workforce. "

Posted by: lee on December 9, 2003 05:05 AM


Some Chicago folks think that the answer is disability. The disability rules were softened, so that now instead of going into the unemployed category, people who are out of work go out on disability.

Posted by: Arnold Kling on December 9, 2003 07:51 AM


The windfall wage benefits of the boom overexpanding the # of workers is surely one important issue. But what about illegals? I don't see how any statistics can collect reliable employment & unemployment info on the illegal workers, who are as likely or more to be fired (no severance pay, etc.), despite them being cheaper to keep.

Posted by: Tom Grey on December 9, 2003 07:51 AM


" I remember a few news articles - and have some personal anecdotes - about "marginal workers" that became regular employees for the first time, well, ever, in the late 1990s boom. "

You should remember them, as they have posted on Semi-Daily Journal's comments section several times.

Posted by: Patrick R. Sullivan on December 9, 2003 08:12 AM


I also wonder how many people fall into my category. After looking for full-time employment unsuccessfully for 3 months, I moved back in with the parents at age 28. I managed to be reemployed by my former employer, but on a part-time basis. At this point, I'm mostly thinking about getting back to school (I have a B.A. in Fine Art and Humanities) to upgrade my education. Since I entered undergrad with an "any degree helps" attitude, but spent most of my time in the workforce under different conditions, returning (and never leaving) academia seems like the best plan.

Posted by: QuickSauce on December 9, 2003 08:23 AM


I'm curious. I wonder what effect 'would be' retirees that lost their 401Ks in the stock bust of 2000 have had on these numbers. Perhaps they could/should be even lower than this.

Posted by: Christo on December 9, 2003 09:36 AM



I don't know maybe there is something wrong with this rather simple argument but could anyone point it out?. I belive this "apparent" change is precisely produced by the increase in the the active population in relation to the work force, produced, as Bard says, by the incorporation of women to the labor market in the last half century.

let's remember this is a correlation we are seeing not causality "the employment-to-population ratio has fallen by 2.7% while the unemployment rate has only risen by 2.1%" can also be seen as an an incease of unemployment now produces a bigger increase in the fall of employment to work force. And that is exactly what happens when the % of people activly seeking employment, and also employment is bigger in relation with the working age population.

Are we confused by the way we express things?


Posted by: RL on December 9, 2003 10:00 AM


"Some Chicago folks think that the answer is disability. The disability rules were softened, so that now instead of going into the unemployed category, people who are out of work go out on disability."

That would be a rather lot of people wouldn't it?

And it doesn't explain why the unemployment rate didn't go down as Krugman predicted.

Posted by: Jim Glass on December 9, 2003 01:54 PM


Why hasn't anyone mentioned the baby boomers? The single largest demographic group is approaching or has hit retirement. This bubble affects overall demographic trends through the "growth" decades, 60s, 70s, 80s, etc., and are now similarly affecting the employment to population ratio.

But I'm just guessing. We could look at the retirement age (60+?) slice of US population by decade (from 1960s to 2000) and see if the percentage share has held constant or shrunken.

There's probably an answer in there.

Posted by: wondering on December 9, 2003 02:18 PM


"Prisoners. That's another component of this. We've added a huge number of people to the prison population."

I don't see how that would work to increase the unemployment rate up until 2000 then reduce it afterward.

BTW, I'm not up on the national numbers but here in NY the plunging crime rate has resulted in prison population falling in recent years and excess prisons now being sold off.

"I wonder what effect 'would be' retirees that lost their 401Ks in the stock bust of 2000 have had on these numbers."

If more people were pushed into the job market that would seem to increase the measured unemployment rate rather than reduce it

Posted by: Jim Glass on December 9, 2003 02:22 PM


The increase in the % of women working accounts largely for the increase in employment as a share of working age population over the past thirty years. The number of two earner families rose substantially in this period. Having two earners in a family (wife and husband usually), does provide some cushion against unemployment. If one person loses her job, the other income earner may be able to make enough money to pay the bills, although the family will probably have to decrease its spending. The "out-of-market-work" person may take over childcare, cook more meals, or do more house cleaning than before thus reducing the family's expenses and providing in this way "extra income". So, maybe some of this change in due to an increase in "unpaid labor" that is provided by homemakers who were once employed and are now out of the work force. Traditionally virtually all of these "homemakers" were women and they still are but more men are now homemakers because of the relative rise in women's incomes to men's incomes. Now it is possible for many men to stay home and take care of the kids.

The big question that needs to be answered is why was one person in the fifties and sixties(typically a man without a college degree) able to earn a family wage that was capable of supporting (with a decent life style that including owning your own home) a large family (having 3 or more children was not uncommon at that time) and why this is not so possible today?

Posted by: Nick on December 9, 2003 02:28 PM


Nick, I think I can answer your big question. I believe a confluence of trends have put the hurt on the blue-collar middle-class of the 50s and 60s:

The cost of housing has risen substantially because the population grows and the land mass of the United States does not.

It is the nature of the Information Age economy that new technology crushes demand for unskilled and semi-skilled laborers. According to Sherwin Rosen's "superstar effect" thesis, technology has turned economic competition into something more like a winner-take-all tournament. A couple of winners . . . and a whole lot of losers.

There are known cases where free trade has depressed wages for people in certain areas and industries, but studies suggest that free trade can't even account for a fifth of the declines we've seen. Rosen's "superstar effect" is probably the dominant phenomenon.

Unions have been obliterated. I think there are a number of reasons for this. Factory work is unusually amenable to unionization, and that kind of work has been dwindling for decades. The absorption of women into the workforce has diminished union power. Finally, the brutal 1981-2 recession was primarily a blue-collar recession, with unemployment rates of more than 20% in some places. I believe it was a mortal blow.

I also think that the price of a "decent lifestyle" has risen, and not just the housing part of it. Blue-collar workers know all too well what has happened to them. A lot of them believe that education is a way for their children to escape the agony of the declining standard of living they have endured. It clearly hasn't helped in the aggregate, but hope springs eternal and they continue to pay and pay and pay. Practically everyone believes their children need *some* college, and that's expensive.

Posted by: Ann Nonymus on December 9, 2003 11:20 PM


AAH35 said it: "Clearly, this number [labor participation] cannot grow indefinitely. It has a ceiling"...

That's what I was thinking. It's like there are three trends. One, short-term, business-cycle driven. The second, more structural, driven by changing values, technology, labor market effects. And the deepest one, which pulls the other two down (or up) when they're too far above. The decision to work must become very inelastic (to the incentives offered, or penalties applied) as you get near the top. A reversion to the third trend-line. Hence, business cycles get longer on the downstroke (for labor participation) relative to the unemployment change (these reluctant workers aren't on the rolls).

As usual, probably a trivial observation to all ya'll.

Posted by: andrew on December 10, 2003 05:51 PM


Housing and the "decent lifestyle" -- in most Black & Latio areas, a 3 bedroom, 1 bath house is still quite affordable. Ann prolly doesn't want to admit that most whites think most decent homes are segregated (my white Mom lived in E. LA, City of Commerce, for years).

But there's plenty plenty plenty of land in the US -- in the Midwest, for instance. Where people are moving away from. (Remember those silly 80's horror stories of the Japanese buying up all the US farmland???)

The home mortgage interest deduction should be reviewed, and prolly revised. I'd suggest an honest 50% income tax credit on house payment (interest AND principal), up to a lifetime maximum of some $300 000, or 2 times the national average price sold. (But housing is different than unemployment)

Posted by: Tom Grey on December 11, 2003 02:12 AM


Surely there was some 50s - 60s benefit to being on the winning side of WWII without having to rebuild most of the country? There was certainly a lot of want left over from the Depression and semi-forced savings left over from the war - unwinding that would just naturally be pleasant for the people in well-organized jobs.

IIRC the 60s were more boomlike than the 50s *because* employment could be lower - more women worked in the 50s than had to in the 60s. Also, expectations were still lower than they are now - Levittown isn't the "good life on TV" any more.

Posted by: clew on December 11, 2003 02:23 PM


I see you all going to extremely complicated explanations.
Let's see
Situation 40 years ago :(I am making up the numbers)
100 is total working age population
50 is work force(people actively seeking a job)
48 is employment
2 unemploid

the rate of unemployment is 2/50=4%
the rate of employment to working age population is 48%

Now let's travel in time and put in the job market lots of new people.

100 working age population
90 work force
86,4 emploid
3,6 unemploid

the rate of unemployment is 3,6/90=4%
the rate of employment to working age population is 86,4%

I intencionally exagerated numbers to show that a 4% unemployment back then supposed a employment to working age population 48% and when increasing the work force this raises to 86,4%.


Posted by: RL on December 15, 2003 05:21 AM


Post a comment