In our relentless attempt to get at the Truth, we often spill pixels showing how government data makes things appear rosier than they are in the real world.
This morning’s data may have made things appear worse.
Providing a glimmer of hope that the U-3 unemployment rate isn’t as bad as it appears, an unexpected surge in teenagers and 20-25 year olds is responsible for a chunk of the unemployment jump.
Also possible — a seasonal adjustment that was expected in June failed to pick up more teens applying for jobs in May. Rex Nutting of Marketwatch notes that some of this is "Statistical noise":
"The
government cautioned that the scope of the increase in the unemployment
rate in May could be a statistical distortion. Month-to-month changes
from April through July can be hard to adjust seasonally."There is a substantial
flow of workers, particularly young workers, into the labor force
during these months," said Phillip Rones, deputy commissioner of the
Bureau of Labor Statistics. The unemployment rate for teenagers surged to 18.7% from 15.4%. A broader measure of unemployment that includes so-called discouraged workers rose to 9.7% in May from 9.2%.
Here’s a chart of teen workers employment:
click for bigger chart
via Jake
>
UPDATE June 6, 2008 11:07am
Real Time Economics notes two factors — the first being the jump in Teen unemployment; the second factor being a catch up to already existing employment pressure. Morgan Stanley, Nomura Securities, and Insight Economics all noted that the "unusually large increase" may have exaggerated the deterioration in the unemployment rate that occurred for last month, but "probably reflects the weakness that has developed since the end of last year."
Insight Economics adds that "In the post World War II period, every time the unemployment rate has jumped by a full percentage point in the course of a year, the economy has slipped into recession."
That is consistent with my economic perspective.
>
Source:
Jobless rate soars to 5.5% in May
Biggest rise in unemployment in 33 years; payrolls fall 49,000
Rex Nutting
MarketWatch, 9:19 a.m. EDT June 6, 2008
http://tinyurl.com/4zp5vh
Economists React: ‘Teen Angst’ or Playing ‘Catch-Up’ in Jobless Rate?
June 6, 2008, 10:39 am
http://blogs.wsj.com/economics/2008/06/06/economists-react-teen-angst-or-playing-catch-up-in-jobless-rate/
Here in Vermont teens are having a hard time finding jobs this summer. They typically work in consumer discretionary type jobs and those are weak. Also, a lot of adults are taking the jobs teens would normally take.
I think this is a canary in the coal mine not a one off. So I call bullshit on Rex.
Are you just responding to the criticism that was directed your way by Jeff Miller?
http://oldprof.typepad.com/a_dash_of_insight/2008/06/ism-interpretation-watch-that-decimal-point.html
~~~
BR: VT — you are discussing 2 different issues: 1) are teenagers having a hard tme finding jobs (I think yes); and 2) Is the BLS accurately reporting the data?
Since this was the biggest jump in teenage unemployment since 1948, that implies it was likely a bit aberrational.
Of course, it has long been my position that this was the weakest job creation post recession cycle since WWII, and that unemployment is vastly understated.
I hadn’t seen Jeff Miller’s post, but — my previous reads of his work puts him in the camp that thinks inflation is low, unemployment moderate, and the BLS data fairly reliable. I don’t take his critique of my analysis as anything more than confirmation bias on his part (and he can say the same about me).
I don’t recall the BLS cautioning us that last month’s anomalous 0.1% decline in the unemployment rate could be a statistical distortion. But it probably was.
If last month’s rate had been reported as 5.1% or 5.2%, today’s 5.5% number would be seen for what it is: a slight acceleration in an established uptrend.
While the absolute value of the U-rate may be questionable, nearly every rise of 0.6% or more from a low has pointed to a recession. This time round, as of today, we have a cumulative rise of 1.1% from the 4.4% low. Verdict: recession.
I don’t recall the BLS cautioning us that last month’s anomalous 0.1% decline in the unemployment rate could be a statistical distortion. But it probably was.
If last month’s rate had been reported as 5.1% or 5.2%, today’s 5.5% number would be seen for what it is: a slight acceleration in an established uptrend.
While the absolute value of the U-rate may be questionable, nearly every rise of 0.6% or more from a low has pointed to a recession. This time round, as of today, we have a cumulative rise of 1.1% from the 4.4% low. Verdict: recession.
I don’t recall the BLS cautioning us that last month’s anomalous 0.1% decline in the unemployment rate could be a statistical distortion. But it probably was.
If last month’s rate had been reported as 5.1% or 5.2%, today’s 5.5% number would be seen for what it is: a slight acceleration in an established uptrend.
While the absolute value of the U-rate may be questionable, nearly every rise of 0.6% or more from a low has pointed to a recession. This time round, as of today, we have a cumulative rise of 1.1% from the 4.4% low. Verdict: recession.
I don’t recall the BLS cautioning us that last month’s anomalous 0.1% decline in the unemployment rate could be a statistical distortion. But it probably was.
If last month’s rate had been reported as 5.1% or 5.2%, today’s 5.5% number would be seen for what it is: a slight acceleration in an established uptrend.
While the absolute value of the U-rate may be questionable, nearly every rise of 0.6% or more from a low has pointed to a recession. This time round, as of today, we have a cumulative rise of 1.1% from the 4.4% low. Verdict: recession.
Last year at this time we didn’t see unemployment rate go up. The job market is very weak. I think this report shows how hard it is to get a job these days.
How do you measure unemployment for teens? They aren’t even eligible for benefits unless they work full time for a full year. Unless they’re highschool dropouts or something.
oeirgjre,
Today’s data have nothing to do with benefits.
The payrolls number is derived from a survey of employers. The unemployment number is derived from a survey of households. Both of these surveys are tortured monthly to confess to whatever pundits want to hear. The details of this have been the subject of much bit-flipping on this blog.
This is the problem with seasonal adjustments in general. I don’t know the exact methodology, but from the broad descriptions I’ve heard it sounds like they assume a normal distribution of variability for the specific month over the last X years (is it 3 or 5?). In essence, they are specifically designed in a way that fails to capture trend changes.
As you’ve noted, when the behavior changes at the “wrong” time, it can lead to extreme overcorrections in seasonality that have to be made up later, so a lot of data is going to be made worse in the latter half of the year. The best way in my mind would be not to compare month-to-month directly, but how the YOY changes month-to-month.
Alternatively, the Calculated Risk graphs that just show the raw data series grouped by months definitely give an intuitive insight into what’s really going on.
Could it be that unemployment filing data are finally ticking up because severance packages for those who have been getting laid off over the last six to nine months are expiring for masses of workers? I don’t believe that one can claim unemployment until severance packages expire, no? If that’s the case, then the uemployment numbers have been artificially LOW and are slowing becoming more accurate.
As an executive recruiter in the financial services industry, I can personally attest that the market in this industry is EXTREMELY weak. Most employers are only looking to fill extremely hard to fill needs and are only moving forward with “9” or “10” candidates that have most or all of the “must-haves”. It’s brutal out there. Trust me.
if there is an uptick in teenagers seeking summer employment, might it be because their parents have told them they have to get jobs becuz the parents themselves are not doing well and need the kids to be helping out (or at least not sponging as much) ?
Estragon
Would this not imply that the government is capitulating, as a matter of growd control to release the details of the plight of the economy at a measured pace so as to not generate a cascade of confidence loss. So as that at least the strong hands can manuever themselves into a defensive postion to prepare for the comming tempest.
Solution: remove from the calculation anyone 18-25. These people should not be working. Also, if you think about it, who wants to work if they are over 50? Remove them from the calculations, too. If we really try we can get unemployment below 5% again. Work with me here, people!
Here’s what Stephen Stanley of RBS Greenwich Capital wrote:
“The unemployment rate surged by over 1/2% to 5.5%, the largest one-move rise in over 20 years. This is a jolting headline I suspect there is some noise element to the spike. The BLS noted that a late May survey reference period could mean that more of the students entering the work force after the school year got picked up in May instead of June (at odds with what the seasonals would assume). Indeed, the unemployment rate for 16-19 year olds spiked from 15.4% to 18.7%, the largest monthly jump on record, corroborating the hypothesis. The 16-19 year old group accounts for only 7% of the labor force but contributed nearly half of the May spike in entrants. These data would help to explain why the labor force jumped by over 500,000 at a time, with job demand weak, that you would expect people to be giving up their job searches. Still, even if the labor force had been flat, the 285,000 drop in the household measure of employment would have been sufficient to take the jobless rate up by almost 2 full tenths. More realistically, a moderate increase in the labor force, given the employment reading, would have taken the figure to somewhere around 5.3%. Even if the June reading retraces to that level, the jobless rate will be noticeably higher than we had been projecting at mid-year. Up to now, we were thinking that the rate might peak around 5 1/2% and do so late this year. Obviously, our “target” has been reached considerably earlier than we had expected, so we will have to construct a new (higher) path for the rate over the balance of the year, perhaps peaking in the 5 3/4% area by the end of the year.”
And Goldman Sachs:
2. The eye-popping 1/2-point jump in the unemployment rate overcorrects what had struck us as an understated level of unemployment based on the behavior of continuing claims for jobless benefits. With the labor force reported to have increased 577K, about three times as much as would be implied by growth trends, one has to discount about half of the increase in the unemployment rate. That still leaves the plenty of weakness in the survey of households; for example, employment in this survey fell by 285K.
OT: Barry, did you catch Joe Battipaglia’s smackdown of the Goldilocks fantasy on Monday’s Larry Kudlow show? It was Joe B at his best. Joe personally wrestled Larry to the ground and hog-tied him – all with a smile on his face. Joe must use Freon for his deodorant.
Is Paulson standing by his promise of an increase of 500,000 new jobs in the second half of the year?
The market is taking this rather good. I think for the most part it’s been priced in on 50% of the NYSE stocks I would say.
Is it possible we are getting to an age where people don’t sell stocks during bad economical future because they realize there will eventually be light at the end of the tunnel?
Larry’s on CNBC still spouting his “Goldilocks” theme. Good grief. Talk about delusional.
Even though Estragon explained how unemployment numbers are derived, there still appears to be some confusion:
Unemployment rates are determined by surveying households. They call and ask if you are working. If you aren’t, and haven’t been lately, they ask if you have been looking for work for at least a few (I think two) weeks. (Not everyone wants a job, and these are not counted as unemployed). If you wish you had a job and don’t have one and have been looking for at least a little while, then you are classified as unemployed.
They figure out your age, gender, race, etc., and feed it into their little database, and out pops a number–overall, 0.5% higher this month than last–a remarkable increase. The unemployment rate is almost always much higher for certain segments of the population, like teenagers, than it is for say, females over age 25. The 5.5% is the ratio of all those that want a job but can’t find one to the number of people in the working age population.
labor force participation in the entire 16-35 yo cohorts has been anomolously low thruout the Bush adm. compared to previous administrations. Leaving aside the manipulation arguments, other explainations:
529 plans underperform..gotta work way thru school.
via NYT, banks are not writing loans for community/or for profit collages…gotta work thru school.
Mom or Dad worked for Bear Sterns… gotta
work.
The economy sucks…all I need to do is drive up and down Central Avenue, through Hartsdale and Scarsdale NY (two affluent sububrbs)…everywhere I see out of business furniture stores…thousands of square feet with for rent signs (Barnes and Noble left)…cars ride up and down advertising liquidation sales.
When I was a kid my summer job was out in the corn and soy bean fields of Indiana. I suspect that teens today probably don’t want to work as hard as I did for $1.10/hour, but even if they did, many of those farm jobs are filled by illegal aliens. The same can probably be said for restaurant jobs.
DonKei,
Good summary. I’d add that the household survey is relatively small, the margin of error is wide (~ +/-430,000), and the series can be noisy. I wouldn’t imply much of anything (including government manipulation) from a single month’s data.
I don’t know why employment isn’t measured using withholding tax info. Wouldn’t that be both more accurate and more timely?
4.25 million 19 year olds this year, 4.5 million 18 year olds graduating high school this year, the largest cohorts in many years now. This is the “boomlet” bubble hitting the job market just when there are no new jobs to be had… they were unexpected by the schools when they hit, now they are unexpected in the work force.
Estragon,
Yes. There is a well-recognized bias for the unemployment survey to understate unemployment when times are bad (people quit looking if they know they can’t find a job). Given that the survey requires truthful answers that are anyway subject to interpretation, and only polls less than 0.5% of the work force, I think it is a very weak measure of economic performance.
Payroll data is a much better measure of economic performance trends in my view, as BR has shown here before.
It is much easier to figure out in real time whether aggregate payrolls are growing, flat or declining by just looking at the withholding data.
Mike M……you’re joking?…right….
If I recall correctly, the accepted number for job replacement was 150,000 per month just to remain even with the population. I don’t think in the last 7 years have we ever had a number close to or exceeding that. Kudlow was always using employment number from 2001 to “prove” the job market was growing. Guess we’ll have to await the election of Obama before Kudlow gets back to railing about equilibrium in the job market.
BR – This is a head scratcher for me…perhaps you can clarify?
Today’s employment report indicates 8.8m total unemployed, up from 6.9m a year ago. By my math, that says 1.9m more unemployed than 12 months ago (avg. of ~ neg 158k/month).
I went back and added up the “net change in employment” figures from the last 12 reports (132k, 92k, -4k, 110k, 166k, 94k, 18k, -17k, -63k, -80k, -20k, -49k), and I get 379k jobs *created* over that same 12 month period according to the headline numbers. Even after subtracting out net revisions from the original statements of -96k over those 12 months, I still get 283k jobs created according to monthly employment reports, vs. 1.9m jobs lost.
Now, I get the arguments about seasonal adjustment, but over a 12-mo period, those should cancel out, right? So why the hugely different numbers?
I’d be interested in your thinking here.
When the BLS calls, the interview goes something like this:
(Click on name/link)
the bls numbers are phantoms as is the cpi numbers real unemployment including underemployment is 12%.