The EPI's Jared Bernstein gives his take on the September unemployment report. The continued divergence between largely unfavorable employment news and estimates of current rates of output growth seems to me that one of two things is likely:
Posted by DeLong at October 09, 2002 02:27 PM | TrackbackOctober 4, 2002
Jobless recovery continues unabated
The unemployment rate unexpectedly edged slightly downward last month from 5.7% to 5.6%, but payrolls contracted by 43,000, according to today's report from the Bureau of Labor Statistics. In fact, the fall in the unemployment rate was statistically too small to be distinguished from no change at all—adult unemployment rates were unchanged, while the more volatile teenage rate fell 1.5 percentage points.
At the same time, job creation, which had been slightly positive, stalled in September. Large revisions to the survey of the nation's firms suggest more jobs were created in August than previously thought: between May and August, employment growth averaged 54,000 per month in total (but only 30,000 per month in the private sector, a better indicator of the impact of the economy on the labor market).
September losses, however, were widespread. Trouble spots included continuing large declines in factory employment (down 98,000 over the last two months), continued loss of jobs in retail, and significantly smaller gains in services last month. Through August of this year, services were adding an average 59,000 jobs per month; September's growth was only 28,000.
In recent months, taking the pulse of the labor market has been made more challenging by two factors: the discrepancy between the two main surveys that produce these data (particularly regarding employment growth), and the larger-than-average revisions to the payroll survey. The first of these issues is more easily resolved: the payroll survey, which showed a decline of 43,000 in September, is a much more reliable source of monthly changes in job growth. It derives from a much larger sample and provides much less erratic monthly estimates, thus making it the primary source of information on job growth used by labor market analysts.
Regarding recent revisions, today's report notes that employment losses over the recession were larger than previously thought. Future employment reports will reflect "approximately 284,000" fewer jobs between April 2001 and March 2002, which will bring the job losses over this period from 1.8 million to over 2 million.
Numerous important indicators show just how difficult it is to find a job for today's unemployed. The share of the unemployed who were laid off rose from 55.1% to 56.0% last month, the highest level during this recession/slow-growth recovery and as high as it was in June 1992 when the labor market was also experiencing a jobless recovery.
Long-term unemployment also rose last month. The average number of weeks spent unsuccessfully seeking a job rose from 16.2 in August to 17.8 last month, the highest level reached thus far in this downturn. Since unemployment's low-point in October 2000 (3.9%), the average time spent unemployed has increased by over five weeks. The share of the long-term unemployed also rose last month, from 18.2% to 19.5% (comparable to the recent peak of 19.6% in May 2002). The number of long-term job seekers is 2.5 times its October 2000 level: 1.6 million compared to 622,000.
The weak labor market continues to place downward pressure on hourly wages, which grew 3% over the past year, the slowest annual growth rate since March 1996.
While most analysts expected unemployment to rise in September, the lack of job growth and the lengthening of the time spent unemployed should still send a clear signal to policy makers that the economy remains ensconced in a jobless recovery. Taking the needed steps to both protect the long-term unemployed (through another temporary extension of unemployment insurance benefits) and stimulate faster growth must remain at the top of the economic agenda.
—Jared Bernstein,
with research assistance by Brendan Hill
>>The unemployment rate unexpectedly edged slightly downward last month from 5.7% to 5.6%, but payrolls contracted by 43,000, according to today's report from the Bureau of Labor Statistics.<<
Uh? I thought payrolls had to increase for the unemployment rate to stay constant... (by, I have in mind, something like 450k jobs - please correct me if I have this figure wrong.) One of these two claims has to be false.
How long are we going to officially endorese these rigged unemployment numbers? The fact that workers fall off unemployment benefit programs or stoped being accounted for in the active population statistics because they have despaired in looking for a job, just tells us nothing about their willingness to take a job at the prevailing wage rate.
What it brings people to think, though, is that Dubya is doing a fine job for the economy given the circumstances (59% think the economy is 'doing the same or better'... than something...). It's just that he should spend a little more 'attention' to it...
Two different surveys: a survey of businesses, which produces job growth numbers; a survey of households, which produces the unemployment rate.
They ought to produce consistent numbers. They don't. It's a problem.
Posted by: Brad DeLong on October 9, 2002 09:24 PMIn addition to that, these are miniscule changes.
What is the margin of error in those surveys?
Can they really spot a change of 0.1%?
A change of 43,000 jobs out of (guessing) 100 million is 0.043%
I'm sick of the phrase "jobless recovery". It hides the misery of the situation. I remember looking for a job during out last "jobless recovery" and for me it might as well have been a "recession" 'cause I couldn't find a thing.
Posted by: Ian Welsh on October 10, 2002 08:08 AMThe household survey (used to measure the unemployment rate) has an estimate of employment changes that is supposed to be accurate to within 292,000 nine times in ten. It's been strong for two successive months---and its therefore tough to dismiss the 1.1 million job gain reported over that period. But of course, there's always the usual problems with seasonal adjustment, particularly tricky in Aug-Sept as students return to school. Teenagers accounted for 363,000 of the jobs gained in the last two months in the household survey... maybe they will disappear in the Oct data if they were just capturing a slightly later return to school.
Note that the household survey, unlike the payolls data, include "self employed" jobs. But they might also capture the job creation from net new business formation, a figure that the payrolls survey essentially has to guess at.
Usually the payrolls data (where the sample, weighted to large employers, covers a total of 39million workers) is more accurate that the 60,000 sample in the household survey. But some argue that the payrolls report can miss a turning point if job creation picks up through new business formation.
Bottom line: wait a couple of months before drawing any firm conclusions about how the labor market is faring.
Posted by: Avery Shenfeld on October 10, 2002 08:11 AMI like Avery's comment, mostly because it sounds a good bit like something I mentioned to our clients last Friday. The establishments' survey is, in general, thought to be more accurate. However, it misses start up firms and probably does a poor job on small firms, which don't have the staffing to answer surveys regularly or accurately. That is why the Labor Department adds a "plug" factor for smaller firms. One problem with the plug is that it adjusts slowly to changed conditions in the labor market. The suspicion is that it overestimates labor market strength as employment growth slows and underestimates labor market strength as hiring picks up. Hiring is picking up. There is no reason to think the quality of responses to the households survey vary by the size of the employer, so the 225,000 or so average monthly job rise households are reporting over the past 6 months may be capturing something that the establishments' survey is missing. Another reason to think that might be true is that larger firms have the flexibility to vary operations, hire temps and any number of other things to avoid hiring in the short run. Smaller firms have less flexibility. Plunging stock prices and hiring do not go hand in hand for publicly held firms, but a family-owned operation that gets more orders than they can handle might feel free to hire. The problem with the "plug" can be large, as job creation is mostly among small and medium sized firms.
I know that visions of sugarplums are not all that common in economic discussions right now, but I'd love to hear any thoughts on why this idea might not be true.
By the way, I suspect that this business of the establishments' survey missing turns in the labor market due to the "plug" might also explain some of the quarterly variability in productivity stats.
K
Posted by: K Harris on October 10, 2002 08:47 AMGood posts but is anyone really under the impression that a lot of new businesses are being created these days? Or that small businesses are doing so well? (Tell me wrong if necessary.)
Surveys show that economists have a tendency to fail to predict downturns as well as upturns, so I always feel unconfortable to make gloomy statements at this point. Nevertheless, I think it's important not to give into wishful thinking because that's going to condition our political and policy response to the current state of affairs.
On the bright side, the fourth quarter should look good as people start X-mas shopping and stuff. So, we're probably in for a round of relatively good news. Year on year comparisons should look good given what last winter looked like. It's the war on Irak that is the wild card here...
Posted by: Jean-Philippe Stijns on October 10, 2002 09:34 AMAfter introducing sugarplums to this conversation, I feel obliged to pass out the castor oil. Christmas has a high risk of being rotten, for 3 reasons. Something that happened last year may happen again. Caution on the part of retailers meant shelves were understocked with the items consumers wanted most last year. Unlike in the late 1990s, when consumers were willing to take what was left (at the right price) just to make a bigger stack under the tree, Christmas of 2001 was less jolly than it might have been because consumers appeared to want what they wanted, and nothing but.
Second, the port closure, as widely reported, has cut down on imports of goods for the holiday season. A supply problem, like problem number 1, except that the cause of the supply problem is a labor dispute, rather than retailer caution.
Finally, there is economic gloom, which seems likely to cut into spending. The fourth quarter will only look good this year before seasonal adjustment, I fear.
Posted by: K. Harris on October 10, 2002 12:42 PM>>Trouble spots included continuing large declines in factory employment (down 98,000 over the last two months), continued loss of jobs in retail, and significantly smaller gains in services last month<<
If I'm right then the tendency is to lose jobs in manufacturing (presumeably in part at least this is structural through jobs moving out of the US via outsourcing), and in retailing (through presumeably a combination of greater efficiency and lower sales), but to gain in services. Could this be related to the fact that the US population is both aging AND having more children (after 10 years of rapid immigration) at the same time. Thus service type areas like education, health, child and elderly care are growing on a secular basis, regardless of the general downturn. Of course what this does for the overall productiveness of the US economy is another question.
Somehow I don't share the 'rising costs of education and health care as the last bastion against deflation' view.
Posted by: Edward Hugh on October 11, 2002 03:12 AMIndeed, Edward, it would be like looking forward to rising oil prices as a way out of deflation threats. I have a hard time formulation my preferences between stagflation and deflation. I admit I have a blindspot on this one. Anybody has some wisdom to share on this? Professor DeLong?
Posted by: Jean-Philippe Stijns on October 11, 2002 09:28 AM
