December 31, 2003

At Long Last, Good Unemployment News

At long last, some unambiguously good unemployment news:

WSJ.com - Initial Jobless Claims Drop To Lowest Level Since 2001: ...The number of workers filing new claims for unemployment benefits in the week that ended Saturday dropped by 15,000 to 339,000, the Labor Department reported Wednesday. It was the lowest level since the week ended Jan. 20, 2001, Mr. Bush's inauguration day. The four-week average, which smooths out weekly fluctuations, decreased by 6,500 to 355,750.

Economists had expected claims to increase by 2,000, according to a survey by Dow Jones and CNBC. The decline, particularly in the four-week average, was encouraging to most economists...

Posted by DeLong at December 31, 2003 08:48 AM | TrackBack

Comments

Could it be that they've finally downsized all the people that they can. There's only one guy left in Ohio that is still working.

Call me when there is an huge number of people going back to work at jobs that are full time, with bennies that actually pay a living wage.

RC

Posted by: RC on December 31, 2003 09:19 AM

____

How can weekly numbers, and especially numbers for the week that includes Christmas, be *unambiguous*? Or for that matter, even monthly numbers for December? What does a downtick indicate beyond the fact that 1) many employers are unlikely to lay off folks going into Christmas and 2) some of the people who actually just got laid off MAY have had other irons in the fire other than filing during that period? The economics-impaired want to know.

Posted by: scylla on December 31, 2003 09:23 AM

____

This piece of data doesn't refute Paul Krugman's NY Times column from yesterday -- no single unemployment factoid would, given the noise in the series, and doubly so given that this data doesn't talk specifically to Krugman's point about workers abandoning employment searches. But it does suggest that Krugman's "no-boom" column was entertainingly ill-timed.

Posted by: Paul K. on December 31, 2003 09:28 AM

____

I'm afraid scylla has you on this one. This is among the least reliable weeks in a very unreliable series. The week before new claims fell 15k, existing claims rose by 81k (both were reported today). Is that bacause firms nearly simultaneously ending layoffs (new claims down) and clamping down on hiring (old claims up)? Or could it be because it rained really hard in one of the most densely populated parts of the country the day before a holiday, inhibiting many new jobless workers from filing claims that week? Couldn't file on Friday, they were shopping.

Posted by: K Harris on December 31, 2003 09:33 AM

____

Take a look at the unadjusted numbers


UNADJUSTED DATA

The advance number of actual initial claims under state programs, unadjusted, totaled 516,501 in the week ending Dec. 27, an increase of 91,785 from the previous week. There were 620,929 initial claims in the comparable week in 2002

http://www.dol.gov/opa/media/press/eta/ui/current.htm

Posted by: Dave on December 31, 2003 11:03 AM

____

Usually, You get one "good news" article along with one "bad news" lately. I noticed that the dollar is now at 1.26 to the EURO.
Why is the good news of the employment numbers not improving the dollar?

Posted by: Jean on December 31, 2003 11:33 AM

____

Adjusted vs. unadjusted numbers:

Dave, bear in mind that the exact timing of layoffs and unemployment filings is not perfectly distributed, and so the measured numbers are subject to "jitter" in time, regardless of adjustment. Thus weekly variations contain relatively little information. Taking a moving average of a few weeks should even some of this jitter out. (Please note that this type of jitter is completely independent from the seasonal patterns that seasonal adjustment is trying to neutralize.) Although we should never forget that we are talking about 91785 unfortunate individuals, regardless of whether their measurement is jittered.

Regarding seasonal adjustment, the BLS and DOL websites are not very specific. What I was able to learn was that they try to estimate a seasonal pattern from past data, and adjust current data by multiplying them by some coefficients. I conjecture what they are trying to do is to adjust for known events (rain seasons, winter, xmas etc.) so that if the unadjusted measured number exactly follows the estimated pattern, the adjusted number will appear flat, and variations should indicate economic events outside the known pattern.

Check out these two websites for their explanation of the adjustment:
http://www.ows.doleta.gov/unemploy/wksaqna.asp
http://www.bls.gov/news.release/empsit.tn.htm

While I believe that their (and the BLS) efforts to even out known events are genuine, and not intended to fudge the numbers, I have to wonder whether the result is helpful, or whether it falls in the category of "fooling yourself and others".

For example, a seasonal event based on weather patterns may occur at a different time due to a shifted rain or snow incidence, winter setting in earlier or later, etc. The result of trying to outsmart nature may be, as you may remember, that people come up with more excuses why some surprise is happening -- store sales declined because of colder than usual weather, or hotter than usual weather, heavier than usual snowfall, and so on.

If you are adjusting for a high-unemployment event (e.g. winter, when presumably unemployment is higher because of reduced construction and agricultural activity), and the event happens earlier (later), then your adjusted number will show an increase (decrease) of unemployment at the start of the event, and a decrease (increase) at the end of the event which are both fake, because you are over- and undercompensating at the edges.

Can somebody offer more details on how the seasonal adjustment is done? Apparently the specific "coefficients" are changing once or twice yearly.

What I would like to do is this moving average thing, but it would be useful to know the adjustment model.

Posted by: cm on December 31, 2003 01:49 PM

____

http://www.brm.org/howlong.asp

Unemployment numbers are pure bunk! Count the new faces with grocery carts down by the bus station. Count the rapidly rising prison population. Count the number of people with three jobs, or 'working from home', sitting by the phone. Count those who work cash, barter or trade to avoid Uncle Sam. Count those who went from $11 an hour store jobs to $7 'independent agent' at WalMart. Count their husbands who are out in the shed, brewing meth. Most of all, count the fools too proud to file for UE, who think there's a new job out there, and are treading water part time, minimum wage, while the value of their paycheck drops 25% YOY.

You don't judge the harshness of a winter by how many birds are crowded around the feeder, but by how few nesting pairs there are next spring.

Soon they won't be able to give away cars, houses or appliances, but hey, you can always buy lotto tickets or share of penny stock with that fiver!

Posted by: Harry Possue on December 31, 2003 02:00 PM

____

Hm. CM, thanks for trying to explain some of this. The fact remains though that the main reason we come up with all these adjustment models is so we can predict what's going to happen next. And by its very nature, the urge to predict is not dispassionate. I guess that's where the "fooling yourself and others" part comes in.

Posted by: scylla on December 31, 2003 02:01 PM

____

The season adjustment process uses what is called
the x-11 seasonal adjusment program. It calculates what the seasonal norms are in different ways and selects what appears to be the best way to make the adjustment. It calculates a factor to adjust the reported number to determine the seasonally adjusted data. The calculation uses the previous few years data to calculate the correct seasonal adjusmnet and the program is run at the end of the year to recalculate this years data. The BLS in particular, but also other agencies publish what the seasonal adjusment factors will be for different data series at the start of the year. Moreover, the X-11 progam is
publically available.

The seasonal adjument data for montly series is generally very good. But when you get into weekly data it is easy to find problems.

One thing I watch is the CPI where the Not Seasonally adjusted data can be of value. In a
low inflation world firms tend to raise prices once a year, typically at the start of the year.
Consequently, over half of the core CPI -- excluding food, energy & tobacco -- now occurs is the first quarter of the year. Moverover, over the past about 15 years since this pattern emerged
if the 1st Q NSA CPI is greater than(less than) in the 1st Q of the prior year the annual cahnge in the Core CPI will be greater than(less than) the prior years annual increase. There has only been one exception. The change in the not seasonally adjusted CPI after the 1st Q is dominated by education and home owners equivalent rent.

Posted by: spencer on December 31, 2003 02:07 PM

____

Harry:

I agree with all of your points. There is certainly cultural and ideological bias in what numbers are measured to begin with. I think your bird analogy is a good one. Also considering that the prison population is excluded from the work force leaves a stale taste, as the unemployed population is probably vastly overrepresented in it. (Anyway I'm always a bit uncomfortable talking about people's fates in such abstracted and technical way.)

One general problem with any indicators (and I don't know a solution to that; anybody?) is that even if you are honestly measuring, you can measure and interpret only those things that lend themselves to reliable measurement. Even unemployment numbers, while being well-defined, suffer from the methodological flaw that certain categories of unemployed (e.g. homeless) cannot be reached by the survey (?).

The effects of paycuts, lack of raises or cost-of-living adjustments, shifting to lower-paid jobs, involuntary unpaid overtime, job "reclassifications", etc., and generally more demands made at workers cannot be reliably measure dand interpreted. Is some shift to part-time work or a transfer to a different job description voluntary etc.

The result, even if everything is done honestly (but maybe too naively), is that some measurable number is declared to represent a larger category. Of course the official unemployment number does not represent the state of the labor market, although it is a relevant indicator. However, there are more measures than the official national rate. On http://www.bls.gov/news.release/empsit.t12.htm consider the difference between U-3 and U-6. I think the latter (unemployed plus marginally attached plus part-time) comes closer to the truth. That's 9.7%, not the official 5.6%. And there is probably a large hidden number of "discouraged" workers that is not even surveyed.

Posted by: cm on December 31, 2003 03:45 PM

____

While there's no way to adjust other than basing on experience, I wonder if this unusual "recovery" would cause this to distort the adjustment. Perhaps the adjustment expects that lots of people are laid off as Christmas shopping season ends, and subtracts 150,000 or so as its "adjustment". But what if that many weren't hired in the first place, because of the unusual labor conditions.

Posted by: Dave Johnson on December 31, 2003 03:51 PM

____

"Perhaps the adjustment expects that lots of people are laid off as Christmas shopping season ends, and subtracts 150,000 or so as its "adjustment". But what if that many weren't hired in the first place, because of the unusual labor conditions."

I hope the adjustment is not as trivial as that. I try to find my way through what I find on the X-11 method (thanks spencer). According to my layman theory alluded to earlier, I would expect a multiplicative adjustment (along the lines of if 10% more people are laid off after Christmas than the yearly average, adjust the number down by dividing by 1.1). Of course that is not the correct thing to do as well, but I think it's closer to the principle.

But the point you are making seems to be in line with the shifted-event fallacy that I described -- if you try to compensate for an event that does not happen, or does not happen at that time, you will overcompensate. If let's say the event (temporary employment peak) does not happen or is smaller in scale, then the adjustment will probably show increased unemployment at the start and throughout, and reduced unemployment at the end (yes! economy is picking up!).

Posted by: cm on December 31, 2003 04:15 PM

____

The best thing to do is to take, say ten years data and compare the same month/month or quarter/quarter. You can create your own crude seasonal adjustment but with a chart like that you don't really need to - you can see.

In generaly - always compare quarter/month to the previous years quarter/months - using the the unadjusted numbers (for amusement you can do them as percentages of the labor force or the total working age population whether in the labor force or not.)

I stopped bothering about March 2003 - but it's not hard to do if you want to do it. And it's very revealing.

Posted by: Ian Welsh on December 31, 2003 05:09 PM

____

Whoa! I got curious, once I read Harry Possue posting, and look I found, which is not a good sign:

"On a per capita basis, according to the best available figures, the United States has three times more prisoners than Iran, four times more than Poland, five times more than Tanzania and seven times more than Germany."

US has over two million people in prisons. That makes about 700 inmates per 100,000 population -- highest in the world. US leads again!


Link:


Land of the free and land of opportunity and highest rate of imprisonment in the world.

Something is very wrong here.

I think Howard Dean and Democrats should take this up.


Link:


Posted by: Bulent on December 31, 2003 05:36 PM

____

Huf! I hit the post button before I pasted the link. Here it is:

http://www.charleston.net/stories/060103/wor_01jailbirds.shtml


Posted by: Bulent on December 31, 2003 05:38 PM

____

BTW, if I recall correctly, someone in another thread was advocating less public spending on education. I wonder if that contributor knew about this sad state of affairs in US per capita rate of imprisonment?

Posted by: Bulent on December 31, 2003 05:41 PM

____

Dave Johnson has, in fact, described pretty much what happened in December of 2002 and January of 2003. In December of '02, a 211k job loss was reported in the seasonally adjusted data (the largest loss in a year), with a 158k job gain reported in January '03 (the biggest gain since November of 2000). At the time, a roughly 100k drop in retail employment in December was followed by a 100k retail rise in January. Oddly, those changes appear to have been revised away in the interim, but the swings in the employment total have not. I have no explanation for that little oddity.

CM,

Your worry about a "shifted event" is pretty easy to accommodate in two complimentary ways. One is to keep an eye on reality, figure out what might be different from what will be assumed in the seasonal adjustment factors, then see if your assessment accounts for the data when they are released. Next, along the lines suggested by Ian Welsh here and lots of folks who watch data for a living, smooth out any series over a couple of months or so. The December 2002/January 2003 swings were mostly offsetting, with the swing in the 2-month average no big deal, while the 3-month average hardly wiggled between December and January (though it did between November and December).

The thing to keep in mind when figuring out whether all this seasonal adjustment stuff is worth it is the magnitude of swings in the raw data. There is usually something like a 2% decline in employment between December and January surveys, before adjustment, and a 1% decline between June and July. After adjustment, the swings in July and January are indistiguishable from those in other months. That is quite an accomplishment, from a statisticians point of view, even if we can quibble with individual results. Does it help in making policy? Very likely it does, but policy makers habitually adjust figures in their head along the lines that Spencer and Ian suggest. Same thing for economists working for investment houses and banks.

Posted by: K Harris on December 31, 2003 06:05 PM

____

(...thought I'd best chip in, being as how that one guy who is stil working in Ohio is me)

Dust mites are highly unambiguous, but you need an electron microscope to see them.

On a firmer note, I was amused to see the spin that Yahoo! News gave to this story; their headline was something like "Unemployment Claims Drop To Lowest Level During Bush Tenure". What do you suppose they meant by that?

Posted by: Frank Wilhoit on December 31, 2003 06:43 PM

____

Frank: that was what I was referring to as "fooling yourself and others". I generally do not presuppose "evil" bias in what media and government _agencies_ are reporting and doing, but somtimes I'm wondering, you know, where is the line between interpreting "cooked" (e.g. seasonally adjusted) data, and cooking them on purpose to show a desired result (please note, no suggestion of evil intent here). Sometimes people are just subconsciously biased to desire certain results.

But I give a lot of credence to people for having honest intent, even though I may appear cynical. It is easy for laypeople to just trust experts and misinterpret numbers when you don't know how they are produced, and for experts to not see certain "obvious" things because of professional bias.

I have some background in industrial pattern recognition and some in statistics (little in economic modeling), and I know how easy it is to bullshit even yourself by adjusting and tuning models, all with the best intentions and no dishonesty.

Posted by: cm on December 31, 2003 07:27 PM

____

Bulent: I have been reading several times already that US-American black males have a "1/3 chance of being incarcerated in their lifetime". This is probably based on some statistic(s), and I'm not sure whether it means that 1/3 of all currently living black males in the US have ever been incarcerated, and whether it only includes prison sentences or also being held in city/county jail after arrests. Whatever it means, think of it! One third!

Aside from the human tragedy, two statistically relevant effects are (1) these people are removed from the unemployment statistics, and (2) they are removed from the voting population (not sure for how long after the incarceration, but one thing that comes to mind is that in the 2000 presidential election, in the pivotal state of Florida there were quite a few complaints of people being turned away from voting stations based on claims they were convicted felons, rightly or wrongly).

Posted by: cm on December 31, 2003 07:51 PM

____

This seasonal adjustment looks like total crap to me. [What it really looks like is data fitting, which it seems is endemic to economics for reasons I don't understand.] What are they using for the adjustment models--a few years of data? There's no way that you can get a useful seasonal cycle from this; the signal-to-noise problem must be pretty bad. How much does the seasonal cycle depend on the actual level of employment--that is, how nonstationary are the statistics? And of course, when you allow for nonstationarity (as it looks like their revised model does), then the data fitting problem is that much worse.

It would be much better to use something really simple, since there is probably not enough information for anything more, but of course then there'd be a need for much fewer economists. If there's one thing economists know about (especially since that one thing doesn't appear to be a working knowledge of statistics), it's (their) job security.

Posted by: Matt Newman on December 31, 2003 08:16 PM

____

One third is large no matter how you look at it. The article in the above link reads that one out of every eight black males in 20s and early 30s were in jail last year and that over 40 percent of black male high school dropouts aged 22 to 30 were jailed last year.

Some thing is being done terribly wrong here.

I wondering how come this has not yet become a human rights issue on international platforms.

Posted by: bulent on December 31, 2003 08:27 PM

____

700 inmates per 100,000 population - another number to throw into the heap that is going to grow and share the same space as the unemployment numbers.

When Congress refused to extend Unemployment benefits 80,000 to 90,000 workers a week fell joined the uncountable.

At the same time nothing is being done to promote the creation of decent jobs and the contiued bleed of quality jobs is still going on unabated.

A lot is made of the 9.6 million people who are supposedly self-employed (see below) but if they were technology or business consultants before they became unemployed they are probably lying through their teeth because if you're in the category you know better than to tell anybody you're out of work if you do want to work.

As for the "out of the labor force" mentioned below, the underground economy will grow as long as "quality jobs" are unavailable - the jobs that have been created are eithe temporary, low pay, no benefits, or any combination of the above.

there is no place in here for Bush's overtime killing antics but there probably should be

from: LA Times:Jobless Count Skips Millions, December 29, 2003

The nation's official jobless rate is 5.9%, a relatively benign level by historical standards. But economists say that figure paints only a partial and artificially rosy picture of the labor market.

To begin with, there are the 8.7 million unemployed, defined as those without a job who are actively looking for work. But lurking behind that group are 4.9 million part-time workers such as Gluskin who say they would rather be working full time the highest number in a decade.

There are also the 1.5 million people who want a job but didn't look for one in the last month. Nearly a third of this group say they stopped the search because they were too depressed about the prospect of finding anything. Officially termed "discouraged," their number has surged 20% in a year.

Add these three groups together and the jobless total for the U.S. hits 9.7%, up from 9.4% a year ago.

...

One statistic proving particularly perplexing is the percentage of the adult population that is employed. This number rises during good times, as people are lured into the workforce, and falls during recessions as companies falter.

True to form, the percentage of adult Americans with jobs dropped from a high of 64.8% in April 2000, just as the stock market was cresting, to 62% in September the lowest level in a decade. If past recessions are any guide, those 5 million people who found themselves jobless should have driven the unemployment rate up to about 8%.

Instead, the rate never went much above 6%.

"More than half of the additional people who would have reported themselves as unemployed in a previous big recessionary period aren't," a puzzled UC Berkeley economist, Brad DeLong, wrote on his website. "They're reporting themselves as out of the labor force instead."

"Out of the labor force" means you're not working for even one hour a week and don't want to, either. It's the traditional category for students, married women with young children, flush retirees and idle millionaires.

A new way that people seem to be joining this category is by getting themselves declared disabled. This designation makes them eligible for government payments while removing them from the unemployment rolls.

From 1983 to 2000, economists David Autor and Mark Duggan wrote in a recent study, the number of non-elderly adults receiving government disability payments doubled from 3.8 million to 7.7 million.

The scholars present a case that the sharp increase isn't because the workplace suddenly became more dangerous. Instead, it has been prompted by liberalized screening policies, which make it possible to claim disabled status for, say, several small impairments as opposed to one big injury. Government examinations also have been downplayed in favor of the disabled's own medical records and the pain he or she claims to be experiencing.

At the same time, benefits have been sweetened. As a result, millions of individuals who lost jobs now have an attractive and permanent alternative to searching for work.

Autor and Duggan concluded that if disability payments weren't so appealing, many more people would be unemployed, boosting the jobless rate two-thirds of a point.

Another way in which people forgo an appearance on the unemployment rolls is if they decide to go into business for themselves. There are 9.6 million people who say they are self-employed full time, a number that rose 118,000 last month. Without the recent increase in self-employed, the jobless number would look much worse.

2004 is going to be a bitch.

Posted by: Jim on December 31, 2003 09:16 PM

____

Here are the links taht got swallowed:

80,000 to 90,000 workers a week: http://www.aflcio.org/yourjobeconomy/jobs/ns12182003.cfm

LA Times:Jobless Count Skips Millions, December 29, 2003: http://www.latimes.com/la-fi-jobs29dec29,1,6472624.story

Posted by: Jim on December 31, 2003 09:22 PM

____

Yeah, thanks. Remembered to add disabled to the not-counted UE list while riding home on the bus today, surrounded by the disabled and marginally retarded, those for whom a vehicle is a luxury. Why doesn't the Fed UE count permanent disabled?

And don't forget to add those who died on the job, or retired and weren't replaced! At least a dozen people at my work early-outed last year, and no sign they're gonna replace anyone soon.
Why doesn't UE factor in the dying and retiring?

Heck, 42,850 died on the highway getting to their jobs, and another 2,500 pedestrians or bicyclists get run over and killed every year. 30,000,000 people are admitted to hospital.
Why doesn't UE factor in those dying on the
way to work, or sitting in a hospital bed?

That's why I stated the UE index is bunk. Also, read somewhere some institute was tracking UE on their own, and their number and the Fed number started to diverge during Clinton, estimating the real number was more nearly 4% higher. The late 90's was really just a fluff job anyway.

Should have posited an alternate to reliance on a Fed UE statistic in an election year: what's the rate of jobs growth on Monster.com? The rate of growth on CareerBuilder.com? The rate on USAJobs, is it? What's the delta on seventeen largest newspapers' classifieds? It can be DHTML-charted as easy as a Unisys weather forecast, if someone built the e-links. "Jobs growth = xx%" Nobody would waste an ad if they weren't hiring.

Then, double-blinding the jobs stat, we could manually add together Fed's delta UE, delta prison, delta DSHS, delta morticians, delta hospitals, delta golden years and so on, and use that number to modulate the jobs report, in a lead-lag analysis.

"Last month's jobs.com + classifieds index rose 1.8%, but aggregate UE index lagged a persistent 9.7% over the prior month's 9.4%, suggesting those new jobs haven't hit the ground, as yet."

Wouldn't that way be a whole lot more rational, than relying on the Fed UE for prognosticating?

It makes me shudder to think a Fed spun-up UE number is 1/4th of the economic growth index, and that GDP, so artificially affected by DoD, and deaf, dumb and blind to deficit spending, is also in that index. It has no "real" meaning, yet everyone is betting their capital spending on this highly manipulated wishful thinking.

Couldn't we all work together in the next few months, pitch in through Dr. Brad, build the e-links, load the manual reports and start real-time charting what's really going on? That's what Internet is all about, free data access.

I mean, we can send a probe to Saturn's Titan, why the heck can't we determine real job growth?!

Posted by: Harry Possue on January 1, 2004 12:14 AM

____

"Couldn't we all work together in the next few months, pitch in through Dr. Brad, build the e-links, load the manual reports and start real-time charting what's really going on? That's what Internet is all about, free data access.

I mean, we can send a probe to Saturn's Titan, why the heck can't we determine real job growth?!"


A-ha! People are already talking direct democracy! :)

Posted by: bulent on January 1, 2004 12:55 AM

____

I wonder what our unemployment rate would be if the unemployable young men in prison were added in, and compared to other westernized countries.

Posted by: northernLights on January 1, 2004 10:39 AM

____

That could reveal some hair raising fact like this:

Unemployment rate in US is about the same as in Western Europe, but it looks lower in US because they throw some of the unemployed in jail in that country, instead of covering under social programs as Europe does.

Posted by: bulent on January 1, 2004 11:01 AM

____

That's a good point. When you're comparing you need to take a number of things into accounjt - and incarceration rate is one of them. Which group is most highly incarcerated in the US? Black males. Which group is most likely to be unemployed if not incarcerated? Black males.

But I'm a big believer in looking at unadjusted numbers. I'm also a big believer in looking at the unadjusted number of people employed over the working age population. That gets rid of a hell of a lot of BS and tells you quite a bit all at the same time. It's not the number that gets reported - but it is, in my opinion, the bottom line number. (Though, of course, it has to be compared to previous numbers. Different countries have different social setups - take a look at Japan, for example.)

Posted by: Ian Welsh on January 1, 2004 02:29 PM

____

"Unemployment rate in US is about the same as in Western Europe, but it looks lower in US because they throw some of the unemployed in jail in that country, instead of covering under social programs as Europe does."

When you do look at the BLS news release, they report numbers U-1 to U-6 in "Alternative measures of labor underutilization" "http://www.bls.gov/news.release/empsit.t12.htm". The official reported number is U-3 (5.6%), but U-6 (9.7%) comes closer to the truth, as it includes people labeling themselves as discouraged workers. It also is more in line with European unemployment measures.

This is one thing to bear in mind -- the unemployment numbers are based on a survey of 60000 somehow chosen households where people have to label themselves. I don't know the survey questions, but I'm not sure how many people would label themselves as "self-employed" out of pride or optimism. What "discouraged" means may also become a value judgement: you may say that they have lost hope, or that they don't even try, depending how you look at other people. Apparently the official choice is the latter, as discouraged workers are not included in the official rate.

There is also the issue of survey coverage; how are the 60000 selected, is there a bias for certain people not to appear in the underlying databases or being excluded (felons?), and as this survey is probably done by mail, people without a mailing address (homeless?) can never be reached.

Regarding manipulation of the statistic, it is known that e.g. in Germany there have been early-retirement programs for people 55+ especially during times of mass layoffs, and both the German DOL and a number of employers have been known to actively encourage people to use it. For the DOL, it cleans up the statistics and removes some administrative load, and for employers it avoids the hassles that come with involuntary terminations. But is it manipulation? That's not so clear-cut. For many people at this age whose job prospects are greatly diminished it is a valuable social program. So it is not just about pushing people off the statistic, that's perhaps more of a side effect.

There are many such gray areas, also exemplified by the people getting themselves disabled issue etc. It is too cheap to say they are really unemployed; they technically may be, but they make use of a social program that provides supplemental income and that is intended for this purpose (the occasional abuse aside). You cannot classify somebody unable to work, and also count them as unemployed. (It is a different question whether the benefit income is actually enough to conduct a dignified basic lifestyle.)

Posted by: cm on January 1, 2004 03:55 PM

____

We ought to keep straight, however, the difference between the data and their implications. A falling jobless claims mean a healthier economy. That much is pretty unambiguous, whatever we may think of one week's data. That is especially true when the jobless claims data are reenforced in a very strong way by ISM data. The ISM headline just hit its highest level since 1983, orders the highest level since 1950, and the ISM employment index its highest level since December 1999, when factory employment was not falling. The steady fall would start in about 6 months.

Posted by: K Harris on January 2, 2004 07:53 AM

____

Post a comment
















__