Brad DeLong's Semi-Daily Journal (2004)

It appears that 195,000 of the May job increase, and over 600,000 of the jobs created over the last three months, were estimated from the BLS Net Birth/Death model(see:http://www.bls.gov/web/cesbd.htm)
rather than from the CES survey. Any comments about the accuracy/manipulability of this model?

It's a weakness, but there hasn't been much wrong with the net firm birth/death model over the past five years...

I have a very basic question about the meaning of another number, and I apologize for sounding like an economic newbie (but I am).

When it's reported that "X jobs were created last month", does that mean "X people were hired", or "X positions were made available", or something else?

Thank you; I promise to pay more attention in class from now on.

Of course we know that all the tax-cut zealots will point to this as PROOF POSITIVE that tax cuts WORK.

Man, it must be sad to be what passes for a "conservative" these days.

I know that the administration's numbers have changed several times, so I am not sure how to judge this report. All in all, it appears to be a good number. But still, I wonder, did it meet or even beat projections? And how does the growing workforce factor into this report?

Unemployment for teens is up over 17%. I wonder how high it will go once all schools are out?

Workforce participation is still under 66. These are not good numbers going into an election.

http://pep.typepad.com/public_enquiry_project/2004/06/prof_delong_ill.html

CraigR,
This particular number reports the net increase or decrease in nonfarm payrolls. Let's say corporations lay off 500,000 people in one month and another million people quit or are fired nationwide, while all employers nationwide hire 2 million people. Nonfarm payrolls increase by 500,000 during that month.

Brian,
I don't know what the Bush administration's projections were, but Briefing.com polls economists every month to get their consensus on what the upcoming payroll number will be. This month, the consensus was that nonfarm payrolls would grow by 225,000. Generally, the low estimate among economists was 200,000 and the high was 250,000. So the May number of 248,000 was a little higher than the consensus but within the range of what economists had expected.

Holden, that's good to know.

But weren't there some signs in the last few weeks that employment number might change a little, as in become worse? Or are those signals, should they be what I think they are, going to affect June's employment numbers?

Three more months of good job numbers and we'll break into the black:
http://anonymous.coward.free.fr/temp/employment_growth_may04.png

By Thomas R. Keene
June 4 (Bloomberg) -- The 248,000 jobs created in May and the 346,000 generated in April move the trend in U.S. employment growth above forecasts made by Treasury Secretary John Snow last October. The reports, showing the biggest gains in manufacturing employment in almost six years, show the U.S. is on the way to creating the 2 million jobs by September that Snow predicted seven months ago.

`Doom-And-Gloomers’
``Even doom-and-gloomers who have focused on apparent weak spots in prior reports should find plenty to cheer in this report,'' wrote Christopher Low, chief economist at New York-based FTN Financial in a morning note to clients. Former U.S. Labor Secretary Robert Reich and other supporters of the presidential candidacy of Senator John Kerry of Massachusetts are focusing less on job creation than on the quality of the new jobs. Reich called the new jobs mostly low-paying service positions. ``The quality of jobs is not nearly good enough,'' Reich, now a professor of social and economic policy at Brandeis University near Boston, said in an interview with CNBC television. Drew Matus, senior financial market economist at New York-based Lehman Brothers Inc., disagreed. ``Today's report shows continued strength in the labor market and is good news for a consumer that is slightly depressed by news of inflation, higher interest rates and a never-ending election,'' Matus wrote in a note to clients.

I would be impressed if 1.7 million jobs were created over the next 2 months. But with the numbers of jobs lost since March 2001 now below 1.3 million, an average 180 thousand new jobs per month for the rest of the year is all it will take to have Bush43 not repeat Hoover's record of losing jobs on net during his tenure as President. Of course with the labor force growing by about 4 million during his tenure, this would still be an awful record.

Is there a convenient way to find out how Social Security contributions are trending relative to the number of jobs? If one $80K/year job goes away and two $40K/year jobs are created, the number of jobs goes up but SS contributions are unchanged. Similarly, if average income goes up but SS contributions are unchanged, the income growth must be either (a) for people making more than the SS cap ($97K this year?) or (b) unearned income. Since there's no cap on Medicare income, could Medicare contribution trends be used to differentiate between (a) and (b)?

'But with the numbers of jobs lost since March 2001 now below 1.3 million, an average 180 thousand new jobs per month for the rest of the year is all it will take to have Bush43 not repeat Hoover's record of losing jobs on net during his tenure as President'

But its not the rest of the year thats crucial so much as the time till November. Thats 4-5 motnhs away. The November jobs report probably wouldn't be out by the election. In order to make 1.3 million jobs before then, jobs growth would have to be around 260- 320,000 per month for the next few months. I doubt it'll be that high, what with interest rates rising. So Kerry and Co. can still use the net job loss since Hoover claim, or even worst job creation record since Hoover (which would still hold in any case).

If we add 850 jobs over the next 2 months, that means the average rate of job creation is 412/month

159 83 353 346 248 850 850 412.7142857

But at the begining of the year Brad told us that
"the administration Troika has implicitly forecast that during the rest of 2004 the U.S. economy will gain 470,000 320,000 payroll jobs a month. "

http://www.j-bradford-delong.net/movable_type/2004_archives/000253.html

So why has the Brad's required average rate mysteriously increased by 100,000 a month - from 320 to 410? He's still a bit confused about average rates and average levels I fear.

I'm an economic newbie, too, but why do I never really hear what KINDS of jobs are being created? Doesn't that matter? I mean, a real job with benefits and a future is different than a temp job with no benefits and no future. If most of these jobs are in the "service" sector, does that mean temps or part-timers or even full-timers with no benefits? You see what I mean--there are jobs and then there are jobs... is this info readily available somewhere?

Alan says "Of course we know that all the tax-cut zealots will point to this as PROOF POSITIVE that tax cuts WORK.

Man, it must be sad to be what passes for a "conservative" these days."

Why would we or why wouldn't we attribute the job growth to the tax cuts? If not the tax cuts, can we point to a cause or causes? Also, any thoughts on why the workforce participation rate isn't reported more widely? Seems to me like it tells a more comprehensive story. Thanks.

Drew Matus, senior financial market economist at New York-based Lehman Brothers Inc., is full of crap. I want to see where REAL jobs with REAL benefits are at. They are NOT out here where we are looking. Nice to sit in some office and say whoosh, we'll add 195,000 and say oh, 248,000 jobs created. That is not the kind of job like mine that was sent toi India.

At the end of the day it will be found that when Dell creates 10000 jobs for the survey , reality will tell us these were not in this country.

Just look at the Mann study. If you use the revised BLD Outlook you will see all the IT jobs to be created by "cheap" IT disappeared by half. That's right, from December when her studcy came out until February's new numbers her whole thesis is blown out of the water. The IT jobs she claimed were in the fastest growing category are replaced by menial jobs requiring "on the job training".

Drew Matus, senior financial market economist at New York-based Lehman Brothers Inc., is full of crap. I want to see where REAL jobs with REAL benefits are at. They are NOT out here where we are looking. Nice to sit in some office and say whoosh, we'll add 195,000 and say oh, 248,000 jobs created. That is not the kind of job like mine that was sent to India.

At the end of the day it will be found that when Dell creates 10000 jobs for the survey , reality will tell us these were not in this country.

Just look at the Mann study. If you use the revised BLS Outlook you will see all the IT jobs to be created by "cheap" IT disappeared by half. That's right, from December when her study came out until February's new numbers, her whole thesis is blown out of the water. The IT jobs she claimed were in the fastest growing categories are replaced by menial jobs requiring moderate "on the job training".

Mark, the relevant metric is not what Snow said in October; it's what Bush et al said in pushing for the 2003 tax cuts, namely, that they would result in 5.5M new jobs between 7/1/03 and 12/31/04.

darryl, this partly addresses your questions. The fact is, the only tax cuts that would directly make hiring cheaper would be cuts in the payroll tax. Since we didn't have those, the impact of tax cuts can only be seen in the indirect affects they produce - more consumption, more investment, things like that. Since all recessions end eventually, and since job creation tends to be a lagging indicator, you need some sophisticated analysis to figure out what level of job growth would have been produced without the Bush tax cuts, analysis beyond me (although perhaps not beyond some readers).

But i can say with certainty that by the metric of the bush folks themselves, the tax cuts have not delivered what was promised for them.

And you're right, darryl, labor force participation is a highly relevant indicator, but who among us can hope to understand why the media settles upon certain numbers and not others to fixate upon?

Thanks for addressing the basic soundness of the net birth/death adjustment model, Brad. I've got another followup question or two:

1) Am I reading the BLS table right, that the model says there's a net gain of 1.1 million jobs since the March 2003 benchmark survey, among businesses that have been born or have gone bust since the last time the Census updated its sample frame? (In other words, does one properly sum the monthly net gains/losses to get the gains/losses over time, which is what I've been assuming, or are the numbers already cumulative?)

If so:

2) Is it a matter of concern that that's all but a very small share of the net gain of 1.3 million jobs since March 2003? I've gotta admit it seems kind of shaky to me that of the apparent 1.3 million added jobs since March of a year ago, we only know about 200,000 of them due to statistical sampling, and we know about the other 1,100,000 only due to economic models that tell us they're out there.

You're the expert, you know about these models, and I don't. But to me, it seems like that's an awful lot of weight to put on models. I'm used to using models to try to predict the future, because there's no better alternative. But I'm a bit less comfortable putting that much weight on models to tell us what's supposedly already happened. To me, it seems to say we're making a projection about the present.

Presumably, the net birth-death model can tell us how many net jobs from businesses whose births and deaths haven't yet registered with the Census Bureau will occur this month or next, but we would treat that as a projection, rather than as fact, simply because we would then be talking about the future - even though we wouldn't have any less evidence to base our numbers on.

With regard to the birth/death numbers, it seems to me that what makes the April 2003 through May 2004 numbers more 'reliable' than those for June or July 2004 is solely that May 2004 is now in the past, but June 2004 isn't - even though the transition of May 2004 from future to past hasn't yielded new information about it. And I think that's what's bothering me the most.

My apologies for being so long-winded.

Brad-- why are you confident that the Birth/Deeath adjustment is accurate.
Over the last 4 months it accounted for 733,000
of the 1,010,000 jobs created, or 72.5%.

While I know the payroll survey has always had a problem capturing job gains in small & start up firms that ratio does not seem right. It is saying small, startup firms accounted for 72.5%
of the job gains while large, established firms only accountet for 27.5% of the job gains.
That does not seem reasonable to me.

There is plently of other evidence that employment is rising -- I am even seeing help wanted signs in the local stores in the Boston, route 128 area. But I find an adjustment of this magnitude hard to accept.

I no longer have the old adjusment data in my data base, but does anyone have any data to compare the new adjusment to the old adjustment data?

Brad-- why do you say there has not been an a problem with the Birth/Death adjusment for 5 years? It is my understanding that this is a new methodology that the BLS has just started using.

"Doesn't that matter?"

Of course it does, but consider two things. First, the fact that jobs are finally starting to appear takes a fair bit of the Democratic argument against Bush out to pasture. People, as one can imagine, will simply be happy that things are turning around. We already know that, despite what bad news may come, voters still like Bush on a personal level, so when you combine that with the good economic news, it gives him an advantage. Second, can't some lower-paying jobs get the ball rolling, for lack of a better phrase, so that we may get some higher paying jobs, or something along those lines?

spencer - looking at just the February through May period might be interpreted as cherry-picking, since the net birth/death model showed a net loss of over 300K jobs in January. But since the numbers look even worse if you go back as far as the BLS shows them (i.e. March 2003), the ratio's even worse: an eye-popping 85% of the new jobs since March 2003 are attributable to the net firm birth/death model.

I find Adrian Spidle somewhat disturbing. I mean, Donald Luskin I understand, because Paul Krugman really is a bigshot, with millions of readers. That doesn't mean that Luskin's coherent or sensible, but I understand why, in the grand scheme of things, he's there. But some guy who serves as the anti-Brad DeLong pointman? That's just weird.

It doesn't matter politically what the projections were or even whether teenagers have 17% unemployment. In fact, the avg voter doesn't even care that it takes more than 100k jobs/month to meet breakeven given the number of people potentially entering the workforce. The only things that matter are that the job creation numbers stay positive and greater than some low-ball whisper number.

>It doesn't matter politically...

I think what matters is when people sense a turnaround in their local job market. I don't where these jobs are, in the Bay Area it is still awful.

"Why would we or why wouldn't we attribute the job growth to the tax cuts?"

For the first tax cut, BushCo said, "This will create X jobs!" Over the next 12 months, net jobs were down. Same with the next tax cut.

Despite their orgy of cutting -- and expanding the government (why don't we look at non-farm and non-governmental jobs?), there are still fewer jobs in the US than when BushCo came into office -- and millions of new people in the job market.

If you let me turn a record surplus into recrod deficits, I promise you *I* would be able to create jobs!

Brad :Maurine Haver of Haver Analytics -- my data vendor & a very, very knowledgeable person on data-- says the Birth/Death model has only been used since last June when they switched to the NAICS.

If you are going to attribute the job gains to the tax cuts you also have to explain why the first two tax cuts did not work. Soemthing working one out of three times makes it very hard to argue cause and effect.

The Birth/Death model projections are visited in revised ( next 2(?) months) numbers, no? The idea being that the payroll may go undiscovered in the first month but not after 2 or 3. So these projections are "subjective" as some point out but the size of the revisions indicate that they are not too far off base (so far).
The genuine tempering detail for me is the ( duly noted by Mark reporting Thomas Keene) advance in manufacturing. After 41 consecutive monthly declines, the numbers show 3 months of increases. The telling detail is the size of these increases.
To write "The reports, showing the biggest gains in manufacturing employment in almost six years, show the U.S. is on the..." is shamless cheerleading. On a per capita basis we are still losing manufacturing jobs. The numbers are improving, but not by much.

> I don't where these jobs are, in the Bay Area it
> is still awful.

Sure, but you're talking about California. Bush would love to win in CA, but he's only running there pro forma, to force the Democrats to spend ad money there; there is no chance of a demagogue (or in this case Republigogue?) like Bush winning in states where a large fraction of voters is non-farm, urban, and educated.

Frankly, all that Bush needs to do is to provide a net of a few thousand jobs in OH, KY, PA, LA, FL, and WVa. Game over.

good site

ciao

nice site

nice

good

thanks