January 09, 2004

Bush Employment Forecasts

Back in February of 2003 the Bush administration began handing around a piece of paper from the Council of Economic Advisers that purported to forecast that--if the Bush administration's 2003 tax cut was enacted--payroll employment in December 2003 would hit 134.3 million (up from its January 2003 value of 130.4 million.

Nobody I talked to could figure out how the Bush CEA had arrived at forecasts of baseline employment growth plus effects of tax cuts that could possibly produce such a forecast: such payroll of employment growth averaging 354,000 a month seemed out of the question. The only theory advanced was that back in January 2003 somebody inside the White House had demanded a forecast showing huge honking employment growth starting now, and that the CEA had failed to resist.

I'm told the piece of paper actually had some influence in persuading some members of Congress to vote for the tax cut...

And, of course, today we don't have 134.3 million payroll jobs. We have 130.1 million. Economic forecasting is a black and inaccurate art. But to be off by 4.2 million in a ten month-ahead forecast of employment is truly remarkable.

Posted by DeLong at January 9, 2004 01:02 PM | TrackBack

Comments

I cannot help but notice the graph's scale...

Posted by: Bruce Cleaver on January 9, 2004 01:11 PM

____

The missing jobs game is still unfolding.

The president was not predicting 4 million actual jobs, but rather programs for 4 million jobs, or perhaps intentions for 4 million jobs programs.

Posted by: Asymetric Inflammation on January 9, 2004 01:15 PM

____

Brilliant Graph - Terrible Reality

Posted by: Mats on January 9, 2004 01:17 PM

____

"The president was not predicting 4 million actual jobs, but rather programs for 4 million jobs, or perhaps intentions for 4 million jobs programs."

Wonderful!

Posted by: lise on January 9, 2004 01:32 PM

____

see the next item.

Posted by: spencer on January 9, 2004 01:34 PM

____

oh well. Note that the trajectory for Dec 03 jobs WITHOUT the tax proposal also grossly exceeds where we actually are today. An expression of something seriously amiss in the models, or just some framing, to make the "jobs with the tax proposal" projection look more reasonable?

Posted by: David on January 9, 2004 01:40 PM

____

oh well. Note that the trajectory for Dec 03 jobs WITHOUT the tax proposal also grossly exceeds where we actually are today. An expression of something seriously amiss in the models, or just some framing, to make the "jobs with the tax proposal" projection look more reasonable?

Posted by: David on January 9, 2004 01:42 PM

____

It has been a tough week for Greg Mankiw! First the Krugman's review of Rubin's AEA paper, and now this embarrassing little graph.

Posted by: David on January 9, 2004 01:43 PM

____

If I read the graphs correctly, just a few months ago the CEA was projecting current employment to be 133M if Congress DID NOT pass the tax cuts. Current employment is actually 130M. Does that mean the tax cuts are responsible for REDUCING employment by 3M jobs? (Bush apologists would certainly say the tax cuts were responsible for CREATING jobs if they could - does it work both ways?)

Posted by: joe on January 9, 2004 02:07 PM

____

Bruce wrote: "I cannot help but notice the graph's scale..."

Yeah, it’s scaled so as not to mislead you into thinking being off by four million is a minor detail.

Posted by: David Moles on January 9, 2004 02:14 PM

____

"The president was not predicting 4 million actual jobs, but rather programs for 4 million jobs, or perhaps intentions for 4 million jobs programs."

4 million sims online jobs.

Posted by: bryan on January 9, 2004 02:34 PM

____

The CEA report says that its near-term projections were on the conservative side - in other words, the CEA expected actual current employment to be HIGHER than 134M.

Posted by: joe on January 9, 2004 03:23 PM

____

If I didn't misunderstand the Francophone channel TV5, Levi's is shutting down plants in US. So they want more tax cuts or what?

Posted by: bulent on January 9, 2004 03:32 PM

____

Does this mean there are 4,000,000 more jobs in India?

Posted by: masaccio on January 9, 2004 03:37 PM

____

Obviously, the expansionary effect of the tax cuts is way overstated. But there's a lot more that's odd about this picture.

Why would they assume that the largest impact would come within two months and would level off thereafter? Whatever modest employment gains the tax cuts would product would surely be expected to kick in after several months, no? Were they mischievously producing a forecast that would be wrong in the worst possible way as soon as possible? If I was ordered to produce a forecast with a wildly implausible end-point, and I had civil service tenure, I think that's what I would do.

Posted by: Dan Ryan on January 9, 2004 03:54 PM

____

"I've seen John Kitchen work. I'd bet on Kitchen. " (See Sept 2003 Archive)

Wont this all get revised away ten years hence ???

Posted by: 49eels on January 9, 2004 06:06 PM

____

Bulent: Levi's

No, that's correct (the plant closings, that is). Or I should say I read the same thing in a US publication. I think it's not to extort tax cuts. Don't be so cynical. Denim pants (which is probably a large (?) portion of what they are manufacturing) have become a commodity (if they were not before). I don't know how strong the brand awareness is in the US. Don't be fooled by the "coolness" factor of US brand jeans in Europe (at least a few years ago in Germany US brand-name jeans were highly regarded, and expensive to the point of being almost luxury items for people of modest means, and a number of people that I knew including myself brought some back from US trips). They supposedly have a hard time competing, and I even believe it.

Posted by: cm on January 9, 2004 09:49 PM

____

I guess I don't understand. The admin's forecasts show, at the time of the report (Feb.2003), that employment was 132 million. Your graph shows that, at Feb.'03, employment was 130 million.

Is one using household data and the other payroll?
Or am I missing something?

Posted by: andrew on January 9, 2004 09:52 PM

____

IS there really any way to produce such a graph/projection except to take a magic marker (the nice thick red kind) to somebodies carefully calculated projection and draw a sweeping curve that goes essentially straight up? Can't you just see somebody in the Bush Whitehouse doing just that?

I mean come on, can anybody even imagine some model that actually yields the historic 02 data then produces the projection?

Actually there is an explanation. Take the previous ridiculously optimistic projections. Assume that they represent the actual labor market equilibium, and that the actual situation in Jan 03 is a severe disequilibrium that will immediately correct .... what was that bit about symmetry between explanation and prediction?

Posted by: Dave Richardson on January 9, 2004 10:30 PM

____

Perhaps the administration will produce some chicken-scratch on notebook paper which will show how the jobs WOULD'VE BEEN created.

Posted by: Luke Francl on January 9, 2004 11:00 PM

____

"...I think it's not to extort tax cuts. Don't be so cynical."

No, no, I'm sure they want to extort further tax cuts! In fact I suspect that they're in cohoots with them neo-cons and so they conspired to make the point that even the tax cuts as is are not enough and there should be more more more or else the economy would just go pfffff!

:)

Posted by: bulent on January 10, 2004 12:17 AM

____

brad, what's the source of the data for the graph. I should know but can't find it.

Posted by: Bobby on January 10, 2004 01:31 AM

____

bls says that level of employment (thousands) was:

Jan 2003: 137790
Dec 2003: 138479

So this says employment increased, which is not right. Where is the data for the three million jobs lost?

Series Id: LNS12000000Seasonal AdjustedSeries title: (Seas) Employment LevelLabor force status: EmployedType of data: Number in thousandsAge: 16 years and over

Posted by: Bobby on January 10, 2004 01:39 AM

____

When Brad DeLong posts graphs on the great disparity between employment forecasts and harsh reality, he fails to mention that we're going to have a bold, exciting lunar base in the near future.

Posted by: Norbizness on January 10, 2004 09:47 AM

____

Bobby

Look at PAYROLL employment:
Jan 2003 - 130,356
Dec 2003 - 130,124

Posted by: Ronald McJob on January 10, 2004 01:59 PM

____

"I cannot help but notice the graph's scale..."

Such obvious efforts to distort often serve no purpose but to undermine the proponent's credibility. If you're trying to question another's credibility (I'm not arguing the point, just the method of point delivery here), you must be extra careful not to live in a glass house.

Brad, please don't become Krugman...

Posted by: Pragmatic Idealist on January 12, 2004 08:51 AM

____

PI — who are you saying is distorting what?

Posted by: David Moles on January 12, 2004 10:53 AM

____

David Moles -

um, not sure if you are making a sarcastic remark or serious inquiry, so I'll treat it as serious inquiry.

The graph Brad presents is distorted and exaggerated. The range presented is 5% of the midpoint. We can argue endlessly about information presentation (and yes, every job is important), but I find if very difficult to argue that placing the graph within such a narrow frame is unbiased.

The point can easily be made (and proven) without an attempt to use graphical scale to exaggerate it and induce an emotional response in the reader by exploiting framing bias.

Of note, I am neither pro nor anti-Bush... when it comes to politics and other celebrity industries I'm the political equivalent of a nihilist - I need evidence to make conclusions and being the Pragmatic Idealist I am, since I don't have that information, I reserve judgement.

My point is about credibility. My example of Krugman is simply this: he has no credibility with me as an unbiased "seeker of truth" as his columns increasingly seem to be just a forum for Bush bashing. I have enjoyed Brad's observations and do not mind the occasional partisan bent in writings, be they Brad's or others.

But when my bias-detection meter goes off, then I don't listen to people anymore. Because then to me, they just become another cog in one side of the spin machine. And spin, to use a (at least) double-entendre, is all around me. I come to Brad's page for thought-provoking discourse, not another shrill shill sermon.

Brad is certainly not Krugman, but this graph scale is just... too much for me to refrain from comment. My point is simply that rather than hammering home a point, the graph makes the point less credible as it puts into question the author's motives. And if I question the author's motives, then I question the facts and conclusions and whether or not the article is based purely on reason or on emotion.

Posted by: Pragmatic Idealist on January 12, 2004 01:24 PM

____

I can’t agree, PI — I don’t think the scale is an exaggeration at all, if you think about what the numbers really mean. The point of the graph isn’t absolute employment, but change in employment, and if you’re measuring that on this timescale, a range of five or ten million is completely appropriate.

What would be misleading would be to scale the graph from, say, 50 million to 200 million, when there’s no way on God’s green Earth that January 2004 employment could ever have been anywhere near the top or the bottom of the graph. The gap between reality and the Administration's projections would then seem to be just a little wiggle in the graph, hardly worth fussing over — but so would the actual drop in employment since the January 2001 peak. Try telling the American voter that drop is hardly worth fussing over.

Imagine for a moment that the title of the graph was “Change in Payroll Employment From January 2000 Baseline” and that the Y axis was labeled “Millions of Workers”, scaled from -2 to +4. Would you say the graph was distorted and exaggerated then?

Posted by: David Moles on January 12, 2004 03:06 PM

____

(If anything, that’s the problem with the graph — that the way the Y axis is labeled causes false alarms on the bias-detection meter.)

Posted by: David Moles on January 12, 2004 03:09 PM

____

Post a comment
















__