Brad DeLong's Semi-Daily Journal (2004)

I am confronted with the "best fit" slope to the shot-gun blast of data. So although the analysis seems to start well and end badly as you point out, the dots defeat me.
Good thing we don't have such scatter plots for, say, deciding whether the next flight will arrive safely.
Can we speak about the confidence levels of the forcasting if we must use these gun shot blasts?
I can not even engage in the number crunching with a slightly different number than 0.13. Why bother? The data exhibits no such promise and the exercise is going to be scrubbed by the politcal faction in the WH anyway. You take the time to correct them and it seems to me the data alone does not warrrant the effort.

One also wonders what the omitted factors are here. The forecast could be considerably different if, for example, those below-the-line dots on the chart are associated with the current stage of the business cycle.

As of January 2004, the number of jobs on nonfarm payrolls in the United States was 2,166,000 fewer than should have been the case according to the projection of what job growth would be if George W. Bush's 2003 "Jobs and Growth" plan were passed (forecast made by the Council of Economic Advisers--not by the administration Troika).

Note that 2.166 million = (10 months to election)*
2.6 million jobs growth forecast 2004)/(12 months in year) precisely.

The jobs forecast is probably tongue in cheek

An interesting graph might be a plot over time of the error of this model relative to actual employment. I.e., how has the predictive value of this model varied over time?

What I wanna know is: What's the p-value? The problem is that I would imagine there's a lot of autocorrelation, which is going to bring down the significance.

The data reflects overlapping periods. Regardless, I had always wondered what the relationship is. Thanks for posting this.

The chart is a pretty good example of why every single piece of marketing material my firm produces has "past performance is no guarantee of future results" plastered all over it.

Pity the CEA doesn't have an SEC/NASD compliance officer.

It's an extremely crude model. The point is that the White House is releasing information that can't be combined to form a consistent picture.

calmo -- Best fit should refer to a least-squares (OLS) estimate. The estimate is invariant with respect to changes in units. I think you want to say that it's too close to zero to be meaningful, but there's absolutely no way to tell that from looking at the estimate alone. It's pointless. At minimum, you need the parameter variance to conduct a t-test.

don freeman -- If the dots below the line are associated with our current position in the business cycle, that only serves to make Prof. DeLong's strong conclusion more conservative.

What liberal said: autocorrelation.

Maybe I don't understand how temporary workers are counted. Are they considered employed when not on assignment for the temp service? In which case there shouldn't be any gain in employment when they move from temp to permanent, and this graph might actually measure something. If temp workers are considered UNemployed when not on assignment, then when they get an assignment, a job is created. And this graph then looks like one measuring the correlation of sales of new car seats to sales of new cars. Surprise! A very strong relationship.

I think the relationship between temp services and aggregate employment growth has probably been broken by post-bubble structural change.

In a typical demand-driven cycle, businesses respond to the initial signs of strength (or weakness) in sales and production by running up (or down) temp employment. It is only after the recovery (or downturn) has progressed and become more widely acknowledged that full-time employment picks up (or runs down) more forcefully.

Accordingly, temp services demand appears to "lead" aggregate employment growth. But logically the real action is not in temp services. It is in the serial correlation of aggregate demand itself; i.e., in the dynamics of the traditional business cycle.

If I am right about that, then there is no reason to believe that the temp services relationship will hold in this cycle, which is being driven by NON-auto-correlated demand pulses interacting with post-bubble structural forces, such as the heightened focus on cost control. Indeed, I would say that the recent concentration of employment growth in temp services is much more a reflection of these structural influences than of the first stirrings of demand.

It is a bit late to speak of first stirrings when the recovery has been on -- in fits -- for a couple years now. And let's face it: this rule of thumb has had a very bad track record in recent quarters. Since the recession ended, we have had two episodes of temp services enthusiasm both of which have amounted to nothing.

Of course, employment will eventually pick up, and at that point someone will be able to point back and see that temp services "led". But what will really have led will have been the Fed's insistence on providing enough aggregate demand to eventually absorb the labor supply. Temp services employment will not likely have provided much insight into the timing.

There is a simpler calculation for what the CEA said before a message guy (generic Rove) told them that they had to predict 2004 avg employment at least equal to Jan 2001 employment. 1.4% productivity growth is calculated back from the conclusion (typical Bush admin style). The quote shows that the lowest level they can calculate from data is 2.1. The serious projection was non farm business sector output growth = 4.4 - 2.1 (a little higher than your number). Wait where did the extra 0.4 come from ? "Predicted" GNP growth = 4 -2.1 is almost exactly your number.

I guess they nudged an additional 0.4 on composition (no real guess on that) then tossed in an additional 0.7 on "hell if they push us like that we don't care how silly they look in the end. I'm sick of this and going back to my teaching job anyway."

To me the interesting question is why did generic Rove insist on something so stupid. I mean this is February. It doesn't matter yet. Stories now on how the Bush admin admits that it will be the first since Hoover to see negative job growth will be old news by election day. Stories about employment lower than when Bush was sworn in (can't be helped) And 2000000 lower than predicted in the most recent Economic Report of the President wont't be old news. The only story is that they really have forgotten that there is a real world out there that they can't spin.

Outsourcing and job losses have nothing to do with Bush. It is the economic era we live in that's to blame. For years the US and Europe were more prosperous than other nations, because they made products other nations could not make. Either because of the advanced technology or because of the closed markets in the West, which took away export prospects and profitability for other nations. It took a long time, but Korean engineers can now build a technologically advanced ship as well as American engineers. With the difference in wages, this means a whole industry moves out of the country. How do you stop outsourcing? By stopping other nations from gaining advanced technology? This is impossible. By protecting our market, so foreign companies can not hope to export and regain their investments and therefore will not invest in certain industries? This is impossible. If we do not go with the time and buy the best you can get, other countries like China might get an advantage over us. Foreigners are not stupid. American power is sustained by staying on top. If you discard advancement, we will wake up one morning and find "communist Russians being the first to put a man in space", or an equivalent of that. The only thing you could do to stop outsourcing is to pay American workers less money. But how low can you go? A chinese worker gets a dollar an hour. Economists know that a country like China could manufacture all goods all the 6 billion people on Earth need. With their low wages. This might happen. What should American, European and other workers do? Nothing. Kapitalism is by it's nature, the best economic system, but also a suicidal system. Kapitalism dictates that all manufacturing jobs go to China. Other countries better start a 1 child a family population reduction policy, because not too long from now there will be a lot of unneeded workers everywhere. It's either population reduction or poverty on a massive African scale. And poverty in the US will lead to a Weimar like Republic and we know what follows after that. An American Third Reich.

BTW, guys, did you see the Jagdish Bhagwati op-ed in the New York Times?

http://www.nytimes.com/2004/02/15/opinion/15BHAG.html

I'd put my own signature under Jagdish Bhagwati op-ed except that he sort of becomes vague and shies away from exactly what should have been said in there in the last paragraph: The Bush policies are wrong wrong wrong vis-a-vis "the end" as in "... In the end, Americans' increasing dependence on an ever-widening array of technology will create a flood of high-paying jobs requiring hands-on technicians."

He is also completely ignoring the socio-political aspects of what is going on.

Database Administrators and Network Administrators are going overseas. The "hands on" techwork is lower value wire pulling, while the abstract knowledge work is going overseas.

In short, the US is doing the reverse - it isn't shipping the low value jobs elsewhere, it is shipping the high value ones. This makes economic sense, a high value job is much more profitable to ship overseas.

But it isn't good long term for the economy to be shipping high value jobs overseas while creating low value ones.

Even more so when the economy is, basically, dependent on the housing sector for future growth.

"Starting from today's 130.1 million, that would imply a forecast not of 132.7 million payroll jobs for June-July and not of 134.5 million payroll jobs for December, but of 131.3 million for June-July (and for the annual average) and 132.4 million payroll jobs for December. "

I will note that temporary help dropped in the most recent jobs report - clearly, whatever hiring the temporary help pointed to has started to occure. 100,000 jobs a month, mostly in the lower end of the spectrum. Whoop it up people, then get back to flipping burgers.

The bagwagti article is a little sobering but not much. Have to agree that Kerry's "Benedict Arnold CEO" view is another triumph of politics over economics. But find the last line: "In the end, Americans' increasing dependence on an ever-widening array of technology will create a flood of high-paying jobs requiring hands-on technicians,..." unconvincing.
So the current ebb will eventually turn and we will have a vibrant high tech sector.(?) After all, we have an increasing dependence that will need servicing. But are the "hands-on technicians" the brains here --they are offshore, no? There's a distinction between the mechanical engineer and the mechanic (an analogy which may not be appropriate). So the 'mechanics' will be 'flooding' back?
Moreover, the increasing dependence on technology seems to be generating less servicing by this cadre, not more ( looking at electronics for example. I am thinking about modular fixes as opposed to component repairs.)
Sorry but this one doesn't fly very far for me.

The data should control for stage of the business cycle and remove overlapping periods. Or at least test for whether employment % gains last 6 months correlates with employment % gains next 6 months to see whether temp employment predicts employment better than employment predicts employment.

"unconvincing."

Of course it is unconvincing, it is a flat out lie.

Software was an area where America had such a large competitive advantage that we could pay very high salaries for a period of time. That kind of wage gap is going to generate huge economic pressures to find a cheaper source of labor. Now that that source has been found, it is absolutely natural for the jobs to flow to it. In fact, almost no amount of government intervention could stop it, since software costs are generally built into the price of something else - few people load up their shopping carts with hours of UNIX system adminsistration, and watching people write EJBs isn't a spectator sport that people turn to on ESPN. Software is, mostly, a cost for the production of something else people buy. If the US were to artificially constrict the movement of jobs overseas, then someone else would hire the Indian programmers and use the cost advantage.

Paying everyone in BushBucks helps ease the relative wage gap, but only for so long. It isn't natural for programmers to make twice what other comparably educated professionals make - at least not for very long. More and more people trained in computers - because of the wage advantage, and supply is now pushing price down because it is closer to demand.

If people want to have high paying jobs in the US, then we have to be creating demand for goods that no one else can make. Either that, or get used to being less well off.

This is the most intellectually dishonest Economic Report of the President ever issued. Never have there been election year manipulations this widespread.

Elson, thanks for posting Bhagwati's article. Bhagwait probably had to write it for the NY Times because Krugman is so tendentious that he won't agree with even an economic truism if it indirectly offers slight support to anyone afilliated with Bush.

Stirling Newberry, "It isn't natural for programmers to make twice what other comparably educated professionals make - at least not for very long."

And why is it natural for MDs to do so?

What about the money supply? If its growth is slowing or it's dipped below inflation then we could see a double-dip recession. It has been declining since about September.

Not that this came from the administration, but Business Week (don't remember the issue date, but it's the one with the AMT cover story) had a short bit about why productivity growth might drop: it turns out that, historically, productivity growth has slowed after the dollar has dropped. The rational is that since imports become relatively more expensive, US producers have less incentive to increase productivity. The person doing the research predicts that productivity growth will be something like 1.6% this year.
There's no article on their website I can point you to - wish I could remember more of what it said.
Dave

Oldman (nice detour)
I just checked the M1,2, and 3 numbers and you are mostly right (M3 is in doubt). Will the mortgage rates adjust to enable more equity extraction to cover this short fall? I mean how much room do Fanny and Freddy have in manipulating say, even lower ARM rates? There are also tax cuts that come on-stream shortly, no? The site that provided me with the previous numbers also had some scary forcasts for oil : http://www.marketvector.com/leading-indicator/m3-money-stock.htm

In today's NYT (2/18/04):

Bush Officials Offer Cautions on White House Jobs Forecast
By EDMUND L. ANDREWS

Published: February 18, 2004

ICHLAND, Wash., Feb. 17 — Treasury Secretary John W. Snow distanced himself on Tuesday from the Bush administration's official prediction that the nation would add 2.6 million jobs by the end of this year.

That prediction, which is far more optimistic than that of many private sector forecasters, was part of the annual economic report released last week by the White House Council of Economic Advisers and was immediately echoed by Mr. Bush himself.

But on a tour through Washington and Oregon to promote the president's economic agenda, Mr. Snow and Commerce Secretary Donald L. Evans both declined to endorse the White House prediction and cautioned that it was based on economic assumptions that have an inherent margin of error.

"I think we are going to create a lot of jobs; how many I don't know," Mr. Snow said, adding that "macroeconomic models are based on a lot of assumptions" and are "not without a range of error."

Unemployment and the nation's surprisingly sluggish pace of job creation has become a significant political weakness for Mr. Bush, who is on track to be the first president since Herbert Hoover to end his first term with fewer jobs than when he started.

. . .