Full Site Articles EconLog EconTalk Books Encyclopedia Guides

# Making Astrology Look Respectable

 Cesar Chavez on Immigration... Some Macro to read and to thin...

On the Internet, it is easy to find the John Kenneth Galbraith quip, "The only function of economic forecasting is to make astrology look respectable." Does anyone have a genuine citation for it? A page number somewhere?

Anyway, they're back at it. Mark Zandi assures us of a significant increase in jobs and GDP if we just "pass the bill," so to speak.

Steve Allen looks at Zandi's analysis and says,

The numbers are sobering: \$233k per job for the payroll tax cuts and \$350k per job for the infrastructure spending. And these jobs would only be around for the duration of the new stimulus package!

Thanks to Greg Mankiw for the pointer.

Here is some old-fashioned back-of-the envelope. You take the stimulus and assume a multiplier of 2. That gets you, with a little upward rounding, \$2 trillion. Compared to GDP of \$15 trillion, that is about 1.3 percent higher. Pretty much what Zandi gets.

[UPDATE: oops!! Thanks to Daniel Kuehn for catching my error. A multiplier of 2 would get you to \$1 trillion, which still is more like 6 or 7 percent of GDP than 1.3 percent. So Zandi's multiplier is a lot lower. Less than 0.5, in fact. ]

Zandi says you get 1.3 million new jobs. If you get \$2 trillion \$200 billion more in GDP with 1.3 million new jobs, then at the margin, these 1.3 million workers are adding over \$150,000 per worker to GDP. This is quite a bit higher than average GDP per worker, which is just over \$100,000.

Now, read Steve Allen's next paragraph.

My plan for zero unemployment: There were 14m unemployed workers in August. The \$447b stimulus package could be used to generate a check of almost \$32,000 to each and every one of them. As a condition of receiving that check, they would be asked to work at some organization, for profit or nonprofit, for one year. These jobs would last just as long as the stimulus package and some of them would no doubt turn into real jobs. Isn't this a plan everyone could support?

Steve Allen is saying that we could get to full employment if we could get people to work for \$32,000 a year. Note that this does not include any benefits. If they wanted the insulation that we call health insurance, it might cost half of their salary.

The average GDP per worker is \$100,000. If \$100,000 is the bogey, is it any wonder that a lot of people cannot get jobs?

Zandi says, in effect, that using stimulus to put people to work requires that they produce \$150,000 of GDP per worker. But that is an even higher bogey. [note: the "cost per job" is higher because the multiplier is less than 1. Unless I am making another mistake.

Tyler Cowen says that the marginal product of today's unemployed workers is zero. That is needlessly melodramatic. But considering the costs of health insurance, taxes, and so on, the marginal product probably has to be at least \$50,000 before you could say that entrepreneurs are leaving money on the sidewalk by not hiring. In any case, I think it is unlikely that there are lots of unemployed workers walking around sporting a marginal product of \$150,000, and that is just one reason I find Zandi's astrology unconvincing.

writes:

First and foremost - how is \$2 trillion 1.3 percent of \$15 trillion.

Second - the econometric task of identifying an impact estimate for a particular policy and the econometric task of forecasting are two very different tasks, with the former being on considerably firmer ground than the latter (Galbraith's point). I don't think we should confuse these two.

It's one thing to say "based on the evidence, we think that GDP will be 1.3% higher than it would have been". It's another thing to say "GDP will be \$15.3 trillion with the stimulus but it would have been \$14.9 trillion without the stimulus".

I think the latter statement is comparable to astrology (or at least comparable to a meteorological forecast two months in advance). I think the former statement is one that we can cautiously make.

Martin writes:

"The average GDP per worker is \$100,000. If \$100,000 is the bogey, is it any wonder that a lot of people cannot get jobs?"

vs.

"I think it is unlikely that there are lots of unemployed workers walking around sporting a marginal product of \$150,000"

Their (individual) marginal product does not have to be \$ 100.000 or \$ 150.000 for the average (economy-wide) product to be \$ 100.000 or \$ 150.000. With a division of labor the marginal product is always (far) below the average product.

In essence you're arguing that the crisis is structural and not cyclical: you can only come to said conclusion by the assumption that the division of labor destroyed in the crisis was not sustainable in the first place.

In other words you disagree with Zandi not because of his prediction but because of his assumption that the crisis is mostly cyclical.

writes:

The multiplier is how much GDP changes, given a change in government spending.

Y = C + I + G +/- NX

If G increases and C, I and NX don't change at all, the multiplier will be 1. If G increases are offset by lower C, I and NX, the multiplier will be 0. If the multiplier is higher than 0, jobs will be cheaper to the economy as a whole than they are to the government, since GDP will rise by some amount to offset the cost.

writes:

Martin - "...the division of labor destroyed in the crisis was not sustainable in the first place."

I don't exactly understand what the word "division" means in this sentence, but generally the statement that the employment level we had (or the economic growth we experienced) in the years preceding the crisis was not sustainable is pretty plausible, given the levels of financing it received from phantom wealth (in housing particularly) and \$600b annual trade deficits.

writes:

Arnold wrote:

Steve Allen is saying that we could get to full employment if we could get people to work for \$32,000 a year. Note that this does not include any benefits. If they wanted the insulation that we call health insurance, it might cost half of their salary.

An average family PPO health insurance plan for a small employer in the U.S. ran about \$9,300 in 2010. That would be about 50% of the annual pay for an individual earning \$8.25 per hour and working full time with no paid holidays, sick leave or vacation. (See here for a more complete breakdown of the costs an employer has to pay to maintain someone on their payroll.)

A key thing to note however is that for 25-50% of the people who are employed at or below this wage level, no health insurance may be necessary because they would likely be covered through their parents' health insurance.

writes:

We need to find a cheaper way for gov. to encourage job creation. Why can't we get the politicians to try getting rid of minimum wage and replace it with a wage subsidy.

Martin writes:

@ziel

the division of labor is the system of specialization underlying the pattern of production. I am not going to argue whether it was largely structural or largely cyclical, but the fact that people have to change their consumption pattern whilst re-balancing does not mean that the pattern of production that does not match the pattern of consumption during that period will also not match the pattern of consumption after that period.

It's akin to arguing that because a bridge collapsed, the city after bridge is unsustainable and should no longer be supplied.

mark writes:

Mark Zandi has one model, it is very basic Keynesian. It assumes all stimulus gets spent, the multiplier is > 1 and each 1% increase in GDP creates 1 million jobs. That's it. He's been endorsing every stimulus idea using this model since the change in Administrations. GIGO.

I should also note that he called the housing bottom in June 2009. You can look it up.

writes:

I think \$50,000 is closer to the bogey. Your point still holds.

Cryptomys writes:

Arnold,

this site
gives US News & World Report, Jan 11, 1988, without a page number, for the Galbraith quotation. I don't know any way to check the citation from my home.

Comments for this entry have been closed