How the government fudges job statistics

By Felix Salmon
February 17, 2010
Marketplace letters segment yesterday, Representative Peter DeFazio (D-Oregon) took issue with me saying that infrastructure investment is an extremely expensive way of creating jobs and "costs a good $200,000 per job". Just as well I didn't use the $1 million figure here, which I stand by, and which was fact-checked by the Atlantic!

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

In the Marketplace letters segment yesterday, Representative Peter DeFazio (D-Oregon) took issue with me saying that infrastructure investment is an extremely expensive way of creating jobs and “costs a good $200,000 per job”. Just as well I didn’t use the $1 million figure here, which I stand by, and which was fact-checked by the Atlantic!

The host, Kai Ryssdal, had no time to read out the letter in full, but has allowed me to reprint it:

Dear Mr. Ryssdal:

I have always enjoyed your show and have enjoyed past opportunities to discuss issues as your guest. However, I was distressed last Friday to hear a purported expert guest, Reuters blogger Felix Salmon, state with great certainty that infrastructure investment is an inefficient jobs creator because those jobs are so expensive to create. To back up his argument he claimed that it costs $200,000 to create one infrastructure job. However, he provided no source for this claim and you failed to challenge his assertion.

The Council of Economic Advisors has estimated that $92,000 in direct government spending creates one job-year, regardless of the sector of the economy. The U.S. Department of Transportation, arguably the most knowledgeable government agency when it comes to transportation spending and the resulting job creation, states that an investment of $35,941 creates one infrastructure-related job. Those two confirmable estimates are a far cry from the dubious $200,000-per-job claim from Mr. Salmon. Unfortunately Mr. Salmon’s assertion went unchallenged while the other guest, Heidi Moore of The Big Money, seemed to tacitly agree with him.

I am particularly sensitive about this issue since the AP ran an article last month on an “analysis” by AP reporters that used incomplete information to draw inaccurate and misleading conclusions about the success of the transportation infrastructure component of the American Recovery and Reinvestment Act (ARRA). The article claimed that “a surge in spending on roads and bridges has had no effect on local unemployment ” based on the reporters’ finding that “local unemployment rates rose and fell regardless of how much stimulus money Washington poured out for transportation.” However, instead of examining the impact of ARRA’s transportation investment on jobs in the transportation industry – an appropriate comparison – the reporters compared the transportation funding, which comprised only 6 percent of spending in the Recovery Act, to the overall unemployment rate. This led to a specious conclusion and ignores the fact that transportation projects funded by ARRA have created or sustained more than 250,000 direct, on-project jobs, with payroll expenditures of $1.3 billion.

I continue to support infrastructure investment as both a justified investment that future generations will benefit from as well as one of the most efficient creators of jobs, contrary to the beliefs of purported experts like Mr. Salmon and the so-called investigative reporters from the AP. I hope you will set the record straight. If you would like to discuss this further you can contact me directly at [redacted].


Peter A. DeFazio, M.C.

Chairman, Subcommittee on Highways and Transit

(In case you were wondering, the “M.C.” just means Member of Congress.)

I have no dog in DeFazio’s fight with the AP. But his attacks on me are just plain wrong. Infrastructure investments are simply not “one of the most efficient creators of jobs”, no matter how much DeFazio might want them to be, and the sources he cites to back up that claim don’t support it.

What’s at issue here is a ratio: I’m talking about dollars per job created. To get that number, you take the number of dollars spent, and divide it by the number of jobs created. DeFazio, by contrast, subtly tries to change the denominator when he says that “$92,000 in direct government spending creates one job-year”: he’s taking dollars, dividing by jobs created, and then dividing again by the number of years that each job is expected to last.

In the real world, of course, if you spend $300,000 to create a job which lasts three years, then that’s one job created with your $300,000, not three jobs. Only in DC would people attempt to claim that their $300,000 had created three “job-years”.

What’s more, the $92,000 estimate covers government spending in general, not just infrastructure spending. Infrastructure spending gets you low bang for the buck, in terms of job creation, compared to other kinds of spending — my example on the show was arts subsidies. A lot of government spending goes on creating new federal jobs: you get much more job creation per buck that way than you do building infrastructure.

And if you look at the CEA report, you’ll see that it carefully fudges the difference between jobs created, on the one hand, and jobs saved, on the other; in fact, it seems to used “created” and “created or saved” as synonyms. So if you’ve had a job for years, and you’re still in that job, you can still be counted in these job-creation statistics if the government somehow determines that you might not be in that job had the stimulus bill not passed.

The fact is that if you move away from vague country-level statistics and start drilling down to the actual number of jobs created by actual infrastructure projects, you never get anywhere near $92,000 per job. For instance, have a look at the job-creation statistics on this page.

A 5-mile stretch of highway, costing $50 million, creates a total of 79 jobs. That’s over $600,000 per job. Even if you divide that by two on the grounds that it’s a two-year project, that’s still $300,000 per job-year. In railways, a $15 million investment creates 12 jobs — that’s $1.25 million per job, and it’s a one-year project.

I’ve seen similar numbers surrounding hospitals, and higher numbers surrounding nuclear power stations — basically, infrastructure investment is an incredibly inefficient way of creating jobs.

But what of DeFazio’s $35,941 figure? I finally tracked it down to here — a report which does not say that spending $35,941 “creates one infrastructure-related job”. Again, it’s talking job-years, not jobs, but more importantly, it says this:

The FHWA analysis refers to jobs supported by highway investments, this includes ‘new jobs’ to the extent unemployed labor is hired; ‘better jobs’ as currently employed workers move into jobs with better compensation and/or full time positions; and ‘sustained jobs’ as current employees are retained with the expenditure.

This is an even looser definition than “created or saved” — it also includes substantially everybody who just gets a promotion as well, along with that ill-defined definition of “sustained jobs”, comprising people who just stay in the same job they’ve had all along.

Finally, what is DeFazio talking about when he says that “transportation projects funded by ARRA have created or sustained more than 250,000 direct, on-project jobs, with payroll expenditures of $1.3 billion”? Simple division here would seem to imply that each worker is earning no more than $5,200 a year, which can’t be right. But again, look at the footnotes — specifically in this report, which is the source of DeFazio’s statistic:

Consistent with the U.S. Department of Transportation’s reporting requirements, the number of direct jobs is based on direct, on-project full-time-equivalent (FTE) job months. One person working full time or two people working one-half time for one month represents one FTE job month. FTE job months are calculated by dividing cumulative job hours created or sustained by 173 hours (40 hours per week times 52 weeks divided by 12 months = 173 hours).

Yes, for the purposes of this report, the government has calculated the number of jobs created by taking the number of hours worked and dividing by 173. If you pay a man to wield a shovel for one year, working 40 hours a week, then hey, you’ve created 12 jobs! If you pay him overtime, and he works 60 hours a week, then you’ve created 18 jobs! If he keeps on working at that pace for three years, then you’re up to 54 jobs! All from one man earning one paycheck.

So it’s not just DeFazio, then: everybody in the government seems to be happy fudging job-creation statistics, especially by using job-years or even job-months rather than actual jobs, and also by eliding the distinction between jobs created, on the one hand, and jobs improved or saved, on the other. That’s worth remembering, next time you hear a politician kvelling about how the government is creating millions of new jobs.


Comments are closed.