Postrel, highways and jobs

One of the great things about blogs written by professional journalists is that they often contain a lot more information than gets printed. Newspaper columns, by their very nature, have to be a certain length and accessible to a wide audience. Blog entries, on the other hand, can provide lots of extra information which didn't make it into the paper. Daniel Radosh regularly regales us with things which got cut from articles of his in print; when Dan Drezner wrote his Foreign Affairs piece on outsourcing, he blogged all his sources.

Virginia Postrel is completely different. She has a column in the New York Times, and generally gives us much more information there than she does into her blog entries, which she uses basically for expressing opinions.

Recently, Postrel has devoted both her NYT column and many blog posts to the subject of federal highway funding. We first heard about it in this post, entitled "Gullible Reporter Alert", in which she excoriated the AP's Jim Abrams for printing nothing more than "propaganda" on the subject of road-building. The article was certainly one-sided, and seemed to accept as an article of faith that spending more money on US motorways is a Very Good Thing. But Postrel didn't really fisk it: she just complained that Abrams didn't provide any sourcing for the claim that 47,500 jobs are created for every $1 billion invested in federal highway and transit programs. Then, she told us to wait for her own article on the subject, forthcoming in the New York Times.

Right on schedule, the article arrived. It was headlined "Does Highway Spending Really Pay Off?" and it asked how much economic benefit the US actually receives from this expenditure. Based on an article in The Journal of Urban Economics, Postrel claimed that the rate of return on infrastructure investment has fallen from 15% in the 1970s to less than 5% today.

It's a good argument, and a good column: it provides ammunition for fiscal hawks on both sides of the political spectrum who hate to see government spending spiralling out of control even as the Bush administration continues to push further tax cuts.

It doesn't, however, say anything at all about jobs.

But Postrel was on a roll at this point, and soon published an email she received from a reader, bringing up that jobs number once again. The anonymous correspondent cited an even more anonymous "expert" at the Congressional Research Office, saying that "the figure was basically a garbage number that was cooked up for a DOT study in the mid-90s".

Then, yesterday, Postrel got an email from Arthur Jacoby, the alleged number-cooker. She blogged it, she said, "in the interests of fairness": not, one notes, in the interests of actually getting to the bottom of the jobs question. Jacoby attached a Word document which is precisely the source for the 47,500 jobs figure Postrel was complaining that Abrams didn't provide. It makes for interesting reading, and breaks the jobs down quite explicitly: in the first round, 12,453 jobs are created in the highway construction sector, along with 7,132 jobs in equipment and materials supplying industries. Then, in the second round, 6,939 jobs are created "because of the additional demand for inputs needed to expand output in industries that supply highway construction materials". Finally, in the third round, 21,052 jobs are created, reflecting "producer’s response to an increase in consumer demand for all types of goods and services".

Of the 47,576 total jobs created, then, a good 44% are connected to highway construction only in the most peripheral way. And, of course, all of these figure come from a computer program – something called the JOBMOD income and employment estimation model – which, like all economic models, will have internal weaknesses and may or may not reflect what actually happens in the real world particularly well.

But Postrel's response to Jacoby's email is fascinating: she doesn't really respond to anything he says at all. Rather, she snarks that "taxpayers paid experts to come up with the calculation" – er, yes, Virginia, would you rather that they hadn't? and continues with this gem:

But it still doesn't pass the smell test. Federal construction jobs pay more than $20,000 each, and this isn't the Great Depression; most people hired would be doing something else if they weren't building government roads.

$20,000, for those of you who are a bit weak on mathematics, is $1 billion divided by 50,000 jobs, and it's an utterly disingenuous figure. The fact that Postrel printed it at all makes me doubt everything else she writes: it's pure rhetoric, with a "la la la I can't hear you" relationship to the facts that Jacoby provided.

For one thing, Jacoby makes it clear that the $1 billion is not really $1 billion at all: by the time that state matching funds are added in, total government expenditures are $1.25 billion. Then, the number of jobs created directly when that $1.25 billion is spent on building roads is not 50,000, it's 19,585. So if you're going to do any division here, try dividing $1.25 billion by 19,585: the result is about $64,000. Which is much closer to how much jobs cost these days, and which does "pass the smell test".

Moreover, Jacoby is surely right to include more than just the first-tier jobs in his calculations. After all, if I'm spending billions of dollars on equipment and materials, it stands to reason that demand – and jobs – will be created in the industries which supply those materials. The third-tier jobs, which are far more trickle-down, we might argue about, but (a) Postrel doesn't; and (b) they're surely non-zero, in any case.

Postrel, however, doesn't buy any of this. Rather, her "most people hired would otherwise be doing something else" argument seems to imply that there can never be any job creation in an economy with a relatively low unemployment rate. If I'm hired by a construction company because they got a government contract to build a road, has my job been created? Not if you're Virginia Postrel, it hasn't: if I'd otherwise be flipping burgers or designing software, then my job laying tarmac somehow doesn't count.

What's more, there seems to be an implication in Postrel's argument that every job created building roads means one less burger flipper or software developer elsewhere in the economy. Let's say that I stop flipping burgers and start laying tarmac when the government contract hits town. OK, if you're just looking at me, I had one job before and I have one job now, so employment hasn't gone up. But when McBurgers hires my kid sister to replace me – and it surely will, given the $1 billion boost my local economy has just received – total employment has, in fact, gone up.

Postrel doesn't stop there, however. Check out her very next sentence:

Keep in mind that these job projections are not based on the assumption that highway spending is investment that increases productivity. (Her emphasis.)

This is just hilarious. The implication, to those people who haven't been following the story on her blog over the past couple of weeks, is that the "highway spending increases productivity" argument is better than the "highway spending creates jobs" argument. She neglects to mention, of course, that she's just written an article in the New York Times comprehensively destroying the "highway spending increases productivity" argument, while she hasn't, up until this point, addressed the quite obvious "highway spending creates jobs" argument at all.

So how does she do it? Any ten-year-old can see that if the government comes into town and gives thousands of people jobs building roads, then, well, jobs are being created. Only an economist could possibly disagree:

They assume that the spending is jacking up employment directly through the hiring of construction workers and indirectly through their spending. That Keynesian story only works if you assume lots of slack in the system.

It certainly looks here that Postrel is characterising "jacking up employment directly through the hiring of construction workers" as a "Keynesian story". Of course, it isn't at all. If I open one of Postrel's beloved nail salons, and hire half a dozen manicurists, then I have a feeling she would agree I've just created six jobs. She wouldn't tell me that those jobs are a Keynesian story which only exist if there's some mysteriously-defined "slack in the system" (I think she means unemployment, but it's far from clear).

When the government hires people, on the other hand, or gives money to construction companies who hire people, then suddenly the spending is "Keynsian" and therefore suspect. Yes, it's possible to argue about how much of a stimulus general government expenditure really gives to an economy and to the employment numbers. But you can't simply say that government spending never creates jobs in both the public and private sectors: just look at the economy of the Washington DC conurbation, where Postrel lives.

At this point, I feel I ought to say that, as an opponent of government pork in general, I think that Postrel is right about the highways bill. What's more, I particularly dislike federal roads spending, because it hides the real cost of America's car culture and does enormous damage to the environment. I daresay that spending $1 billion on the arts, say, would create even more jobs, with much less collateral damage.

Do I think that the government should increase employment? Yes. Do I think the government should do that directly, by spending money on programmes which are labour-intensive? No. But does spending money on labour-intensive projects create jobs? Of course. To think otherwise is to be blinded by idealism.

PS Virginia, if you're reading this, can you please date your individual blog entries? They're timestamped on the main blog page, but not on the permalinks.

UPDATE: Postrel has responded to this, and I've responded right back.

Posted by Felix at 18:04 EST