Minimum Wages, Immigration, and the Rise of the Machines


A Washington Post editorial today takes up the question of whether (and to what extent) illegal immigrants depress wages for low-skilled workers by flooding the market with cheap low-skill labor. It cites Harvard's George Borjas, who argues [PDF] that wages for high-school dropouts are 7.4 percent lower than they'd be without immigration, as a "pessimist" on the question, though as Ezra Klein and Matthew Yglesias have both argued, even when you're just looking at workers (as opposed to consumers and employers, who obviously benefit), that still makes it a massively winning proposition on net unless you decide to just not count in your calculus the welfare of people with the bad luck to be born on the wrong side of the border.

On the more optimistic end, the Post cites a paper by Berkeley economist David Card [PDF] which compares high-immigration and low-immigration cities and finds a negligible wage effect. Here's how the Post sums up his findings:

In low-immigration cities, it seems, employers don't necessarily respond to a paucity of low-skilled workers by bidding up wages to attract more of them. Instead, they may respond by investing in machinery that allows three low-skilled workers to do what six might do in a high-immigration city. Construction workers get extra trucks and power tools; gardeners get electric trimmers instead of manual shears.

Now, what's interesting here is that Card is most famous for being the co-author, with Alan Krueger, of a pair of controversial studies comparing fast food restaurants in New Jersey and Pennsylvania and concluding that Jersey's minimum wage hike hadn't lowered employment. (This is typically cited as proving that "minimum wages don't cause unemployment," which is a ridiculously broad claim no sane economist would make: Of course minimum wages at some level will cause unemployment. But the studies did purport to find that you could get away with moderate hikes from current levels without an unemployment effect.)

The curious bit is that the immigration and minimum wage questions seem like they ought to be mirror images of each other. In other words, if a relative shortage of unskilled labor leads employers to substitute capital for labor rather than bidding up wages to the level necessary to do the job with a higher number of unassisted people, then why don't they do the same thing in the face of minimum wage hikes? Presumably, employers don't care why labor is costlier, just that it is.

The simplest explanation is that one study or the other is just wrong. Another possibility is that the magnitude of the difference in wages that would be necessary for employers to attract enough workers is greater than the hike Card and Kruger looked at. If that seems unlikely in light of the relatively small effect even Borjas claims, bear in mind that employers in most cases are going to have to raise wages for all their workers, not just the marginal ones, to attract that marginal worker. That is, imagine an employer can do the same job with either 75 unskilled workers, or 50 plus machines. At $5/hr, the employer can only fill 50 jobs; to attract 75, she'd have to pay $6/hr. But since it'd probably be hard to wage-discriminate here and only pay $6/hr to those additional 25 workers, she's going to have to raise the wage she's paying the original 50 workers by $1/hr too. That makes the labor/capital substitution that much more attractive. But now assume the minimum wage is raised to $6/hr. That $1/hr raise for the 50 workers becomes, in effect, a sunk cost. Which means that the additional cost of going with the 75-workers-no-machines option becomes lower, so depending on the cost of the machines, that might now make it the more attractive option.

A simpler possibility, though—and one I find both the most plausible and the most interesting—is that you've got significantly more leeway for labor/capital substitution in sectors like landscaping and construction (to pick the two examples the Post uses) than in fast food restaurants. There might just not be that many cost-effective ways to automate a short-order kitchen so that fewer employees can do more work, beyond what most restaurants of that sort have already done. That's a possibility worth investigating, because it would imply that even if Card and Kruger's results for New Jersey restaurants are correct, they're of limited application. The same wage hike that has relatively little effect on fast food restaurants might cause significant labor/capital substitution in other sectors.