No, Krueger Didn’t ‘Prov[e] that Raising the Minimum Wage Doesn't Increase Unemployment’
By Thomas Firey
News of Princeton economist Alan Krueger’s recent death prompted well-deserved tributes and reflections on him and his academic and public policy work. (David Henderson offers some kind words here.) The eulogies underscore Krueger’s sharp and inquisitive mind as well as his reputation for integrity, careful scholarship, and concern for his fellow man.
However, some of the tributes go a bit too far when they claim that among his accomplishments was “proving that raising the minimum wage doesn’t increase unemployment,” to borrow from Bill Clinton. Barack Obama wrote something similar in a lengthy tribute. (In 1994–1995 Krueger was chief economist at the Labor Department under Clinton; in 2009–2010 he was assistant treasury secretary for economic policy under Obama and in 2011–2013 headed Obama’s Council of Economic Advisers.)
Krueger’s best-known work on the minimum wage, co-written with Berkeley economist David Card (another excellent scholar), was this 1994 American Economic Review article. In it, they employed a clever solution to the standard problem with empirical studies of the effects of a policy change: how to determine what would have happened in the counterfactual world where the policy didn’t change?
To overcome that problem, Card and Krueger exploited a natural experiment. New Jersey was set to raise its state minimum wage to $5.05 on April 1, 1992, while the minimum wage would stay the same in neighboring Pennsylvania. In February and March of 1992, Card and Krueger surveyed fast-food chain restaurants (specifically, Burger King, Wendy’s, Roy Rogers, and Kentucky Fried Chicken) in the New Jersey and Pennsylvania parts of the Philadelphia metropolitan area, asking staff size, the mix of full-time and part-time workers, and other information. They repeated the survey of those restaurants in November and December. Because all of the restaurants were in the same metropolitan area, Card and Krueger assumed that if there was a change in employment on the New Jersey side but not on the Pennsylvania side, that change could be attributed to the minimum wage change.
Sure enough, they did find a change: employment increased in the New Jersey restaurants while it fell in the Pennsylvania restaurants, contrary to what one would expect under standard economic theory. That finding shook what had become a broad consensus among economists that raising the minimum wage reduced employment. Hence, they concluded, in this case at least, monopsony forces in the labor market for low-skill fast-food labor appeared to have been repressing wages, and the minimum wage could be raised without disemployment effects—indeed, with some employment benefits.
I confess that I’m little-persuaded by Card and Krueger’s research, because it suffers from what I think is a serious flaw. Though they carried out their first-wave survey a few weeks before the minimum wage increase, the New Jersey law mandating the increase was enacted back in early 1990. That is, the fast-food chains had two years to prepare for the policy change. Employers typically don’t like to fire workers, so a rational strategy to prepare for the policy change would have been to reduce employment through attrition in anticipation of the policy change, rather than issue morale-crushing pink slips on March 31, 1992. Indeed, Card and Krueger’s first-wave survey data show the Pennsylvania restaurants averaged 23.3 full-time-equivalent employees while the New Jersey restaurants averaged 20.4.
Card and Krueger might respond to this objection by asking why, then, did the Pennsylvania restaurants have staff declines (to 21.2 FTE workers) while the N.J. restaurants had (very slight) staff increases (to 21.0)? But as they note in their article, the mid-Atlantic experienced a recession that summer. Thus, I’d argue, the Pennsylvania restaurants responded to the tough economy by contracting their staff while the New Jersey restaurants took different action to stay afloat. (Indeed, it’s interesting that the Pennsylvania and New Jersey restaurants converged on the same staff size.) Card & Krueger’s article may tell us what that different action was: the New Jersey restaurants experienced a much sharper increase in menu prices than the Pennsylvania restaurants between the two survey waves. That phenomenon poses a problem for Card and Krueger’s monopsony hypothesis; apparently, the restaurant owners weren’t sitting on excess rents they had wrested away from the workers. The phenomenon is a good fit with my hypothesis, suggesting there weren’t many excess rents to begin with. And, it’s worth noting, price increases are an odd behavior in a recession.
So, are Card and Krueger’s conclusions certainly wrong? No, no more than their paper proves standard economic theory on the employment effects of the minimum wage is certainly wrong. These arguments and research are simply bits of evidence in the effort to determine the effects of the minimum wage. If you remember Bayes’ Theorem, they are tiny pulls in the intellectual tug-of-war to accurately predict the outcome of a minimum wage policy change. And there are plenty of other tugs out there. My reading of the literature is that there are more, and stronger, tugs on the side that says minimum wage increases hurt employment as opposed to the side that says increases don’t hurt (or even increase) employment. If you want a sense of that literature, here’s a link to Charles Brown, Curtis Gilroy, and Andrew Kohen’s classic review for the late 1970s/early 1980s Congressional Minimum Wage Study Commission; here’s David Neumark and Bill Wascher’s giant review of subsequent research through the mid-2000s, and here are a review by Paul Wolfson and Dale Belman and another by Neumark of research since then. In the coming weeks, the Cato Institute will release a paper by UCSD economist Jeffrey Clemens discussing the latest research on the employment effects of the minimum wage.
This literature can be interpreted as suggesting that minimum-wage labor may not be perfectly competitive. However, this may be the simple result of “search” costs for workers who could move on to slightly better paying jobs, as well as other “sticky prices” phenomena, and not to large monopsony power by employers. (Hence it’s the minimum wage itself that causes the problem.) Regardless, it wouldn’t be surprising if small increases in the real minimum wage would have negligible disemployment effects, while larger changes—such as the “Fight for $15”—would have worrisome effects. Krueger seems to ascribe to this idea in this 2015 New York Times op-ed.
However, even if minimum wage increases contribute to disemployment, that doesn’t decide whether a minimum wage increase is good policy. That decision is a matter of values, not economics. Reasonable and honorable arguments can be made both for a low minimum wage (or none at all) that promotes a large number of low-paying jobs to complement the many higher-paying jobs in the economy, and for a higher minimum wage that forces a shift to higher-paying, albeit fewer, jobs.
I take a radical position: I not only think the minimum wage should not be raised, but it should be abolished and Congress should use its power under the commerce clause to prohibit state and local minimum wage laws. I explain this position here; in a nutshell, I think society is better off having lots and lots of entry-level jobs (to complement the many, many more above-minimum-wage jobs) so that new workers can more easily gain experience and build work histories that lead to better-paying employment. Still, I appreciate the values choices of people who believe differently.
That’s one of the reasons I’m saddened by Krueger’s death. Not only is it the loss of a decent human being—a loss that must be especially hard for his wife and children—but also the loss of a thoughtful and insightful contributor to that intellectual and values debate.
Thomas A. Firey is a Cato Institute senior fellow and managing editor of Cato’s policy journal Regulation.