Almost 30 years ago, New Jersey was raising its minimum wage and two Princeton University economists, David Card and Alan B. Krueger, decided to study the impact. Surprisingly, they “found no indication that a rise in the minimum wage reduced employment” in the state. Immediately, this result became hotly debated, and it has helped shape our political discourse ever since.
On Monday the Nobel committee awarded this year’s prize in economic sciences to Card, reigniting the controversy (Krueger died in 2019 and so was ineligible for the prize). The committee also named two other laureates for their work on developing techniques to measure the effect of minimum wage increases and other policy changes, Joshua D. Angrist and Guido W. Imbens. Card is getting most of the headlines because of his famous study, but when it comes to what he and Krueger found, count me among the skeptics. A look at the work of Armen Alchian, a pioneering economist at the University of California, Los Angeles, who died in 2013 at nearly 99, explains why.
Economics 101 has always been clear about the impact of setting the minimum wage higher than the prevailing rate in a competitive market: It raises the pay of some low-skilled workers while reducing the pay of others—those who lose their jobs or see their hours cut. The higher the minimum wage is raised—whether to $10, $15 or $100 an hour—the worse the negative results become.
Yet, the minimum wage remains a divisive issue, even among economists. For one, they disagree on the effect on employment. Presumably, a $100/hour minimum wage would grind the economy to a halt. Equivalent to $200,800/year for a full-time worker, a legally mandated wage that high would make it impossible for most people to find jobs outside of black markets. But the effects of a $10/hour ($20,800/year) or $15/hour ($31,200/year) floor on wages are less clear.
Countless academic studies have tried to measure and explain the consequences of raising the minimum wage, but this isn’t easy. While we can observe the change in the economy after a minimum wage is raised, we cannot observe how the economy would have changed anyway. For example, suppose employment falls sharply after a rise in the minimum wage. Is that because the increase had the expected result? Or was there a confounding factor, such as a recession or another policy change, that was the true cause? We can ask similar questions if employment doesn’t fall. This difficulty is common throughout economics because we can’t run controlled experiments on the economy.
“Difference-in-difference” studies attempt to square this circle. In these natural experiments, researchers identify a “control group” as a baseline to compare with the “treatment group.” Angrist and Imbens developed this technique and launched the “credibility revolution” in economics. It quickly became an alternative to the highly mathematical methods that many economists use. These often make large assumptions about how people behave and how the data are generated. But studies using natural experiments make fewer or smaller assumptions. In general, studies relying on smaller assumptions produce more credible results.
Crossing the Delaware
The minimum wage study by Card and Krueger provides an illustration. In April 1992, New Jersey raised its minimum wage. Across the Delaware River, Pennsylvania left its minimum wage unchanged. The two economists surveyed fast-food restaurants in both states shortly before and after the new law took effect. Using eastern Pennsylvania as the baseline, their idea for how to measure the impact of the minimum wage hike was to take the change in New Jersey’s employment and subtract the change in Pennsylvania’s employment—the difference in the difference. They concluded that increasing New Jersey’s pay floor didn’t have any impact on employment. The state’s employment fell, but Pennsylvania’s fell by more.
Other economists quickly raised questions about Card’s and Krueger’s results. Was eastern Pennsylvania a valid control group for New Jersey? Did New Jersey restaurants, anticipating the higher wage, lay off workers before the increase became official? Did the increase have little impact because the prevailing wages for low-skilled workers were already higher than the new minimum wage? Some economists countered the objections and ran improved studies. Like a tennis ball, economists have batted the issue back and forth for decades.
Alchian’s work suggests a more fundamental critique of the Card and Krueger study. In his 1950 paper, “Uncertainty, Evolution, and Economic Theory,” he argues that it’s impossible for companies to maximize profits, as economic models assume, because in the real world there is too much uncertainty. Companies rarely have all the information they need to make the best decisions, and even if they did, there’s usually no unambiguously best course of action. For example, one strategy might produce higher profits but carry a higher risk of a big loss. A second strategy might mean lower profits but also a lower risk.
Markets Force Companies To Evolve
On its face, dropping the assumption that companies can maximize profits seems to explain Card’s and Krueger’s conclusion. If companies are not profit-maximizing, then we have no reason to assume that they consciously reduce their use of labor in response to higher minimum wages.
But, Alchian argued, companies face the pressure of competitive markets and are always adapting to changing circumstances, often by chance and with limited foresight. Companies that adapt well will have higher profits than ones that adapt poorly. Less-profitable companies will imitate more-profitable ones, copying their “best practices.” Companies losing money will go out of business.
The process is like biological evolution, Alchian noted: “The economic counterparts of genetic heredity, mutation and natural selection are imitation, selection and positive profits.” In the case of the minimum wage, his evolutionary framework predicts that most jobs will be lost only after a delay. It might take years for companies that don’t adapt to higher minimum wages—such as those that continue to employ large numbers of low-skilled workers—to go out of business. Yet, to guard against confounding factors that might affect the experiment, Card and Krueger measure the treatment and control groups shortly before and after the policy change. As a result, their study and similar ones miss the longer-term effects.
To the extent that Alchian is right, the costs of minimum wage hikes are not easily measured with difference-in-difference studies. His framework suggests that labor economists should be skeptical of strange statistical results that contradict tried-and-true economic theory.