Have You No Respect for the Law (of Demand)?!

The law of demand is a bitter pill for defenders of labor market price controls. Noted economic theorist Matt Yglesias has grown weary of appeals to “Economics 101” in the minimum wage debate. “After all,” Yglesias writes, “there’s a reason they offer more economics classes and you don’t get your degree after taking just one.” His American Prospect colleague Ezra Klein says of the law of demand that “It’s a good guideline, but it’s got no end of exceptions.” The minimum wage, of course, is one those exceptions.

They’re both right in general, if not about the minimum wage in particular. There are more economics classes, and they do teach exceptions. However, let’s not imagine that there is some advanced economic class in which you learn that the law of demand is false. (Well, no doubt there is somewhere. There is contradiction-friendly “paraconsistent logic,” after all.)

To use Yglesias’s misapplied example, Econ 101 principles do not stand to higher-level economics in the way that Newtonian physics stands to relativity and quantum machanics: as a useful, but literally false, simplification of reality. Economic laws are not strict laws of nature, codifying ineluctable relationships of necessity, and they do not pretend to be. So counterexamples are not ipso facto falsifying, and the law of demand is never replaced with a better, more empirically adequate, law. The law of demand is very, very empirically adequate as it is: It captures a ubiquitous regularity of human behavior that is abundantly comfirmed every moment of every day, and without which there would be no science of economics.

But it is just a regularity, like people flinching involuntarily when they hear a sudden, loud sound. It doesn’t have to happen, but it’s pretty surprising when it doesn’t. (“Is he deaf? Paralyzed?”) And when it doesn’t, there’s need for some special explanation.

Economic laws, like the principles of all the “special sciences,” are ceteris paribus generalizations: generalizations that are true other things being equal. Econ 101 lays out the basic laws and explains what follows from them ceteris paribus. Later, students learn about cases when other things are not equal – when there are exceptions to the generalization. So, it is always possible to argue that the law of demand does not apply in this or that kind of circumstance. A certain necessary auxiliary condition, which is almost always present, may be absent in a certain kind of case, causing the regularity to break down. But then, in order to predict an exception to the regular pattern, you need to cite the absence of the relevant auxiliary condition (e.g., “He can’t hear; that’s why he didn’t flinch”). I hope Yglesias is not also tired of Philosophy of Science 101.

Now, let’s note two things. First, you will be utterly hopeless in reliably identifying exceptions to a ceteris paribus law when you never grasped its logic in the first place. Exhortations to mind your Econ 101 generally aren’t exhortations to stop being so darn advanced. They are exhortations to actually comprehend the principles upon which advancement depends. And, second, the fact that a law is ceteris paribus does not mean you can deny its applicability whenever you want to. Political convenience tends not to be an appropriate auxiliary condition. You can’t wave your hands and just hope that a good argument is in some upper-level textbook you haven’t read.

If you want to say that a wage floor is not going to throw some low-wage workers out of their jobs (or prevent them from getting jobs), you’ve got to say, in a principled way, why not. The burden is on those who predict an exception to an immensely reliable regularity. The most popular principled explanation for the failure of minimum wage increases to create unemployment is a story about monopsony conditions for low-wage labor, i.e., imperfectly competitive labor market conditions in which there is a single buyer of low-wage labor (or a colluding band of buyers), that is able to set wages that workers have little choice but to accept. A simple model (Econ 101, even!) shows that under such conditions, an increase in the minimum wage, within a certain range, could even increase employment and raise efficiency.

Card and Krueger’s famed statistical work on minimum wage, which wage-increase advocates wave around as if it were proof of the Resurrection, tells a kind of very complex monopsony story. They recognize, unlike many of the people who abuse their findings, that their statistical results on minimum wage hikes, in isolation from further theory, can at best establish that other things were not equal at a certain place and time (e.g., fast food restaurants in Pennsylvania and New Jersy in the mid-1990s). This provides no basis for predicting the effects of future minimum wage increases unless it is accompanied by a principled theory of what general features of the situation were not equal.

Their theory, in a nutshell, is: “Turnover costs, imperfect information, search frictions, commuting costs, and inertia generate short-run, and possibly long-run, monopsony power for individual firms.” This is not exactly a simple condition, likely to apply uniformly across a huge, diverse country. That an increased minimum wage might not cause unemployment in some places where certain conditions apply does not provide a strong argument for raising it everywhere. And the monopsony story, as far as I understand it, establishes only that there is a range up to which the wage floor could be raised without creating a disemployment effect. In order to use Card and Krueger to support an increase to $7, you would need to provide evidence that the federal minimum is not already at the top of that range, and that $7 will not exceed it. Maybe somebody has done this, but I haven’t seen it.

Card and Krueger’s empirical work would constitute some evidence in favor of their monopsony hypothesis. But how well does the evidence support the theory? For a taste just from the Cato archives, try Douglas K. Adie and Lowell Gallaway’s review of C&K’s book Myth and Measurement in the Cato Journal, or “Sense and Nonsense on the Minimum Wage ” by Donald Deere, Kevin Murphy, and Finis Welch, in Regulation Magazine. Yglesias cites a petition of economists in favor of raising the minimum wage. Probably more informative is this serious 2002 NBER research summary by UC Irvine economist David Neumark, in which he reports that “although there may be some outlying perspectives, economists’ views of the effects of the minimum wage are centered in the range of the earlier [than Card & Krueger] estimates, and many of the more-recent estimates, of the [significantly positive] disemployment effects of minimum wages.” That is, C & K’s position is an “outlying perspective.”

But consensus tennis is a very silly game. Better is Neumark’s analysis of the state of play (in 2002) regarding the C & K studies:

More recent studies have used panel data covering multiple states over time, exploiting differences across states in minimum wages. This approach permits researchers to abstract from aggregate economic changes that may coincide with changes in the national minimum wage and hence make difficult untangling the effects of minimum wages in aggregate time-series data.

Evidence from these “second generation” studies has spurred considerable controversy regarding whether or not minimum wages reduce employment of low-skilled workers, with some researchers arguing that the predictions of the standard model are wrong, and that minimum wages do not reduce and may even increase employment. The most prominent and often-cited such study uses data collected from a telephone survey of managers or assistant managers in fast-food restaurants in New Jersey and Pennsylvania before and after a minimum wage increase in New Jersey.

Not only do these data fail to indicate a relative employment decline in New Jersey, but rather they show that employment rose sharply there (with positive employment elasticities in the range of 0.7).

On the other hand, much recent evidence using similar sorts of data tends to confirm the prediction that minimum wages reduce employment of low-skilled workers; so does earlier work with a much longer panel of states. Moreover, an approach to estimating the employment effects of minimum wages that focuses more explicitly on whether minimum wages are high relative to an equilibrium wage for affected workers reveals two things: first, disemployment effects appear when minimum wages are more likely to be binding (because the equilibrium wage absent the minimum is low); second, some of the small or zero estimated disemployment effects in other studies appear to be from regions or periods in which minimum wages were much less likely to have been binding. Finally, a re-examination of the New Jersey-Pennsylvania study that I conducted, based on payroll records collected from fast-food establishments, finds that the original telephone survey data were plagued by severe measurement error, and that the payroll data generally point to negative employment elasticities.

That is to say, C & K’s findings have been challenged.

Meanwhile, studies continue to appear emphasizing the hazards of minimum wage laws. I find Neumark’s recent paper with Olena Nizalova especially unsettling. They find evidence that minimum wage laws discourage teenagers and young adults from acquiring the human capital they need in order to get better jobs and higher wages later in life. That is, minimum wage laws work to ensure that those who already have the fewest opportunities to develop their capacities, have even fewer still. They say this baleful effect is strongest for young blacks.

Progressives find grand, symbolic political importance in the minimum wage. But isn’t their most important concern the welfare and prospects of the poor?