Elegant, Simple, Coherent, and… Oh Yeah, Totally Wrong

In a Labor Economics class, I had a great TA named Peter. He taught me a deep truth about labor markets: namely, that TAs sometimes teach better than professors.

If people looked like bad drawings, he’d look like this:

He also taught me one of the most enduring lessons I’ve learned about economics, modeling, and the limits of theory to explain the social world.

But the lesson wasn’t about those things. Not explicitly. It was about the minimum wage.

A model is a simplified representation of reality. A helium atom is reality, but this little picture is a model, because it captures some key features while leaving out others.

The classical model of the minimum wage goes something like this. Start with a world that has no minimum wage. The market should stand at a nice equilibrium. If you’re working a job, then the wage you earn must be worth more to you than the time you spend. Otherwise, you’d quit.

Meanwhile, if you’re an employer, then the workers you hire must contribute more to your business than they cost in wages. Otherwise, you’d fire them.

For example, say Bob is working at McDonald’s for $4 per hour. It’s worth it to him—in fact, he’d work for just $3 per hour. And it’s worth it to McDonald’s, too—Bob contributes $6 per hour to their bottom line. So both Bob and McDonald’s benefit from the exchange.

But now the government institutes a minimum wage of $7. What happens?

Well, unfortunately, Bob is toast. He’s only worth $6/hour to McDonald’s, but they’re required to pay him $7 per hour, which is a bad tradeoff for them. So they fire Bob instead.

Bob is willing to sell his labor for $4 per hour. And McDonald’s is willing to buy it. But the government won’t let them. That minimum wage law dooms Bob—and anyone else unable to provide $7/hour of value to a business—to unemployment.

So goes the classical model. It predicts that raising the minimum wage will drive low-productivity workers into unemployment.

But Peter showed me another possibility.

You’ve heard of a monopoly, which occurs when there’s only one company selling a certain product. Similarly, a monopsony occurs when there’s only one company buying a certain product—in this case, people’s labor.

So, let’s suppose that McDonald’s is the only employer around, and each new employee provides them with $10/hour of value.

Anita is willing to work for just $5/hour, so McDonald’s happily hires her.

Biff is willing to work for $6/hour. McDonald’s would like to hire him, but if they do, they’ll have to raise Anita’s wage too. (They can’t pay a new hire more than an experienced veteran!) So hiring Biff really costs them $6/hour, plus the $1/hour raise for Anita. That’s a total of $7/hour, which is still worth it, so they hire Biff.

Carmen is willing to work for $7/hour. By the same logic as before, hiring her will cost McDonald’s $9/hour. Still worth it, so they do it.

Diego is willing to work for $8/hour. But hiring him costs McDonald’s $8/hour + $3/hour = $11/hour. That’s not worth it to them. So Diego stays home.

Again, Diego is willing to work for $8. And McDonald’s is willing to hire him for $8. But it doesn’t happen—not because of government interference, but because of the structure of the labor market itself.

Now, introduce a minimum wage of $9. As before, McDonald’s hires Anita, Biff, and Carmen. But now, hiring Diego only costs them $9, because they don’t need to raise their other employees’ salaries simultaneously. That’s a good tradeoff. So they do it.

In this model, a minimum wage doesn’t drive up unemployment. It actually drives it down.

As Peter finished, I began mentally picking apart this model. My classmates did the same aloud: “Is that monopsony a reasonable assumption?” “Are starting wages really tethered together in this way?” “Doesn’t this depend on the government magically setting the minimum wage in the sweet spot between worker productivity and workers’ willingness to sell their labor?”

Peter handled the questions amiably and fairly. He acknowledged the model’s flaws, affirmed its strengths, and elucidated its predictions.

“But what makes this better than the traditional model?” someone finally asked.

Peter blinked. “Nothing, of course,” he said. “Nothing intrinsic to the model.”

“Then why teach it to us?” someone else asked.

“Because when one model predicts A, and another model predicts B, then it’s time to set aside the models and go gather some data. You can have the world’s most elegant model, but if it can’t predict what happens in real life, then it’s just a pretty piece of mathematics.”

To me, it’s a lesson that reaches beyond introductory economics. It’s about dogma. It’s about resisting the rigid principles of any ideology that prompts us to dismiss the counsel of lived experience.

Not to get too political, but let’s take the Tea Party. My problem isn’t necessarily that I disagree with their small-government principles. It’s that they fling those principles outward like a javelin, into a reality they haven’t full investigated. Let’s deregulate the economy! Let’s revive nullification as a doctrine! Let’s default on the national debt! And let’s do this all with utter conviction, never mind that no one can possibly know for certain what will come from any of these untested policies!

The Tea Party is all model and no data.

In the real world—the one out there, with air and trees and seven billion real people—any cleanly-stated principle is bound to be, at best, a crude first-order approximation of wisdom. The world’s too subtle for easy answers. Principles, like all models, are simple; the world, full of data, is not. So we’d best roll up our sleeves and learn what we can.

Oh, by the way: the research suggests that raising the minimum wage has a borderline-significant (read: rather small) negative effect on employment, though some argue (and this is a more fun punchline) that it may have virtually no effect at all.