Friday, March 22, 2013

Gene schools me (although he graciously does not note that I conceded Bob's point)

I am too quick, sometimes, to focus on the point where I think the other guys is definitely wrong and concede the points that sound more reasonable to move past them.

I don't think minimum wage earners en masse are producing $22 of output for their employers and getting screwed. Monopsony power involves some exploitation but I can't imagine that much or else the marginal effect of the minimum wage would be far more positive.

But that does not mean the phenomenon has to be all that rare. Gene ably reminds us that asymmetric information's a bitch if you're on the right tail of an expected productivity distribution and don't have access to any signaling options.

Three points before sharing Gene's thoughts:

1. Bob suggests it ought to be easy to develop a business plan to take advantage of this - of course that need not be the case if it's costly to get information on all applicants' skills and only picking the undervalued ones.

2. Gene is defensive about being considered anti-marginalist. He need not be. He need not even invoke the "frictionless physics" analogy. Frictions in economics are marginalist - they just add new margins to think about.

3. This still might not be a wildly common occurrence (although probably more common than Bob or I originally suggested), because of internal labor markets. Once you observe an employee's productivity, given the high turnover in these industries, you're probably going to promote him and share some of those rents before you lose all of those rents.

Alright, here's Gene in his entirety because it's so good:


How Labor Markets Really Work

Bob Murphy asks Daniel Kuehn:

"Do you think there are at least 1000 workers in the United States who produce more than $20/hour for their employers, yet they are only earning minimum wage?"

My guess would be: "Certainly."

Too many economists have no idea how companies hire people. I have done it at several places. Here's what happens:

Someone higher up in the company decides some task needs to be done. They give you the task. You say, "Well, OK, but I'll need a couple more programmers."

"How much will they cost?"

"Good Java programmers get around $100,000 per year."

"Hmm... can you get by with one full-timer and one part-timer?"

"Yeah, I guess so..."

"OK, do it."

Most likely, no one in the company has any real idea how much this project is worth. No one has any real idea what the value of the output of these 1 and 1/2 programmers is. All that matters is whether the higher up is pleased with the outcome of the project. And so long as her superiors are pleased, and the company as a whole makes money, things can go on this way forever.

So, my guess is that there certainly are well over 1000 workers in the United States who produce more than $20/hour for their employers, yet they are only earning minimum wage. As well as whole boatloads who earn the minimum wage but produce far less for their employers. Bob attempts to dismiss the possibility of such failures of wages to equal marginal products by saying that, if they existed, we could create "a business plan for how we can hire these 1000 (at least) people at $10/hour, and still make $10,000 in pure profit per hour."

There are at least two problems with the above attempt at a reductio:

1) Since the employers don't know (for the most part) who these people are we certainly can't either!

2) It is quite possible someone could make employer A $21 per hour, but make far, far less if employed by B. The factors of production are not homogenous and are not perfect substitutes for each other. 

(Ah, [some] Austrians! This point is in the forefront of their minds when they wish to critique some interventionist scheme, but when it comes to defending the near-perfect efficiency of the market, it goes walkabout.)

The same thing applies to other factors of production as well: Where I have worked, when someone "needed" a new computer, no one in the chain of approval asked "What will its marginal product be?" And no one would have known how to answer if they had. No, we justified it because, "Carl has been here two years, and yet he has the slowest computer in the group, and he's really griping about it." And the request would get approved in good times and turned down in bad times, which is absurd in a marginalist theory of how prices are set.

Final note: I am not "anti-marginalist," any more than I am against physics models that feature frictionless surfaces. I am just noting that it is an idealization, and the real world only resembles the model in a very rough fashion.


  1. Obligatory modified advertising cliche: "Half of all workers have a higher marginal product than their wage, it is just that no one knows which half."

  2. Gene's explanation can account for individual cases of over/under-payment. It doesn't work as an explanation for the aggregate being biased in one direction.

  3. So ... another dead end in your attempt to account for a wages diverging from value-productivity?

    1. I think you're confused about the issue at hand here, Silas.

    2. The wages/productivity divergence question is a different question and a very hard question given that the productivity figures published aren't even exactly what we'd be interested in.

      The question at hand here is somewhat more narrowly circumscribed.

  4. Daniel: going back to the monopsony case: In Peter Diamond's 1971 article on search theory, he has a little thought experiment that is relevant here. Imagine a bunch of sellers with identical costs selling an identical good with consumers facing no search costs, one buyer per seller. Buyers have downward sloping demand schedules. Imagine Bertrand competition, so the price will equal marginal cost. Now introduce a small search cost. What Diamond argues is that now the only symmetric Nash equilibrium has each seller charging the monopoly price. Imagine any price other than the monopoly price as an equilibrium; every seller could increase profits by raising the price by the amount of the search costs, so it can't be an equilibrium. It doesn't matter how small the search costs are! So this tiny friction actually completely changes the model. Search frictions aren't like the "frictions" physicists talk about, where a small friction makes for a small divergence from the frictionless world - imagine dropping objects of different weight in what is almost a vacuum: they will drop at almost the same speed. Anyway, couldn't we make a similar argument, mutatis mutandas, in the labor case? If workers have no search costs and we have many employers, wages will equal marginal products. But even a small search cost leads to the monopsony outcome, no matter how many employers there are.


All anonymous comments will be deleted. Consistent pseudonyms are fine.