Friday, May 13, 2016

Michael Strain and James Kwak debate Econ 101

Very interesting debate over Econ 101, between Michael Strain and James Kwak. Strain attempts to defend Econ 101 from the likes of Paul Krugman and Yours Truly. He especially criticizes my call for more empirics in 101:
Critics suggest that introductory textbooks should emphasize empirical studies over these models. There are many problems with this suggestion, not the least of which that economists’ empirical studies don’t agree on many important policy issues. For example, it is ridiculous to suggest that economists have reached consensus that raising the minimum wage won’t reduce employment. Some studies find non-trivial employment losses; others don’t. The debates often hinge on one’s preferred statistical methods. And deciding which methods you prefer is way beyond the scope of an introductory course. 
As you might predict, I have some problems with this. First of all, I don't like the idea that if the empirics aren't conclusively settled, we should just teach theories and forget about the facts. I agree with Kwak, who writes:
I don’t understand this argument. The minimum wage may or may not increase unemployment, depending on a host of other factors. The fact that economists don’t agree reflects the messiness of the world. That’s a feature, not a bug.
Totally! This clearly seems like the intellectually honest thing to do. It seems bad to give kids too strong of a false sense of certainty about the way the world works. When a debate is unresolved, I think you shouldn't simply ignore the evidence in favor of a theory that supports one side of the debate.

As a side note, I think the evidence on short-term employment effects of minimum wage is more conclusive than Strain believes, though also more nuanced than is often reported in the media and in casual discussions.

Strain also writes this, which I disagree with even more:
Even more problematic, some of the empirical research most celebrated by critics of economics 101 contradicts itself about the basic structure of the labor market. The famous “Mariel boatlift paper” finds that a large increase in immigrant workers doesn’t lower the wages of native workers. The famous “New Jersey-Pennsylvania minimum wage paper” finds that an increase in the minimum wage doesn’t reduce employment. If labor supply increases and wages stay constant — the Mariel paper — then the labor demand curve must be flat. But if the minimum wage increases and employment stays constant — New Jersey-Pennsylvania — then the labor demand curve must be vertical. Reconciling these studies is, again, way beyond the scope of an intro course. (emphasis mine)
Strain is using the simplest, most basic Econ 101 theory - a single S-D graph applying to all labor markets - to try to understand multiple results at once. He finds that this super-simple theory can't simultaneously explain two different empirical stylized facts, and concludes that we should respond by not teaching intro students about one or both of the empirical stylized facts.

But what if super-simple theory is just not powerful enough to describe both these situations at once? What if there isn't just one labor demand curve that applies to all labor markets at once? Maybe in the case of minimum wage, monopsony models are better than good old supply-and-demand. Maybe in the case of immigration, general equilibrium effects are important. Maybe search frictions are a big deal. There are lots of other possibilities too.

Strain's implicit assumption - that there's just one labor demand curve - seems like an example of what I call "101ism". A good 101 class, in my opinion, should teach monopoly models, and at least give a brief mention of general equilibrium and search frictions. And even more importantly, a good 101 class should stress that models are situational tools, not Theories of Everything. Assuming that there's one single labor demand curve that applies to all labor markets is a way of taking a simple model and trying to make it function as a Theory of Everything; no one should be surprised when that attempt fails. And our response to that failure shouldn't be to just not teach the empirics. It should be to rethink the way we use the theory.

Anyway, I agree with what Kwak says here:
People like Krugman and Smith (and me) aren’t saying that Economics 101 is useless; we all think that it teaches some incredibly useful analytical tools. The problem is that many people believe (or act as if they believe) that those models are a complete description of reality from which you can draw policy conclusions [without looking at evidence].


  1. Man when I teach 101 we don't talk much about the minimum wage at all. We talk a lot about price discrimination, the plusses and minuses of intellectual property rights, externalities, unintended consequences, the returns to human capital, and the folly of active investing, all with examples and in some cases basic data analysis.

  2. If your goal is to raise future economists and researchers, or even intellectuals, then of course you want to teach economics more scientifically, and Noah is right. If your goal is to raise future voters, most of whom will probably never manage much scientific thinking and certainly won't learn it for the first time from an economics class in college, then Strain has a more realistic understanding of what education should aim for. The fact is that their beliefs will be better matched to reality if they leave 101 class thinking the textbook statements are the likely truth, rather than with a notion that it's all inconclusive empirics requiring perpetual research and scientific progress, which is what would happen otherwise.

    1. Don't make the high school math curriculum error of trying to teach a course that is appropriate for everybody but actually valuable for nobody.

    2. Excellent way of putting it.

    3. @Saturos: Wait. So you think the republic is better served by having voters have an unjustified confidence in models known to be incomplete or wrong, rather than voters who approach expert but interest-motivated advice with some critical skepticism? You have an unusual view of democracy.

  3. Noah,

    There is an interesting debate between Menzie Chinn and commentator Rick Stryker in the comments section of the econbrowser post you reference. That exchange is continued in the following post:

    I have difficulty following it, but it looks important.

  4. Worse than you described, Noah: If monopsony is the right model, then there isn't a labor demand curve at all (just as there is no supply curve for a monopolist). If there is no demand curve, then obviously there is not a vertical demand curve. The Strain quote you provide thus assumes away, as an analytical matter, the best theoretical explanation for the empirical result he doesn't want to believe. He's welcome to his analytical priors, of course, but I don't see why everyone else has to adopt them.

  5. Actually, the theory taught in econ 101 is ill-equipped to deal with the minimum wage. Competitive equilibrium works under the assumption that everyone can trade all they want at the market price. If the government puts a floor under a market price, the model has nothing to say about what we will observe people doing.

    1. This can't be right. Econ 101 certainly discusses rent control. If the model can say something about what happens when the government imposes a ceiling, why can't it say anything about what happens when the government imposes a floor?

  6. The interesting part of economic analysis is not 'what effect' but 'how much effect'. For the defenders of current Econ 101, treating models of (usually a selected) effect as a complete description from which policy conclusions emerge without the need to look at evidence is a feature, not a bug.
    The comment by Saturos provides an example. Econ 101 can only be aimed at voters, and at fact-free a priori conclusions, if its purpose is propaganda. Voters, as much as professional economists, need to understand that it's the size and direction of an effect that matters and that only facts can ground an estimate.

  7. Frankly I think most intro texts are more open to bringing in real world stuff that may not fully fit the standard poop than most of the intermediate undergrad theory texts, especially the macro ones that seem to be overwhelmingly still extreme RBC orthodoxy despite the more or less total discrediting of this by the Great Recession. At least most intro texts will say some things about Keynes and multipliers and such a little bit, whereas he barely exists at all in these intermediate texts, which are for suckers who are buying into the story rather than the intro students who are more concerned with other things, including the real world.

    Of course at the grad level, math becomes more important, although all kinds of exceptions and so on are more readily admitted than at the intermediate undergrad level, although these exceptions are more likely to be theoretically based rather than drawing on empirical studies.

    Barkley Rosser

  8. Anonymous8:40 AM

    Why not teach economic history?


  9. Econ101 — worse than useless
    Comment on Noah Smith on ‘Michael Strain and James Kwak debate Econ101’

    You summarize: “People like Krugman and Smith aren’t saying that Economics 101 is useless; ... The problem is that many people believe ... that those models are a complete description of reality from which you can draw policy conclusions ... “

    Where the heck do people get these silly beliefs from?

    Imagine a person has seen the promotion clips for a new car and decides to buy one. He jumps into the driver’s seat, starts the roaring engine, rolls a few meters, and then the car breaks down in a cloud of black smoke. The person takes the dealer to task and he explains that there is no problem with the car but ‘The problem is that many people believe that those promotion clips are a complete description of reality.’

    It is exactly the same with economics. Economists advertise their stuff as science, well knowing that their theories/models break down at the next corner: “suppose they did reject all theories that were empirically falsified ... Nothing would be left standing; there would be no economics.” (Hands, 2001, p. 404)

    In fact, there IS NO economics, there is merely Walrasianism, Keynesianism, Marxianism, Austrianism, and all four are provably false.

    Given the proto-scientific state of economics, Noah Smith’s pedagogical advice “It seems bad to give kids too strong of a false sense of certainty about the way the world works.” is bordering on absurdity. Obviously, it would be much better not to teach the kids this scientific garbage in the first place.

    The very purpose of Econ101 is to teach students the true theory and not a false theory with the disclaimer not to take it seriously. “In order to tell the politicians and practitioners something about causes and best means, the economist needs the true theory or else he has not much more to offer than educated common sense or his personal opinion.” (Stigum, 1991, p. 30)

    And this is the crux: economists do not have the true theory. And the root cause is that standard economics is built upon this set of foundational propositions, a.k.a. axioms: “HC1 economic agents have preferences over outcomes; HC2 agents individually optimize subject to constraints; HC3 agent choice is manifest in interrelated markets; HC4 agents have full relevant knowledge; HC5 observable outcomes are coordinated, and must be discussed with reference to equilibrium states.” (Weintraub, 1985, p. 147)

    Methodologically, these premises are forever unacceptable because constrained optimization, rational expectations, equilibrium etc. are nonentities. This invalidates the whole theoretical superstructure from distribution to employment theory (2014a; 2014b). And this in turn fully explains why empirical tests run into a vacuum. Nonentities simply have no real counterpart. What has to be done first is to fix the theory. This means, the axioms HC1|HC5 have to go out of the window for good. Economics has to be rebuilt from scratch.*

    Right policy depends on true theory, everything else merely amplifies the confusion.

    The necessary paradigm shift, though, is not to be expected from people who have not realized until this day that supply-demand-equilibrium or IS-LM are methodologically indefensible constructs that cannot be taken seriously for one minute. Economics, though, is caught in secular stagnation: “... most economists neither seek alternative theories nor believe that they can be found.” (Hausman, 1992, p. 248).

    Econ101 is not entirely useless, it provides ample evidence that one can sell any green cheese behavioral assumption and any absurd claim and any scientific garbage and any silly excuse to economic students.

    As far as the future of economics is concerned that much is certain: it will take place without hand-waving half-wits like Strain and Kwak and Krugman and Smith or, for that matter, of useless Walrasians, Keynesians, Marxians, and Austrians.

    Egmont Kakarot-Handtke


  10. Maybe it is time to bring in more empiricism. Things like supply-demand curves are emergent. They are the result of individual decisions. Most college kids taking Econ 101 have access to computers. Start with Leontiff's input-output model and see what supply-demand curves can be derived from it. As with most of physics, everything is locally linear, but then the interesting stuff happens.

    Another approach was taken by Joan Robinson who taught the history of economics. I had a great American history course taught as a study of American history through American history (historiography). Look at what the economists have said, test their assumptions, follow their arguments and see where they hold and don't hold.

    The problem with Econ 101 as it is taught is that it is not only wrong, but no one is being taught how to reason about economics to figure out what is right.

  11. The empirical debate should be shaped around decent theory. If you use the same theory for the employment as you do for the market for applies, you will never make sense of the empirics. The key is introducing moral hazard and employer-employee relationship early on. These are not hard concepts - any student knows what it's like to shirk. Unit 6 of CORE does this brilliantly in my opinion:

  12. This comment has been removed by the author.