Monday, December 05, 2011

What to teach intro economics students


Greg Mankiw has a good response to the walkout of his introductory econ course, which basically agrees with what I wrote. Peter Dorman, however, is not satisfied:
[T]here is a central narrative at the introductory level that has hardly changed in at least a generation, perhaps longer.  It presents a system of perfectly competitive markets composed of rational, unconnected agents as the benchmark, from which specific deviations, like externalities, behavioral anomalies, sticky prices, etc., are considered one at a time.  Most of the interesting and important work in economics is about these deviations.  If you added up all of this innovative research, you would have a composite picture that is exciting, relevant—and light years away from the introductory narrative. 
A huge gap has opened up between the introductory course and the work professional economists are actually doing.  Each departure from the narrative is considered one at a time, even though research has chipped away at all of them...Thus the introductory course still looks like a distillation of the research frontier, even though, if you put all the research results together, you would have something quite different.
First of all, I agree with Dorman completely regarding the research frontier. The whole notion of thinking of each interesting feature of the economy as a "friction," and then of considering only one or two "frictions" at a time, has been very detrimental. For one thing, it makes it hard to develop a useful model of the economy, since the actual economy contains many, many "frictions" (so many that the "frictions" together are usually more important than the "frictionless" dynamics that supposedly "underlie" them). Also, the "one friction at a time" approach makes it very difficult to generate any alternatives to the classical "core theory" of Walrasian general equilibrium. In fact, my main reason for disliking the DSGE modeling framework is that it is so unwieldy that it makes it prohibitively hard to introduce more than two "frictions."

But when it comes to intro economics - or, at least, intro macro - I don't see this dynamic in operation as much as Dorman does. Of course, I only know the classes I've taught (I've never taken an undergrad econ course). But I taught right out of Mankiw's book, and there was nothing particularly unusual about the curriculum. 

Most of what I taught was not based on a system of perfectly competitive markets. For business cycle theory, we taught the AD-AS model (beloved of montarists like Scott Sumner), the supply and demand for money, the New Keynesian Phillips curve, and the old Keynesian Cross. RBC theories were never even mentioned in the lectures (although I gave students a brief overview of them in discussion section, much to their annoyance since RBC was not on the test!). In fact, there were nothing but frictions. The theories were so different that students often asked me "How do we know which of these models to use?" (provoking a laugh from me, since that is such an excellent and often-ignored question). The focus was always on market failures, and how governments should use policy to correct them.

In sum, intro macro really doesn't look like a distillation of the research frontier. Which is a good thing.

Actually, my beef with intro macro is different. I think there is a big chunk of the research frontier that intro courses completely ignore. I'm talking, of course, about empirics. The courses I taught had nothing to say about how we know if and when a theory is right. (Note: anyone who read that sentence and immediately started typing "But ALL theories are wrong!", please think very carefully about the concept of a "domain of validity" before you start spamming the comments. Thank you.)

In natural science courses, much of the focus is on empirics. It is critical to know when a theory is a good approximation of reality and when it is not, and looking at evidence is the only way to know this. So in high school physics, you roll a ball down a ramp and measure its position at different times to see how gravity works. In chemistry you dump some silver nitrate into some potassium chloride, and you watch the silver chloride precipitate out of the solution. In biology you look at cells under a microscope.

But in introductory macroeconomics this is not done. I never once gave students a data set and said "Here, regress Y on X". There was no reason I couldn't have. OLS is easy to do, and easy to explain (simple least squares is very intuitive and can be shown with a picture). Sure, intro students aren't going to be coding DSGE models in Matlab or converging MLE routines for structural models. But doing some simple empirics would give students a feel for how economists test their theories. It would give them a hands-on feel for data. And it would allow lecturers to explain why certain statistical techniques can lead to false certainty (i.e. the Lucas Critique).

As it stands, students in introductory economics courses walk away feeling that econ theories are "received wisdom" - that a theory is just something that a smart guy dreamed up, and then concluded was right because it sort of seemed plausible (which, sad, to say, describes some econ theories all too well). And - fortunately for American society - we have somewhat of an aversion to received wisdom. We call it by the name of "bullshit." And rightly so! As Feynman said, "Science is a belief in the ignorance of experts."

So I say, if you want to give introductory economics students a better picture for what the science is really good for, teach them the part that links theory to reality. 

20 comments:

  1. "For one thing, it makes it hard to develop a useful model of the economy,"

    Especially since the economy, and the world it is in, keep changing. I think Samuelson's 1960 Economics had the right basic approach: a macro overview first, circulation of inputs and outputs in one direction, of money and purchases in the other. Then micro more or less in its historical order of development. He came up short on international trade at the end however, if memory serves.

    ReplyDelete
  2. I couldn't agree more strongly with this. I did my undergraduate in biology and economics. I loved the theory aspect of econ, I was always turned off by the almost complete absence of empirical verification, which was a major factor in picking grad school in biology. I even got told, by a professor I held and still hold in extremely high regard, that it would take too much time away from teaching theory to incorporate empirical data examples.

    The one thing I would say is that, even more useful than having students analyze a data set, would be including concrete data, with analysis, along side every theoretical idea in texts. Learning how to do a statistical technique is handy, but the most vital thing is to show your hand: supply as much supporting evidence as possible beyond theoretical argument.

    ReplyDelete
  3. Anonymous1:39 AM

    Hey, I mentioned the Labor empirical debates on immigration and the minimum wage in my discussion section in a intro to micro course today. But I was really only able to do so because it fit well into what I was supposed to teach; in general I don't get to talk nearly as much about empirics as I would like. So I agree pretty strongly with what you've written here.

    ReplyDelete
  4. Anonymous5:47 AM

    It isn't enough to state what you would like to see taught. You must also state what you would take out to make room. Time constraints are binding!

    ReplyDelete
  5. Herman5:54 AM

    No disagreement re macro, but my impression was that the Mankiw kerfuffle, and Peter Dorman's comments were mostly about intro *micro*.

    Also, in case of modern macro (not intro macro) a view is that most of its problems stem from treating it as a trivial extension of the most simple micro models. In fact the basic narrative is the same as for intro micro, which is what Dorman is criticising.

    ReplyDelete
  6. Anonymous6:25 AM

    For several years, I generally (but not completely) believed my intro econ professor's assertion that "government doesn't produce anything".

    Then I learned of List's theory of Productive Powers. His critique of Smith's materialism seems rather prescient in the age of austerity.

    ReplyDelete
  7. As a physics undergraduate and econ grad student, I agree whole-heartedly. I tell myself that someday I'll write an evidence-centric intro textbook.

    ReplyDelete
  8. Anonymous9:31 AM

    I'm just a lowly TA, but one prof. that I TA for is an adjunct that works as a research economist for the Fed. His intermediate class involves a lot of regressions and data, and he strays away from "standard" models and techniques.

    Some of the subject matter brings up concepts that I personally never encountered until graduate school, albeit in stripped down ways more appropriate for undergrads. Models and theories are regularly tested with data. He does so in a way that shows where the models succeed, and where they fail so that students can see how such models do lead to insights, but where they also fall short. In some cases, it's actually the ONLY time even I've ever seen some theories put to data, including all the time I've thus far spent in a Ph.D. program.

    It's a very hard course for nearly every student, and students looking for an easy A avoid it like the plague. Those who do take it find it hellishly hard. TAing his course involves a lot more work due to the amount of extra help the students require. Students who pull it off leave with a impressive understanding of economics.

    What I find encouraging is despite all the complaints about how unfairly hard his course is (compared to other intro courses), he's regularly voted the top professor by students. It just goes to show that, in the end, students really do like seeing how their textbook work relates to the real world.

    I think of lot of his methods stem from the fact that he's not a pure academic, that his job at the Fed does involve a lot of real world data analysis (as well as theoretical modeling). He passes that perspective on to students. They find it really hard, but they end up being very thankful

    Undergrads who take his course often end up with a better understanding of "real world" economics than many students I've met even in a Ph.D. program. Sure, the Ph.D. kids can tell you about eigenvalues and the determinacy of DSGE models and how to do linear approximations and what not, but ask them what they think of a current Fed policy and they'll give you a frighteningly bad analysis of it. Conversely, ask an undergrad who has taken this prof's intermediate course, and you'll get a response better reasoned and thought out than anything you'll read by any "experts" in most newspapers.

    There is indeed a better way to teach undergrad econ. I know because I've seen it. But it's hard and requires much more of students than anything I've ever encountered in one of the "standard" undergrad textbooks.

    ReplyDelete
  9. Erm, as far as I'm aware all undergraduate economics students generally take a module on econometrics. That would certainly be the case in Harvard, so there is no point in teaching OLS in macro when it is already being taught in econometrics. You can cite papers with empirical verification (even though I thought Mankiw already did that), but literally making them do econometrics coursework in a non econometrics module seems like a waste of time, and economics undergrads don't have that much time.

    ReplyDelete
  10. Anonymous12:54 PM

    I guess the 64 thousand dollar question is why you didn't introduce any empirics in your own class.

    ReplyDelete
  11. Herman1:04 PM

    The point is not about teaching econometrics in a macro or micro course - running an OLS regression is trivial.It is about evidence based economics.

    This is standard practice in physics. I've never understood why economists are so terrified of it.

    Here is Herb Gintis, a game theorist, writing about it:

    -----
    "Economic theory has been particularly compromised by its neglect of the facts concerning human behavior. This situation became clear to me in the summer of 2001, when I happened to be reading a popular introductory graduate text on quantum mechanics, as well as a leading graduate text on microeconomics. The physics text began with the anomaly of blackbody radiation, which could not be explained using the standard tools of electromagnetic theory. In 1900, Max Planck derived a formula that fit the data perfectly, assuming that radiation was discrete rather than continuous. In 1905, Albert Einstein explained another anomaly of classical electromagnetic theory, the photoelectric effect, using Planck’s trick. The text continued, page after page, with new anomalies (Compton scattering, the spectral lines of elements of low atomic number, etc.) and new, partially successful models explaining the anomalies. In about 1925, this culminated with Heisenberg’s wave mechanics and Schr¨odinger’s equation, which fully unified the field.

    By contrast, the microeconomics text, despite its beauty, did not contain a single fact in the whole thousand-page volume. Rather, the authors built economic theory in axiomatic fashion, making assumptions on the basis of their intuitive plausibility, their incorporation of the “stylized facts” of everyday life, or their appeal to the principles of rational thought. A bounty of excellent economic theory was developed in the twentieth century in this manner. But, the well has run dry. We will see that empirical evidence challenges the very foundations of both classical game theory and neoclassical economics. Future advances in economics will require that model-building dialogue with empirical testing, behavioral data-gathering, and agent-based models."
    ------------
    Bounds of Reason
    Herbert Gintis
    pg xvi

    ReplyDelete
  12. @most recent Anon:
    Simple. It would have made the students mad, since empirical evidence and techniques were not on the exams.

    ReplyDelete
  13. @earlier Anon (get screen names, you peoples!):

    Yes, time constraints are binding. But what material I'd take out would vary from course to course.

    One thing to note is that many homework assignments in all intro courses I've taught have had a lot of simple but redundant homework problems. Instead of calculating the Keynesian multiplier for the umpteenth time, why not try estimating it from data? (And yes, I know spending is endogenous; but in practice you can instrument for it.)

    ReplyDelete
  14. Econometrics has no place in an introductory curriculum in my judgment, for the following reasons: (a) that there are no functional relationships and no true variables governing economic behavior either of the individual, any collection of individuals or of society as a whole, owing to the lability, immeasurability, and theoretical “unknowableness” of human behavior; (b) that therefore the mathematical = sign is never appropriate in the so-called “mathematical” description of economic relationships, even though certain logical relationships remain; a, in consequence of (a) and (b), algebraic equations are an illegitimate feature of economic analysis except as an informal guide to thinking along the lines described by Marshall, with all equal signs replaced by “is roughly proportional to” notation; (d) that in consequence of a,b,c, and d, it goes without saying that the machinery of the calculus is inappropriate to economic model building, the use of integral signs, Lagrangians, derivatives, etc., being the chief sin of this fallacy; (e) that DSGE models suffer from the same fundamental flaws of current Climatological models with one added twist: namely they attempt to estimate non-existent parameters for simultaneous systems of equations which do not exist and use as an empirical verification procedure so-called “post-diction” which amounts to nothing more than curve fitting; (f) the added twist is that Climatology at least has the theoretical advantage of dealing with matter instead of real human beings, and is therefore in principle governed by the principles of statistical mechanics (thermodynamics) and/or the Hysenberg uncertainty relationships, and, furthermore is amenable to a theory of measurement error using Baysean and frequentist techniques and assumptions of normality; (e) that the uncertainty of economic actors, prices, measurements of quality, volume, income, unemployment, inflation, and all other economic factors and entities of interest is different in kind from the uncertainties of physics, being incalculable in principle as well as in practice, this being a consequence of the unmodelability of the lability of human behavior (not normally distributed) together with the immeasurability of all factors in both time and space whether absolutely or statistically.

    ReplyDelete
  15. Another riff on why economics is not a science suitable for econometric analysis (which I would have thought Noah would agree with):

    One of the reasons economics is not such a science is that the world is always changing. Each generation faces new problems, with new factors coming into play: innovations in technology, new definitions of money, alterations in the law, trade, politics, international relations, demography, customary standards of living, family patterns -- the list, if not endless, is certainly indeterminate.

    Unlike the worlds of physics, chemistry, and biology, where the laws of nature are ever the same, both in time and space; where fundamental particles are always identical, the constants of nature always constant, functional relationships precise, continuous, and well-defined (f=ma, The Heisenberg uncertainty principle), in the world of economics there is nothing of the kind. Such general laws as do always apply are qualitative, not quantitative, capable at best of a rough geometric illustration, not algebraic formulation, for which no equal sign is ever justified, no tilden unjustified, no two objects identical, or even the same object identical at two different moments in time.

    In other words economics is to be conceived as a historical science the sense that the trained practitioner must be educated not only in general principles but also in the history of economic ideas and policies as they have evolved in relation to changes in the economic and political circumstances of the societies in which they arose, and to which they were applied. He must, in short, be cognizant of modern historical experience, as well as to the changing contemporary realities of the world he lives in.

    For these reasons economics is not a cumulative body of knowledge, except to a very limited extent. A student of the hard science need scarcely bother himself with the history of the discipline he is learning; with economics such knowledge is indispensable. For the competent practitioner of the art recent history may be more relevant than distant history, though not always, and no history is so remote as to be completely irrelevant at one time or another.

    ReplyDelete
  16. Noah said...

    " @earlier Anon (get screen names, you peoples!):"

    From my casual estimation, at least half of the Anon's are not worth reading (what I'll call 'anonymous for good reason!').

    I suggest requiring authentication. It's far from sound, but it cuts down on the drive-bys.

    ReplyDelete
  17. The thing I've never understood about micro, ever since tutoring my wife and her classmate through the core-curriculum grad-school micro (Kreps was the prof), was that it assumes stuff about humans that we know is often false -- for example, independent utility functions, and perfect knowledge, and rationality. Yet we know that we have dependent utility functions ("keeping up with the Joneses"), we know that we overestimate our relative skill/luck/etc (Lake Wobegon Effect), and we know that we have information asymmetries (market for used cars, market for private health insurance).

    And wearing my mathematician's hat, if your assumptions are false, you've got nothing.

    "Friction" gets discussed at great length in intro physics, so it seems that these "imperfections" would be appropriate here (interesting that we call them "imperfections", isn't it? Friction isn't an imperfection in physics, it's just more physics).

    ReplyDelete
  18. If I were taking Econ 101, I would be looking for the prof to explain why mainstream economists 1) did not see the crisis coming, 2) cannot agree on what caused it, 3) don't know how to fix it, and 4) have not come up with a plan to prevent another crisis.

    Some people did get it, but they are not mainstream and their ideas are neither being taught nor listened to. See James K. Galbraith, Who Are These Economists, Anyway?, which was written in response to Paul Krugman:

    Of course, there were exceptions to these trends: a few economists challenged the assumption of
    rational behavior, questioned the belief that financial markets can be trusted and pointed to the long history of financial crises that had devastating economic consequences. But they were swimming against the tide, unable to make much headway against a pervasive and, in retrospect, foolish
    complacency.
    —Paul Krugman, New York Times Magazine, September 6, 2009

    ReplyDelete
  19. dr2chase: "And wearing my mathematician's hat, if your assumptions are false, you've got nothing."

    A comment - we're not talking about mathematics here. The big trick is that the assumptions are false; it's how false and what effect this has.

    ReplyDelete
  20. @Barry - what it means is that "a free market is welfare-maximizing" is a belief, not something that is proved for the real world. I notice, also, that people are most quick to declare that wonderfulness of the free market when applied to those things where we know that the assumptions are particularly false -- for instance, health care.

    Recent article in the New Yorker on apples points out how the market even fails in the case of apples -- apparently, left to their own devices, given two apples marked "Gala", humans will pick the more attractive one. Over time, this leads to genetic drift in the "clone" (it isn't, over time) towards prettier, crappier apples, because farmers know this and will propagate more attractive sports under the original brand. And lo, "Red Delicious". Market "failures" come in many flavors (or lack of flavor, as the case may be).

    And again, note the use of "failure", to indicate that a normally functioning market "works".

    ReplyDelete