Bryan Caplan  

Escape from the Ivory Tower: Obscure Academic Economic Concepts Worth Knowing

Bruce Bartlett on Debt Default... Zingales Joins the Swamp...
In econ grad school, you learn a dozen good ideas and a hundred seemingly irrelevant ones.  Most of those hundred ideas deservedly languish in obscurity; no one uses them to explain the real world, even in conversation.  But once in a while, you realize that one of those obscure concepts is genuinely enlightening.

Trembling hands perfection is my favorite example.  A typical explanation:
A trembling hand perfect equilibrium is an equilibrium that takes the possibility of off-the-equilibrium play into account by assuming that the players, through a "slip of the hand" or tremble, may choose unintended strategies, albeit with negligible probability.
When first described, the concept seems like a mere make-work project for game theorists.  But I've gradually noticed that it's a big deal.  It explains, for example, why imposing harsh punishments for small infractions isn't nearly as smart as it seems: People sometimes accidentally break the rules.  Automatically imposing harsh punishments imposes needless costs on well-meaning people, and gives incentives to avoid valuable actions with above-average accident rates.  In a noisy world, forgiveness and second chances are common sense, not sentimental folly.

The trembling hands concept also explains the value of trying to exceed others' expectations.  In the real world, it's not smart to apply the minimum acceptable level of effort, or pay others the smallest amount you can get away with.  Accidents happen - and if you cut everything close, those accidents will have needlessly bad consequences.

OK, that's my nominee.  My challenge for other econ bloggers: Name yours.  What's your favorite obscure-but-genuinely-enlightening academic economic concept?

Comments and Sharing

COMMENTS (24 to date)
Steve Sailer writes:

Thanks. I'd never heard of that term before, but it's helpful.

david writes:

Why stop at trembling hands? Look for the evolutionarily stable strategy.

Troy Camplin writes:

The trembling hand concept explains how one avoids settling into local minima. If the hand is "trembling," it can jar you out of the local minima, allowing you to find the system minimum. It's from systems theory (which also states that, in fact, equlibria are utter nonsense in a constatntly changing system).

Charlie writes:

I'll think about more obscure concepts, but one that really shouldn't be obscure that seems to be among older economists is commitment in repeated games. I hear over and over that the U.S. should adopt a no-bailout strategy or that bailouts create moral hazard, but the economists never grapple with commitment to a no bailout strategy. The lesson of a NE in a repeated game is that if the gov't makes an off-equilibrium move (for instance, no bailout) it doesn't change the NE. Whereas, if the player (gov't) can move first and have commitment, it can change the game and change the NE.

Bailouts are almost never discussed in that light among bailout critics. If bailouts occur, why? And if they shouldn't occur, the incentives need to be changed, the game needs to be changed. One group of austere, principled politicians is meaningless if it's an off-equilibrium path move. I think Tyler Cowen brought up the idea a lot, but could never get much traction with it. I think it's a hard concept for people that haven't solved simple infinitely repeated games.

Is that obscure, Bryan? I was going to name the names of two older GMU economists to make my point, but maybe that's in bad taste. It certainly rarely makes it to real world conversation, where commitment is too often assumed.

agnostic writes:

Economists... this is otherwise known as simply a "stable equilibrium" from differential equations. Shorter and more informative name, and doesn't make it sound like economists discovered it.

For example, there are two equilibria for a person to be at rest on a swingset -- in the swing at the bottom and standing on top of the whole thing. One is stable because if you unintentionally jerk your head around when someone calls your name, you return right away to rest; the other is unstable because when you do this, your departure from rest grows and grows until you fall off.

The idea of an equilibrium that is stable to perturbations shouldn't be obscure in econ. If it is, it explains a lot regarding why people worry about economists in their role as engineers:

Did you test this bridge to see if it's stable to perturbations by the wind?

"Ummm, what's thing you call 'stable'?"

Quick everybody -- get off that death bridge right away!

Jody writes:

economists in their role as engineers

Engineers doing a little economics are not much better. For example, I can't count the number of papers related to game theory and wireless networks I've reviewed where there's an assertion that because the network has a Nash equilibrium it's stable.

I fight the good fight, but I fear it's a rear guard action.

William Newman writes:

agnostic, precise mathematical translations of the notion of "stable" may be limited to infinitesimal perturbations (e.g., "Lyapunov stability") or may accept infinitely large perturbations (e.g., "global minimum"). From my reading of Bryan Caplan's original post, economists use "trembling hand stability" when they are referring not to either of those precise extremes, but to an intermediate property defined in terms of some imprecise plausibility cutoff: something like "stable within those finite perturbations which might plausibly happen by accident." It's often useful to consider that sort of stability, and I don't know any universally accepted term for distinguishing it from infinitesimal or infinite perturbations. (I do know the somewhat-related terms "[chemical] activation energy" and "[a] boil the ocean [business strategy].") I suspect if you pushed your bridge designer (or a control theorist, or a macromolecule chemist, or a population biologist) to express clearly his concern for stability under plausible perturbations, he'd consider it a relevant distinction, he'd either coin a phrase or come up with a phrase as field-specific as "trembling hands stability," and he'd be likely to consider the economists' "trembling hand stability" a fairly reasonable phrase.

MikeDC writes:

Does the winner's curse count as an academic obscurity? Certainly most of the kids I teach in intro to econ classes have no conception of it, but when I start to point it out, they see it all around them.

Bob Lawson writes:

My nominee is Baumol's "unbalanced growth" (aka cost disease) concept. I see it everywhere but few economists I know seem to appreciate it.

Philo writes:

". . . albeit with negligible probability." Shouldn't that read: ". . . albeit with small (though not negligible) probability"? If the probability is negligible, we can safely *neglect it*; that is, we should not bother with the notion of "trembling hands."

azmyth writes:
Nate writes:

Is it just me, or do Caplan's examples not at all involve a specific "trembling hands" concept? Accounting for performance error is a highly commonsensical idea--either "trembling hands" means something other than "performance error," in which case the concept is nowhere near as ubiquitous or useful as Caplan makes it seem, or it's not, in which case it's not obscure.

afinetheorem writes:

Without fail, the physicists seem to think that economists are completely ignorant of mathematics. Trembling hand perfection is *entirely* unrelated to stability, in the Lyapunov or asymptotic sense. Indeed, trembling hand perfection is an equilibrium concept - it involves no dynamics at all! The principle reason to use thp is to eliminate "weakly dominated" strategies, or actions that are strictly worse or equal to some other strategy, but never better. Indeed, in a two player game, the concepts are equivalent. My reading of thp, though, it that it is most useful in dealing with uncertainty about the utility functions of other agents. That is, in the real world, I don't know exactly how you map outcomes into utility, and so I should be "cautious", in some sense.

In infinitely repeated games, we do have concepts like Lyapunov, but unsurprisingly, we call them Lyapunov stability.

The best "obscure" (though not to theorists, of course) economic theorems for me are the various information theorems due to Aumann and friends, which formalize precisely in what ways we can disagree, when trade can happen, etc.

JLA writes:

I'm not much of a macro guy but the Chamley-Judd result is pretty cool.

agnostic writes:

"In infinitely repeated games, we do have concepts like Lyapunov, but unsurprisingly, we call them Lyapunov stability."

OK, but both examples that Bryan gave for why the concept is enlightening involve more or less infinitely repeated games.

I.e., there's a potential punisher and punishee, or a potentially late person and a person who potentially would scold them. There's some kind of long-term supervision, formal or informal, of the one over the other. It's at that long-term level that Bryan's talking about not giving the minimal required effort or punishing small infractions, seems to me.

Mesa writes:

Le Chatelier's Principle

Jeff writes:

The problem you're talking about is known as the time-consistency problem, and the classical example is the government that says "people who build in a flood plain will not be eligible for flood assistance". When the flood actually happens, the government's incentive is to provide the assistance anyway. Everybody knows this, so the government's original warning is not credible. This idea has been discussed in the context of monetary policy at least since the famous Kydland and Prescott paper (1984?).

Scott Sumner, on his Money Illusion blog, has discussed the idea that bailouts will always happen due to this problem. The only way to really prevent them is to have capital requirements and lending restrictions severe enough that banks never go bust.

Robert writes:

Does the Dumb Agent Theory count? It always seemed interesting.

Nate above identifies the problem with the post: the examples have little to do with trembling hand perfection.

Kudos to David in the second comment for bringing up evolutionarily stable strategies. More generally, there are many definitions of stability in games. Fortunately, they mostly predict the same outcomes.

In my experience running experimental games in the laboratory, stability is key. People almost never learn to play an unstable equilibrium in the lab, while they often learn to play a stable equilibrium if one exists. If there is no stable equilibrium, behavior in the lab never converges to any equilibrium.

If there are multiple stable equilibria--which there often are--trivial differences and simple random chance can determine which equilibrium gets played. I would put that fact high on the list of important esoteric facts. Most games have multiple equilibria, so economists who claim that such-and-such must happen because it is an equilibrium are overclaiming.

CIP writes:

Could you explain some real world situation where the concept actually applies? How does it relate to "forgiveness," for example.

Henry writes:

I always thought of the trembling hand perfect equilibrium as applying most conspicuously to mutually assured destruction.

Popeye writes:

IMHO this post shows just how cloistered in the ivory tower Bryan truly is.

Economists and game theorists often assume perfect information, or perfect rationality. These are often useful ivory-tower concepts but sometimes economists end up assuming that both players in a game can rationally look 100 moves ahead and that they can each count on the other to rationally look 100 moves ahead. Equilibrium!

Now the obvious criticism of such an analysis is "um, I don't think your assumptions are sound, and therefore I don't trust your conclusions." No fancy economics training necessary.

But in his post Bryan conflates this common-sense criticism with the idea of "trembling hand equilibrium." Trembling hand equilibrium is actually something a bit different; it is an equilibrium concept that attempts to address this common-sense objection. That is, it is a tool for game theorists that responds to a particular criticism.

Giving economists credit for an obvious insight that they were probably the last people on earth to appreciate is rather strange. Moreover, it's not even true that economists depend on trembling-hand-equilibria to understand optimal punishments or efficiency wages.

John Fast writes:

How obscure does an economic concept have to be?

Bryan, you know how ignorant the average person is of economics -- worse than ignorant, because they often "know" things which are actually false.

I'd say that the theory of comparative advantage would be really, really helpful for people to know. Ditto with rational ignorance and "the logic of collective action" and regulatory capture; and externalities and how to solve them (via Pigou and/or Coase).

$#!%, I'd be happy if the average person simply knew about the First Fundamental Theorem of Welfare Economics!

Glen writes:

I nominate quasi-rent.

Comments for this entry have been closed
Return to top