Is economic theory simply absurd?

Ariel Rubinstein wonders.  His conclusion:

As in the case of a good fable, a good model can have an enormous influence on the real world, not by providing advice or by predicting the future, but rather by influencing culture.  Yes, I do think we are simply the tellers of fables, but is that not wonderful?

Hat tip to Peter Klein.


"I'm not sure whether you will take this as a confession or a boast, but we are basically story-tellers, creators of make-believe economic systems." - Robert Lucas in What Economists Do

"... but some metaphors are useful." -Box (sort of)

No, there's no worthwhile way to do economics without models. In fact, there's not even any way to think about economics without a model, regardless of whether you're a professional economist or not. That's because as soon as you suggest any answer to the question of "What would happen if we did X?" then you've cooked up a model. There's a quote by Keynes that's appropriate: "Practical men, who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist." That is, there is no such thing as the "rest of us".

Naturally, most models are wrong, because they're stories about how the world works, and the story may or may not have any connection with reality. So you need a way to tell. That means two things. First, the model has to be rigorous enough to be testable. Does it make predictions? Does it tell a causal story you can test? A necessary condition for this is casting it into mathematical form, because natural language is too imprecise to even tell if the story being proposed is inconsistent or not.

This isn't a sufficient condition, though. Deirdre McCloskey points out that in addition to being in mathematical form, you're model must make predictions that are specific enough that you can unambiguously tell from the data you have whether they're true or not. Mathematical economics is not nearly as careful about this as it ought to be. But being careful in that way would require more math and more stringent modelling, not less.

"Why on earth is there such a concern in econ with

You basically understand a economic argument if and only if you can build a model about it.

"This isn't a sufficient condition, though. Deirdre McCloskey points out that in addition to being in mathematical form, you're model must make predictions that are specific enough that you can unambiguously tell from the data you have whether they're true or not."

There is almost no such model. That´s why one has to make "identifying assumptions" when doing econometrics.

"In many ways, Rubinstein's Jungle Economy is an equally useful way of viewing the world, one that is rather similar to some Marxist views."

There is no production in Rubnsteins Jungle economy, so the explanatory power is kinda limited.

First, I think that someday we'll likely end up dispensing with each of the separate social sciences and they'll all fuse into one "social science." This fused social science will probably borrow the modeling and empirical methodologies most currently practiced in formal economics today, but it will use these methodologies to test a wider variety of theories than those implied by the straight-up rational choice model. The rational choice model will still likely provide a good base case. We're already seeing such work in the field of behavioral economics.

In the meantime, economics does a pretty good job giving advice about economic development. Following export-led growth works better than import substitution, for instance. Cost-benefit analysis can identify good investments for nations to make, especially those that supplement and generate robust market growth.

I think economics is at its most useful when it comes to teaching people the basic insights of opportunity costs and sunk costs and cumulative compund effects and good basic economic decisionmaking. Based purely on anecdotal things I've heard and witnessed, a lot of small businessmen learn economic decisionmaking through a lot of effort and trial and error, when people could learn economic decisionmaking the easy way if they'd be better about listening to economists.

This also brings me to problems with the practice of economics. The insights from things like opportunity cost and sunk costs and other things are not as redily understood by people as economists might think, or at least as economists might model.

The distributional consequences of this could be highly underrated. Even in an economic system where economic theory does a good job of predicting the market equilibrium, if the "smart economist" agents are few in number and the "dumb non-economist" agents are many, the "smart economists" may end up very well off and the "dumb noneconomists" may do very poorly. So the distributional consequences of differences in the ability to "think economically" could be very important.

In addition, in an economic system with high returns to imitative behavior, there may be a wide deviation from the rational choice economic equilibrium.

"No, there's no worthwhile way to do economics without models. In fact, there's not even any way to think about economics without a model, regardless of whether you're a professional economist or not. "

Yet Adam Smith didn't seem locked into any models ....

The Keynes quote doesn't mention models.

"You basically understand a economic argument if and only if you can build a model about it."

Couldn't that have more to do with how you understand an ecnomic argument than with economic arguments (and understanding) generally?

"Is there any way to think about anything without models? Seriously, I think not."

An amazing assertion!

The opposite also happens confirming economic models.For example in Bangalore city in India there are highly educated second hand book sellers-some of them are retired officials-who perfectly know which books are running out of print,which are scarce,which is the rarest first edition ,and the like.They price books according to scarcity, oldness of the edition etc so that you need to pay a higher price."Select Book House" off Brigade Road Bangalore, "Sudha Book House" near Rajaji Nagar entrace Bangalore and the like are examples of such second hand rare book shops. Here seller retains better knowledge and the 'Lemons theory' works.

The laws of demand and diminishing marginal returns are absurd? I think not.

Kind of like the brilliant Karl Rove and his model that showed a permanent Republican majority, Michael?

Ten years ago or so, a former editor of the AER asserted that "game theory is THE paradigm of economics". I asked him: But what about demand and supply (aka the laws of demand and dimiishing marginal returns)? His response? "Not anymore"

Now that I'd call ABSURD.

"How about we just dispense with formal economics altogether? After all, what real value to the world do economists add? It's scientists and engineers and inventors who advance technology ... In many ways, Rubinstein's Jungle Economy is an equally useful way of viewing the world, one that is rather similar to some Marxist views."

You answered your own question. We need economists to keep debunking wrong economists - like Marx. We would have no entrepreneurs or inventors if we all still believed in Marx's economic theories.

Until totalitarianism and sloth of bureaucracy are eliminated, we need economists to help us understand why we should get rid of them.

So the propositions we get are:

1. People are going to study economics and economies.

2. Incorporating some basic insights like opportunity cost generates conclusions, like the fact that trade acts like a productivity-enhancing technological innovation, that many people find very surprising.

3. It's very difficult very quickly to use language in a precise enough way to logically explore these issues further, especially if you also want to be clear and concise.

4. That makes it desirable to use mathematical models. In addition, mathematical models enable communication and understanding across languages.

Jim, you're absoutely right.

One reason why behavioral economics will overtake game theory as the real useful tool of the future, in my opinion, is because behavioral economics can readily incorporate insights from other social sciences into simpler models and predict behavior, rather than using some of the more rococo mathematical optimization to try to explain human behavior. The latter method just doesn't seem to generate good predictions.


Marshall had an interesting perspective, that cdertainly influenced Don Godon's thinking about the negative impact of mathematical complexity upon the generation of operational propositions in economics.

In 1952, Paul Samuelson wrote of Marshall that his disdain for "long chains of logical reasoning" was because "Marshall treated such chains as if their truth content was subject ot raioactive decay and leakage -- at the end of n propositions, only half truth was left, at the end of 2n propositions, only half of half remained, and so forth in a geometric multiplier series converging to zero truth."

Without referencing Don Gordon's hypothesis, Leontief in 1971 reasoned similarly that: "Uncritical enthusiasm for mathematical formulation tends often to conceal the ephemeral substantive content of the argument behind the formidable front of algebraic signs."

More recently (in 1997), discussing the literature on theories of consumption, Frank Akerman wrote: ". . . it is remarkable how little of Veblen shines through the dense mathematical thickets of contemporary economics."

The points you are making have long been a concern of SOME economists.

On some topics, some economists have gotten pretty far without the use of mathematical models. Coase isn't a great example, because his 1960 article is full of little numerical examples that are basically toy models with varying parameters. Demsetz might be a better example. Adam Smith got there early and strip-mined a lot of the easily extracted ore without math (although he certainly got some things wrong partly due to a lack of modeling rigor). But for many, many topics and for most people, it's just too hard to see what the heck is going on without a formal model of some sort. And math is often the best way to describe a formal model.

If Michael Blowhard thinks he can explain how tax incidence works without algebra or graphs, or explain why high marginal income tax rates cause disproportionately greater deadweight losses without math, or even explain why Burger King headquarters wants lower prices than Burger King franchisees without math, he's welcome to try. I'm pretty good at explaining stuff, but I'm not that good; and successful efforts to explain economic arguments verbally often are parasitic on mathematical understanding, in that the explainer has a model in mind and translates it to the audience.

Indiana Jim: I'm not sure that the length of a chain of deductions is a stable measure, because different frames and proofs are employed over time. And I'm not sure length of chain is the biggest problem with mathematical models--sometimes it is, I guess, but there are lots of other ways for mathematical models to lead one astray. For example, introducing variables into models without some conceptual idea of what units they would be measured in presents risks. Using non-cooperative game theory when the information structure and move order assumptions are not well-grounded is another area where things can get sticky. And there are lots of other ones.

No matter how worried I get about these issues, however, (and I try to assess them on a case-by-case basis in my own work and that of others), I see zero evidence that attacking problems in purely verbal fashion is safer from error or more likely to lead to operational propositions. I suppose it's easy to generate operational propositions if you don't require any logical basis for them, but I don't think that's what you have in mind. And a long chain of verbal propositions should be just as prone to whatever disabilities you have in mind.

In my opinion, this sort of thing is hard to debate on a general level. You have to look at particular examples of economic reasoning to get a sense of where the problems and opportunities are.

Indiana Jim: You keep talking as though the length of a chain of reasoning and the intensity of mathematics are the same thing. They aren't, either marginally or inframarginally. There are long verbal chains and short mathematical chains, and subchains of both types strung together at various lengths. Adam Smith's points about the invisible hand and the role of specialization and why mercantilism is a bad idea come from much longer chains than, say, Akerlof's analysis of the lemons problem. Also, as I said before, I'm not sure the length of a chain is a well-defined concept, since we can endlessly subdivide things into smaller steps, especially when they are stated verbally. Incidentally, your point of view may be reflected by a quote from Keynes (I don't know the reference, unfortunately) where he compared economics to physics and talked about how in physics we start from a small number of assumptions we know with great certainty and pursue long chains of reasoning while in economics we know a lot of things imprecisely and pursue short chains of reasoning.

I'm also not sure what you mean by "operational." For example, if I saw a market where experience goods of varying quality were being sold under conditions of asymmetric information, I might look to see if there is a lemons problem. The way some people keep score, the "operational" proposition from the lemons model is zero trade. I wouldn't predict zero trade. Knowledge of the model would cause me to search out institutional or customary features that mitigate the problem, or look to see if greater asymmetric information led to more transactions safeguards and/or fewer transactions.

So a lot about your proposition about operational propositions strikes me as non-operational. Maybe the chain of reasoning is too long.

Comments for this post are closed