What modern art and economic theory have in common

Economic theory and modern art are a lot alike.

They're abstract. When Gainsborough painted a horse, he tried to represent it as realistically as possible. But Deborah Butterfield's Jerusalem Horse #1 looks at first glance like a tangled piece of rusting metal. Then one sees the neck of the horse, and the head, traced out with two lines of steel.

Modern artists, like economists, understand that, to portray a horse, it's not necessary to paint every strand of hair in its mane. Let x=horse and p=grass eaten. A symbolic representation can be much more powerful than a literal one. In both art and economics, abstraction sacrifices literal truthiness in order to capture radical (as in "at the root of") truths. Combining all of the costs of horse ownership into a single abstract concept --“price” -- makes it easier to understand the decision to acquire horses, and reveals the similarities between horse ownership and other consumption decisions.

Yet, because they're both abstract, economics and modern art are vulnerable to the same criticisms. My son looked at the pile of tangled metal in the art museum and said "that looks like a bear not a horse" (it depends on how you look at it).  In the same way, judgments about whether or not a particular theory captures the relevant aspects of a problem in a meaningful way are subjective.

For example, a colleague dropped by my office the other day to talk about a model of minimum wages that he was working on. It was a trade model where everyone in the economy was paid the same wage, w, where w was greater than the market equilibrium wage.  My immediate reaction was "that's not a good way to model the minimum wage." Why? To me, the key issue is the equity/efficiency trade-off: raising the minimum wage (probably) increases equity and (probably) decreases efficiency. But if everyone has the same wage, there is no inequality, so no possibility of increasing equity. There's no equity/efficiency trade-off, and the model fails to capture what matters. Others, however, might be concerned only with the efficiency impacts of minimum wages in a global economy, in which case my colleague's approach might be the way to go.

There's a cautionary lesson here for would-be theorists. Modern art seems so easy - a canvas painted in one or two or three colours or splattered with paint blobs, a retouched photo of some famous person. A lot of really good economic theory appears easy too. Take, for example, George Akerlof's work or Curt Eaton's. All they do is take an interesting problem, distill the essence of that problem into a few key variables, capture the relationship between those in a meaningful yet tractable formula, and solve it out.

I think that's why many graduate students and academics choose to do theoretical work. "That model's so simple, I could write one like that." Trivial - until you try to do it.

Even so, setting up and solving economic models is fun. It can be done in the comfort of your home. There’s no need to find data, learn statistical software - so much less messy than, say, gluing together 30 years of census data for a pseudo-cohort study. And great theory - like great art - has a timeless, universal quality. Human capital theory changed the collective weltanschauung, the way economists thought about the world.  

Yet while producing paint blobs is easy, convincing someone that paint blobs are great art is not. There are hundreds or thousands of artists hoping to get their work in the Palm Springs Art Museum. Likewise, there are hundreds or thousands of economists hoping to get their theories published in the American Economics Review. I suspect the odds might be slightly better for the economists, but in either case the odds of becoming great, influential or recognized are pretty low.

Indeed the two evaluation processes – for art and for theory -- have much in common.  They are aesthetic judgments. When presented with theory or art, how do you decide if it’s good? Perhaps you can judge it on its objective merits – but that’s hard.  

So people use shortcuts. Is the paper by a big name, someone from a high ranked department, published as an NBER working paper? All signals of quality.  Is the paper new? An original paper might be worth reading, a reworking of established theories is boring and derivative. But with hundreds or thousands of economists writing one, two, or several papers a year, the odds that your paper will be truly pathbreaking are low. So selling matters – convincing other people of the merits of your research is as important, if not more important, than doing good work in the first place.

Don’t get me wrong. Some modern art makes me think, or laugh, or is deeply moving, while Gainsborough does nothing for me.  But if one of my kids was contemplating becoming an artist, I would do my best to give them a realistic estimate of their chances of success. And I'd do the same for any doctoral student contemplating a theoretical thesis.

(thanks to Linda, Ingela and Dave for all of your input).

Comments

[Adam P]

really great post. Spot on.

[Nick Rowe]

Yep. The analogy works.

("But is it art?")

[reason]

So Theory is mostly useless?

[Declan]

And I thought *I* was pessimistic about economic theory...

[Frances Woolley]

Reason, Tim Worstall on a previous post recalled Sturgeon's Law, that "90% of everything is crap." I'd say that mediocre theory is more useless than mediocre empirical work, for two reasons

- great theory scales. This is particularly true of economic theory. Nick can explain almost anything using just a few basic theories. Who needs a mediocre theory when you could read a great theory instead? I was thinking of blogging on a simple theory about why it's hard to make money from growing strawberries far from the equator, but really, who cares? (Well, actually I do, so I'll write the post anyways because theorizing is fun and I enjoy it, which is actually the point I'm trying to make here.)
- empirical work is about the particular. There are a handful of theories about why education pays (signalling, human capital, etc). There are thousands of studies that one could do estimating the return to human capital investment in particular times and in particular places and for particular groups of people. It's always conceivable that there's *someone* who's interested in that obscure empirical topic.

Theory is not useless. Neither is art. But becoming a theorist or an artist is a risky career move.

[Nick Rowe]

Nick's simple theory of why it's hard to make money from growing strawberries far from the equator: (well, Adam Smith's simple theory of why it's hard to make money doing almost anything almost anywhere): if it were easy, lot's of other people would already be doing it, driving the price of strawberries down, and driving the price of inputs up, until it weren't easy any more. It's only easy if you've got something very few other people have: like a special skill, knowledge, a patent, government monopoly, or luck!

See: theory is not useless. It helps us understand the world. Even really simple, abstract theories. Even when it totally ignores (abstracts away from) the fact that strawberries are coloured red and taste good!

[Jim Rootham]

Anything that scales is a risky career move.

[Wonks Anonymous]

A lot of modern art seems purposefully ugly in order to repel the rubes. Is the same true of economics?

Nick, that sounds like what Scott Aaronson dubbed a "Malthusianism":
http://scottaaronson.com/blog/?p=418

[Frances Woolley]

"See: theory is not useless. It helps us understand the world. Even really simple, abstract theories. Even when it totally ignores (abstracts away from) the fact that strawberries are coloured red and taste good!"

Or perhaps *especially* simple, abstract theories, because they allow us to focus on what really matters. Radical truths.

Wonks - good question, I hope someone better qualified than me to answer is reading this. When it comes to both art and theory my approach is "I know what I like when I see it."

[Nick Rowe]

Wonks: (I'm not more qualified than Frances to answer this, just more willing to take a chance); Economists usually aim for beauty in theories. They don't deliberately aim to repel the rubes, but sometimes in aiming to appear technically sophisticated, they have the same effect. But mostly we try as hard as possible to be widely understood. The more people that read and understand (or think they understand) your stuff, the greater your fame and career. *Provided* you don't come across as dishing up trash to pander to the unwashed. Friedman managed to pull this off; Galbraith didn't.

Those "Malthusianisms" are great! That's how a lot of economists reason, a lot of the time. "Why is the world the way it is? Suppose it weren't; then it would be in disequilibrium because there would be a force making it change towards where it is." But I think of these as Smithianisms.
Smith, to Malthus, to Darwin.

[Frances Woolley]

Wonks and Rowe - one thing that both Linda Welling and Ingela Alger (who are theorists) said when we were discussing this post is that often, once you have a theoretical model worked out, someone will say to you "well, given your assumptions, that result is pretty obvious and trivial, isn't it?"

So I think the vice of theorists is not ugliness, but obfuscation - making something appear much more complex and sophisticated than it really is, because without spurious complexity, it would be obvious that the results followed in a trivial way from the assumptions of the model.

But then again, obfuscation is ugly.

But doing something that's both interesting and elegant is fiendishly difficult.

[pcle]

Rick Szostak at University of Alberta wrote a book about a decade ago that showed how cubism and surrealism had influenced economic theory.

[Phil Koop]

I think there is a critical difference, which was expressed admirably on the Psi-Fi blog: "Although economists are generally a fairly harmless group there’s the ever present danger that they’ll get someone or other with some real power to actually listen to their beliefs. Once that happens we’re all potentially in trouble ..."

Similar (or even more grandiose) claims are made for fine art: that it provides a non-discursive expression of ideas. Such expression is said to be particularly persuasive, and in our world, the power to change the opinions of others is the greatest power of all.

But I do not believe these claims are true. Modern art is consumed by a narrow range of specialists; it is only after it has been mediated by those specialists who produce mass art - architects, graphic designers, advertisers etc - that it truly affects the rest of us. But such mediation is seldom accomplished without the interpolation or outright reversal of the perspectives of the original artists by the mediator. Artistic debts can be traced by experts, but modern originals never have the iconic power today that, say, a religious painting hanging in a renaissance cathedral had in its time. The ideas and models of 20th century economists, debased and mangled as they are in our popular understanding, nevertheless have a far more direct and material social influence.

[Min]

"Yet, because they're both abstract, economics and modern art are vulnerable to the same criticisms. My son looked at the pile of tangled metal in the art gallery and said "that looks like a bear not a horse" (it depends on how you look at it). In the same way, judgments about whether or not a particular theory captures the relevant aspects of a problem in a meaningful way are subjective."

So economics is not testable?

[westslope]

Sure economic theory is testable. The same applies to modern art though the criteria for judging success are a little different.

In the case of economic theory, we usually assume preferences or tastes are given. Modern art is a larger competition for aficionados. If the art is new, tastes are inevitably treated as malleable.

[Jim Rootham]

Is not the main difference between Friedman and Galbraith who's interest is being served by their writings?

[Frances Woolley]

Min - excellent point about testability. Testability - or falsifiability (can you prove that something's wrong?) - is a good way of separating good theories from bad theories.

So modern art and economic theory aren't exactly alike in that I don't know that modern art can be disproved.

Today when someone says "I'm an economic theorist" often they *don't* mean "I come up with theories about the way the world works, derive testable hypotheses and then go out and test them." They mean "I come up with a theory that explains an interesting stylized fact (e.g. growing strawberries in Canada is less profitable than growing strawberries in California)." Basically if you tried too do the theory part and the empirical part in one paper it would be impossibly long, no one would publish it, and researchers are so specialized that they can't do both theories and empirics (some of us can do neither).

The problem arises when you have about five plausible theories that explain why it is less profitable to grow strawberries in Canada, e.g. labour costs, cost of land, tax structure, distance to markets etc. How do you distinguish between them?

Perhaps aesthetics. Perhaps some other criterion such as methodological individualism (the best explanation is the one that explains things in terms of individuals making rational choices). Perhaps ideology.

By the way, I have no idea whether in fact it actually is less profitable to grow strawberries in Canada.

[Adam P]

Shouldn't somebody point out here that for the most part economics is not in fact testable?

Virtually all tests suffer from some version of a joint hypothesis problem. (The joint hypothesis problem usually referes to tests of the efficient market hypothesis, to see if prices are right you need a model to tell you what they should be. So how do you know if the model is just wrong or if markets are inefficient).

Another way to say this is basically all economic theories have key parameters that are entirely unobservable, things like expectations.

[Nick Rowe]

Jim: "Is not the main difference between Friedman and Galbraith who's interest is being served by their writings?"

In short, no. For example, I would say that, as an economist, Paul Krugman is much closer to Friedman than to Galbraith. Not everything is politics.

[Frances Woolley]

Adam P's answer to Min is better than mine.

[Tim Worstall]

"Shouldn't somebody point out here that for the most part economics is not in fact testable?"

Sure economics is testable. It's just that it tends to get done in reverse. East v. West Germany, N. v S Korea, Hong Kong v Mainland China (even Taiwan v both)are great tests of economics and economic theories.

The problem if you like is that the tests have happened and now we've got to try and tease the theories from the facts we've got.

As PJ O'Rouke pointed out, what the hell is it with socialism that it managed to make even Germans poor? That's a pretty good test of a socio-economic system really.....

[Wonks Anonymous]

Phil Koop, Ronald Reagan supposedly developed a great fear of nuclear war from movies like "The Day the Earth Stood Still", leading him to pursue arms agreements with the Soviets.

Adam P, W.V.O. Quine would say that all science is plagued by the joint-hypothesis problem.

Tim Worstall, a contrary take on the natural experiment of communism:
http://charleskenny.blogs.com/weblog/2008/02/what-does-the-e.html

[Wonks Anonymous]

The general term for Quine's position is "confirmation holism".
http://en.wikipedia.org/wiki/Confirmation_holism
You can find his "Two Dogmas of Empiricism" from the link.

[Nick Rowe]

Wonks: neat link on "confirmation holism". I too have found Quine's view make a lot of sense when applied to economics (and presumably to other sciences(?) too).

[Min]

On the question of testability in economics, let me bring up something that has bugged me for a while. From time to time I see economics graphs online that show something that makes me say Wow! And then I notice that the striking part of the graph is for several years in the future. :(

My first reaction is, why don't they use dotted lines for projections? And my second reaction is, where are the error bands? How can I make sense of these projections without an indication of their error? Sometimes the projections go so far into the future that I suspect that any reasonable error bands would reveal them to be absurd.

Not long ago a noted economist went on at some length on his blog comparing a couple of alternative projections for 50 (!!!) years into the future. My reaction, then and now, is how do we know that the degree of difference between them isn't just noise?

Now maybe I have gotten the wrong impression from all this. But it does make me wonder how much economists value testability.

[westslope]

The 'joint hypothesis' problem, eh? Biologists experience the same difficulties. For example: fisheries and marine biologists use pre-season forecasts to test for the negative impacts of salmon farming.

From what I'm seeing so far, the peer-reviewed articles do not even bother to discuss the accuracy of the pre-season forecasts, which are used as benchmarks and which are notoriously inaccurate. The claims of epistemological holism aside, I'd settle for an explicit, intelligent discussion. In a policy setting, and in the absence of that explicit, intelligent discussion, I call this sloppy science: 'bio-fraud'. Do we have similar term for sin-by-omission economics?

As for theory generating testable hypotheses, much theoretical development offers folks new ways to think about a problem, and in the initial stages, not much more.

Incidentally, I note that applied economists are just as reluctant to publish 3rd and 4th moments as well as diagnostic tests these days as they were several decades ago. Replication has not advanced much either.

Science.....

[PGD]

East v. West Germany, N. v S Korea, Hong Kong v Mainland China (even Taiwan v both)are great tests of economics and economic theories....As PJ O'Rouke pointed out, what the hell is it with socialism that it managed to make even Germans poor? That's a pretty good test of a socio-economic system really.....

maybe that's a test of Hayek but it's not a test of neoclassical economic theory. Economics the edifice of mathematicized speculation erected in American universities is not the "hey, I like capitalism!" economics of country-club opinion.