Economists: They're Just Modelers

I'm reading a lot of criticism of economists by the usual suspects and it reminds me of when people criticize politicians for 'arrogance' and 'not listening to the people'. Now, we tell children that politicians are 'leaders' who 'implement the will of the people', when in reality they are naive dupes who genuinely believe what the masses believe or cynical hypocrites who are willing to make any sacrifice of conviction and self-respect in order to hold their jobs. So, the criticism works, but only for someone who believes the kiddie version of the job description. Similarly, some people think economists are those who understand the economy, who study the economy, and indeed, many are. But for the best, the most prominent, this is their avocation.

Fundamentally, economists don't know much more than your average undergraduate economics major about economics or the market. Prominent economic theorists are 'modelers', they make little models, akin to Bohr's atom or Newton's inverse square law, that are meant to explain and predict (of course not nearly as successfully), using mathematics. Economists are perhaps better understood as 'those who model economic phenomena', as opposed to 'those who understand economic phenomena'.

Famous economists like Bob Lucas, Joe Stiglitz, or Greg Mankiw, got famous for creating models useful for teaching economics to grad students, and publishing by other economists. The model that made them famous, by the time they are famous, is usually forgotten.

For example, Lucas' seminal model was a model that explains why unanticipated inflation causes business cycles. In the 1970's when he wrote it, it seemed apt as inflation uncertainty and recessions were prominent, but it doesn't really hold water, and it is mainly discussed now because it is a darn neat model, with an intuitive little functional form derived from intuitive--if implausible--assumptions: you push up this parameter (assumption), you get this nice implication.

Mankiw's model tried to show how 'menu costs', the costs of changing pricing (eg, signage, menus), led to large business cycles. Again, plausible at first, now, not so much. Great economists are given a pass if their models fail empirically, because they are primarily judged as modelers, not describers of reality. A neat model has its own virtue.

One could go on and on. An economics or financial theorist, if prominent, is not really someone who really understands the economy, but rather, someone who takes some current stylized facts and creates a model that seems capable of explaining these facts. By the time they get tenure or their Nobel prize, everyone understands either the model merely begs the question (eg, what causes productivity growth in Solow's model? What causes increasing returns to scale in Krugman's model?), or it is empirically useless (input-output models, dynamic programming).

The pedagogical value of a model can not be underestimated, because for a professor, if you can use a model to generate a lot of g-loaded logic puzzles that appear relevant to economics, it is surely much better than being forced to read a bunch of essays on how and why tax cuts work. Good models have correct answers that are easy to grade and work through; real life doesn't.

Economists aren't wrong, they are idiot savants, and there's nothing wrong with that. It's depressing at one level, but we have enough people trying to explain and predict the big picture, taking into account the political-historical-sociocultural-global warming context. That is, the world is a vast, interconnected place with lots of important things going on, and reality has to take this into account. People who try to capture this big picture are called 'journalists', and while there are many good ones, this field does not seem to be advancing, and more than one could say legal theorists are using more discriminating logic than 100 years ago.

Academics place an inordinately high premium on novelty and elegance, which often makes it irrelevant, as there is little correlation between originality and the usefulness of an idea. That they fail in describing reality, is about as shocking as discovering a Chicago politician was engaging in quid pro quos.

Comments

[Smarty_Pants:]

Good article.

Using models isn't the problem, everyone does that to some extent or another in most aspects of their lives.

The problems come from failing to know or consider the limitations on the model and what that might mean when it is applied to real life.

All models have underlying assumptions. When those assumptions are violated, the model can become dangerous to use.

This is seen in the hard sciences all the time. When I got my engineering degree, scientific calculators were still expensive and not widely available. We had to learn and understand the theory behind the material in order to minimize the amount of number crunching we did.

Engineering students today have easy access to computers and complex software to do the same calculations and what is 'taught' is more the use of those software packages than the principles being applied. Computer model use greatly speeds the completion of design problems where 'standard' conditions apply, but the students often don't realize the assumptions that come with the software or they don't realize the impact violating those assumptions might cause.

I can't tell you how many times I've seen 'answers' provided that to me by younger engineers that are obviously incorrect because a fundamental engineering tenet is violated and not properly handled by the software or accounted for in the work.

It's not the model's fault. The lack of comprehension of the model's limitations is where the shortcoming falls.

Many times the same can be said for economic models. Assuming that the price of real estate "always goes up" is probably a violation of many economic models, yet it went unquestioned for several years before the oversight became obvious and the impacts cost many people a lot of money.

Other potential assumptive oversights:

1) 40+:1 leverage isn't dangerous as long as we're making a profit
2) counterparties won't go under
3) large packages of junk MBS are uncorrelated in behavior
4) we can always borrow in the overnight market
5) Level 3 assets don't really matter
6) Don't worry, we're hedged

I'm sure SA posters can probably generate a few more to add to the list.

Jan 13 12:39 PM

[Tradememe:]

The quality of the economic models is as good as the underlying hypothesis. So, the mathematics is not the most complex, but rather it's how close the basic hypotheses are to reality.

Jan 13 01:23 PM

[drewriders:]

I would listen to economists most than stock analyst. Economists aren't trying to sell you something.

Jan 13 02:02 PM

[Joseph Gold...:]

Great article, and I am a (non-theoretical) economist. But here's a thought: If we can't rely on neat elegant models to predict economic outcomes say three months from now, how come so many seem inclined to accept modelled results on what the world's climate is going to be 50 years from now? After all, which prediction is demonstrably more complex, if one were to actually get it right?

Jan 13 05:40 PM

[BrianSJ:]

How many economists does it take to change a light bulb?
8. 1 to change the bulb and 7 to hold everything constant.

Jan 13 05:41 PM

[speculari:]

"If all economists were laid end to end, they would not reach a conclusion" - G.B. Shaw.
Excellent article! Thanks!

Jan 13 08:24 PM

[driftwood2:]

Excellent. Sheds light on the real world of economists and why we shouldn't regard them as we do.

Jan 13 10:55 PM

[Al Capital:]

Economists attempt to explain the system. Those that publish models are putting their ideas to the test, to be validated or invalidated. With either result, we learn something. Improving on economic models, or pointing out that despite the existence of a mature academic discipling there is no functional model of capitalism, has some value. Indeed, being critical of economic models is a vital part of the process of learning about them and improving our understanding of what those models are trying to simulate. But bashing economists, well that is simply pointless.

Jan 14 09:19 AM

[raytayzmd:]

...good article but not nearly scathing enough...economists KNOW the limitations of their profession; they KNOW they no clearer a crystal ball than any Jane, Dick, or Harry when it comes to predicting the future; they KNOW that their advice running an economy has about as much merit as that of a witch doctor...but do they ever admit that whenever an reporter asks for an interview or a politician asks for advice -- nope!...their faces pop up again and again in the news with them being treated as authorities but with no disclaimer pointing out the fallacy of such a claim...the public and, unfortunately, many politicians -- both groups,in general, lacking any training in critical thinking -- are prone to fall for whatever they say...as Erasmus pointed out: "There is an infinite supply of fools."

Jan 14 10:04 AM

[Smarty_Pants:]

"do they ever admit [the model's limitations] whenever ... a politician asks for advice[?]" - raytayzmd


Ah, there's the rub. Politician's aren't stupid, foolish yes, but not stupid (well, most of them anyway). It goes without saying that politicians are only going to ask for advice from economists who tell them to borrow and spend (ie. Keynesians) because that's how they buy votes with other people's money. Other views need not apply.

There's a reason why the only place you'll see debate on the shortcomings of Keynesian economics as applied is the blogosphere. Other views are not allowed to be openly debated anywhere else, only presented and ridiculed as a show in support of the current (and broken) system.

Political corruption in a broken system goes a long way toward explaining why our models aren't very robust. That's what politicians want.

Economic models are the tools politicians exploit for personal gain

Jan 14 11:21 AM

[derryl:]

The humble economists who recognize the limitations of their understanding make for poor headlines. Politicians prefer economists whose blind certainty supports an economic policy agenda. The blindly certain get all the press so those are the ones we hear about. But there are still good economists around who can reliably lay out if-then scenarios of possible futures. (I'm not an economist so I'm not saying this out of ego preservation.)

In 1973 my Econ 101-102 prof's favorite answer was, "It all depends." When I went back to school in the mid 90s my economic history/economic policy prof believed it was a mistake to take the 'political' out of political economy. By that time economic theory was all math, plugging numbers into assumptions and using those assumptions to build models that were taught as 'real' with real predictive powers. But they only modeled economic factors, whereas historically, political and economic factors work in tandem to generate actual outcomes.

Mathematical economics is good at graphing outcomes after the fact so we can understand why things happened as they did. But as for predicting the future, 'it all depends' on political and economic factors that are only dimly foreseeable and highly uncertain.

Jan 14 11:28 AM

[carey_jim:]

To paraphrase a famous philosopher: If economists didn't exist, we would have to invent them.

However, I think you 'give away your game' when you say:

'Fundamentally, economists don't know much more than your average undergraduate economics major about economics or the market.'

But as you said, it is a question of models and not 'the economy', which is vast. Models are abstractions of very complex parts of reality, which we construct in order to predict or control those small parts of the world.

Most people can't navigate through a city, for example, without a map but they don't mistake the map for the city.

One of the failings of economists is that they often mistake their maps for reality by saying such nonsensical things as 'socialism doesn't work' or 'capitalism is the exploitation of the poor by the rich.'

Most philosophy professors admit that, with all their models of wisdom, they have no more wisdom than the average layman who seeks wisdom on his/her own.

No president would think of gathering a council of philosophical advisers around him/her, but they all have a council of economic advisers.

And yet, most of us know that economists don't know more 'about economics and the market' than the average serious investor.

Jan 14 11:42 AM

[Chris B:]

Consider that if you have ever taken a prescription medicine, that medicine was probably developed based on a theoretical model built by an academic or scientist. The actual functioning of the medicine (molecules interacting in a living body) has never been observed, only the effects. Because the effects concur with the model's predictions, the model is thought to probably be true.

Now suppose some guy comes along and says that model-building is a bunk way of advancing knowledge, because people still get sick and still sometimes die. He says, without any supporting evidence, the people who built the model "don't know much more than your average undergraduate" and that they all "fail in describing reality." He says the researchers are only "creating models useful for teaching economics to grad students, and publishing by other economists" such as E=MC^2 I presume.

Of course, if that guy could come up with a method for developing knowledge that was more effective or accurrate than the model building and testing method that has generated almost all of the technology in human history, he would be the greatest genius since John Locke or David Hume.

But no, the guy's point is just to criticize as he sends bytes across an internet that started out as a model drawn up by an academic.

Jan 14 02:11 PM

[raytayzmd:]

...say what:

On Jan 14 02:11 PM Chris B wrote:

> Consider that if you have ever taken a prescription medicine, that
> medicine was probably developed based on a theoretical model built
> by an academic or scientist. The actual functioning of the medicine
> (molecules interacting in a living body) has never been observed,
> only the effects. Because the effects concur with the model's predictions,
> the model is thought to probably be true.

...hmmm, as I recall, insulin was discovered after noticed a swarm of flies feeding on the dog's urine...penicillin was discovered after someone noticed bacteria wouldn't grow around a mold colony in a contaminated petri dish...statins were discovered while looking for enzymes in fungal extracts for improving the quality of certain foodstuffs...discovery of calcium channel blockers occurred by chance in 1963 while scientists were working on developing adrenoceptor antagonists....in fact, off hand, I can't think of any new medicines developed from a model developed by a scientist...too bad serendipity can't play a bigger role in economic research -- economists might finally make some progress to something productive.

Jan 14 03:04 PM

[carey_jim:]

That's not the way medicine or science developed.

Medical molecules were not built in biochemistry labs and then tested on humans or animals. Most medicines came (and still come) from nature (herbs) and were tried on animals and people without any idea of why they worked. It is only recently that scientists have synthesized these molecules in the lab.

Modern biochemistry and nanotechnology are completely new methods of science (constructing nano machines for medical purposes) but don't have any applications in economics as far as I know.

Science is is still based on experiment. Theories are developed which best fit the experimental findings and these theories are then used to help scientists discover new facts.

When the world is too complicated (such as the human body or the economy) experiment usually trumps theory and models tend to be ad hoc and temporary.

In the past, doctors killed as many people as they cured with their completely stupid, if sometimes useful models. The Chinese had their fanciful models of yin and yang and ch (acupuncture) ... and the Europeans from the time of Hippocrates and Galen had their theory of the four humors.

Shakespeare thought the liver was the center of love but managed to say some nice things about love anyway, and Aristotle thought that the purpose of the brain was to cool the blood but he managed to use his own brain quite well in spite of what he thought it was.

Mathematical models are best used for the simplest physical processes such as those described by Euclid and Newton but don't get very far with diseases of the body or economy.

I wish more economists understood this simple fact.

George Washington was essentially bled to death by his doctors and I'm afraid the American economy was almost bled to death by the economists.

On Jan 14 02:11 PM Chris B wrote:

> Consider that if you have ever taken a prescription medicine, that
> medicine was probably developed based on a theoretical model built
> by an academic or scientist. The actual functioning of the medicine
> (molecules interacting in a living body) has never been observed,
> only the effects. Because the effects concur with the model's predictions,
> the model is thought to probably be true.
>
> Now suppose some guy comes along and says that model-building is
> a bunk way of advancing knowledge, because people still get sick
> and still sometimes die. He says, without any supporting evidence,
> the people who built the model "don't know much more than your average
> undergraduate" and that they all "fail in describing reality." He
> says the researchers are only "creating models useful for teaching
> economics to grad students, and publishing by other economists" such
> as E=MC^2 I presume.
>
> Of course, if that guy could come up with a method for developing
> knowledge that was more effective or accurrate than the model building
> and testing method that has generated almost all of the technology
> in human history, he would be the greatest genius since John Locke
> or David Hume.
>
> But no, the guy's point is just to criticize as he sends bytes across
> an internet that started out as a model drawn up by an academic.

Jan 14 03:25 PM

[Chris B:]

I think you guys miss the point. In the cases of accidental discovery a very simple theoretical model is still being built, perhaps as simple as X causes Y. As simple as this is, it is still a model or theory. Economic or scientific models are not built from a tabula rasa state, they are inspired by the existing information, and then further research is done to confirm or disconfirm the model.

In the case of penicillin, the model or hypothesis was: The mold is doing or producing something that kills bacteria but not itself, perhaps producing a waste product. Further research confirmed this simplistic model and led to another model: Perhaps penicillin could be used to cure disease in humans. Further research also confirmed this model. Nobody just stumbled into a herb patch and knew what to do.

The medical opinions held by Aristotle, Shakespere, and Washington's doctor obviously had no basis in fact or observation. This is the alternative to creating, testing, and periodically discarding models.

The author claims that economic models are too simplistic because economists do not use evidence to support their theories and have other motivations. Yet, a cursory review of an economics journal shows that compliance with observed reality is the only way economic models gain acceptance. The author does not explain why complex, ungainly models should be considered more desirable than simple, elegant models. Nor does he tell us if he believes that understandable models cannot be used to predict economic or behavioral events. Does hyperinflation occur when a government prints too much currency? Can the prevent value of a bond investment be calculated? Does E=MC^2? The author might say we can never know because such theories are too simplistic.

Thus, this article is only a Luddite complaint about the slow pace of knowledge development with no solutions offered.

Jan 14 05:14 PM

[raytayzmd:]

On Jan 14 05:14 PM Chris B wrote:

> I think you guys miss the point. In the cases of accidental discovery
> a very simple theoretical model is still being built, perhaps as
> simple as X causes Y. As simple as this is, it is still a model or
> theory. Economic or scientific models are not built from a tabula
> rasa state, they are inspired by the existing information, and then
> further research is done to confirm or disconfirm the model.

...I think you're confusing a model with an hypothesis...a model is built based on existing data and it's "quality" is determined by how successfully it recreates the exisiting data...an hypothesis is a speculation derived from an observation that and whose verity can be evaluated by statistical methods...with penicillin, the observation originally was lack of bacterial growth around the mold...the speculation or hypothesis then was that the mold was producing something that inhibited bacterial growth...an experiment was then designed to compare plain agar with agar collected from moldy dishes...the results confirmed the speculation with some degree of statistical signifficance...a major problem with economic models and theories is that it is basically impossible to generate testable hypotheses from them...the only other way of assessing validity is for the model or theory to generate some probability of future events -- in other words, they're actually able to predict future consequences for a given input...I'm confident there's not an economic model or theory in existence that can do that...consequently, economists opinions are nothing more than that -- opinions!...they have no more validity than opinions from "Joe the plumber...like children, economists are better seen than heard.

Jan 14 08:41 PM

[carey_jim:]

The definition of 'scientific method' is hotly contested and science itself can be defined in many ways.

We know, however, that during the twentieth century science attained such a high degree of respectability that everyone wanted to call what he/she does science. There is a science of cooking, a science of lovemaking and a science of literary criticism, for example.

Sociology, psychology and economics are not hard sciences, even if they make use of the results of mathematics, statistics and hard sciences such as biochemistry, ecology, geography, agriculture, anthropology and meteorology.

To bring home the point that economics is not a science, consider:

Economists usually define 'homo economicus' as a 'rational, perfectly informed and self-interested actor who desires wealth, avoids unnecessary labor, and has the ability to make judgments towards those ends.' en.wikipedia.org/wiki/...

But this definition is not based on modern psychology and sociology and is, in fact, absurd!

In the Soviet Union, Marxism which was taught in all Soviet universities, was also considered scientific, but it collapsed along with the Soviet Union.

My first sentence was 'If economists didn't exist, we would have to invent them, so I won't echo Shakespeare's famous cry 'The first thing we do, let's kill all the lawyers (economists.)' (Unfortunately we need lawyers too.)

But we can't have the same confidence in economists and their models as we have in science because economists are not doing science.

I just wish the economists would take off those pointed hats with stars on em and fess up.

Jan 16 02:24 AM