Friday, October 10, 2008

Why Economics Will Never Be Like Physics

I try to post something that often reflects the "times". And the times of lives right now has mostly been about the economic meltdown, especially driven by the spectacular collapse of the US financial markets.

It is during this time that, as scientists, when we try to make sense out of how things occur in economics, that we continually realize that it doesn't make sense, and that many of the things that occur often do not have clear rules and patterns. So I guess an article like this one, arguing why economics will NEVER be like physics no matter how hard it tries, really is a no-brainer.

My bet is that economics will always remain as squishy as anthropology, political science, history and other social sciences. The physical sciences address phenomena-ranging from electrons, elements and the nuclear forces up to stars, galaxies and gravity-that are relatively stable and well-defined. In contrast, economies vary wildly across space and time. The U.S. economy today is radically different than it was a decade or even a month ago. Economists are chasing a rapidly moving and mutating target.

It gets worse: Protons, plasmas, planets and other strictly physical systems don't care what scientists think about them. Social systems, on the other hand, consist of objects that read newspapers, journals, books and blogs and change their behavior as a result. Newton's model of planetary motion did not affect Jupiter's course, but an initially persuasive model of dollar-yen exchange rates may affect those rates in a way that soon renders the model totally obsolete.

In other words, the effort to understand economies or indeed almost any human social system changes the system! I'm not talking about some subtle quantum-observer effect but something much more dramatic. Marx creates Marxists, Keynes Keynesians, Hayek Hayekians, leading to wars and depressions, bubbles and recessions.


It is because of this that it irritates me whenever politicians (and even economists) can talk about many policies with such utter certainty. You see politicians arguing about this policy versus that policy which can (or cannot) create jobs, or doing such-and-such will cause taxes to fall, or people to lose jobs, etc, etc. And of course, the public will buy it, details unseen as always. It certainly is ironic that fields in "hard" sciences are more concerned about uncertainty and "error bars" in our work and experimental data, while "soft" science such as politics, economics, and social sciences that deal with less-verified principles are presented with such definite certainty.

Go figure.

Zz.

10 comments:

Daniel Lecoanet said...

I'm not sure I agree completely with your analysis. I personally know an economist whose work is much more mathematically rigorous than my work in physics. However, just as in physics, the predictions are only as good as your assume in your model. I think that ``real'' economists are just as truthful and forthright about these assumptions as members of the physics community. However, there is some relatively large group trying to influence economic policy who try to make it sound like everything is well understood and known exactly. Parallels to this could probably be made to how physics and other nature sciences are covered in the media.

ZapperZ said...

But there is a distinct difference here, though. In physics, no matter what the assumptions and simplifications are in a model, there is one thing that must be applicable. It must be testable and verified via experiment. The same thing can't be said about economic model. Where exactly is such a thing testable, and more importantly, reproducible? If such a thing is possible in economics, then one can make valid predictions under the same conditions over and over again.

One can argue that the same identical situations never occurs very often, or ever. But that's the whole point of the article, isn't it? The situation is never the same twice and thus, one can never make any kind of a valid model in the first place.

As for popular media coverage, I would say that if anything, physicists tend to make it sound LESS certain than it really is. The often bastardized Feynman quote that "no one understands quantum mechanics" is one prime example. In fact, we DO understand quantum mechanics (see my earlier blog entry on this), enough that we can make popular electronic devices using it. So if anything, we should convey the impression of more certainty that what we are conveying now.

Instead, it is those people in the less certain areas who are acting more certain than they should.

Zz.

Anonymous said...

I agree with Mr. Lecoanet, I know several economists who work with very rigorous mathematical models to explain their work. Not to mention I feel your assertion is very subjective and anecdotal (it is your blog though so that's your choice). As for testing their models, have you ever read/heard/done anything with econometric work? My final point is that many people who have graduated with bachelors and masters in physics who have gotten their doctorate in economics have done great work. If they were true believers that physics was a much more certain and testable area where logic and reason dominate, why would they switch fields?

ZapperZ said...

Because they needed a job?

I'm sorry, but no one majors in physics with the ambition of working in economics after they graduate. That just doesn't happen. They may either be forced to go into it out of financial necessity, or they develop an interest in it later on when they realize that their knowledge of models can become useful.

Furthermore, just because one can come up with some model doesn't mean it is valid. If it is, then it would be as CERTAIN as physics, wouldn't it? What do you test it on? Can you experiment with it to make sure that it is valid?

Zz.

Anonymous said...

Yes you can, especially through the use of regression analysis many economists have been able to build successful models. The testing in the regression analysis will be able to tell whether each variable is statistically significant and how well the model predicts reality of your data set. Also I was not arguing that people who have switched from physics to econ did so with the original intention of switching. I was just merely making a point that people who already had a highly scientific background felt no qualms about switching over to economics which you were arguing was less exact than a "hard" science.

ZapperZ said...

But just because they switch doesn't make the field they switch to AS RIGOROUS as the field they left. There's no logical connection here at all.

As far as economic model, can you point to me one economic model that has the SAME degree of certainty as any of the established physics model?

Zz.

taz5005@gmail.com said...

Just my two cents, I am a physics major and I have been planning on going into economics since early in my sophomore year.

craig said...

"As far as economic model, can you point to me one economic model that has the SAME degree of certainty as any of the established physics model?"

Thats impossible due to the complexity of the system you are looking at, most complex systems in physics suffer chaotic problems which lead to it not being predictable practically. Planetary orbits and quantum effects are exceptions rather than the norm due to the fact they are easy to examine as isolated entities.

When you look at other systems like the weather or like thermal efect with feedback you get problems.

Thats what it comes down to really you get very little feedback in physical systems due to the low number of variables. As of such economics is more complex and so will take longer to generate good formulae.

Also when taking data there is much greater statistical error from the difficulty of deciding of which data to measure as you can not cancel variables like in physics experiments.

Physics has had over 1000 years of development alongside mathematics economics is relativly modern and so undeveloped. Hopefully economics will develop faster however due tocomputational power now available.

I'm a physics graduate btw.

ZapperZ said...

I disagree. Complexity is NOT the issue here. Take the human population of the earth. That number is STILL ridiculously TOO SMALL when compared to condensed matter system that deal with multiple orders of Avogadro's number! We know how to deal with large numbers of interaction via many-body physics, and we know what "emergent" phenomena are based on such many-body ground state.

What isn't considered in many of these economic models is that human beings are known to not follow any type of "collective" model. A single, or a small group of people can, in fact, cause a major change. And there are plenty of unpredictable behavior that often defy any rational explanation. No type of modeling will account for any of those.

Zz.

Unknown said...

I'm an undergraduate economics student. On the most recent term, my teacher stated that economics only deals with people whose behavior is considered rational, and that specialists from other sciences "bully" economists because of the several unconfirmed assumptions that economists are forced to accept due to their inability to comprehend irrational behavior. I see that you are pointing out that same problem, but have you heard about the field of neuroeconomics? I'm not very sure about how neurosciences work, but I think it's pretty much the employment of tools to determine which brain areas are active whenever a subject does especific tasks. I suppose that neuroeconomics might be able to leave aside the concept of rationality and build models according to patterns related directly to brain cells, in which case we would start dealing with particles, not with completely erratic values.

I have the conviction that the biggest mistake that could be made about economics is to not think of it as an interdisciplinary field. Words such as "econophysics" and "neuroeconomics" reveal, I guess, a slight rejection to include concepts from physics, neuroscience and other disciplines into "regular" economics, but economics, just as it is done in the present, includes already a considerable amount of knowledge from other disciplines: mathematics, history, philosophy and more. To be an economist it is needed to be a polymath. I just don't understand why most economists haven't already adopted neurosciences as part of their field, and I think that the attitude of economists of sticking to mainstream economic theory and dismissing the posibility of expanding their knowledge is the core of all the problems that can be found in economics.