Learning from econophysics’ mistakes

by
Lars Pålsson Syll
My articles My siteAbout meMy booksMy videos
Follow on:Twitter

Learning from econophysics’ mistakes

By appealing to statistical mechanics, econophysicists hypothesize that we can explain the workings of the economy from simple first principles. I think that is a mistake.

To see the mistake, I’ll return to Richard Feynman’s famous lecture on atomic theory. Towards the end of the talk, he observes that atomic theory is important because it is the basis for all other branches of science, including biology:

“The most important hypothesis in all of biology, for example, is that everything that animals do, atoms do. In other words, there is nothing that living things do that cannot be understood from the point of view that they are made of atoms acting according to the laws of physics.

Richard Feynman, Lectures on Physics“

I like this quote because it is profoundly correct. There is no fundamental difference (we believe) between animate and inanimate matter. It is all just atoms. That is an astonishing piece of knowledge.

It is also, in an important sense, astonishingly useless. Imagine that a behavioral biologist complains to you that baboon behavior is difficult to predict. You console her by saying, “Don’t worry, everything that animals do, atoms do.” You are perfectly correct … and completely unhelpful.

Your acerbic quip illustrates an important asymmetry in science. Reduction does not imply resynthesis. As a particle physicist, Richard Feynman was concerned with reduction — taking animals and reducing them to atoms. But to be useful to our behavioral biologist, this reduction must be reversed. We must take atoms and resynthesize animals.

The problem is that this resynthesis is over our heads … vastly so. We can take atoms and resynthesize large molecules. But the rest (DNA, cells, organs, animals) is out of reach. When large clumps of matter interact for billions of years, weird and unpredictable things happen. That is what physicist Philip Anderson meant when he said ‘more is different’ …

The ultimate goal of science is to understand all of this structure from the bottom up. It is a monumental task. The easy part (which is still difficult) is to reduce the complex to the simple. The harder part is to take the simple parts and resynthesize the system. Often when we resynthesize, we fail spectacularly.

Economics is a good example of this failure. To be sure, the human economy is a difficult thing to understand. So there is no shame when our models fail. Still, there is a philosophical problem that hampers economics. Economists want to reduce the economy to ‘micro-foundations’ — simple principles that describe how individuals behave. Then economists want to use these principles to resynthesize the economy. It is a fool’s errand. The system is far too complex, the interconnections too poorly understood.

I have picked on econophysics because its models have the advantage of being exceptionally clear. Whereas mainstream economists obscure their assumptions in obtuse language, econophysicists are admirably explicit: “we assume humans behave like gas particles”. I admire this boldness, because it makes the pitfalls easier to see.

By throwing away ordered connections between individuals, econophysicists make the mathematics tractable. The problem is that it is these ordered connections — the complex relations between people — that define the economy. Throw them away and what you gain in mathematical traction, you lose in relevance. That’s because you are no longer describing the economy. You are describing an inert gas.

Blair Fix

Interesting blog post. Building an analysis (mostly) on tractability assumptions is a dangerous thing to do. And that goes for both mainstream economics and econophysics. Why would anyone listen to policy proposals that are based on foundations that deliberately misrepresent actual behaviour?

Defenders of microfoundations and its rational expectations equipped representative agent’s intertemporal optimisation frequently argue as if sticking with simple representative agent macroeconomic models doesn’t impart a bias to the analysis. They also often maintain that there are no methodologically coherent alternatives to microfoundations modelling. That allegation is, of course, difficult to evaluate, substantially hinging on how coherence is defined. But one thing I do know, is that the kind of microfoundationalist macroeconomics that New Classical economists and ‘New Keynesian’ economists are pursuing are not methodologically coherent according to the standard coherence definition (see e. g. here). And that ought to be rather embarrassing for those ilks of macroeconomists to whom axiomatics and deductive reasoning is the hallmark of science tout court.