Lars P. Syll — The ergodicity problem in economics (wonkish)

Less wonkishly, the basic problem here can be viewed in terms of the logical fallacy of hasty generalization. Hasty generalization involves extending one's one's position, or that held by one's group, universally. In philosophy this result in claims of naturalism to humanity as a whole. For example, natural law is often reducible to a particular set of Western values that is generalized. The "laws" of economics are largely of this sort, and homo economicus as a rational agent that carries them out is basically a reflection of the economists that posited them, assuming all to be like them.

This fallacy has been a temptation from ancient times, but it culminated in the scientific age with the discovery of invariant laws of nature, in physics and astronomy in particular, in that these discoveries could be rendered universally using mathematical expressions. Subsequently, this formalism became a criterion of truth that prevailed for formalists above empirical observation. Owing to the success and prestige of the natural sciences, would-be scientists in other fields, and philosophers as well, sought to emulate the formalism of the natural sciences.

However, the great success of the natural sciences in discovering invariant lays in the ergodicity of the subject matter, which is rendered the mathematical expressions and formal models time-invariant. Lacking ergodicity of subject matter, this would not apply strictly. There is a significant difference between a general case and specific cases. In Economics Rules: The Rights and Wrongs of The Dismal Science, Dani Rodrik argues that of the art or craft of economics is being able to discern which model applies in which case. The natural sciences are not concerned with this kind of decision in the same way. There is a clear difference among theoretical science, experimental science, and engineering.

There is an old joke about some engineers and an economist shipwrecked with nothing to eat other than canned food. The engineers set about trying to figure out how to open the cans by applying their theoretical expertise and practical experience. The economist chimed with, "Let's just assume a can opener."

This actually happened in a less dramatic way. In effecting his synthesis of Keynesian and neoclassical thought, Paul Samuelson was confronted with Keynes having posited future uncertainty at the foundation of the "moral sciences," which we now call social science, including economics.

Samuelson solved the difficulty by assuming ergodicity as a methodological convenience for tractability, as had neoclassical economics in assuming equilibrium. This view became orthodox in conventional economics. The follow-up retort to heterodox objections then became, "The methodological debated is already settled." As Paul Krugman asserted, equilibrium and maximization as a framework.

This doesn't mean that economics or the other social sciences are not scientific or cannot be scientific. It just means that they are not the same as natural sciences and that making claims that approach this are unjustified.

Moreover, there is a difference between the meaning of being a science and being scientific. Being scientific just means observing the scientific method. Engineering is scientific in its approach, but this is applied science.

Being a science assumes a framework in terms of which theories can be compete. For example, the framework of physics includes the conservation laws, which are universal and independent. Of course, there is change over time in physics owing to motion and entropy, for example. But these phenomena are explained using models that data supports. The explanation (formula) is time-invariant, even though the data change.

Economics has no such framework, which is why there are competing views of how to approach economics in the first place. "The law of supply and demand" is not the same as the conservation laws in physics, and the assumption of equilibrium is not ergodicity.

Not being ergodic, economics is not a natural science, which is not the same as saying that economics cannot discover universal invariances that data support regardless of time series, economics, like the other social sciences, being historical. Moreover, social systems are complex adaptive system subject to reflexivity (learning from feedback) and emergence (change that is unforeseeable based on priors).

So the next time someone says, "Where's your model," ask them, "Which one?" 😀

Not one in business or finance takes forecasts as the same as or similar to the physics, and no one confuses weather forecasts to astronomical invariances. But economic forecasts and the reasoning which they are based are often treated as dogma in policy circles. That's a problem.

Lars P. Syll’s Blog
The ergodicity problem in economics (wonkish)
Lars P. Syll | Professor, Malmo University