As mentioned in the previous post, Post Keynesians argue with mainstream economists (Neoclassicals, New Keynesians, etc.) over the ergodicity assumption in mainstream models. The ergodicity of economic data is, however the wrong argument to be having because it presumes the use of improper tools for economic analysis.
Ergodic processes allow their parameters to be deduced from a statistically significant sample of data. Human action, or any subset of it, does not have constant parameters because humans are not predictable in the way that particles, for example, are, due to subjective and ever changing preferences. Particles in water or moons orbiting a planet are ergodic, human behavior is non-ergodic.
Put another way, non-ergodic processes are path dependent, meaning their history, the path the system took to reach its current state, has some bearing on its future development and that there are many possible equilibriums for a system. This seems to be a valid assumption and provides an important place for the history of economics. Mainstream theory denies path dependency and places man, at least in its method of analysis, in the realm of automatons or particles, always seeking some predetermined, “natural” equilibrium. Mainstream economists want economics to be a quantitative discipline like the hard sciences, devoid of historical analysis.
This application of mathematics to the study of human action leads to many problems. The Post Keynesians take the non-ergodicity of human action a step further and proclaim that ontological uncertainty exists in economics, a philosophical step towards nihilism where even the use of general economic axioms is questioned. The possibility of economic coordination would even be in doubt.
The mainstream economists deny the existence of ontological uncertainty and maintain that the goal of economics is to reduce our epistemic uncertainty through the use of better models containing better mathematical techniques.
Both sides are wrong: the Post Keynesians in their flirtation with nihilism and the mainstream with their belief in the resolvability of the problems with mathematics.
There is another way, however, which does not rely on mathematical techniques and in which the problem of ontological or epistemic uncertainty does not even need to be decided. Whether or not future psychologists will be able to predict human behavior, whether or not free will exists, is not important now. Extreme uncertainty, not the mild uncertainty of a random walk, exists for us now, regardless of our future ability to tame it (doubtful though this is, and undesirable as it would reduce man to a predictable physical process).
The Austrian School uses deductive, a priori reasoning as the only valid means of coming to certain conclusions regarding human action, and an Austrian view of uncertainty is described in Chapter 6 of Mises’ Human Action. Class probability, where the frequency of a large number of events is used to assign a probability on future events, is applicable to many areas of the natural sciences, despite its use by mainstream economics. Case probability, where each event is unique, is not open to numerical analysis, since some or all of the relevant factors are unknown, but is the proper method of analyzing economic events.
The outcomes of some types of human action are able to be known with certainty, though not with mathematical certainty. The law of supply and demand holds, but the magnitude of the change in one due to a change in the other remains unknown. Constant relations do not exist, utility curves are not continuous, and a change in one factor can change the relative positions of many other factors. The use of mathematics in these questions adds nothing to our understanding of economics. Instead, it can be harmful by providing a false sense of scientific certainty where none can exist.
Sources
http://ineteconomics.org/blog/inet/paul-davidson-response-john-kay