Comparison of Mises and Keynes on explanation and prediction in natural science versus social science, including economics. Being ergodic, natural science can produce causal explanations that yield precise predictions, whereas being non-ergodic, social science — including economics — tells stories. These narratives are not (necessarily) trivial, however, depending on the craft of the storytellers.
Social Democracy For The 21St Century: A Post Keynesian Perspective
No Constants in Human Behaviour?
Lord Keynes
(h/t Auburn Parks in the comments)
13 comments:
See this is what I just cant get with these people...
Rothbard writes this:
"In fact, one lesson above all should be kept in mind when considering the claims of the various groups of mathematical economists: in human action there are no quantitative constants. As a necessary corollary, all praxeological-economic laws are qualitative, not quantitative.” (Rothbard 2009: 845)."
Which I think I understand and probably agree with... but then these people go off the rails when like for instance govt policy is proposed to be utilized to correct for these qualitative differences/variations...
So we have a huge Cat 5 hurricane hit the coast and we propose to send fiscal agents in to provide for the increase in local transaction settlements for the recovery/cleanup of these now damaged/chaos "qualities" of the local destroyed area and these Austrians lose their minds...
So what are these people all about? How do their brains work?
Here they admit that economics is all about "quality" and there is no sort of "Darwinist" hard fast quantitative "Laws" that we must be subject to in our economics, but then when we propose to access this authority to correct qualitative deficiencies, they go crazy.... are they really this insane?
I cannot understand these people AT ALL...
rsp,
To me its like they are saying "we have complete flexibility to operate our economics as we humans see fit, but we just cannot allow ourselves to do this...."
This is just irrational or perhaps insane to me I'm sorry... I feel very sorry for these people all caught up in these philosophies that leave them more or less "brain dead" (to me) ... too bad for the rest of us that people with these philosophies seem to be occupying almost all of the policy positions...
rsp,
I don't think the practical problem Keynes is talking about has much to with ergodicity or non-ergodicity. It's just a consequence of the complexity of human social systems. Even if human behavior is entirely deterministic, the number of possible states and outcomes is such that many properties are not amenable to prediction.
A non-ergodic system could be very simple and relatively predictable, while an ergodic system, on the other hand, could be very complex and admit of only long-run statistical outcomes. For some reason, ergodicity has become a trendy concept, but I don't think it's the important one to focus on.
The issue is timeless v. historical. Scientific laws are independent of time and conditions. They assert that the future will closely resemble the past barring exogenous shock. We know where the planets were, are and will be at time relative to the solar system based on an ephemeris. While not deterministic, QM is stochastic in that the data are normally distributed.
However, human behavior is historical and influenced by changing circumstances and attitudes over time in ways that not presently possible to know in advance. Even if ontological certainty is assumed underlying epistemic uncertainty (Positivism), that assumption remains to be established, and so it is moot for present purposes.
Conventional economics tries to circumvent this limitations by assuming cet. par. Neoclassical economics holds that economic data is quantitative and time-independent, as in classical physics, so that causal explanation and prediction are possible with a great degree of accuracy. Before the backpedaling begins, that is.
Austrian economics holds that economic data is qualitative; yet, it is time-independent owing to the principles of human action that are known to be true a priori. So while precise quantitative prediction is not possible, relatively good predictive results can be derived from the causality that is a priori, e.g., based on his law of human (purposeful) action that Mises sets forth in Human Action
Keynes denies both that economic data is timeless and also that there are a priori laws governing human behavior qualitatively that overcome uncertainty about the future wrt economic behavior.
Neither history, nor social science, nor psychology has produced a theory based on law-like behavior. Economics is the only social discipline that has claimed that not only is this possible but it has also been accomplished.
This means that economics is contingent on circumstances and attitudes rather than necessary. Different stories must be told about these contingencies.
For example, we know that spending/saving desire is changeable and we also know from the sectoral balance approach what changes in the level of spending/saving imply for the possible scenarios, so that contingent narratives can be put forward. But we cannot know in advance the spending/spending curve. Therefore policy needs to be adaptable meet changing conditions, as well as preventive to reduce the possibility of untoward scenarios developing.
Take the recent crisis. It was not an economic crisis to begin with, but rather a financial one (Minsky) and as it turns out also a forensic one (Black, Tavakoli). The economy was doing quite well and both inflation and employment were sufficiently balanced for the Fed to be comfortable with maintaining the policy rate, even though the minutes show that there was apparently unnecessary agitation over "potentially rising" inflationary expectation.
The problems that led to the "exogenous shock" lay elsewhere and were foreseeable and avoidable. Unfortunately, economists were telling the wrong story and policy makers were listening. Since finance and forensics are not their bag, economists excused themselves, or tried to explain it away based on "rates too low too long" or government policy (CRA leading to subprime explosion). How could have seen that coming from the economic data?
I think we can say the same thing about other complex but, non-human systems. The future geological and meteorological history of a planet that contains no life would be just as difficult to predict, even though its causal history supervenes entirely on the timeless quantum mechanical laws governing the stuff of which it is made. Timelessness of underlying laws doesn't rule out an anisotropic evolutionary trajectory over historical time and a high degree of chaotic context dependence among macro phenomena.
See Lars Pålsson Syll, Rational expectations – a fallacious foundation for macroeconomics in a non-ergodic world
In the end this is what it all boils down to. We all know that many activities, relations, processes and events are of the Keynesian uncertainty type. The data do not – as models assume – unequivocally single out one decision as the only "rational" one. Neither the economist, nor the deciding individual, can fully pre-specify how people will decide when facing uncertainties and ambiguities that are ontological facts of the way the world works.
Some macroeconomists, however, still want to be able to use their hammer. So they decide to pretend that the world looks like a nail, and pretend that uncertainty can be reduced to risk. So they construct their mathematical models on that assumption. The result: financial crises and economic havoc.
How much better – how much bigger chance that we do not lull us into the comforting thought that we know everything and that everything is measurable and we have everything under control – if instead we would just admit that we often "simply do not know," and that we have to live with that uncertainty as well as it goes. Fooling people into believing that one can cope with an unknown economic future in a way similar to playing at the roulette wheels, is a sure recipe for only one thing – economic catastrophy. The unknown knowns – the things we fool ourselves to believe we know – often have more dangerous repercussions than the "Black Swans" of Knightian unknown unknowns, something quantitative risk management – based on the hypotheses of market efficiency and rational expectations – has given ample evidence of during the latest financial crisis.
"I think we can say the same thing about other complex but, non-human systems. "
By definition, complex systems are non-ergodic owing to emergence if emergence takes place in an unorganized way e.g., involving jumps and surprises. But organized emergence is regarded as weakly ergodic since it follows a pattern. So there is strong (static) and weak (dynamic) ergodicity in addition to non-ergodicity.
The question is how closely the data approaches a normal distribution. Outliers, OK. Fat tails, not. Assuming normal distribution where there may not be one is the issue.
Randomness, fat tails and ergodicity — a Keynesian perspective on Knightian uncertainty
Human behavior exhibits strongly ergodic, weakly ergodic and non-ergodic aspects to the degree that that it is highly predictable, less predictable, or unpredictable. It is not correct to equate predictability with rationality and unpredictability with irrationality. Although this sometimes occurs, it is not always true.
For example, disposition and habit are make for strong ergodicity, tendencies for weak, and feedback and learning for non-ergodicity. Most people's mindset, attitude and preferences are fairly stable, e.g., in preferring utility to disutility. They are also changeable but somewhat predictably, e.g., trend reversal where relative utility is not constant, as in liquidity preference. But this is somewhat predictable, with liquidity preference rising with uncertainty or fear. Sometimes a lot of people act unpredictably at the same time, as in bubbles. It's where they are unpredictable that gets interesting, especially when this involves crowd behavior.
Here are some relevant comments by Phil Pilkington
Ergodicity Versus History: A Critical Commentary on the Work of Ole Peters
Rational expectations is a very unrealistic assumption in itself; but I don't think it implies ergodicity. As I understand them, rational expectations models allow for multiple equilibria. They assume that the agents inside the model have probabilistic expectations for relevant outcomes that match the probabilities the model itself generates for those outcomes. But that is consistent with the model having multiple solutions, and with the dynamics of the model being non-ergodic. You just have to assume that the expectations of the agents in the system are just as time-sensitive and path dependent as the system itself.
"Rational expectations is a very unrealistic assumption in itself; but I don't think it implies ergodicity."
Lars Syll: "REH only applies to ergodic – stable and stationary stochastic – processes." ibid., p. 36.
Well, that's what Lars says, but I'm not sure that it's true. I'd like to see a theorem.
I would say that neoclassical economics dealt with simple, static, ergodic systems in the search for laws comparable to the laws of classical physics as the standard of scientific knowledge.
Keynesian attacked that view successfully and Keynesian economics became the new standard.
Subsequently the neoclassical approach was modified and Keynesianism compromised by Samuelson.
The result was Post Keynesians going their own way and the mainstream drifting back toward the neoclassical vision through emphasis on econometrics. That developed into DSGE modeling, which uses a dynamic (changing) system rather than a static one like the early neoclassical economists. That implies weakening the prior strong ergodicity but still being able to factor out randomness sufficiently to arrive at useful and reliable results. But this meant juggling many more balls in the air. However, the empirical results were not entirely satisfactory because of oversimplifying in choosing from a huge humber of factors. The answer has been to add more balls to the mix at the cost of increasing intractability. Then there is the Lucas critique that DSGE models cannot anticipate policy changes and their effects. Is this approach looking like Ptolemy's addition of epicycles?
See Richard P. Holt, j. Barkley Rosser and David Collander, The Complexity Era in Economics
This article argues that the neoclassical era in economics has ended and is being replaced by a new era. What best characterizes the new era is its acceptance that the economy is complex, and thus that it might be called the complexity era. The complexity era has not arrived through a revolution. Instead, it has evolved out of the many strains of neoclassical work, along with work done by less orthodox mainstream and heterodox economists. It is only in its beginning stages. The article discusses the work that is forming the foundation of the complexity era, and how that work will likely change the way in which we understand economic phenomena and the economics profession.
"Well, that's what Lars says, but I'm not sure that it's true. I'd like to see a theorem."
Did you read his paper?
Post a Comment