Tuesday, September 29, 2009

Alan Kirman on What's Wrong with Economics

Alan Kirman, one of our contributors, was unable to post this himself for technical reasons:

The first thing that comes to mind, as one reads through the various postings about the crisis, is that economic theory was locked into a bubble which has now burst--and that the reactions have either to ignore this (John Cochrane) or to herald the return of one of our old heroes, Keynes (Paul Krugman). Economists seem to have been victims of extremely short memories and an incapacity to anticipate what the next theoretical developments will be. We periodically come to believe that we have hit upon the "right model" and that all previous efforts can be consigned to the waste basket of history. When the current model turns out to be completely at odds with reality, the reflex reaction seems to go back to the previous model and to chide the modernists for having lost sight of it. Dave Colander has tried to add a little historical perspective, but the debate remains very short sighted and ideologically motivated.

All of this seems to be misplaced. Supposing that we accept that economic theory, like the economy, is a complex adaptive system. Then we should expect to see it continually evolving and being modified to take into account both new theoretical insights and the evolution of the economy itself. We will not see theory evolving toward a given model which more closely represents the economy, since the economy itself is changing.

However, we might expect theory to evolve to at least be able to envisage the occurrence of the major crises that periodically shake the economy. We might then avoid the usual habit of falling back on the standard equilibrium notions and claiming that some major exogenous shock has hit the system. The latter sort of claim rarely identifies the shoc,k and as Bouchaud has shown, almost every significant turning point in all of the major stock price indices was accompanied by no notable news, and hence shock, at all.

Thus these large and abrupt movements must be due to the endogenous dynamics of the system. What has become the standard macroeconomic model--Dynamic General Stochastic Equilibrium (DGSE), for insiders--is justified by its proponents on the grounds that it has more "scientific" foundations than its predeceesors. By this is meant that it is based on rational maximising individuals. But there are two problems with this. First, we have known since the mid-70s that aggregating the behaviour of lots of rational individuals will not necessarily lead to behaviour consistent with that of some "representative agent." Second, the axioms that are used to define "rationality" are based on the introspection of economists and not on the observed behaviour of individuals. Thus, we have wound up in the weird position of developing models which unjustifiably claim to be scientific because they are based on the idea that the economy behaves like a rational individual, when there is a wealth of evidence to show that the rationality in question has little or nothing to do with how people behave.

Why do I say we do not look back far enough? Consider the Efficient Markets Hypothesis which has ruled the roost for some years in finance. Its originator was, by common accord, Bachelier, who developed the notion of Brownian motion at the turn of the twentieth century. His argument that stock prices should follow this sort of stochastic process, after years of being ignored, was acclaimed by economists both for analytic and ideological reasons. Yet, shortly after having written his report on Bachelier’s thesis, the great French mathematician Henri PoincarĂ© observed that it would not be sensible to take this model as a basis for analysing what happens on financial markets. As he said, individuals who are close to each other, as they are in a market, do not take independent decisions, they watch each other and it is always herd behaviour that persists. Thus PoincarĂ© clearly envisaged one of the most prevalent features of financial markets long before modern economists took this theme up to explain "excess volatility."

With regard to modern macro models, the same Poincaré wrote to Walras and chided him for his assumptions of infinite farsightedness and infinite selfishness. The latter he could believe at a pinch, but the former he found dubious, to say the least. Yet, while in other areas of economics we have moved on from these assumtptions,we are still faced today with macro models in which these two assumptions are central.

This brings me to my second point. Why are we so reluctant to envisage different models and different tools? As somebody said, we went through the twentieth century developing and perfecting a model based on nineteenth century physics; perhaps in the twenty first century we could move on to a model based on twentieth century physics. But as Paul Krugman has pointed out, the vested interests are strong and to ask economists to take up a new set of tools is probably asking too much. To discard equilibrium in the standard sense and to move on to study out of equilibrium dynamics is perhaps too big a step. To place externalities, the influence of one person’s actions on another, at the centre of the action rather than to regard them as "imperfections" in our equilibrium model is a necessary step. But if we argue that the interaction between individuals is important, then we have to specify the structure of that interaction. This means that we have to study the structure and fragility of the networks which govern the interaction between individuals and again to make this central in our analysis and not just a peripheral, albeit fascinating, topic.

Such changes are essential if we are to progress, but the inertia in the economics profession is strong and whilst the economy has shown that it is capable of sliding rapidly into a new phase, economists may well self organise to prevent this from happening to them in the immediate future. But in the end we will move on, for as Max Planck said, "A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it."

--Alan Kirman

1 comment:

Michael F. Martin said...

First, we have known since the mid-70s that aggregating the behaviour of lots of rational individuals will not necessarily lead to behaviour consistent with that of some "representative agent."

I couldn't (and didn't) have said it better.

The underlying problem is the misinterpretation of mean-field approximations.

I see that everybody is already on-board with the "complex adaptive system" label for this new and growing body of theory. I hate the "complex." What's the point of a theory if it doesn't make things simpler? Besides there were people doing work on systems theory or "general systems theory" at MIT in the 1960s, albeit without the help of computers. Why not latch ourselves to them? Lately, I've developed a great fondness for Douglas McGregor who did his work in management and organizational theory, but who anticipated most of the applications of CAS to economics at large.