Saturday, June 1, 2013

Complexity and Economic Policy

I am reading a fascinating book by Alan Kirman, titled Complex Economics: Individual and collective rationality Routledge: 2011 – being The Graz Schumpeter Lectures of 2007 (which interestingly had the different and more engaging title: The Economy as a Complex Interactive System: Theory, Empirical Evidence, and Experiments. I know of Kirman’s work over the last thirty years exploring the inadequacies of the “representative agent” in economics; his critique of the lack of connection between the individual agent, who interacts directly with other agents, and the macro outcomes that result. The book is something of a summary of much of the work he has done over the years on this issue and the main chapters are case-studies – of fish markets, financial markets, public goods and segregation in real estate markets (a la Schelling) respectively. His analysis uses mathematical modeling and other ideas from various disciplines that have investigated complex interactive phenomena.

This is a theme that will be familiar to Austrian economists. The Austrians, after all, ever since Menger, have wondered about these questions, and certainly Hayek has been credited with anticipating many of the insights that have emerged. Yet, it is not until his summary concluding chapter that Kirman even mentions the Austrians; but he does so in a very interesting and challenging way. Here is the relevant text (page 214ff).

My idea and aim has been to suggest that we should re-center economics on different themes than the traditional ones of efficiency and equilibrium and that we should place interaction and coordination at the center of our interests. In particular, we should focus on interaction and its consequences for aggregate behavior. … We should accept that individuals are suspended in a web of relations and linked directly and indirectly to others. … one is faced with a system which is in evolution with no special reason as to why it might converge to some stationary state.

This sort of idea as I have said before, is far from new, and it has strong echoes of the Austrian School, which fell from favor because of its ‘lack of rigor’ and also because of its ideological associations. Yet reading Schumpeter on ‘creative destruction’ and von (sic) Hayek on ‘self-organization’ one is struck by the insights which preceded, by far, the appearance of what we are pleased to call complexity theory. The Austrians were not too preoccupied with efficiency or the progress towards efficiency as are modern economists; indeed von Mises argued that it is impossible to determine and meaningless to suggest that the real economy is closer to the final state of rest and therefore manifests a superior coordination of plans and greater allocative efficiency, at one instant of time than it was at a previous instant (Mises, Human Action 245-6). He had a vision of a market in continuous restless movement drifting back towards a state of rest but constantly perturbed by the arrival of fresh information or by the discovery of opportunities by the actors.

Thus von Mises has the idea that individuals look for opportunities but often do not have the information necessary to be able to profit from them. He does not view individuals as optimizing but rather, as opportunistic. Individuals never possess the information necessary to fully optimize but this does not prevent them from trying to seize opportunities as they arise. Out of this constant search and adjustment some sort of consistency of actions arises. This is what von Hayek argued, since he had the view that markets self-organize but, in so doing, make internal adjustments as their participants learn new modes of behavior or simply how to get around the existing rules. Once again, as I have said, learning is important but individuals may be attempting to learn about moving targets, in particular other individuals who are themselves learning. There is no good reason to believe that such a process will converge, as we have seen [from the models he presents].

And here comes the kicker:

But once we recognize this we can put paid to a standard myth. The idea that markets do self-organize has been used to justify the injunction to leave them to their work. The less interference there is the better. Yet since the self-organization may not be stable, such a view is not justified. While there is a constant restructuring of the economy and the relations within it, there is always the possibility that this process will lead to a sudden and possibly catastrophic change at the aggregate level. In the simplified models of financial markets that I presented we saw how the fact that people herd on successful strategies can generate bubbles and crashes. This does not preclude the existence of long periods of apparent stability.

… [our model suggests that] a stock market crash is not the result of short-term exogenous events, but rather involves a long-term endogenous build-up, with exogenous events acting merely as triggers.

He references the work of Hyman Minsky on the ‘disruptive internal processes’ in the economy. He also refers later to the importance of the structures of interrelations between individuals to the networks (qua institutions) these interactions create. And, though he shies away from policy recommendations, he has a few remarks that suggest he rejects government responsibility for the crisis and affirms a particular type of sophisticated intervention based on the informed understanding he is providing.

I found this work to be a fascinating window into the world of a conventionally trained, turned heterodox, but not Austrian, economist – one oriented toward the use of formal quantitative reasoning. It seemed to me a good-news-bad-news story. Here are some reactions.

1.       What I found most surprising (though perhaps I should not have) is how surprising he seemed to think the importance of individual interaction was and would be for the mainstream. I mean how many decades does it take before people wake up? Maybe a lot.

2.       His appreciation of the Austrians is nice, but he doesn’t seem to appreciate the importance of subjectivism consistently applied. Most important, he does not examine the government as part of the society, as part of the social nexus (cf. Richard Wagner). If one cannot go simply from the behavior of individuals in the market to the aggregate outcomes, would not the same apply in those social organizations we group within the sphere of government? Are not the individuals acting within these networks subject to the same complex interactions that render the outcomes unpredictable, sometimes perverse? It seems like such a natural extension.

3.     That being the case, one may wonder about the problems of ideological bias that he sees in the work of the Austrians – their faith in the stability of the market process – and turn to his inability to see instability in the government process. Is this a mental block occasioned by an ideological commitment to the idea of government palliation?

4.      Nevertheless, he does make one important point for Austrians – not a new one, but one that continues to be a challenge. Austrians from Menger to Lavoie affirm that the market process is a complex one that defies reductionist understanding. And they are happy to extend this to the government process. Yet they claim that the market process is stable, more stable than the government process; and that goal-oriented government intervention is likely to fail because of this overall complexity (the knowledge and incentive problems that manifest). This strong policy conclusion is hard to “prove.” (Obviously or else the battle would already be won.)

5.    My own view is that it is possible to make this argument and not simply rely on particular interpretations of historical events – though the latter will of course be the decisive arbiters in the end. One starts with the asymmetry of incentives between market and government (situations in which outputs are sold in markets and those where they are not), and one proceeds to an analysis of the type of interactions that characterize many of the networks we regard as helpful institutions, like the market, the common law, language, conventions, standards, etc. to show how positive feedback produces convergence. By contrast, in situations like central-bank financed government expenditure to counteract recessions the feedback mechanisms from individual interactions between agents of the government, central bank, and large banks are such as to amplify divergence, etc. Network effects deserve more widespread application.  

No comments: