Monday, September 14, 2009

On models (following Krugman)

I already commented on the contribution of Paul Krugman on the GFC. There has been a flurry of activity since.

Our own Alan Kohler has contributed the view that economists just need to catch-up on the bit they missed, namely the behaviour of real people, writing;

The collapse of Lehman was neither the end of capitalism nor the end of economics; the capitalists have bounced back and it’s imperative that economists do the same and come up with a new theory that accepts that economies are run by flawed humans and that they are buffeted by a financial system that is subject to the madness of crowds. .

A different take was in a later article in The New York Times. In this a reseracher from the Santa Fe Institute (home of complexity theory of all kinds) was quoted as saying;

You don’t need a model of human psychology to see that there was a danger of impending disaster. But economists have failed to make models that accurately model such phenomena and adequately address their couplings.

So this is the theory that we can get it right if we just build in more terms to our model. (For the technically minded the failure in the modelling was an assumption that derivatives were spreading risk rather than compounding it. The theory worked so long as the risk in the underlying assets weren't correlated. In fact, the risks were correlated in the event of an overall slump in the housing market. The housing market could experience an overall slump because the terms of credit kept getting relaxed because the new markets in derivatives seemed to be diversifying risk. Hence, not only was the assumption about correlation wrong, the assumption about correlation was a contributing factor to it being wrong).

Also Glenn Dyer writing in Crikey noted a Professor Figlewski saying it was about not measuring the right kind of risk.

What wasn’t recognised was the importance of a different species of risk -- liquidity risk. When trust in counterparties is lost, and markets freeze up so there are no prices, really showed how different the real world was from our models.

And finally we had also in Crikey Steve Keen returns to explaining why it is still hard being a bear - his theme being that underlying all the rest is a failure of the maro-econmomic models to allow for total debt and that even if we recover through short term stimulus the underlying problem (tumour as he calls it in his analogy) is still there.

So what should we conclude? I think there is an awful lot here to note. The first is that simply adding more variables to the overall equations won't permanently solve the problem - because we are still likely to leave something out. More specifically the overall economic structure is so complex that it can't really be perfectly modelled. Our imperfect models will, however, give us an illusion of control and encourage us to pursue some extreme strategy till we have an encounter with the next flaw. This suggests a policy like the Roman one of eveything in moderation - regulatory systems should be designed to always dampen excesses.

The second is that at the very least the theory space is contended and policy makers should seek out the implications of a wide variety of advisers, not merely the established school.

Finally, thre is the position Eva Cox outlined in Crikey today - which was really in response to the Howard-Rudd stoush. She notes;

Neo-liberalism, or neo classical economics, assumes that human nature is predictable as individuals act rationally on the basis of self interest. Equations therefore can predict and explain our choices and there is no need to explore more complex bases for human behaviours such as relationships, optimism, irrational exuberance and idiocy.

These are excluded as tastes as are the pressures and pleasures of belonging to family, society, community or other groupings. Market theory is problematic to politicians because it is value free and creates winners and losers as power and resources are unevenly distributed.

This is not totally correct as an assessment of all approaches to economics - but it certainly is of the orthodox school of markets. There is an interesting conundrum in fact faced by market theorists that they cannot admit of "free will" as the perfectly rational decision making individual never has a "choice" to make, they follow a course of action determined by an array of factors around them but ultimately when faced by alternatives can only follow the one course that represents maximising their utility.

But the twin elements Cox describes are the focus of the school of behavioural economics (the psychology of decision making) and institutional economics (the behavioural rules and norms of a social grouping).

Cox goes on;

I am not arguing against markets, where appropriate, but against the ideological assumption that competition and markets are always preferable. This is what needs urgent reviewing to create more equity, efficacy and therefore efficiency.

The concession to markets begs the question of determining when they are appropriate. One of the outstanding errors in the reasoning of the orthodox is that markets are to be preferred because they are "efficient" which is deemed to be measured by the condition of a Pareto optimum. That condition is that no person can be made better off without making one person worse off. The matching concept is a Pareto improvement in which at least one person can be made better off without making one person worse off. But this also gets fudged, to the concept of a potential Pareto improvement, wherein at least one person can be made better off and could compensate those made worse off. A lot of policy is pursued to achieve potential Pareto improvements without compensation - and indeed the same market theorists argue that such compensation blunts the incentives of both the winners (the rewards are less) and losers ("rewards" for doing nothing).

Economics can do a lot better than it currently is, but should never be relied upon as the sole definitive tool for policy analysis.

No comments: