Dodgy games

Economics, like most disciplines, is subject to trends. The latest fashion is behavioural economics, whose empiricism is portrayed as a cure-all for the the failings of neoclassical rationalism. Before that, a major trend was game theory, which plugged a gap in mainstream economics by purporting to show how economic actors are likely to behave in particular sets of circumstances. Until then economics was very much based on aggregation, with few clues as to the real behaviour of consumers or producers given certain sets of motivations. Game theory remains central to mainstream economics.

But all movements have their shortcomings. In this short paper Ariel Rubinstein expresses doubts about the usefulness of game theory and the validity of its predictions. Near the beginning he says that:

…we need… to deal with a fundamental difficulty of predicting behaviour in the social sciences, where prediction itself is part of the game and forecasters themselves are players.

The problem of the forecasters themselves being players is what is generally referred to as the problem of reflexivity, which Keynes talked about in the General Theory. He likened stock market investment as akin to a newspaper beauty contest in which the winner is the person who predicts the most popular entrant. A competition entrant should try to predict others’ entries, not the woman he himself finds most attractive. Other entrants will, if behaving sensibly, base their entries on others’ likely entries.

In the same way stock market investors are should base their predictions of future price movements not on what they think the stock is fundamentally worth, but on the likely behaviour of the crowd. Any notion of a fixed, underlying price becomes extremely wobbly.

Reflexivity refers to the relationship between the object of examination (humans) and the theories devised by people to examine human society. Humans are examining themselves, so to speak. This two-way relationship casts doubt on the existence of an independent truth.

In a similar vein Rubinstein allows for the possibility that game theory alters the culture of economics, and that it might in turn change people’s behaviour.

I view economics… as culture… Game theory changed the culture of economics… When we teach game theory we may be affecting the way people think and behave in economic and strategic interactions. Is it impossible that the study of game theoretical considerations in economics makes people more manipulative or more selfish?

Rubinstein also says that:

…game theory does not have normative implications and its empirical significance is limited…game theory does not tell us which action is preferable or what other people will do.

This is pretty radical stuff, but I think he’s probably right to place limits on the use of games. I remember a friend saying that two American political scientists tried to use game theory predict the precise behaviour of Yasser Arafat at a particular point in the Middle East conflict. What they didn’t take into account was the fact that Arafat had, at the time in question, left the country! Game theory is no substitute for sensible analysis based on all the available evidence.

Rubinstein concludes that although it is a privilege to play games:

…we need to keep in mind that the challenges facing the world today are far too complex to be captured in any matrix game.

Economics has a liking for cure-all solutions based on tricky procedures. Like game theory, behavioural economics will soon have been superseded by a new method of the moment. But the most useful analysis usually takes into account all the available evidence, historical, practical and theoretical, and it may use a variety of methods. There’s no such thing as a cure-all.