Homo Economicus or Homo ... bounded rationality?

This week's reading:
Camerer, C. F., and E. Fehr. 2006. When does "economic man" dominate social behavior? Science 311:47-52.

It has become quite trendy in some circles to reject economics for its adherence to the quaint and seriously flawed idea of "economic man," the rational utility maximizer, hell-bent on material, quantifiable gain. Well, last week the MRG read yet another paper - this one by Camerer and Fehr - that takes up this discussion. But I think we agree that this one brings some quite valuable nuance to the topic of human decision making, or "models of man," in both theoretical and practical terms.

I should say that I forced this somewhat mathy piece on my colleagues because it ties together a few major ideas that have tormented me of late:

• rational decision making,
• game theory, and
  
• bounded rationality

The third of these is an idea that I find quite appealing. It basically asserts that people make decisions with limited information and a limited investment of time and effort. Herbert Simon, in pieces like Reason in Human Affairs or Sciences of the Artificial, elegantly makes the case for this theory, providing compelling examples to back it up. But, while he effectively argues against rationality in modeling human behavior, he does little to provide us with a quantifiable or predictable alternative.

This is where the Camerer and Fehr paper comes in. They add two concepts to this list that were new to me: strategic complementarity and strategic substitutability. The analogy here is to substitutable or complementary goods, the latter being mutually reinforcing, and the former in opposition to one another. The authors show with both experimental and real-world examples that an understanding of whether strategic complementarity or substitutability is at play can be very useful in determining whether a rational model is appropriate, or whether "boundedly-rational" behavior is more likely to explain results.

This is where game theory comes in. My frustration (perhaps due to a very primitive understanding) with game theory has been what seems like an effort to generalize, at some level, the primary motivating factors that can explain human behavior in particular decision making environments. The beauty of this paper is that this endeavour can be brushed aside. Instead, the authors accept that in any situation you may have unexplainable behavior that might be rational or boundedly rational. What is more important is to understand that in some situations (strategic substitutability) the rational behavoir is dominant in determining the outcomes, thus rendering boundedly rational actors irrelevant. In situations of strategic complementarity the situation is reversed.

Of course we would lose all credibility if this blog entry was a totally reverent of the piece without some brilliantly cogent criticsim. We noted that the definition of rationality becomes a bit problematic in some of the cases presented in this article, especially once the discussion shifts to questions of strategy. For example, it is noted that in some games, the presence of boundedly rational players causes more rational players to mimic that behavior. But at this point, the decision to adopt boundedly rational behavior actually appears to be rational, at least in terms of our expectations of utility maximization. So, while behavior is homogenous, some players are genuinely boundedly rational, while others are simply acting that way to maximize their utility, but it is impossible to tell which players are which.

Our question is: in this context, what would constitute rationality? It seems that now the term refers only to what is classically expected in situations with all rational players, rather than to our more traditional notion of maximization. It would help to clarify this once the discussion gets into these complicated situations in which rational players adapt their behavior to to presence of boundedly rational players.

That aside, perhaps the most important recognition here, with respect to neoclassical economics, is that aggregate behavior does not reflect homogenous behavior or even homogenous motivations. In some situations the homo economicus model fails to predict aggregate outcomes, and in others it does not. But in neither case does it provide an accurate account of individual decision making.