Rational Herds

日期:2013/01/20

The book Rational Herds: Economic Models of Social Learning by Christopher Chamley has many stories, models, and algorithms, that are helpful for gaining insight to markets. The stories start with the penguins standing on the edge of the ice, needing to get food but not knowing whether a killer whale or seal is waiting for them underneath. The first penguin to dive in provides much information for all the others. But it's not advantageous for him. The asymmetry between what's in the interest of the individual and the group and the advantages of social learning are readily seen by this example. The solution is for the other penguins to push the unlucky one in. The analogy of running the stops in markets with the first one to do so possibly losing money, but the others all gaining from the information is seen.

Another story is based on yellow cabs being 90% probable in a city. But an accident happening and the observer saying it was a red cab that caused it. Problem is that the observer's is only right 4/5 of the time. Bayesian analysis shows that after the first observation it's 9/13 that the yellow cab hit him. But after two reports the probabilities drop to around 48. The rate of convergence to red versus yellow follows a definite process which leads to all sorts of implications for cascading, herding, randomness, and social learning. Many examples of investment decisions based on following the leader and false decisions making from random events are given.



One wishes that the author would have followed some of the stories that motivated the book and shown how all the formulas would work for the simple examples above. The book is intended mainly for economics, social psychologists, finance people, and statisticians. But it's also relevant for anyone interested in how information travels. It's not easy reading and requires pencil and paper and working out a few examples to get much benefit from it.

I alternated reading it with modern times, and books on plants in my recent visit to Chicago. Glad to be back with you.