Friday, 15 June 2012

Ariely on dishonesty

I got the new Ariely book about dishonesty the moment it landed. Now that is an interesting topic, probably one of the most interesting topics in social science. Why? Well, take any game, i.e. any economic or social situation. Almost all of game theory, i.e. of the study of conflict of interest, becomes irrelevant if the players can sign contracts, because they can just agree on one of the possible efficient solutions. So why then is game theory still widely studied and important? Because in lots of important situations -- war, crime, non-contractible outputs -- legal contracts can't be made.

Except that if people are honest, then they will do what they say; and so promises become contracts. Robert Solow makes this point in his review of Titmuss' book on blood donation. Thomas Hobbes understood it earlier, and put it as rule number two of the Law of Nature: "That men perform their covenants made."

So, dishonesty makes game theory interesting, and social outcomes much worse for everyone.

Professor Ariely's book has a really simple, interesting thesis. Most dishonesty, he says, is not bad people cheating; it's good people cheating a little bit. The most powerful example is when he computes the money he's lost by being actually swindled, compared to the amounts he's lost because of insurers, advertisers etc. marginally overcharging, marginally exaggerating the benefits of their product, and so on. Anyway, he pursues that thesis in many clever and interesting experiments, all written up in his witty prose. Top-notch popular science.

Ironically for a book on dishonesty, many of the experiments involve deception of (as well as by!) the experimental subjects. For example, sometimes the researchers hire an actor to pretend to be dishonest, so as to see if others copy him. In the world of social science experiments, this is a controversial topic. Sometimes deceipt can be unfair to be participants. A much bigger problem, in my view, is that deceipt is going to make future subjects suspicious. Suppose this book is as successful as it deserves -- and students love this behavioural stuff right now. Suddenly, all our potential experimental subjects are going to be in our labs going "where's the trick? Who's the actor?"

Of course, once that happens, we lose control and have no way of knowing how subjects see their situation. (Here is the classic example: when a subject in a psychological experiment had an epileptic fit, 3 of 5 other subjects thought this was part of the experiment!)

For this reason, many experimental economists believe, as I do, in making experiments honest wherever possible. Then at least our experiments will have a good reputation. In future, we may even have to announce "In our experiments, unlike the psychology department's experiments, you will not be deceived"! That's harsh, but if psychologists like Dan Ariely keep writing these brilliantly popular books about experiments where they deceive people, it may be necessary.