“Psychology and Economics” Rabin (1998)

by vonNudge

This review paper by Rabin gives an overview of how psychological research can be used to improve the empirical predictions of economic theories. The paper is interesting in that it presents a very biased overview of psychology. The citations are overwhelmingly composed of researchers from the “heuristics and biases” literature started by Kahneman and Tversky. I believe that not only do scientific programmes undergo “paradigm shifts” in the sense of Thomas Kuhn, but that individual programmes almost always push their explanatory claims too far, leading to a backlash from the next programme. Rabin’s article fails to mention the backlash against Kahneman and Tversky that had already started in the cognitive psychology literature.

Backing up a bit, I do really appreciate behavioural economics and heuristics and biases: they have done a lot to falsify the earlier view in economics that we can be accurately modelled as individuals with unlimited rationality. This view comes up with obvious absurdities when forced to explain facts about the real world, such as the famous “equity premium puzzle” (why stocks tend to perform as well as they do). So it’s no wonder that Rabin focuses on this area of psychology, as much of his work is concerned with creating economic models based on more plausible assumptions.

A problem with the heuristics and biases programme is that it focuses almost entirely on cataloguing errors, and not explaining why they exist (which can be done by illustrating scenarios where the bias does not occur). Perhaps ironically, key examples used in two later psychological programmes, the natural frequentist view and the causal models view, were first shown by Kahneman and Tversky, but given little explanation. For example, performance on the famous “Linda the feminist bankteller problem”, where subjects often say that the probability of Linda being a feminist bankteller is higher than her being a bankteller at all (feminist or otherwise), can be improved by asking subjects to consider a sample of 100 people who meet this description. This representation of the problem makes it clearer that the set of feminist banktellers is a subset of all banktellers, and was first displayed by Kahneman and Tversky in 1983.

Perhaps encouragingly for the idea of nudging people into making better decisions, many of the classic Kahneman and Tversky findings have been found to be greatly reduced under certain circumstances. Now that the stick figure view of unlimited rationality has been cast aside, it’s time for behavioural economists to take on a broader view of psychology, and instead of merely cataloguing errors, start understanding how their impact can greatly be reduced. (For example see this early paper by Gigerenzer.) There’s also a good discussion of the Gigerenzer vs. Kahneman debate called “Ending the rationality wars“.