Today I was listening to EconTalk on my way to class and Nassim Taleb, the guest speaker, discussed the issues of “rationality” and “risk”. One of the things that struck me as a misrepresentation or perhaps a “linguistic misunderstanding” is the idea that behavioral economics seeks to label what individuals do as “irrational”. My first instinct is to say that the view of the author may be a notion that is perpetuated by general-audience books like Predictably Irrational, which discuss some of the results of the behavioral economics literature without being careful to remind the reader of the academic diction.
Nevertheless, what seems to be missing in the commentary of the guest speaker is a more critical understanding on the behavioral economics literature. We need to keep in mind that economists have often described the expected utility approach as the benchmark for “rationality” and thus, in this context, deviations from the behavior prescribed by expected utility is “irrational”. Note that this is simply a linguistic faux pas. More importantly, this is not a statement as to whether a set of behaviors – no matter if these deviate from expected utility theory or not – are optimal.
First of all, there is a common understanding in behavioral economics that biased can be optimal. For instance, there is a large literature in psychophysics (which many behavioral economists are acquainted with) that shows that perception of angles is biased towards those that occur more frequently in nature. While a gut response may be to label our perceptive system as “irrational”, if we assume the agent is a Bayesian Observer (see work by Alan Stocker and coauthors), it becomes obvious that a biased perception will reduce the overall variance and thus lead to smaller mean-squared error (which corresponds to the objective function this model seeks to minimize and is one metric used to differentiate among estimators in statistics).
Furthermore, more recent work in economics from well-known academics like Michael Woodford tries to determine whether these “irrational” behaviors (in the sense that they are not part of the classical expected utility framework) can arise as optimal responses to imprecise representations. In this line of work, Michael Woodford is able to explain some of the characteristic behaviors described in the behavioral economics literature such as the fourfold pattern of risk and other “anomalies” when it comes to risky choice (many of which were popularized by Kahneman and Tversky). This line of work largely represents the adoption of “optimal algorithms” and “bounded optimality” from the artificial intelligence literature and trying to bridge these notions with the findings of behavioral economics.
In light of some this evidence and the larger behavioral economics literature, I was struck by what appeared to be a gross misrepresentation, but more likely just a misunderstanding, in the commentary of Nassim Taleb in this episode of EconTalk. If I get a chance, I will upload some links to the literature that I mentioned above.