For anyoneÂ researching or teaching in the field of behavioural economics, Stefano DellaVigna has written the nicest summary paper of the literature that I’ve yet seen.
I had a quick read of the first paragraph of the paper to which you linked and I was already mildly annoyed. You need to differentiate between the standard model in its most general form and the version of the model that is commonly used as a convenient or tractable version of that model. The standard definition of rationality involves an individual with preferences that are complete, reflective and transitive. I agree that rationality is perhaps an unfortunate term to use for such people. Weak copnsistency of choices might be a better term. Nonetheless, rationality is the term that has been used to describe such preferences. See, for example, definition 1.B.1 on page 6 of Mas-Colell, A, MD Whinston and JR Green (1995), Microeconomic theory, Oxford University Press, USA.
There are preference orderings that are not additively separable across time that satisfy these assumptions. Consider a two good world in which the goods are consumption today and consumption tomorrow. A Cobb-Douglas utility function involving these two goods is rational in the sense described above. Similarly, there are preference orderings that are not linear in the probability of events that satisfy these assumptions. Consider a two good world in which the goods are consumption in state of the world one and consumption in state of the world two. A Cobb-Douglas utility function involving these two goods is rational in the sense described above. There are ven preferences that canot be represented by a utility function that satisfy these assumptions. The classic example of a rational preference ordering that cannot be represented by a utility function is lexicographic preferences.
Now, it is true that we do not have a well developed theory of strategic interaction for preferences that are not additively separable over time or non-linear in the probabilities. There is still plenty of theoretical work to be done. But nonertheless, I think the behavioural economists need to be a little more precise in their claims. This is especially the case when they are suggesting things as strong as people discount time using hypewrbolic preferences or people handle choice under uncertainty using prospect theory. Ad hoc solutions are not particularly helpful. A well specified theory is much more useful, even iof it is not accurate in all settings. The reason for this is that it provides a framework within which the impact of deviations can be considered. What happens id we relax the independance of irrelevant alternatives assumption in choice under uncertainty? What happens if people do not consider compound lotteries to be the same as the simple lottery that has the same probability distribution over final outcomes when making choices under uncertainty? What happens if people’s preferences are not time consistent and they have the opportunity to alter some of their decisions at a later date? Does completeness make sense when we are considering choices over time? The answer toi this might depend on whether we use Lancasters approach to consumer choice, in which characteristics enter the preference ordering or the more traditional approach in which commodity bundles enter the preference ordering. Of course, while Lancaster’s approach allows you to handle the introduction of new goods and services in a simpler fashion from a conceptual point of view, all it really does is transfer the ambiguity from the preference ordering to the characteristic production technologies.
I agree the intro was a little sloppy (reads like a dissertation chapter) but the paper is still a solid survey of the literature.
“Ad hoc solutions are not particularly helpful.”
I always go back to Gene Fama’s critique of behavioral finance: there’s always another behavioral theory in the toolkit to justify any given behavior ex post.
What’s even more amusing is the willingness of behavioral economists to resort to the time separable rational expectations model for their normative work…
Pingback: New Economist
Comments are closed.