Adam G. asks:
I’ve always wondered how it’s possible to square claims of irrationality against the notion of subjective preferences. It seems to me that if you take the latter seriously, then nothing is irrational—maybe people don’t keep all the money in the dictator game because the subjective value they assign the stuff they could get with the extra cash is lower than the subjective value of a certain image of themselves, or an image they like to maintain in front of an experimenter. Maybe “seeing big numbers” just happens to be a scenario that increases people’s subjective valuations. Etc.
But it seems that if you go down that road you just end up in a tautology that isn’t very useful.
First, let me get out of the way the fact that it is obvious that we are not fully rational. This is true even if we were to find later in the post that it is hard to operationalize a test of rationality. Some of the reasons for this obviousness are a priori ones. Rationality is a faculty that evolved along with all of our other faculties. We are not perfectly rational just as we don’t have perfect eyesight. Our irrationality and our imperfect eyesight are products of evolutionary neglect: there are diminishing reproductive returns to both rationality and eyesight, and at some point there is just not enough evolutionary pressure to sustain further improvements.
Furthermore, the evolutionary pressure that shaped our rationality was applied in a completely different environment than the one we find ourselves in today. We are an evolutionarily dislocated species. We evolved in small bands, but most of us now live in big cities. We evolved in a time of zero technological change; now we have Moore’s law. Property rights, agriculture, “traditional” family arrangements, and numerous other social institutions we view as commonplace are actually relatively recent inventions in evolutionary time. Consequently, even the rationality that we do have is often poorly adapted to operating in the modern world.
But even if you don’t buy these a priori arguments or the experimental evidence, the full rationality hypothesis does not pass the “look around” test. As Bertrand Russell wrote, “It has been said that man is a rational animal. All my life I have been searching for evidence which could support this.”
Rationality is often used by economists as a simplifying assumption and as a theoretical disciplining mechanism. As a simplifying assumption it works rather well; simple theories that are based on an assumption of rational behavior frequently explain and predict the phenomena they were intended to explain and predict. I often read claims that such-and-such company is behaving irrationally, but usually the author is confused and does not know the relevant model. It is often best to interpret claims of this sort as simply “I don’t understand what is going on.”
As an intellectual disciplining tool, the rationality assumption is misunderstood even by many prominent economists. Take Barro-Ricardian equivalence, the idea that rational agents will respond to deficit spending with greater saving so as to leave bequests net of future taxes constant. As an empirical hypothesis, it seems to be false. This causes some economists to rant about how stupid it is. But as a tool for advancing our understanding of macroeconomics, I think it is invaluable. Why don’t people respond to deficit spending with greater saving? Will people eventually figure out that maybe they should? The Barro-Ricardian thought experiment shows us that to a large extent, fiscal policy is a trick, and it should make us more suspicious of hydraulic, expectations-free accounts of fiscal stimulus.
Now to Adam’s point: there are a lot of frames for the experimental literature that are not very constructive. If the goal is to point out that humans are not fully rational, then we already know that. If the goal is to get economists to stop using rationality as an assumption to simplify or discipline their work, then the experimental project fails. Where the experimental project succeeds is as a for-its-own-sake exploration of human behavior, the limits of human rationality, and human institutions.
Viewed in this light, I don’t think there needs to be any conflict between judgments of irrationality and the understanding of preferences as subjective. (By the way, the dictator game is not really about rationality versus irrationality but about self-interest versus altruism). If we find that subjects primed with religious words are more likely to cooperate, that is an interesting result regardless of whether we interpret it as a strange utility function or as evidence of irrationality (we’re assuming that simply priming people with words does not actually change people’s religious beliefs!). Apply common sense and basic scientific parsimony and you’ll be fine.
Finally, I think that many or most experimental economists are good on subjective preferences. Certainly the ones I have talked to at Mason are aware of the problem, sometimes to the point of caricature. Where you really get in trouble is when the experimental literature trickles into popular writing, law journals, or policy circles. People first need to appreciate the usefulness of the rationality assumption and of insisting that preferences are subjective before they can properly evaluate the limitations of these ideas. To take a handful of experimental studies outside of the context of the broader economics literature would be irrational.