Critics love behavioral economics because they think it gives traditional economics the comeuppance it deserves. Antara Haldar, writing recently in The Atlantic, dishes up the standard fare:
Homo economicus, created to personify the supposedly rational way humans behave in markets, quickly came to dominate economic theory.
But then in the 1970s, the psychologists Daniel Kahneman and Amos Tversky made a big discovery. The academics drew on psychological evidence to show that the actions of human beings deviate from the ironclad rationality of Homo economicus in all sorts of ways. . . . These insights led to the founding of a new field, behavioral economics which became a household name 10 years ago, after Cass Sunstein and Richard Thaler published the best-selling book, Nudge and showed how this new understanding of human behavior could have major policy consequences.
Richard Thaler merited the Nobel Memorial Prize for his work. But Haldar emphasizes discontinuity where there is mainly continuity. “Nudge” is not a theory, and behavioral economics asserts no general “understanding of human behavior,” new or otherwise. It offers an incremental contribution to economics, and doesn’t come close to the paradigm shift its fans assume.
A big part of the problem is Haldar’s constricted styling of traditional economic theory. She paints in stark colors to highlight the discontinuity between behavioral economics and conventional economics, but the starkness misleads. Haldar writes,
[Homo economicus] is infinitely rational, possessing both unlimited cognitive capacity and access to information, but with the persona of the Marlboro Man: ruggedly self-centered, relentlessly materialistic, and a complete lone ranger. Homo economicus, created to personify the supposedly rational way humans behave in markets, quickly came to dominate economic theory.
Haldar holds a Ph.D. in economics, so she must know this assessment caricatures modern microeconomic theory beyond recognition. In her telling, economists assume people are “infinitely rational.” This means they possess “both unlimited cognitive capacity and access to information.” On this view, people make irrational decisions when they do not have full information, or when they face uncertainty, or when they need to search or learning.
As Haldar surely knows, there are entire subfields in economics, all entirely mainstream, devoted to understanding decision making with incomplete or imperfect information, decision making under uncertainty, and when people search and learn. The field of computational economics allows scholars to impose knowledge and cognitive constraints on actors, operating as “finitely rational” as the converse of Haldar’s curious phrase can imply. All of these theories proceed under the standard rationality assumption; all inconsistent with Haldar’s view of the assumptions mainstream economics makes.
Haldar’s suggestion that rationality requires full information conflates making a decision which one regrets ex post with making an irrational decision. But even very smart bets ex ante can fail to materialize ex post. These can nonetheless eminently reasonable. For example, very few people would turn down this gamble: For a wager of $100 you might receive $100,000 with 90 percent probability and $0 with ten percent probability. For most of us that’s an eminently reasonable gamble; the expected payoff is far greater than the wager. Many risk-averse people would be willing to take this gamble. The fact that ten percent of us end up with a payoff of nothing in return for our wager of $100 does not mean the gamble was irrational for those ten percent. It was a rational gamble ex ante, we just happened to lose. If offered the same gamble again, however, we would take it, no question.
Haldar further suggests mainstream economists assume people are “Ruggedly self-centered . . . and a complete lone ranger.” But what about the theory of games, distinctive because it describes interaction between two or more people? The point is precisely that people in groups make decisions differently than the Lone Ranger or Robinson Crusoe. Or the huge literature in economics on altruism?
Consider too Haldar’s assertion that economic theory paints actors as “relentlessly materialistic.” Even if one does not consider rational choice models in political science or sociology, even in economics, scholars model individuals seeking to maximize achievement of policy preferences, which are manifestly non-materialistic. If of interest, economic models can easily accommodate altruistic or spiritual preferences as well. That most economists are substantively interested in understanding economic behavior means they make simplifying assumptions about goals, such as work, earn, and consume, to make analysis easier. Those assumptions are conveniences, however, not ontologies.
When economists assume people seek to maximize income (relative to the constraints and trade-offs they face), they have no need to make assumptions regarding why people do this. While some people may seek to maximize income to spend on themselves, others may seek to maximize income to provide for their families, or to give away to the poor or to save the planet. Economics makes no assumptions about these types of personal goals. As the British evangelist, John Wesley, famously preached, “make all you can . . . save all you can . . . give all you can.” A decent microeconomic model of an individual choosing between labor and leisure works just as well for this person as for the hedonist.
But none of the advances in economics regarding information, uncertainty, or the like flow from behavioral economics. Mainstream economics abandoned the view of human rationality Haldar ascribes to it some time ago, relaxing their views to encompass a wide range of actors’ motives and assumptions. All this work is consistent with traditional economic views of rationality.
In the next post I’ll chat a bit about the modeling process in economics (and rational choice theory more generally) and comment on the introductory classes in economics that Haldar criticizes in light of this process.