Tim Harford had a great column in the FT last weekend in which he discussed some of the limits of what I’ll call micro-behavioural economics (in general I think micro/macro splits are problematic, but this branch is operating as such). Basically, Harford describes the Thaler-Sunstein policy nexus, wherein minor policies will have large impacts because humans behave in ways violating assumptions of neoclassical economics:
Behavioural economists point out cases in which our decisions don’t match neoclassical theory, and thus the “as if” defence fails…
Consider the human response to risk. Neoclassical economics says that we act as if considering all possible outcomes, figuring out the probability and utility of each outcome, multiplying the probabilities with the utilities, and maximising expected utility. Clearly we do not in fact do this – nor do we act as if we do.
Behavioural economics offers prospect theory instead, which gives more weight to losses than gains and provides a better fit for the choices observed in the laboratory…
But, say Berg and Gigerenzer, it is even more unrealistic as a description of the decision-making progress, because it still requires weighing up every possible outcome, but then deploys even harder sums to produce a decision. It may describe what we choose, but not how we choose…
This is tough on behavioural economists, because in order to be taken seriously by other economists they have had to play the optimising game. Switching to Gigerenzer’s rules would mean the end of economics as we know it.
Yet the critique is sobering. If behavioural economists do not really understand why we do what we do, there are surely limits and dangers to the project of nudging us to do it better.
Indeed, it’s unlikely if behavioral economics will ever get at the “why” (perhaps neuroeconomics will some day, but it’s hard to envision how that branch will unfold). The whole nudge policy nexus aims for low-hanging fruit- policies that, for a variety of reasons, will likely work in getting whatever goal is sought. However, this type of behavioral economics will not help us better model an economy, predict crises, et al. In part, it’s because decision-making is fluid, and changes in interaction with other agents in the world. And in part, it’s because decision-making doesn’t necessarily tend to maximize anything in particular.
Indeed, I think we’re better off looking in a direction that Daniel Little points to, wherein real, macro-level complexity reigns supreme.
Axelrod and Cohen make use of three high-level concepts to describe the development of complex adaptive systems: variation, interaction, and selection. Variation is critical here, as it is in evolutionary biology, because it provides a source of potentially successful innovation — in strategies, in organizations, in rules of action. The idea of adaptation is central to their analysis — in this case, adaptation and modification of strategies by agents in light of current and past success. Interaction occurs when agents and organizations intersect in the application of their strategies — often producing unforeseen consequences.
It’s likely that models of the real-world economy with these characteristics will be unsolvable, because these features are difficult to turn into datapoints. Nevertheless, the more ambitious project of seeing how real, unpredictable behavior aggregates into a socially-embedded, uncertain economy, is far more interesting than 401(k) nudges.
The answer lies in the discovery of cognitive science that human beings are not “rational.” Neurologically, it is not possible to think “rationally” because thought is bound up in emotion, and emotion is intrinsically connected with bodily sense. Thus, a fundamental assumption of neoclassical economics is erroneous.
Moreover, even most people that are qualified to think rationally, i.e., in terms of mathematical outcomes, do not actually do this because the “transaction cost” is too high. It simply is not worth it in the vast majority of cases, especially considering that key information is generally difficult to obtain, or not available and would have to be estimated statistically.
The idea that there is a representative “rational agent” maximizing utility based on perfect information through implementing decisions in perfect markets is just daft, even as a simple heuristic.
“Behavioral” economics is as scientific as psychology itself, i.e. not much. Science seeks truth along with cause and effect — being able to say “If ‘A’ happens then ‘Z’ will happen.”
Of course, “A” never happens. Instead an unmeasurable “A” + unmeasurable “B,” “C,” “D” . . . “T” all happen to various degrees and in various time frames, so like snowflakes, no two circumstances are alike. Worse than that, because at least all snowflakes are cold and six-sided.
So the whole thing devolves into an “I think . . . if” exercise, a great deal of sound and fury signifying nothing. If you enjoy futile enterprises, where you never can prove anything (but seldom can be proven wrong), behavioral economics is for you.
Rodger Malcolm Mitchell
Your post is exactly right. My view is that we will one day look back to behavioral economics as a short-lived fad, like pogs and crocs.
The funny part about mainstream macroeconomic models is that although new graduate students can only master them by struggling through problem set after problem set of complicated math, the models assume that ALL economic agents fully understand the model correctly, and that it is the correct model. In response to this, some people like Paul Krugman have abundant faith in the behavioral approach, which does not require agents to be so ‘smart.’
The problem, of course, is that modeling ‘stupid’ agents is even harder. Explaining ‘stupid’ behavior by merely saying agents are stupid hardly explains anything, and does not make for much of a theory.