The Thaler Nobel: A Nudge to Progressivism in a Populist Age
Richard Thaler won the 2017 economics Nobel. Another win for Chicago. Ironic, in a way, given that in many ways Thaler is the anti-Chicago. The fact that the prime critic of homo economicus is on the faculty of the school most associated with neoclassical economics that utilizes homo economicus as its primary analytic engine is an indication of Chicago’s self-confidence, and reflects a belief that intellectual tension is a spur to scholarly innovation. Also, it may well indicate a canny instinct about future trends in economic science.
Insofar as the Nobel is a measure of impact, this one is warranted: there is no doubt that behavioral economics, of which Thaler is the recognized leader, has had an impact on the profession. That said, this area was recently recognized with Kahneman’s Nobel a few years back, and it would have been preferable IMO to have awarded Thaler along with Kahneman.
Insofar as the substance of behavioral economics is concerned, I largely agree with Mario Rizzo’s opinions on the Thaler award. Along with Rizzo, I find it useful to divide things into positive and normative economics.
With respect to positive economics, as Rizzo notes the primary use of the rational actor assumption is to derive predictions about aggregate/market behavior. It is not at all evident how the irrational actor assumption leads to more empirically robust models. I vividly recall Gary Becker’s discussion of irrationality in an Econ 301 class at Chicago, in which he showed that rationality (in the form of utility maximization) is not necessary to derive the law of demand (though it is sufficient). Random choices on the budget line lead to a downward sloping demand curve, meaning that the location of the opportunity set, rather than how agents choose a point on that budget set, is the more important factor in causing the demand curve to slope down.
Indeed, there is a danger of falling prey to the fallacy of composition: even if certain behaviors are observed at the level of the individual does not imply that they will characterize behavior at larger elements of aggregation. Allowing for individual irrationality certainly adds modeling degrees of freedom, but that’s more of a bug than a feature, especially given the now vast number of alleged behavioral biases. There is always the risk of cherry picking this bias or that to explain a particular phenomenon, and then cherry picking another (which could be completely at odds with the first one) to explain another. This creates the risk that behavioral economics is empirically vacuous.
Further, there are already plenty of degrees of freedom even within the standard economics maximizing agent framework. Information environment–note that the most die-hard advocate of neoclassical economics and the exemplar of the Chicago School, introduced costly information in the form of search costs to explain price dispersion, which is inexplicable in the Marshallian costless information, perfect competition framework. Preferences–which raises a question: is habit persistence rational or irrational? Strategic interaction–one of the problems with game theory is that virtually any outcome is possible with rational actors depending on the details of the game, the information environment, beliefs, etc. Many phenomenon that seemed anomalous in one type of model with rational actors have been explained by tweaking one of these features all the while retaining the rational actor assumption.
So I’ve yet to see how deviating from the maximizing agent framework (and maximization is really what rationality means) improves the ability of economics to improve the empirical performance of its predictions regarding aggregate/market behavior. Meaning that the contributions of behavioral economics to positive economics are dubious, in the sense that they are unnecessary, and often subject to abuse.
But what really distinguishes behavioral economics is its avowedly normative thrust. People are irrational, and would be better off in objectively measurable ways if these behavioral biases were corrected. Furthermore, many behavioral economists, and Thaler specifically, are quite confident in their ability to identify and correct these biases–and make people better off–through “nudges”.
I have two major objections to this. The first is the fallacy of composition problem mentioned earlier. Nudged agents interact in markets, organizations, and institutions. Individual behavioral changes will lead to changes in prices and market outcomes. It does not follow that “better” individual behavior will result in “better” market outcomes–that’s the fallacy of composition in action. Economies are emergent orders, and small changes in individual behavior can lead to very different emergent outcomes. The law of unintended consequences is ruthless in its operation in emergent settings.
My second objection is more straightforward. Behavioral economics of the nudge variety is relentlessly progressive, in the political sense. There are the elite nudgers, and the irrational hoi polloi who can be improved by the beneficent interventions of the nudgers. Moreover, the elite are apparently not just benevolent, but also devoid of their own behavioral biases.
To which I reply: one of the major biases identified by behavioral economists is the overconfidence bias. Mightn’t the nudgers be particularly prone to that bias? The likely commission of the fallacy of composition suggests that they are. As does the dreary experience of social and behavioral engineering efforts large and small, where technocratic elites in their overweening confidence wreaked great havoc around the world.
Ironically, I would assert that behavioral economics actually feeds the overconfidence bias among its practitioners. A seemingly powerful intellectual tool has the tendency to do that. In economics, I would proffer Keynesianism as an example.
Shall we consider other biases as well? Given that there are many of them, we could be here for a while. Suffice it to say that once you admit the nudgers are themselves imperfect decision makers, the case for nudging becomes very weak indeed. When you add the fact that even Spock-like nudgers operate with seriously limited information (about outcomes of emergent social processes in particular), the case becomes weaker still.
Behavioral economics therefore is just what a would-be technocratic elite ordered. It provides a justification for their existence, and also for an existence that should be independent of check by popular institutions. For it would be irrational, wouldn’t it, to subject rational, bias-free technocrats to the whims of irrational individuals crippled by various behavioral biases? Decision making elites unconstrained by popular forces is the essence of progressivism (and in its extreme form, totalitarianism).
Behavioral economics is particularly precious to the elite in this populist age when technocratic elites are under attack from the hoi polloi. The Nobel committees are notoriously political, and often make political statements through their choices. I would not be surprised if the Thaler award has a strong political undercurrent, given the palpable elite panic at resurgent populism, and the decidedly elitist, progressive thrust of behavioral economics generally, and its Thaler-inspired nudge variety in particular.
Behavioral economics is very congenial to top-down approaches to social problems. It is viewed by deep skepticism with people like me who believe that the knowledge problem; emergence and the law of unintended consequences; and the deforming effects and perverse incentives of power (to name just three things) make top down solutions disastrous in most cases.
So the Thaler Nobel is accurately reflective of the influence of behavioral economics on the profession, and on the profession’s contribution to policy debates. And that is a disturbing reality.
So,we have the Nannies and the Nudgies running our lives–bleak!
Comment by eric — October 9, 2017 @ 10:31 pm
Wreaked? Don’t you mean wrought? 🙂
Comment by Tim Newman — October 9, 2017 @ 11:34 pm
It is not clear to me whether (positive) behavioral economics rests on more or less shaky ground than other subdivisions of the field. In its defense, it’s building on repeatable experiments. Whether it is necessary for most macro purposes is a different question (what if, to take Becker’s logic further, only hard budget constraints really matter for aggregate demand?) but, by way of analogy, the contribution of nuclear science to macro-level physics wasn’t particularly important until recently.
The rational actor assumption is probably unfalsifiable, like most assumptions about human nature, but it allows for unlimited flexibility. I don’t quite see yet how Becker’s or Levitt’s explanations of human decision-making are inferior to theories accepting “irrationality,” even if none offer much insight into the actual workings of the human brain. Once we posit rational behavior, any contradiction between a behavioral model and empirical data can be explained by our inability to fully comprehend the actor’s true preferences, which may be stochastic, dependent on her past preferences and other actors’ behavior, and extend to a larger set of variables than the model assumes, some possibly unobservable or unmeasurable.
Comment by Alex K. — October 10, 2017 @ 9:30 am
Economics is not a science. Tell me what economist you are talking about, and I will tell you his opinion on any issue. It’s politics, not facts.
Comment by Thomas Jefferson — October 10, 2017 @ 10:41 am
I don’t disagree that behavioral economics practiced by the bureaucratic class is reminiscent of scenes from Animal Farm, 1984, or Fahrenheit 451. However, my gut tells me I have seen it in action in trading pits. I also think that when applied to marketing/consumer behavior it can help marketers sell more stuff to people. Clearly, people are not always rational or econs all the time. Yet, the classical model works a lot better than other models when you want to predict or explain things. Recently in our home of Chicago, the Cook County Board levied a .01 cent tax per ounce on pop and thought it would bring in $200M per year in revenue. In the first month, it only brought in $180k as people switched their preferences and either bought somewhere else-or didn’t buy. I’d love to see the behaviorists take on that outcome. Clearly, the Econs were rational. Or where they irrational because they didn’t cough up the extra .12 cents per can of pop?
Comment by @pointsnfigures — October 10, 2017 @ 7:36 pm
@Alex K and @TJ-
Economics is not an experimental science, for the most part. This is inherent in the nature of the phenomena it studies. It is a social science, and running controlled experiments on societies is either impossible, is profoundly immoral, or usually has horrific results. The inability to run controlled experiments makes it difficult to assess causation in economic/social phenomena, hence the recent focus on “natural” experiments in which individuals are assigned different treatments due to some exogenous intervention.
Given the constraints of being a non-experimental science, economics still strives to follow the scientific method of theory-hypothesis-test, not always successfully or faithfully. But as we’ve known since Kuhn, even physical and biological sciences often fail to follow the classical scientific method. The paradigm/anomaly/change dynamic works in economics as it does in other physical sciences. From a sociology of science perspective, and from an ideal perspective, economics is scientific though it cannot rely on experiment to the same degree as some other sciences.
Alex K remarks about behavioral economics’ reliance on “repeatable experiments.” Yes, but. . . And the buts are very large, and fatal in my view.
First, as I noted (and as Rizzo emphasizes with reference to Vernon Smith, also an experimentalist), even if one believes the experimental results, the fallacy-of-composition problem makes it difficult to derive hypotheses about aggregate/market/social/emergent order patterns from these experiments. To do so requires modeling the interactions of the irrational individuals supposedly proven by the experiments. I do not believe that behavioral economics has done this, and indeed is far behind traditional economics in doing so. I understand the discipline is a relatively new one, but I really don’t perceive the same rate of progress in that endeavor as more traditional economics achieves, despite its more mature state.
Second, as Rizzo also notes, even on the level of individuals, it is not clear that the “odd” behavior observed in the lab actually occurs in the real world, at least with the frequency that it does in the contrived conditions of the lab.
Third, the types of experiments that are common in behavioral economics are very similar to experiments in psychology which are characterized by the “replicability crisis.” That is, we cannot be confident that the biases exhibited in experiments are robust even within the lab environment.
With regards to falsification, specific models embedding the rational actor assumption are clearly falsifiable. It happens all the time. This motivates the formulation of other models which vary certain other assumptions of the model, which can be many. I mentioned a few in my original post–information environment, strategic interaction, institutions, and yes, preferences.
For the most part economists are reluctant to rely on models whose implications are driven by specific assumptions about preferences. More plausible models rely on weak assumptions, such as more is preferred to less, and semi-concavity. Finance has proved to be one area where stronger assumptions about preferences are invoked in order to explain certain empirical asset pricing regularities. That’s why I mentioned habit persistence preferences in my post. They can help address certain anomalies, like the equity premium puzzle–which is itself only a puzzle because it is hard to reconcile actual behavior and the equilibrium implications of this behavior for returns with simple assumptions about preferences. That is, it is a puzzle in the context of particular assumptions about preferences.
There is a certain intuitive appeal to habit persistence (or other forms of time-inseparable preferences), but independent validation of that anyone actually has such preferences seems impossible–which is basically what Alex K is saying more generally. It seems to work tolerably well in explaining some phenomena, but I confess to some reservations about it because I am leery of theories that are dependent on particular preferences. I would be more comfortable if the theory had more testable implications than those it was designed to produce. Habit persistence preferences were posited in order to explain a particular anomaly–the high average rate of return on stocks. They can do that more or less, but since they were effectively reversed engineered to do that, I’d have more confidence if it had predictions regarding things that it wasn’t reverse engineered to explain.
Another approach to anomalies is not to fiddle with preferences, but to posit various frictions that impede risk sharing, consumption smoothing, etc. The limits to arbitrage literature in finance is in this vein. I actually find it more appealing, as we can frequently find good reasons for such frictions, e.g., asymmetric information, imperfect contract enforcement, the inability to write complete state contingent contracts, etc.
I believe all of these more traditional approaches to have been more effective in advancing understanding of aggregate phenomena than behavioral economics, and are likely to remain so in the future.
[…] about Professor Thaler. Here is the Chicago Booth release. I also think it is worth reading Craig Pirrong’s blog post. Craig is a professor and a Chicago Ph.D. He is a classical Chicago School […]
Pingback by Where Disagreement is Embraced | Points and Figures — October 11, 2017 @ 3:58 am
Thank you for taking the time to write this out – an enlightening post and comment, for me at any rate.
To add to this, Mario Rizzo’s and Douglas Whitman’s 2009 paper, The Knowledge Problem of New Paternalism, discusses, in a non-technical language, many of the arguments for and against the policies recommended by Thaler, Sunstein, Camerer and others.
Comment by Alex K. — October 11, 2017 @ 12:27 pm
Great post, including your point that Thaler should have been awarded jointly with Kahneman. But my experience since grad school is that too many of us take our theories literally instead of using them to guide our thinking. For example, I participated in an effort to find a cause of the 1987 crash, and was frustrated on the rigidity and defensiveness of my colleagues in shoehorning their analyses into their rational expectations and efficient markets frameworks. Further, even Richard Roll acknowledged that what we now call behavioral economics can be useful in explaining unique, non-repeat events such as corporate takeovers that sometimes appear to defy economic logic. So it seems to me that the idea that deviations from the formal assumptions of rationality can have significant economic effects can be quite useful in positive economics. Perhaps behavioral economics does not qualify as a theoretical framework in itself, but it certainly help explain some anomalies.
With regard to normative economics, in contrast, you’re right on the money: it’s too easily used as a smokescreen for one’s own political biases. Having worked in both government and business, I see nothing in behavioral economics that should comfort those who reflexively advocate government intervention to solve every “failure” of markets. Quite the contrary, the biases revealed by behavioral economics apply as strongly if not more so to government as to private decisions. “Nudge” does not quite describe the effect of laws like Dodd-Frank.
Comment by DrD — October 12, 2017 @ 9:04 pm
[…] what is rationality? What is rational behavior? In a word, maximizing: choosing the preferable consumption bundle given the budget constraints and certain other […]
Pingback by “Normative theories tell you the right way to think” | The Dilettante's Winterings “Normative theories tell you the right way to think” | At 55°45' N.L. — October 17, 2017 @ 7:31 am