In a recent NY Times column largely devoted to improving soccer in various ways and how those methods might be used to improve financial regulation as well, behavioral economist and Nudge author Richard Thaler writes the following about the Consumer Financial Protection Bureau:
“Above all, I’d urge the head of this agency to devise rules under the assumption that, someday, he or she will be succeeded by a nitwit.”
Very Epsteinian advice. Nothing to argue about there. But in light of our recent discussion of some questionable interpretations of empirical data related to credit cards by Elizabeth Warren (see also Ribstein and McArdle), the widely reported first-choice for the head of the agency (indeed, some describe the decision whether to appoint Warren or an alternative as a test of the President’s backbone — or if not backbone, something else quite serious), Thaler’s comment raises the following tension in my view.
The CFPB as envisioned by Professor Warren is largely based on the insights from behavioral economics. The idea is to identify instances in which consumers systematically deviate from rational behavior, and propose regulatory solutions that will improve consumer decision-making. Next, the regulator has to decide what to do to protect the error-making consumers. There are a number of approaches. One can reduce the number of errors to zero by banning a product, for example. But in the Sunstein-Thaler “libertarian paternalism” or “soft paternalism” form, this is done (in principle) by retaining consumer sovereignty over decisions but framing choices to encourage the “right” choice (that is, the one that is not an error). Error reduction is the name of the game. And the information required to play this game is substantial. One must be able to interpret data and distinguish rational error from systematic deviations from rationality that might be prone to “nudges.” This is the point of the earlier post involving some problems with Professor Warren’s interpretation of some specific data on consumer rationality in the credit card market. But on top of the availability of data sufficient to identify such errors, and measuring their social cost, and interpreting those data, one must also have information about how consumers will respond to the proposed nudges. This is complicated. Some interventions might work. Others might not. Others might have perverse unintended consequences. The welfare effects will be hard to track. For example, calibrating a “sin tax” for cigarettes on the grounds that individuals hyperbolically discount might seem simple (“impose a tax”), but calculating the right level requires information on the discount rate — which is extremely hard to measure, and for which there is little convergence in the data in terms of reasonable ranges for estimates. Take the “plain vanilla” requirement in the original CFPA legislation. Again, it is relatively simple to state the rule: “you must sell a plain vanilla product and introduce it to the consumer before you try to sell alternative products.” But regulator selection of what constitutes plain vanilla, what the disclosures about both products should look like, and measuring the welfare effects of the intervention are all quite complex. Not to mention that consumers might be irrational in multiple ways at the same time, i.e. the problem of conflicting quirks, thus making the exercise of isolating and identifying “errors” truly difficult. Note that none of this is to discourage the science of behavioral economics. Indeed, the idea is to get regulators to take the science more seriously.
The point is that while I suspect that there are some forms of behavioral-economics-based regulatory proposals that might arise out of the CFPB that would invoke truly “simple” rules in the sense Thaler seems to be getting at, i.e. rules that even the “nitwit” that succeeds the Chief might be able to implement. Note that I’m not claiming that regulation based on price theory or even game theory are simple. But abandoning the rationality assumption in economics adds a layer of complexity. Even putting other issues with the behavioral approach to consumer protection and regulation generally aside for the moment (what to do when regulators and judges are systematically irrational too, for example), the behavioral approach makes necessary (1) identifying systematically irrational behavior, (2) distinguishing it from rational behavior in a world with search, transaction, and information costs, and (3) designing Sunstein-Thaler-style “choice architecture” to ensure that the social benefits of the intervention exceed its costs render the approach quite complicated relative to alternatives. For example, “hard paternalism” such as simply banning the credit products at issue involves a much more simple rule. Never regulate is also a pretty simple rule. I’m not saying any of the three is superior — the point is just that the behavioral approach strikes me as inherently more analytically burdensome to do properly. Indeed, much of my complaint about regulators who have attempted to invoke the approach is that they do so sloppily and skip steps required both by logic and methodological soundness.
Thus, I suspect that there is a tension that lies between Thaler’s advice to devise simple rules with the behavioral economic approach that he advocates. Perhaps Thaler has in mind a subset of simple rules that could be implemented by the CFPB. But his advice is almost certainly at odds with the rules that would have to be put into place to implement many of the behaviorally-informed consumer protection regulatory proposals involving financial products. Is it possible to design a “nitwit proof” behavioral approach to consumer credit at the CFPB that follows Thaler’s advice and consists only of simple decision rules? Maybe. I’m skeptical. But is it likely based on the current discussions of the CFPB, the use of behavioral economics in the regulatory world and the legal literature, and the complexities of distinguishing irrational error from error economizing on the costs of perfection? I’d bet not.
UPDATE: Whatever disagreements I might have with Professor Warren over economic theory and interpreting empirical evidence, it should be remarkably obvious both that she is certainly not a “nitwit” and that the word “nitwit” in the title is not a reference to Professor Warren but to Thaler’s use of the term in the article. Emails and comments ignoring the issues raised in the post — which I hope suffice for interesting discussion on their own as per usual TOTM comment standards — will be summarily deleted.