I’ve been, for some time, a behavioral law and economics skeptic. Sometimes this position is confused with skepticism about behavioral economics, as in — believing that behavioral economics itself offers nothing useful to economic science or is illegitimate in some way. That’s not true. Now, I have some qualms about the explanatory power of some of the behavioral models as well — but the primary critique (in my view) has always been the threat that behavioral economics will be used as the intellectual cover for regulation judged by the preferences of the regulators rather than rigorous economic analysis of any sort.
Recall, that much of behavioral economics amounts to a demonstration that individuals exhibit inconsistent preferences. But as Glen Whitman points out, the absence of knowledge about true preferences requires that the policy maker make some decisions about actual preferences. That’s hard to do. It requires a lot of information. In the case of hyperbolic discounting, for example, the conventional approach is to arbitrarily assume that true preferences are reflected by the utility function of today’s “self” rather than future “selves.” The point is that the behavioral law and economic exercise necessarily requires that planners impose some decisions about true preferences — or in the case of firm, optimal decision-making. What will be the basis for those decisions? One can naively dream that they will always be founded on the best economic theory and evidence available. But in the real world, it is apparent that one can and should be concerned that these decisions will reflect the regulators’ priors and own preferences and perhaps also those of special interests. While there are other important methodological debates about behavioral economics, this argument is at the heart of the Rizzo and Whitman “slippery slope” objections to behavioral law and economics — and suggests that behavioral economics will be used in ways that extend beyond its limitations.
One possible response to the high likelihood of this kind of abuse is to play the “I can’t help it if…” card. Here’s Richard Thaler in the Cato Unbound exchange, that one “cannot control how my ideas are used, either by those who advocate similar but more intrusive policies.” But that is hardly satisfying when one is selling the ideas to the general public (now just one-click away at Amazon and sure to “improve decisions about health, wealth and happiness ” all for under $20!). In the exchange, Thaler seems to at least implicitly agree that the ideas have been or will be used to make bad policy. Perhaps the ideas should come with an instruction manual for how to implement policies. But alas, the behaviorists tell us that nobody reads those sorts of disclosures anyway. Maybe a nudge is in order? Of course, I think Sunstein and Thaler ought to be able to sell the book without a nudge. In fact, I bought a copy. But note than many of the policy proposals in the behavioral law and economic literature — restrictions on credit cards and sin taxes for tobacco and soda come immediately to mind — involve products that some folks will use to make themselves better off and some, because of cognitive biases, will not. If cognitively biased regulators’ decision-making processes are skewed toward policies that are consistent with their policy preferences and ideological views rather than what maximizes social welfare — one would think the creator of the “choice architecture” concept could come up with something a bit more creative to “debias” the decision-maker than “I cannot control how my ideas are used.”
Its a bit of an odd moment to decide to that influencing the choices of others no longer makes sense isn’t it? In the CFPA, the so-called “plain vanilla” provision would have required those selling consumer credit products to offer consumers a “plain vanilla” version of the product and disclose the risks of the alternative product before selling any “flavored” variations. Perhaps what is called for is a similar “plain vanilla” provision for behavioral interventions where the regulator or policy maker must show that they carefully thought through the alternative, “standard” economic interventions before choosing the riskier behavioral intervention.
Its also a critical moment. Evidence in support of the suggestion that behavioral economics is being used in ways that extend well beyond its limits, and are quite plausibly welfare-decreasing, are not hard to find. For example, my recent critiques of behavioral antitrust (Nudging Antitrust Part I, Part II) suggest an abuse of behavioral economics to solve policy problems without regard to its limits. I’ve also pointed out (with co-author David Evans) that the intellectual arguments for some of the regulatory interventions in credit markets found in the CFPA/CFPB, explicitly based on behavioral economics, extend well beyond its logic and limits. Claims of legal scholars about the policy implications of behavioral economics for consumer contracting also extend well beyond any intellectual and empirical support the behavioral economic literature can provide. And as readers of TOTM will know, the legal literature in particular has played fast and loose with the endowment effect for quite some time.
Of course, the abuse of behavioral economics by regulators and legal scholars is a danger with any sort of methodological commitment and so, one can equally point out that the same regulators and judges might abuse Chicago School microeconomics, or game theory, or a non-economic methodological commitment, e.g. originalism. In these situations it is especially important for academics to identify those sorts of abuses. But it is especially beneficial for leading figures in the “abused” field to stand up and identify policy proposals do not really fit the model. This is one of the questions that I’ve had about the behavioral economics movement. Why doesn’t one see a prominent leader of that movement saying: “Wait a minute, that’s not what we had in mind” or “no, that is really not what behavioral economics says.”
Asked and answered. Much to their credit, in Wednesday’s New York Times, behavioral economists George Loewenstein and Peter Ubel do exactly that. Loewenstein and Ubel write:
But the field has its limits. As policymakers use it to devise programs, it’s becoming clear that behavioral economics is being asked to solve problems it wasn’t meant to address. Indeed, it seems in some cases that behavioral economics is being used as a political expedient, allowing policymakers to avoid painful but more effective solutions rooted in traditional economics.
While they don’t talk about some of my favorite examples in antitrust and the credit markets, they offer some of their own:
Take, for example, our nation’s obesity epidemic. The fashionable response, based on the belief that better information can lead to better behavior, is to influence consumers through things like calorie labeling — for instance, there’s a mandate in the health care reform act requiring restaurant chains to post the number of calories in their dishes. Calorie labeling is a good thing; dieters should know more about the foods they are eating. But studies of New York City’s attempt at calorie posting have found that it has had little impact on dieters’ choices.
Obesity isn’t a result of a lack of information; instead, economists argue that rising levels of obesity can be traced to falling food prices, especially for unhealthy processed foods. To combat the epidemic effectively, then, we need to change the relative price of healthful and unhealthful food — for example, we need to stop subsidizing corn, thereby raising the price of high fructose corn syrup used in sodas, and we also need to consider taxes on unhealthful foods. But because we lack the political will to change the price of junk food, we focus on consumer behavior.
Loewenstein and Ubel also discuss other interventions where standard economics might provide better results:
Our over-reliance on behavioral economics is not limited to health care. A “gallons-per-mile” bill recently passed by the New York State Senate is intended to help drivers think more clearly about the fuel consumption of the vehicles they purchase; research has shown that gallons-per-mile is a more effective means of getting drivers to appreciate the realities of fuel consumption than the traditional miles-per-gallon.
But more and better information fails to get at the core of the problem: people drive large, energy-inefficient cars because gas is still relatively cheap. An increase in the gas tax that made the price of gas reflect its true costs would be a far more effective — though much more politically painful — way to reduce fuel consumption.
The one thing that is missing are examples where the problem is not just that the behavioral intervention improves things marginally while the standard intervention (or the combination of the two) would be optimal, but that the behavioral intervention reduces welfare. And from some of the examples they use, I suspect I might disagree with the authors on what policy prescriptions “standard” economics would recommend. I believe that the credit examples in the CFPA/CFPB provide exactly that case; I also strongly believe that behavioral antitrust policy of the sort proposed by Commissioner Rosch (Part 3 of Nudging Antitrust will discuss this next week) would make consumers worse off.
Nonetheless, Loewenstein and Ubel should be applauded as “insiders” starting a high-profile discussion on the limits of behavioral economics. From the examples the authors give, and the ones I mention above, it appears inevitable that regulators and will abuse the new tools provided them by behavioral economics. Perhaps its time to talk about what to do about it. Behavioral economists ought to have something valuable to say about this rather than merely punting. The question is how to frame the policy discussion in a manner that “debiases” regulators and provide incentives for decisions that are based grounded in theory and evidence and away from their own preferences. If the plain vanilla proposal or rules like it to encourage serious deliberation about choices are good enough for credit cards, certainly such rules should also sensible for regulatory decisions about credit cards.
RSS - Posts
I am sitting in a coffee shop using their WiFi while nursing a regular iced mocha latte. Their system logs one off every two hours for a ten-minute break. It doesn’t make me buy any more stuff, though. Nudge fail! On the other hand, I come in here a lot and there are usually empty tables so they are probably gaining from my patronage.
On the broader point of the post: Just as the elite of the econ profession has shown little interest in public choice issues with traditional regulatory analysis, so they will do the same for behavioral economics. The problem is that most economists do not start out with a symmetrical attitude toward all the players in the game. They quickly slide from a conditional normative stance (“what would be good policy if we could implement it?”) to an unconditional one (“we should adopt this good policy”). In addition, the role of the adviser or the analyst writing the article is also moved behind the veil–self-referential questions about why I’m telling you this are usually left unexplored, so “talking my ideological book” is hidden as much as possible. To some extent, these conventions are academically productive, but they are dangerous for policy and politics.
First off, I am really disappointed I did not know about this blog before – it is first rate! (And has no replaced the Epicurean deal maker in my book marks list).
I think the problem with the misuse of behavior economics basically originates from the fact that since it’s inception, the policy proposals simply don’t apply the concept.
The first few pages of Nudge (and the academic paper “Libertarian Paternalism is not an Oxymoron”) are fascinating because it shows that choices are necessarily reliant on context which, in turn, is unavoidably set by some third party — and in a significant number of cases — government. Therefore, whoever sets the context of choices unavoidably influences decisions, and it would be a good thing if they did so in a way which left people are better off.
However, in either a demonstration that the this idea has little applicability, or simply a lack of imagination, the policy proposals suggested simple pigovian taxes or regulatory hurdles and calls them nudges. There is nothing new about these ideas.
These recommendations don’t change the context of choices of all — just the weighting of the choices. And that is something completely different.
If I sat down – I could probably come up with a few “nudges” to be used in private industry. (For example – coffee shops complain that one person buys one cup of coffee and uses wifi all day. Solution – just send a barista around to ask people if they want more coffee – they would sell a lot more); but when it comes to government action it’s a lot harder. Perhaps this is because the American people don’t interact with government outside the methods of voting, regulatory compliance/laws, and paying taxes. So the method of interaction (which is very important to the context) cannot be changed.
By way of background, I my Phd thesis was on choice functions and rationality, AK Sen was my external and commented that the work was highly original.
I want to make a few observations.
1. Ordinary classical preference theory cannot make sense of context dependent choice and in effect is only interested in pairwise choices. This is a huge problem area – and not matter what Dan Airely and others say, classical preference theory cannot even say when something like the contrast principle is rational or not. So, classical preference theory is going to be pretty bad at explaining most of consumer behavior.
Are there alternatives? Here are some skeptical responses – all labelled “behavioral” economics.
a) The most important work done by Tversky was to cast doubt on invariance, or what philosophers would call “truth-functionality”. Tversky’s examples of framing showed that the very same two objects a and b could be preferred differently depending on how their features were detailed. This has nothing to do with preferences being inconsistent – there simply are no preferences or rational choices if this thorough going skepticism is true.
b) Next, there the many and sometimes annoying lab experiments designed to show that people suffer from cognitive dissonance, priming, or don’t understand the choice problem. Most of the experiments start with a poorly defined problem and report with glee that the ordinary person finds some other well defined problem to solve instead. Perhaps the most annoying of these experiments are designed to show that people don’t understand intuitively probability. Most of these “behavioral” economic experiments would not even be published if they instead showed that people’s intuitions of static physics were wrong.
c) Now there is another entirely different body of work, also known as “behavioral” economics which seeks out which heuristics work, as opposed to ones which don’t. It would be important to have a better understanding of these good mental shortcuts and which could be described by an optimizing model.
In summary, we know that classical preference theory cannot explain even modest complexity, three outcomes with the contrast feature, and yet we have no good alternative choice theories at this point. Nobody should be very happy about this state of affairs.