Richard Thaler’s Rejoinder to the TOTM Free to Choose Symposium

Professor Thaler was invited to respond to the TOTM Free to Choose Symposium, and he graciously accepted and offered the following response.

I have now had a chance to read through the contributions to this event and have a few thoughts to share.  I cannot, of course, reply to everything that has been said here, and in any case, most of what I would say already appears in print.  Before getting into specifics let me say one thing up front:  take a deep breath!  These posts have a lot of emotion.  I am not sure why.

On to specifics:

1. It is simply unscientific to ignore empirical evidence and theoretical progress.  Take prospect theory, for example. This is Kahneman and Tversky’s descriptive model of decision making under uncertainty published in Econometrica in 1979, and one of the most frequently cited papers in all of economics.  There is simply no doubt that prospect theory is much better at describing behavior than expected utility theory, both in the lab and in the field.  See Colin Camerer’s paper Prospect Theory in the Wild, for example.  At some point I am sure that a better theory will come along, but for now this is the best one we have.  I teach my students that they should strive to obey the axioms of rational choice and thus be expected utility maximizers, but if they are trying to predict what someone will do, use prospect theory.  No one who is really familiar with the literature could seriously disagree with this advice.  Isn’t this just good science?
2. As one of the contributors pointed out, Christine Jolls, Cass Sunstein and I defined behavioral economics as simply economics with a higher R squared.  There is nothing radical or explicitly political about this research agenda.  The question is also not whether people are “rational”, a word I try to avoid. Rather, the question is about the descriptive validity of rational choice models. Cass and I use the terms Humans and Econs to distinguish between “people” and the imaginary creatures who obey the axioms of rational choice.  The evidence is, unsurprisingly, that the world is populated mostly by Humans, not Econs.  For several years I wrote a series of articles in the Journal of Economics Perspectives on Anomalies.  (The early ones are collected in my book The Winner’s Curse.)  I wrote 19 of these columns and quit only because I got tired.  This is simply not a debatable point any more.  I find it amusing to think that the great paper written by Armen Alchian years before any of this evidence became known is still the best reply.
3. The notion that behavioral economics is primarily based on laboratory data was never very accurate and has been wildly offbase for at least 20 years.  For example, of the 19 Anomalies articles, only 5 are based on lab studies, the rest rely on so-called “real world” data, many from financial markets.  This is not to say that the findings in the laboratory should be dismissed, a notion that physicists would find laughable.  Behavioral economics is now a field that relies on a multitude of empirical methods, and I am happy to say that most of the most important findings, such as loss aversion and overconfidence, have been confirmed in numerous different domains using lots of different methods.
4. Many of the posts reflect some confusion about the role behavioral factors in competitive markets.  This is a subject that behavioralists have been writing about for 25 years.  Three early papers published in 1985 set the stage:  Akerlof and Yellen (QJE), Haltiwanger and Waldmann (AER) and Russell and me (AER). These were followed by the very influential series of papers on “noise traders” by De Long, Shleifer, Summers, and Waldmann and then by Shleifer and Vishny.  See Shleifer’s book Inefficient Markets.  The bottom line of all these papers is that noise traders, or Humans, do not disappear in markets, nor do they leave markets unaffected by their presence. An important point that many critics seem to misunderstand is that markets do not tend to eliminate less than fully rational behavior; rather, they will often cater to it.  For example, if people are too risk averse for small stakes, which Matthew Rabin has shown to be inconsistent with expected utility theory, then firms will try to sell them extended warranties.  What happens then?  It may come as a surprise to Judd Stone, but many of us have thought and written about this question.  Stone writes as follows:

“Unless behavioralists can demonstrate – or even articulate! – some reason to believe that firms (1) remain sufficiently rational to predate upon consumer biases, yet (2) despite this rationality will not compete the monopoly profits of those biases away – I echo several other contributors’ insights that the leap from behavioral quirk to public intervention is something of a non sequitur.”

Let’s examine his two questions in the context of the extended warranty question.  Suppose that narrow bracketing plus loss aversion induce consumers to be willing to pay $100 for an extended warranty that has an expected value to them of $50, and take my word for it that Rabin can shown that if you do this you will agree to other exchanges that look very, very dumb (such as turning down a 50-50 chance to lose $1000 or win a billion dollars).  It does not take a stroke of genius for some firms to figure out that you can make some money selling this type of insurance.

Now let’s introduce some competition.  Competition will drive the price of this insurance down to a competitive rate of return for sellers, but this does not at all mean that consumers are prevented from doing something dumb.  Say the competitive price is $90.  (It is very expensive to sell this kind of insurance, one policy at a time, and divvy up the profits among the insurance company, the store where the product is purchased, and the sales clerk who is rewarded for pushing it.)

The point is that we can easily have an equilibrium in which consumers are buying something that economists would recommend they not buy, and that they would not if they were expected utility maximizers.  Now, as I must have written and said thousands of times over the past 30 years, this story does not in any way imply a role for regulation, nor if there is, what form that regulation should take.  All the story does is reveal that, exactly contrary to what Stone and many other of the contributors imply, the case for laissez faire is somewhat weakened.  It is no longer possible to make the lazy laissez fare argument that nothing can go wrong in competitive markets, or that Humans always choose what is best for them.

Consider the recent financial crisis.  As we know, many of the troubles started with sub-prime mortgages.  These mortgages were often sold door to door, much like home vacuum cleaner systems in earlier times.  The mortgage brokers who worked this market were rewarded based on how many mortgages they sold, and how profitable those sales were.  Keeping in mind that in this segment of the market there is little repeat business, what should we predict about the truthfulness of the salespeople?  Clearly one plausible equilibrium is that the most deceptive salesman wins the biggest share of the market.  Competitive markets plus badly aligned incentives create a race to the bottom by mortgage salesmen.  And it is well documented that many borrowers chose badly.

5.  Okay, so now let’s talk about the thing that everyone seems most concerned about, regulation.  Let’s take my mortgage example seriously and ask what the new CFPA should do if the market is as I describe it.  The traditional tools of the regulator are to ban some products and require disclosure for others.  What would I suggest if Elizabeth Warren asked for my advice?  The first principle is to realize that regulators are Humans too.  As I wrote in one of my New York Times columns (in an article about World Cup refereeing), Professor Warren would be well advised to assume that there will eventually be a nitwit heading this office for a while. So, as Larry Summers said recently, we do not want financial regulations that require that anyone has to get any smarter.  The next principle is to recognize that choosing a mortgage is complicated, and to state the obvious, Humans do better at easy tasks than hard ones.  People are pretty good at deciding whether they like ketchup or mustard on their hot dog, in part because they have been able to conduct trial and error learning.  Choosing a mortgage is both harder and done less often. Then the job of a regulator in a “complex” market is to help consumers make better choices (that is the choices they would make for themselves if they were fully informed and capable of doing all the math), while imposing as few costs as possible on both suppliers and sophisticated consumers.  Banning all but the simplest of mortgages would punish sophisticated borrowers who would benefit from some exotic vehicle.  Imposing big costs on lenders will make everyone worse off.  We get this!

My favorite idea (discussed at length in Nudge) for an intervention in this market would be to require that all disclosure rules be made electronic.  Instead of (or perhaps in addition to for a while) giving borrowers 30 pages of fine print that no one ever reads, borrowers would get an electronic file that details all the fine print in a spread sheet.  Consumers would then take that file and with one click upload it to competitive web sites that would analyze, compare and recommend alternatives to consumers.  Shopping for a mortgage would become more like shopping for a plane ticket.  The costs to lenders of complying with this regulation would be tiny, and the benefits would potentially be huge.

One thing to notice about this suggestion is that it is just trying to make the market work more like it would in the imaginary world in which choosing mortgages is as easy as choosing between ketchup and mustard.  It would make markets more efficient, and importantly, would shift the playing field to favor the honest lenders over the deceptive ones.  And it is all based on the important premise that Humans have limited information processing capabilities, a premise that I don’t think anyone can seriously question, especially those of us getting on in years.

6.  The idea that behavioral economists want to create some sort of new nanny state is preposterous.  Many leading behavioral economists, including me, have strong libertarian leanings.  Some are actually card-carrying libertarians.  Still, much of the criticism of the field comes from conservative, free-market economists who just “know” somehow that, no matter what we say, behavioral economists are part of some left-wing, probably communist, and certainly socialist conspiracy.  I am not sure how to convince anyone that this is not true. But here is what I really think.  The philosophy of libertarian paternalism that Cass and I advocate in Nudge, could accurately titled Free to Choose, 2.0.  If people would read with care what we have written, they will see that this is accurate.  We do not advocate a larger role for government, just a more efficient, smarter way of achieving a government’s goals, whatever the democratic process determines that they should be. Does anyone advocate dumber and costlier instead? Tom Brown made this point very nicely with his excellent illustrations of effective and ineffective communications efforts.

7.  Lastly, I can certainly not comment objectively about the role my friends and fellow travelers have played in the Obama administration, or what role behavioral economics has played.  What I will say is that the Obama administration has certainly not gotten enough credit, especially from free market economists and lawyers, for their solid improvements in transparency.  On his first day in office, President Obama issued an executive order mandating transparency as the “default option” for his administration.  It says, in part:

Government should be transparent.  Transparency promotes accountability and provides information for citizens about what their Government is doing.  Information maintained by the Federal Government is a national asset. My Administration will take appropriate action, consistent with law and policy, to disclose information rapidly in forms that the public can readily find and use. Executive departments and agencies should harness new technologies to put information about their operations and decisions online and readily available to the public. Executive departments and agencies should also solicit public feedback to identify information of greatest use to the public.

Additional steps taken in many branches of government have followed this order.  One only has to look at www.data.gov to see a portion of the evidence that in two years we have moved from the most secretive administration in our history to our most open.  I know of one firm, Brightscope, that provides ratings of company retirement plans.  In the past they had to request the data on each company one by one in some office in the Department of Labor.  Within six months of Obama taking office, they received a CD with all the data, allowing them to hire a bunch of new employees and rate hundreds more firms.  And here is a subtle point about just releasing data.  Several firms discovered they had terrible 401(k) plans because of a low Brightscope rating, and they have taken appropriate steps to improve their plans, for example by reducing fees.  Can we all agree that releasing data in machine readable form is better than locking reports in file cabinets?

So as I said at the top, take a deep breath.  The alternative to us is bans and mandates.  Of course I have no control over what people do with behavioral economics.  But if someone bans soda pop, cigarettes, or marijuana for that matter, don’t blame me.  If you appreciate having better data on what the government or your employer is doing, then please feel free to send the President a thank you note.