Thursday, October 12, 2017

Irrationality, Baselines, and Government "Intervention"

by Michael Dorf

Shortly after the news of the Las Vegas mass shooting broke last week, business reporters noted that, as is typical after mass shootings, the share price of companies that manufacture firearms went up. Why? Because in the wake of such events, gun sales spike. Why? Because people who are thinking of buying a gun think they better do so quickly, before the government makes gun purchases illegal. Savvy investors know this will likely happen, so they bid up the share price of the gun manufacturers in anticipation of the increased demand, increased sales, and thus better short-term profits.

But this only makes sense if the investors also anticipate that serious gun control will not actually be enacted. After all, if a mass shooting were to lead to the enactment of serious gun control, then the short-term spike in sales and thus profits would be more than wiped out by the long-term decline in business for the gun manufacturers. And it must also be true that the potential gun purchasers do not realize that gun control won’t actually be enacted; if they did, they wouldn't feel the need to stock up now. Indeed, we must assume even more—namely that some of the people buying guns in the wake of a mass shooting are not merely buying early but buying guns they otherwise wouldn’t have bought at all. After all, if gun purchases are merely shifted from (say) 2018 Q2 to 2017 Q4, that should have a small impact on share price, which, in a perfectly rational market, aggregates the value of the firm out to infinity. If all that were happening was the shifting in the timing of purchases, there should be only a modest benefit to the firm from earlier sales (because of the time value of money), probably not enough to justify the price spike we see following mass shootings.

And the market eventually figures this out. Thus, if you look at the performance of firearms stocks, you see that the gains they make in the immediate aftermath of a mass shooting are usually given up by about a week thereafter. For example, at the beginning of this week, American Outdoor Brands (the parent of, among other brands, Smith & Wesson) was trading for slightly less than it was before the Las Vegas tragedy, after having gone up then down. Storm Ruger stocks followed the same trajectory. Perhaps this means that the market figured out that this time there will be serious gun control, but that seems extremely unlikely, given the talk about, at most, banning bump stocks. Rather, it looks like Wall Street investors who bid up share prices of firearms stocks in the wake of mass shootings are not perfectly rational actors. If they were, they would anticipate the eventual fall of the stocks and so not buy. Indeed, if even a substantial minority of investors appreciated the pattern, they would also appreciate the arbitrage opportunity and bet against the firearms stocks for the medium-term aftermath of a mass shooting, which should eliminate the effect altogether. So these predictable gains-then-losses following mass shootings provide further evidence that the efficient capital markets hypothesis is not perfectly accurate. It appears that collectively, investors buying firearms stocks in the wake of mass shootings are victims of the availability heuristic: they overweight recent salient events in making their decisions.

The availability heuristic is one of the many contributions to our understanding of how people think from the work of Daniel Kahneman and the late Amos Tversky. For his work, in 2002 Kahneman received the Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel, commonly but inaccurately called the Nobel Prize in Economics. (It is not awarded posthumously, so Tversky was ineligible.) This week it was announced that this year's recipient of the Bank of Sweden Prize is Richard Thaler, whose work in behavioral economics builds on and expands the insights of Kahneman and Tversky.

The over-simplified version of behavioral economics goes like this: Conventional economics errs by assuming that people behave rationally; yet often they do not; various heuristic biases like the availability heuristic, the endowment effect, and other mental shortcuts lead to what conventional economists would call irrational behavior.

The slightly less over-simplified version says that behavioral irrationality has an impact on how markets function and therefore on the (in)accuracy of the predictions of conventional economics. After all, conventional economists already knew that people are not always rational. The claim of conventional economics was always that economists could generate useful models of the economy based on the assumption that people acted rationally, even though they don't always do so. So for behavioral economics to matter, its champions had to do more than show that people are irrational; they had to show how that irrationality undercuts the conventional view.

Is that what behavioral economists did? Not at first blush. Many of the policy prescriptions of behavioral economics--and especially of Thaler--aim to help individuals make decisions that are better for themselves. For example, Thaler's book with Cass Sunstein, Nudge, advocates changing various default rules so that people end up with retirement portfolios and other baskets of goods and services that better serve their interests. To use their language, they aim to bring people's choices into line with what their reflective mental system would choose, rather than what their intuitive system does automatically. In response to the charge that their program is paternalistic, Thaler and Sunstein can and did say that there was nothing inevitable about the prior default. It was not even really your preference, because it was an artifact of arbitrary (or worse than arbitrary) framing.

I said above that the sorts of policy prescriptions offered up by Sunstein and Thaler do not directly challenge economic orthodoxy. That's because standard economics treats preferences as exogenous. I can explain what I mean by that with an example. We know that people will save more for retirement if the default rule is that employers withhold salary from employees for retirement savings (401(k) and 403(b) accounts) than if employees have to opt in to such accounts. The difference between an opt-in system and an opt-out system thus affects how much money people save for retirement, but it doesn't affect how the economy deals with what they do save. With an opt-out rule, demand for financial assets will be greater than with an opt-in rule, but whatever the demand is can be plugged into the equations. Other things being equal, opt-out systems will lead to higher financial asset prices, but, one might say, that's just a function of the demand.

If one steps back, however, one realizes that the behavioralist point undercuts the standard model. Neoclassical economics posits that government interventions in the economy typically cause distortions mediated through prices: regulations raise the cost of production and thus raise prices; taxes on goods and services raise the prices of those goods and services; income taxes leave consumers with less money with which to express demand and thus lower prices, which results in less supply; etc. All of that is true in a sense, except that neoclassical economics assumes that prices are being raised or lowered from a baseline that expresses a natural equilibrium. What the work of Thaler and other behavioralists shows is that demand is often, perhaps almost always, a function of how choices are framed. The equilibrium price achieved after a series of government nudges is not a distortion of some socially optimal natural level that would exist without government intervention; at least when designed to correct for the misfirings of our cognitive biases, the nudges result in an equilibrium price and distribution that is better along all relevant dimensions and no less natural.

Note that, as I've characterized it, Nudge intersects with some of Sunstein's solo work on baselines. In that work, he builds on legal realism to challenge the claim that common law rules of contract and property are neutral and that statutes and regulations altering the common law are somehow an intervention into the natural order. Somewhat more controversially, Sunstein has made the same point  about free speech: Laws regulating campaign finance, commercial speech, (certain forms of) pornography, and (certain forms of) hate speech do not intervene in the natural order; they merely shift the allocation of rights from one baseline in which the government is pervasively involved to another such baseline.

Putting aside the merits of Sunstein's work on baselines, note that the forms of regulation it approves (or at least does not disturb) are broader than those that follow from behavioral economics. One can invoke the point that the common-law baseline is not neutral or natural as a ground for overcoming objections to programs that aim at redistribution, the protection of privacy, or virtually anything, not just as a ground for overcoming objections to programs that aim to compensate for cognitive biases.

Once one recognizes that point, it should become clear that behavioral economics does not pose a fundamental challenge to neoclassical economics. It challenges the assumption that people always act rationally, which is important, but it accepts the hidden normative premise of neoclassical economics that people ought to act rationally. Nudges and other corrections that aim to overcome cognitive biases operate from the premise that biases are harmful.

To be clear, I generally share that view. My point is simply that, as the term "nudge" suggests, a nudge is a relatively modest intervention.

The baseline point is much more far-reaching, however, because it encompasses a wide range of potential "interventions" in the economy that need not have anything to do with correcting cognitive biases. Last week, I chaired a panel for a conference in which one of the papers--by my colleague Bob Hockett and Tel Aviv University Professor Roy Kreitner--explored (among other things) how numerous government policies affect prices, sometimes as their goal (as with a statutory minimum wage and as when the Fed targets interest rates, thus affecting the price of credit) and sometimes as a byproduct of some other policy. Hockett and Kreitner end up concluding that there is, or at least should be, nothing sacrosanct about prices as a supposed spontaneous example of market ordering.

There is not yet a publicly available draft of the Hockett and Kreitner paper, but with their permission, I'm going to quote what I regard as a critical passage, which builds on some of Hockett's prior solo work. They say:
The price system as we know it is commonly distorted by recursive collective action problems. These are situations in which individually rational behavior leads to collectively irrational outcomes (the collective action problem simpliciter), with iterative, self-exacerbating structures. The bank run is a prime example, but financial markets in general are especially susceptible to such dynamics. Asset price bubbles and busts and debt-deflation spirals are best analyzed in this frame as well.
The key move for present purposes is the one in the second sentence: even "individually rational behavior" can and often does lead to socially harmful consequences. That's not news, of course. As they recognize, what Hockett and Kreitner call "recursive collective action problems" are merely a subset of collective action problems, and the standard and well-justified response to collective action problems is regulation.

Behavioral economics generates important insights and a sound basis for various government programs that aim to compensate for individual irrationality. But we should not lose sight of the fact that the nudges and other programs recommended by behavioralists are the tip of a much larger iceberg. Irrationality is a sufficient condition for regulation; it is not a necessary condition.


Shag from Brookline said...

Paul Krugman has a recent MYTimes column, followed up with a post at his blog, on behavioral economics. Interesting.

Query: Are these Wall Street investors acting rationally in their view that gun purchases will increase after mass killings because of the anticipated irrationality of people making such purchases? I am speculating/assuming that these investors are not making such purchases. Also, do gun prices rise after mass killings that might impact the higher stock prices? In my mind, irrationality ties into the 2nd A as a result of Heller (5-4).

Krugman accepts behavioral economics with some reservations. Thaler recently expressed concern that less attention has been paid to another writing of his.

Also, regarding the first sentence of Mike's closing paragraph, can we expect that Congress won't act irrationally?

Bob Hockett said...

Thanks so much, Mike. In case there is interest, the original Recursive Collective Action Problems article is here:

Thanks again! Bob

Hashim said...

Mike: I'm not sure your shooting-stock hypo demonstrates irrationality. Why isn't this a simpler, and more charitable, explanation: Rational investors buy gun stocks immediately after a mass shooting because they want to buy low in case there will be a surge in gun purchases from less-rational customers who fear (erroneously) that significant gun control legislation will be enacted; and thus the stock price drops soon after a mass shooting once it becomes clear even to the less-rational customers that significant gun control legislation will not be enacted and rational investors therefore know that there will be no surge in gun purchases.

Indeed, I don't even think the customers have to be "less rational" in a pejorative sense for the above explanation to work: like the investors, they too could realize that gun-control legislation is extremely unlikely to pass, but might be willing to stock up based on even a modest increased likelihood of the legislation's enactment, if they value guns highly enough; unlike the investors, they won't lose money on the back-end if gun-control legislation passes (indeed, they'll make money, because their guns will become more valuable, even if only on a black market). For example, even if all correctly agree that the likelihood of gun-control passing goes up from 5% to 10% after a mass shooting, it might be rational both for gun customers to stock up on guns (they don't want to take the 5% chance) and for investors to purchase the stock (they're willing to take the 10% risk).