Behavioral economics and policy evaluation

Richard Thaler, a leading developer and proponent of behavioral economics, calls it “the combination of psychology and economics that investigates what happens in markets in which some of the agents display human limitations and complications” (Mullainathan & Thaler, 2001, p. 1094).

Revesz and Livermore (2008) reviewed such criticisms in a defense of BCA.

As noted below, BE findings suggest that such reversals may be more frequent and profound.

It is important to be clear that the consistency between the two does not make the positive analysis normative. One could point out that one outcome is more efficient than another, or even Pareto-preferred without claiming that the more efficient outcome be realized. But simply as an observation from teaching in a multidisciplinary department, it has long seemed to me that a methodological distinction of economics compared to other social sciences in analyzing policy is this consistency between the positive and normative side, at least as compared with political science and sociology, however valuable are the positive insights and normative cautions contributed by these other disciplines.

We do not normally recommend citing Wikipedia, but the entry “List of biases in judgment and decision making” contains over 100 different supposed biases in decision making and belief formation. http://en.wikipedia.org/wiki/List_of_cognitive_biases.

The term “loss aversion” here requires some care. It often is presented in shorthand as saying that an individual is more averse to a loss of $X than a gain of $X. However, that is the definition of risk aversion in standard economics. “Loss aversion” in behavioral economics requires an endowment effect. Using the terminology in the text, it says that if someone prefers A to B, her willingness to pay for the “gain” of A starting at B is less than her willingness to pay to avoid the loss of going back to B if she starts at A, or to put it in BE terms, views A as part of her endowment. One might view the goal of advertising as not to persuade a consumer that A is better than B, but to convince a consumer that A is part of her endowment, what she should regard herself as having. If so, she will be willing to pay more to keep it by buying it than she would to add it if she regarded her endowment as including only B. The added and more controversial claim with loss aversion on top of the endowment effect is that people are risk preferring when it comes to avoiding losses, even if risk averse when it comes to evaluating gains.

Of course, conventional BCA requires a claim that there are no biases distorting the benefit and cost information contained in empirically observed demand and cost curves. To be clear, the claim here is not that BE is incorrect, but that if it is correct, the advantages of BCA fail to hold, requiring investigation of replacement methods for policy evaluation.

Executive Office of the President, Executive Order 13563, “Improving Regulation and Regulatory Review,” issued January 18, 2011.

Imagine two new restaurants across the street from one another, and a succession of diners walking down the street. The first may have information that one restaurant is better, or may have none, and make a random choice. The second diner may have information that one is better, but if she sees the first diner in the other restaurant, will assign some likelihood that the first diner knew something as well. Or, if the second diner has no information, she may just decide to follow the first diner, in case that diner was informed. If a third diner sees two in the same restaurant, he may decide that the chance that his information (if any) outweighs theirs is so small as to make it virtually automatic to follow the first two. Any subsequent diner seeing three in one will follow them, etc. Hence, early information, even if incorrect, can lead to an outcome where everyone eats in one restaurant, even if their aggregate information indicates the other is better.

Farrell and Saloner (1985) similar argued that inefficient technological standards can end up being adopted because early choices based on relatively minor considerations make changes down the road difficult. Liebowitz and Margolis (1995) critique the empirical relevance of this theoretical possibility and the inefficiency of “path dependence” more broadly.

Sugden (2008, pp. 241–242) characterizes different, inconsistent choices with different ancillary conditions as a “multiple selves” problem. Jon Elster (1979, 1985) has invoked “multiple selves” arguments to justify self-paternalism – using policies to precommit to restrict future choices to prevent actions the present self views as harmful – and to explain weakness of will.

This is obviously a topic far beyond the scope of this paper. We have found Rawls (1971) a useful starting point.

However, it will suffice a lot of the time, since many if not most of the projects for which a BCA is called for do probably do not confront fundamental rights or have society-wide distributive consequences.

Sugden (2005, p. 154) applies this to the familiar difference between how much people are willing to pay (WTP) or an environmental benefit and how much they would have to be paid to accept its disappearance (“willingness to accept” or WTA). If, as is normally the case, a policy would involve people paying for the environmental benefit, either as taxpayers or consumers from suppliers subject to regulation to provide that benefit, then one should base benefits on WTP rather than WTA.

This is not the same as the related empirical finding that in market settings, revealed preferences are less susceptible to the errors identified in behavioral economics (List, 2006). The similarity is that the explanation for findings such these is that people learn through repeated interactions (see also Sunstein’s pre-Nudge standpoint discussed above). This suggestion broadens how learning may occur, from repeated instruction to a single explanation.

Let us invoke a personal example. I have a number of guitars, a few of which would command a high price in the “vintage” guitar market. I could sell some of these for more than I would be willing to pay for them. This appears irrational, because in foregoing the opportunity to sell them, I am in effect paying a price I claim I would not pay. However, as I know this, it is difficult to contend that this is irrational. Rather, there must be something else going on, e.g., uncertainty about trusting unknown buyers or the real option value of keeping the guitars because I might have a hard time finding a model of that vintage if I want one later on.

The need for judgment may arise in part because advocates of the behavioral economic perspective could argue that the person failed to understand the information and explanations.