In economics, choices, policies, and institutions are evaluated on their ability to serve individual preferences. If everyone in the office prefers whole milk, the Wellbeing Economist would recommend taking whole milk rather than skimmed milk to stock the office refrigerator. This approach is invested with an anti-paternalistic logic. However, orthodox welfare economics faces the challenge of behavioral economics. My milk choices depend a lot on the default coffee machine setting or whether people around me have just reminded me of the need to live healthier. Johanna Thomas asks, if we cannot derive a consistent preference from people’s choice behaviors, can we still live up to the anti-paternalistic ideal?
Traditionally, welfare economics has done something like this. We start from the idea that, on the basis of the observed choice behavior of individuals, we can attribute preferences to them. If the choice behaviors respect various consistency conditions, we can represent these choices by a preference relation that has all the characteristics normally assumed in economic theory: it is stable, independent of the context and respects the axioms of rational choice (Sugden, 2018 , p.7). For example, if I never choose skim milk to put in my coffee when whole milk is available, and always choose whole milk when only skim and whole milk are available, I can be assigned a stable preference and context-independent for whole milk. on skimmed milk. Having thus assigned stable, context-free, and consistent preferences to individuals on the basis of their choice behaviors, orthodox welfare economics proceeds to use these preference relations as the norm of well-being: the choices , policies and institutions are evaluated according to how well they serve these preferences. For example, if everyone in the office shares my milk preferences, the Wellbeing Economist would recommend whole milk instead of skim milk to fill the office refrigerator.
This approach is generally endowed with an anti-paternalistic logic: the preference relation is supposed to capture what an agent takes himself to serve his best interests. This seems like a plausible hypothesis when people’s choice behaviors are consistent, and we have no reason to believe that they have false beliefs about crucial aspects of the options available to them. In choosing policies that seek to serve people’s preferences, we therefore defer to their own views of what is in their interest, rather than imposing an external, objective standard of what is good for them. This is exactly what the anti-paternalist wants.
Orthodox welfare economics faces the challenge of behavioral economics, due to the well-documented and now widely known examples of choice behaviors that systematically challenge representation with stable, context-free, and consistent preference relations. Whether I choose skimmed or whole milk may, for example, depend a lot on the default setting of the coffee machine, or whether people around me have just reminded me of the need to live healthier. If we cannot derive a consistent preference relation from people’s choice behaviors, can we still live up to the anti-paternalistic ideal?
Many behavioral welfare economists are optimistic that we can still do this, at least in the way that matters most: we can still identify people’s subjective interests and overrule their choices only when it is shown that they don’t serve those interests well. In their optimism, many behavioral welfare economists have assumed that people’s true subjective interests must always be representable with a stable, context-free, and consistent preference relation. (Attempts to explicitly reconstruct latent preferences include Bleichrodt et al. (2001), Bershear et al. (2008), KwhereSzegi and Rabin (2007), Manzini and Mariotti (2012) and Salant and Rubinstein (2008). This reconstruction is possible is presumed by much of the wider literature. Famous, Thaler and Sunstein’s (2008) libertarian paternalism claims to intervene only to help agents achieve what is best for them, “in their own judgment”, p. 5.)
What Bob Sugden, in his 2018 book The Advantage Community, calls the “new consensus” is committed to the idea that we can assign true latent preferences to agents even if they choose inconsistently, and that agents who display external choice behaviors inconsistent with classical rational choice theory can still, in this sense, be said to have an “inner rational agent”.
Resume the case where I inconsistently choose different types of milk in my coffee on different occasions. According to the new consensus, even though I sometimes choose skim milk and sometimes whole milk, there is always one fact of the matter that I really prefer, and we have practical ways to verify it. But this example already serves to illustrate the basis on which the New Consensus has been criticized. (Beside Sugden 2018, see also Rizzo and Whitman 2020.)
It is often far from clear which of a set of inconsistent choices most authentically reflects an agent’s true interests. And the mere assumption that she must have, say, truly preferred the healthier of the two options amounts to a more problematic kind of paternalism than most economists want to engage in. In my recent article, I further point out a dilemma for those working with the idea of true latent preferences.
Either these represent the actual best judgments of people… (But then all cases of context-dependent deviation from true preference are cases of weakness of will, of acting against one’s own judgment. This is simply psychologically unrealistic in many cases of context-dependence.)
Or they represent hypothetical preferences, preferences I would have if I were ideally rational. (But it is unclear whether such hypothetical preferences are subjective in the way the anti-paternalist would like them to be, that is, whether they stand in the right kind of relationship to what I, incoherent me, actually want.)
Sugden himself takes the failure of the New Consensus to motivate a more radical overhaul of normative economics: We shouldn’t even try to rank options in terms of people’s subjective interests. On the contrary, we should only try to increase their possibilities of choice. But I don’t think you have to go that far. Consider these two intuitive ideas:
- Preferences are not a primitive mental state that forms the starting point for deliberation. Rather, preferences are the result of balancing various considerations that speak for or against a certain choice. For example, in my choice of milk, I might weigh taste and health considerations.
- Many of the basic desires, on the basis of which we form preferences, for example health or taste desires, are vague. And there may not be a single correct way to aggregate them. Therefore, there may not be a preference relationship that uniquely captures what is in your subjective interest – there may be a number of allowed ways to trade the things that matter to you .
The practical implication of these two ideas for welfare economics is that inconsistency in an agent’s choice behavior may not be the result of error. It may simply be the result of factors in our environment that cause us to trade the things that matter to us in slightly different, but equally acceptable, ways on different occasions. And so, unless we have specific reasons to believe that some of an agent’s inconsistent choices involve error (e.g., are based on false beliefs), we should assume that each of a set of inconsistent choices expresses a permitted manner for an officer to serve. his interests.
Where does this leave the welfare economist, whose goal is to arrive at a measure of subjective interest to use as a standard for evaluating policy? She may have to accept that there may be indeterminacy in her measure of well-being. But that doesn’t necessarily mean she has nothing to do with it: while we may have to consider it undetermined whether skimmed or whole milk is best for me, I can quite consistently choose one or the other. rather than cream. Our best attempt at anti-paternalistic welfare economics is to defer to consistent aspects of people’s choice behaviors when evaluating policies, but to accept indeterminacy where context dependence does not arise. not a clear and demonstrable error. My article argues that a choice-theoretic framework developed by Douglas Bernheim and Antonio Rangel (Bernheim and Rangel 2007 and 2009, under the recently revised interpretation offered by Bernheim 2016) can in fact be interpreted in this way. (Douglas Bernheim also defends his framework along similar lines against Sugden’s criticism.) Adopting this framework will mean that in practice, compared to the New Consensus, there will be fewer occasions where welfare economists will support going beyond people’s choices, even when these violate orthodox rational principles. choice theory. But that’s as it should be if we take the traditional anti-paternalistic commitments of economics seriously.