Nudging and Participation: a Contractualist Approach to Behavioural Policy

Accounts of informed or true preferences, as opposed to actual preferences, existed before the emergence of behavioural economics. See for example: John C. Harsanyi (1977). Similarly, concerns of endogenous or adaptive preferences have been raised elsewhere, see for example: A. Sen (1987, 45–47) or J. Elster (1983). However, as behavioural economics reveals the extent to which preferences depend on welfare irrelevant variables, informed or true preference accounts of welfare become particularly relevant for normative economics.

Note that beyond libertarian paternalism there are different attempts to develop a notion of behavioural welfare economics (Bernheim 2016, 2009; Dold 2017). As they seek to integrate behavioural findings into neoclassical models, however, they remain within the conventional welfarist framework and face similar objections to those I raise against libertarian paternalism here. For some of the most advanced models of behavioural welfare economics, see for example Bernheim and Rangel (2007, 2009); Salant and Rubinstein (2008); Rubinstein and Salant (2012); Bordalo et al. (2013); Bleichrodt et al. (2001); Pinto-Prades and Abellan-Perpiñan (2012); Chetty et al. (2009); Kőszegi and Rabin (2007, 2008).

Infante et al. (2016b) argue that if behavioural welfare economics were to treat the purification of preferences as standardisation on a descriptive and pragmatic basis, rather than as identification of erroneous deviations, their views would not be open to such weighty objections. Behavioural welfare economics would then provide a model to standardise individuals’ preferences to make them consistent with expected utility theory. Such an approach would need to proceed from a choice architect who makes the preferences compatible with rational choice theory according to the chooser’s own judgements about her welfare. However, this is not what behavioural welfare economics understands as preference purification, especially not the version underlying libertarian paternalism. Beyond that, see E. Angner (2015) for a thorough discussion of the (epistemological) status of neoclassical theory within behavioural economics by comparison with Max Weber’s notion of ideal types. Similarly, S. Heidl (2016) describes behavioural economics as de-idealisation of standard economic theory and says that it faces the same methodological limitations (cf. Lecouteux 2017).

Another objection is that determining what it means for an individual to have complete information, unlimited cognitive abilities, and no lack of self-control, is itself a difficult task. It is also inescapably normative, as it involves substantive assumptions about welfare (Fumagalli 2016; Sugden 2009). To put it simply, it may require critical normative hypotheses to infer what, say, complete information might imply for an individual and her preferences.

This leads to another question about the relation of a purified state of complete information, and welfare (Hausman 2016). Even if it were possible to determine what an individual would choose if they had complete information and no reasoning impairments, this choice is not necessarily identical to their welfare – at least there is no guarantee that this is true. As T. Cowen (1993) puts it, perfectly informed preferences might not always be relevant for actual, imperfectly informed choices in the real world. Knowing what an individual would want if endowed with perfect information doesn’t necessarily provide helpful information about what would increase an individual’s welfare now.

Additionally, one may ask what behavioural findings may imply for methods of surveys and questionnaires themselves given such mechanisms as framing, anchoring, or availability effects (cf. Sunstein 2017a). Moreover, see also Tannenbaum et al. (2017) who show that there might be a ‘partisan nudge bias’ in people’s evaluation of nudges (see also Fox and Tannenbaum (2015)).

As Sunstein (2017c) emphasises, there may be clear cases where one might reasonably assume latent preferences. Beyond Sunstein’s examples of apparent antecedent preferences and cases of self-control problems (cf. Sugden 2016), one might also imagine complex tariffs which leave individuals worse off as they do not choose the cheapest alternative (Infante et al. 2016b). However, such an identification of subjective (alternative tariffs) and objective preference rankings (cheapest prices) is not always possible. In fact in many cases there may not even exist such an obvious analogue as objective ranking. This highlights the proximity of a preference purification view to objective approaches of welfare, in sharp contrast to Sunstein and Thaler’s emphatic claim of a neutral and formal kind of mere means-paternalism.

Consider one of Sunstein and Thaler’s most well-known examples. A cafeteria director must choose how different food items are presented (the opening example in Nudge (Thaler and Sunstein 2008). She notices that some customers are inclined to choose those items that are more visibly displayed. Sunstein and Thaler believe she should try to arrange the items such that she highlights those that customers would choose themselves, since this would best satisfy the criterion of ‘better off as judged by themselves’. However, people do not always have stable and context-independent preferences, and the arrangement of items has a significant effect on people’s preferences. In other words, customers’ well-ordered, true preferences may not formally exist (Sunstein and Thaler 2003; Thaler and Sunstein 2008). Therefore, the strategy of choosing what the customer would choose on his own collapses as it is not possible to identify his preferences independently from the director’s arrangement (Sugden 2008). Sunstein and Thaler suggest the only remaining strategy is to make the customers best off, all things considered. While this is clearly paternalistic on their own terms, it preserves freedom of choice insofar customers may still make their own (unhealthy) choices if they wish. However, the normative criterion that guides the director is straightforwardly paternalistic. The director is supposed to arrange the items as to what she thinks would make the customers best off, all things considered. Sunstein and Thaler’s seem to appeal to some self-evident objective value of ‘healthy behaviour’ which suggests that the director’s assessment cannot be different from the customer’s true preferences. This, however, establishes a substantive normative assumption which would need an explicit and rigorous justification. Nevertheless, by referring to obvious examples that no one would reasonably object to, Sunstein and Thaler may seem to overcome the conceptual normative fallacy by implicitly invoking an objective account of welfare based on allegedly common values (cf. Whitman and Rizzo 2015; Sugden 2016). I do not claim that the paternalistic stance of libertarian paternalism is problematic in itself, but it should be explicitly defended in order to provide a valid normative framework.

Prior to McQuillin and Sugden’s (2012a) image of a reconciliation problem, N. Berg (2003) argued for a notion of normative behavioural economics. As this approach does not specifically address libertarian paternalism, it may be neglected here. However, there seems to be no contradiction between the reconciliation perspective which may rather be perceived as behavioural normative economics (Dold and Schubert 2016) and a notion of normative behavioural economics. Instead, it demonstrates the importance of the links between behavioural and normative economics on a theoretical, conceptual, and methodological level. D. R. Just (2017) recently introduced the ‘behavioural welfare paradox’, which addresses the same issue, namely the paradoxical consequences of behavioural economics for standard welfare analysis. Similarly, T. Grüne-Yanoff (2009) described the tension between welfare economics and behavioural economics’ findings within libertarian paternalism as a ‘soft paternalist’s paradox’.

Compare J. Heath’s distinction of micro- and macrocontractualism in which he defines contractarianism as a distinct kind of microcontractualism (Heath 2014, 145ff.). See also Hausman et al. (2017, 224ff.) who despite small differences distinguish between perspectives seeking mutual advantage and those which focus rather on impartiality or reciprocity.

The distinction of values as attached to persons and preferences as ascribed to selves draws on K. Arrow (1963) who distinguishes interests from values, but it relates particularly to J. Harsanyi (1955) who distinguishes an individual’s subjective preferences from her ethical preferences which are those had she an equal chance of being in anyone’s position (cf. A. K. Sen 1970, 66).

Note that while G. Dworkin (1972) rather vaguely refers to some certain kinds of temporary conditions and humans’ limited cognitive capacities due to which individuals tend to make irrational decisions, today behavioural findings are able to provide reliable evidence on how human choices are shaped and depending on normatively irrelevant factors. Interestingly, however, he points to cooling off-periods which have become an integral part of Sunstein and Thaler’s concept of nudge (cf. 2003) and of behavioural public policy in general (cf. Lynch and Zauberman 2006).

See also Lepenies and Małecka (2016; 2015) who focus on the institutional implications of behavioural policies and argue that nudges require different legal measures as institutional safeguards against their possible negative consequences. They explicitly take an institutional rather than individualistic perspective. I adopt their institutional perspective, but aim at the normative justification provided by the framework of libertarian paternalism. Thus, I take my contractualist approach as institutional proposition while being based on (the institutionalisation of) collective choice rules.

The assumption is that conventional welfare economics takes a preference-based utilitarian stance, and a contractualist framework may eventually overcome the utilitarian bias in economics by enhancing rather deontological aspects.

Sugden argues that based on a particular understanding of normative economics, a welfarist approach characterises politics as executive action, while his contractarian model describes politics rather as negotiation to achieve mutually beneficial outcomes (Sugden 2013, 529). A contractualist perspective, I might add, emphasises the deliberative character of the complex reality of politics to achieve fair agreements between individuals as autonomous moral equals.

Guala and Mittone (2015) similarly suggest a political, though contractarian, justification of nudges. However, their approach relies on negative externalities which biases of human decisions may cause and which they argue justify the employment of behavioural policies.

Schnellenbach (2011) points at two empirical case studies of individual self-binding. In the first, employees voluntarily chose a contract option which gave them less money if they failed to achieve a certain productivity goal which they agreed on in advance (Kaur et al. 2010). The second concerns retirement savings based on Benartzi and Thaler’s popular experiments and their SMarT plan (Thaler and Benartzi 2004; DellaVigna 2009). Another example is provided by sticK, a company founded by Yale economist Dean Karlan which enables its customers to engage in individual commitment contracts (The Economist 2008). Additionally, there is one empirical study on individuals’ motivations for engaging in behavioural policies. It finds that paternalism is not demanded by people who need it as a commitment device (Pedersen et al. 2014) which may provide an argument for paternalism rather as a device for collective self-commitment (Schubert 2017).