Preferences versus opportunities: on the conceptual foundations of normative welfare economics

Abstract

Normative welfare economics commonly assumes that individuals’ preferences can be reliably inferred from their choices and relies on preference satisfaction as the normative standard for welfare. In recent years, several authors have criticized welfare economists’ reliance on preference satisfaction as the normative standard for welfare and have advocated grounding normative welfare economics on opportunities rather than preferences. In this paper, I argue that although preference-based approaches to normative welfare economics face significant conceptual and practical challenges, opportunity-based approaches fail to provide a more reliable and informative foundation for normative welfare economics than preference-based approaches. I then identify and rebut various influential calls to ground normative welfare economics on opportunities rather than preferences to support my qualified defence of preference-based approaches.

1. Introduction

Normative welfare economics (henceforth, NWE) commonly assumes that individuals’ preferences can be reliably inferred from their choices and relies on preference satisfaction as the normative standard for welfare (e.g. Bernheim 2009; Hausman 2012; Adler and Fleurbaey 2016). The idea is that individuals are well-off to the extent that their own preferences are satisfied and that policies enhance individuals’ welfare to the extent that they satisfy such preferences (e.g. Salant and Rubinstein 2008; Rubinstein and Salant 2012; Manzini and Mariotti 2014). In recent years, several authors have criticized welfare economists’ reliance on preference satisfaction as the normative standard for welfare and have advocated grounding NWE on opportunities rather than preferences (e.g. McQuillin and Sugden 2012; Infante et al. 2016 a, 2016 b). The idea is that welfare economists should rely on opportunities rather than preference satisfaction as the normative standard for welfare and that individuals’ opportunities have value irrespective of whether individuals’ choices reveal consistent and stable preferences (e.g. Sugden 2004, 2007, 2018: Chs 4–5; also Roemer 1998; Van Hees 2010; Bavetta et al. 2017). ¹

The ongoing debate concerning the merits of preference-based approaches to NWE (henceforth, PREF) and opportunity-based approaches to NWE (henceforth, OPP) has substantial policy implications. For PREF and OPP respectively draw on different primitive concepts and frequently ground dissimilar evaluations of policies’ welfare implications (e.g. Adler 2013, Fumagalli 2016). And even in those cases where PREF and OPP ground similar welfare evaluations (e.g. Piacquadio 2017), PREF and OPP ground such evaluations on dissimilar justificatory principles (e.g. Hausman 2016; Sugden 2018: Chs 4–5). In this paper, I join the debate and argue that although PREF face significant conceptual and practical challenges, OPP fail to provide a more reliable and informative foundation for NWE than PREF. I then rebut various influential calls to ground NWE on opportunities rather than preferences to support my qualified defence of PREF. The paper is structured as follows. Section 2 outlines the main tenets of PREF and explicates some major challenges faced by PREF. Section 3 outlines the main tenets of OPP and explicates some major challenges faced by OPP. Section 4 identifies and rebuts three influential calls to ground NWE on opportunities rather than preferences to support my qualified defence of PREF.

Before proceeding, the following three preliminary remarks are worth making. First, my qualified defence of PREF does not rest on specific assumptions as to whether preference satisfaction is most plausibly taken to constitute welfare or merely provide evidence for welfare. Below I use the term ‘PREF’ broadly to encompass both approaches that take preference satisfaction to constitute welfare (e.g. Adler 2019, on preference satisfaction theories of welfare) and approaches that take preference satisfaction to merely provide evidence for welfare (e.g. Hausman 2012: Chs 7–8, on the evidential account of welfare). These approaches differ in significant respects (e.g. Hausman 2011, for a critical appraisal of preference satisfaction theories; Sarch 2015, for a critical appraisal of the evidential account), but share the aim to vindicate economists’ ‘practice of making inferences concerning well-being from people’s preferences’ (Hausman and McPherson 2009: 16).

Second, OPP are not the only approach alternative to PREF that has been developed in the NWE literature (e.g. Haybron and Tiberius 2015: 714, on value-based approaches, which advocate grounding policies’ welfare evaluations ‘on people’s values … as opposed to their preferences’; Kahneman and Sugden 2005, on experienced utility approaches, which advocate grounding policies’ welfare evaluations on measures of experienced utility rather than measures of preference satisfaction). Below I target OPP (rather than other approaches) since to date OPP are significantly more developed and widely supported than other approaches (e.g. Hersch 2020, for a critical appraisal of value-based approaches; Fumagalli 2019, for a critical appraisal of experienced utility approaches). In doing so, I focus primarily on the influential works by Sugden (e.g. 2004, 2006, 2008, 2010, 2015, 2018) rather than other proponents of OPP because these works provide one of the most systematic contemporary defences of OPP and offer a commendably clear formulation of OPP’s empirical and theoretical presuppositions. I shall expand in section 3 on such presuppositions. For now, I note that OPP sharply differ from PREF not merely because they articulate an alternative view of the currency of advantage (replacing preference satisfaction with opportunity), but also because they rest on a specific account of individuals as responsible agents who ‘at each moment, [identify with their] own actions, past, present, and future’ (Sugden 2018: 106) and presuppose a contractarian approach which assesses policies’ welfare implications ‘from the several viewpoints of individual members of society, considered as potential parties to an agreement’ (Sugden 2018: 14; sections 3–4). In these latter respects, OPP are quite distinct from both PREF and various influential opportunity-based views of the currency of advantage (e.g. Sen 1985, 1991, 1999, 2002: Ch. 20).

And third, the relative merits of PREF and OPP may importantly depend on the analytical and practical purposes for which PREF and OPP are applied. My critical evaluation of OPP by no means excludes that the notion of opportunity has significant analytical and practical applications in normative economic analysis (e.g. Vallentyne 2005; Bavetta et al. 2014, on opportunity-based measures of freedom; Fleurbaey and Peragine 2013; Roemer and Trannoy 2016, on opportunity-based measures of equality). In fact, my evaluation grants that welfare economists may combine specific tenets of PREF and OPP (e.g. Sen 1991, 2002: Ch. 20, for accounts of freedom and opportunity freedom that endorse a preference-based view of opportunity; also Suzumura and Xu 2009, for an axiomatic framework whose primitive concepts are preference orderings over outcomes and opportunity sets from which those outcomes are chosen). Still, as I argue in sections 2–4, one can offer a detailed assessment of PREF’s and OPP’s respective potential to provide a reliable and informative foundation for NWE. ²

2. Preference-Based Approaches

According to PREF, individuals’ preferences for some state of affairs count as satisfied if such state of affairs obtains (e.g. Hausman 2010: 326; also Sobel 1994; Sumner 1996). Knowing that one’s preferences are satisfied often gives one feelings of pleasure or satisfaction. Still, on PREF, preference satisfaction does not have to involve any feelings of pleasure or satisfaction and does not enhance welfare in virtue of such feelings (e.g. Hausman and McPherson 2009; Angner 2016). Two sets of PREF are especially prominent in the NWE literature. On the one hand, actual PREF take individuals to be well-off to the extent that their actual preferences – i.e. the preferences individuals happen to have – are satisfied (e.g. Binmore 1994; Gul and Pesendorfer 2008). On the other hand, informed/ideal PREF take individuals to be well-off to the extent that their informed/ideal preferences – i.e. the preferences individuals would have if they were not affected by limitations of information, cognitive ability, or self-control – are satisfied (e.g. Griffin 1986: part I; Harsanyi 1997). ³

According to the critics of PREF, neither actual PREF nor informed/ideal PREF can provide a reliable and informative foundation for NWE. The critics of PREF typically proceed as follows. Individuals’ preferences can be reliably inferred from individuals’ choices only if these preferences satisfy stringent consistency (e.g. transitivity) and stability (e.g. context-independence) requirements. However, the empirical findings collected over the last few decades demonstrate that individuals’ actual preferences frequently fail to satisfy these requirements (e.g. Sugden 1991; Padoa-Schioppa 2011, on violations of transitivity; Rabin 2002; Kahneman 2003, on violations of context-independence). Moreover, the satisfaction of individuals’ actual preferences often tracks factors that seem prudentially irrelevant (e.g. Tversky and Kahneman 1981; Camerer and Loewenstein 2004, on cases where individuals’ actual preferences are influenced by frames) or even detract from what many theories of welfare regard as individuals’ welfare (e.g. Hausman 2011; Bar-Gill 2012, on cases where individuals’ actual preferences rest on inaccurate information about the available choice options; also Harsanyi 1995; Vromen 2022, on antisocial preferences). For these reasons, the critics of PREF go, the satisfaction of individuals’ actual preferences cannot be plausibly regarded as the normative standard for welfare and actual PREF cannot provide a reliable and informative foundation for NWE.

For their part, informed/ideal PREF grant that individuals frequently have inconsistent and/or unstable actual preferences, but hold that individuals typically have consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare and that welfare economists can reconstruct such preferences. The idea is that despite the inconsistency and the instability of individuals’ actual preferences, welfare economists can take the satisfaction of individuals’ informed/ideal preferences as the normative standard for welfare. In this context, some contributions rely on model-less approaches, which attempt to reconstruct consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare without drawing on any specific models of the neuro-psychological processes that generate choices (e.g. Salant and Rubinstein 2008; Bernheim and Rangel 2009). Other contributions, instead, rely on model-based approaches, which attempt to reconstruct consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare by drawing on specific models of the neuro-psychological processes that generate choices (e.g. Rubinstein and Salant 2012; Manzini and Mariotti 2014). ⁴

Regrettably, the critics of PREF go, it is hard to establish what notion of informed/ideal preferences should be adopted by welfare economists unless one makes substantive normative assumptions about individuals’ welfare (e.g. McQuillin and Sugden 2012: 560, claiming that the concepts of ‘complete information’ and ‘unlimited cognition’ figuring in various informed/ideal PREF are ‘inescapably normative’). Moreover, different criteria have been proposed to reconstruct individuals’ informed/ideal preferences, and different criteria single out different subsets of informed/ideal preferences (e.g. Salant and Rubinstein 2008; Manzini and Mariotti 2012, on various divergences between the sets of informed/ideal preferences that are respectively singled out by model-based and model-less approaches). In fact, reconstructing some individuals’ informed/ideal preferences does not per se enable welfare economists to determine what is prudentially valuable for these individuals (e.g. Sobel 2009; Sugden 2018: Chs 4–5, on putative cases where individuals reveal context-dependent informed/ideal preferences). For these reasons, the critics of PREF go, the satisfaction of individuals’ informed/ideal preferences cannot be plausibly regarded as the normative standard for welfare and informed/ideal PREF cannot provide a reliable and informative foundation for NWE.

3. Opportunity-Based Approaches

OPP hold that NWE should be ‘based on considered judgements about opportunity rather than on considered preferences’ (Sugden 2006: 211, italics added; also McQuillin and Sugden 2012: 563, claiming that according to OPP, welfare economists should ‘use a normative criterion of opportunity rather than preference-satisfaction’). The idea is that welfare economists’ ‘normative theories [should be] based on opportunity rather than preference’ (Sugden 2020: 66), where an individual’s opportunities are ‘the set of options from which he can choose, options being described in such a way that they are mutually exclusive and jointly exhaustive’ (Sugden 2018: 84). According to OPP, ‘a person is well-off by being able to choose whatever she might desire and be willing to pay for’ (Sugden 2010: 55, italics added; also Sugden 2018: Ch. 5). In particular, ‘it is good for an individual to have a wide range of alternative options from which to choose’, and for each individual ‘opportunities to choose between alternative options have value, irrespective of whether or not those choices can be rationalized in terms of [consistent and] stable preferences’ (Sugden 2004: 1014 and 1016, italics added; also Van Hees 2010: 687, holding that the opportunity ‘to make a choice from among a set of options … has a value that is at least partly independent of the actual choice we make or intend to make from that set’).

OPP sharply contrast with PREF. For according to OPP, ‘what ultimately matters [is] freedom of choice’ rather than preference satisfaction, and ‘rather than asking how far economic outcomes satisfy individuals’ preferences, we have to ask how far … those outcomes endow individuals with freedom of choice’ (McQuillin and Sugden 2012: 556, italics added; also Sugden 2015: 299). To be sure, OPP do not deny that individuals may value satisfying their own preferences. In fact, OPP explicitly hold that it is ‘in each individual’s interest to have opportunities to satisfy … any preferences she might come to have’ (Sugden 2018: 99; also Sugden 2006: 209). Still, according to OPP, opportunities rather than preference satisfaction are the normative standard for welfare. Therefore, rather than trying to reconstruct and satisfy individuals’ preferences, welfare economists should aim to increase the range of opportunities available to individuals (e.g. McQuillin and Sugden 2012; Sugden 2018: Chs 5–6). ⁵

OPP have become increasingly prominent in the NWE literature (e.g. McQuillin and Sugden 2012; Sugden 2015, 2018: Chs 4–5; Infante et al. 2016 a, 2016 b). Despite OPP’s increasing prominence, the hitherto proffered calls to ground NWE on opportunities (rather than preferences) face several major challenges. Below I focus on three major challenges in turn, which respectively concern (3.1) the measurement of opportunities, (3.2) the value of opportunities, and (3.3) the notion of agency presupposed by OPP.

3.1. Measurement of opportunities

According to OPP, ‘each individual has the opportunity to do whatever he might desire to do with those things to which he has entitlements’ (Sugden 2010: 56). In particular, a policy P provides more opportunity than another policy P’ when P ‘contains all the options that [P’] does, and something in addition that the individual might conceivably want to choose’ (Sugden 2018: 85). However, one cannot specify an individual’s set of opportunities independently of how other individuals choose to use their own opportunities. For although each individual is ‘free to choose from his own opportunity set’, the contents of this set are ‘determined by the choices that other individuals make from theirs’ (Sugden 2018: 192). For this reason, ‘it is crucial [that] an individual’s opportunities are defined independently of her actual preferences’ (Sugden 2018: 109, italics added). Moreover, ‘often, we cannot specify in concrete terms what a person [has] the opportunity to do’, and opportunity remains ‘an open-ended concept’ (Sugden 2010: 48, italics added).

There are at least two reasons to doubt that OPP’s characterization of the notion of opportunity provides the basis to ground reliable and informative measurements of individuals’ opportunities. First, OPP’s characterization of the notion of opportunity does not enable welfare economists to compare pairs of opportunity sets ‘unless one [set] is a subset of the other’ (Sugden 2003: 804; also Hausman 2022). In fact, OPP ‘make no attempt to compare opportunity sets that are not nested’ (Sugden 2018: 85, italics added). However, many opportunity sets are not nested, and profound divergences remain as to how non-nested sets should be ranked in terms of how many opportunities they provide (e.g. Harless and Pancs 2022). As a result, the proponents of OPP often lack the resources to reliably assess how many opportunities are provided by competing policies. This applies to not only individual opportunities, which are ‘defined in terms of the extent of each individual’s opportunity set’, but also interactive opportunities, which are ‘defined for each set of individuals in terms of what those individuals are able to achieve … by trading with one another’ (Sugden 2021: 425).

And second, it remains unclear what the basic metric for measuring individuals’ opportunities is supposed to be (e.g. Bernheim 2021, on various divergences between measurements of individuals’ actual opportunities and measurements of individuals’ perceived opportunities). In particular, the proffered attempts to compare opportunity sets in terms of the number of options they offer and the related space of characteristics (e.g. Pattanaik and Xu 1990; Carter 1999) fail to take into adequate account the diversity and the complementarity of the options figuring in opportunity sets (e.g. Weitzman 1992; Van Hees 2004). Yet, providing reliable and informative measurements of opportunity in concrete policy contexts requires one to specify ‘whether two putative options should be treated as distinct [and] how significant the difference is between one option and another’ (Sugden 2003: 803). And the proffered methods to measure the diversity and the complementarity of the options figuring in opportunity sets (e.g. Bervoets and Gravel 2007, for measurements targeting the most dissimilar pairs of options; Nehring and Puppe 2002, for multi-attribute measurements; Pattanaik and Xu 2000, for measurements targeting similarity-based partitions) have implausible implications (e.g. Van Hees 2004, on various methods’ tendency to let either diversity or cardinality considerations dominate; also Dowding and Oprea 2022, for a recent review). In fact, it is rather hard to provide reliable measurements of options’ diversity and complementarity without relying on information concerning the preferences of the involved individuals (e.g. Sen 1991; Van Hees 2010). This, in turn, significantly constrains OPP’s potential to provide reliable and informative measurements of opportunity in concrete policy contexts. For OPP avowedly aim to ‘defin[e] people’s opportunities independently of their preferences’ (Sugden 2018: ix). And according to OPP, one ‘should not measure [opportunity] relative to preference [since] the whole point of opportunity is that … you can satisfy whatever preferences you might happen to have’ (Sugden 2020: 66).

A proponent of OPP may object that grounding reliable and informative evaluations of policies’ welfare implications ‘does not require the explicit specification of opportunity sets’ and that welfare economists and policymakers should ‘accept [opportunity’s] non-measurability’ rather than ‘try to measure opportunity directly’ (Sugden 2010: 66, italics added; also Sugden 2003: 785). However, both welfare economists and policymakers frequently need to measure how many opportunities competing policies provide to individuals. In this respect, the open-endedness of OPP’s characterization of the notion of opportunity makes it extremely hard for welfare economists and policymakers to determine what counts as more and less opportunity in concrete policy contexts. To illustrate this, consider the notion of collective (rather than individual) opportunity, which figures centrally in the proffered evaluations of policies’ welfare implications (e.g. Stewart 2005). According to OPP, ‘each group of individuals has the opportunity to do whatever its members unanimously desire to do with those things to which, taken together, they have entitlements’ (Sugden 2010: 56, italics added). This unanimity condition ‘introduces another element of open-endedness into the definition of opportunity’ (Sugden 2010: 56). And this additional element of open-endedness, in turn, makes it prohibitively difficult for welfare economists and policymakers to reliably measure how many opportunities competing policies provide to distinct groups of individuals (e.g. think of policies having statistical targets or targeting vast population segments). ⁶

3.2. Value of opportunities

OPP are ‘premised on the idea that it is good that people are free to do what they want, when they want, even if their preferences are unstable’ (Sugden 2007: 678–679, italics added; also Sugden 2018: xi, proposing ‘the principle that, as viewed by each citizen separately, more opportunity for that person is better than less’). According to OPP, ‘any expansion of a person’s opportunity set promotes her interests, irrespective of her actual preferences and independently of [what] she has reason to prefer’ (Sugden 2018: 99, italics added; also Sugden 2006: 217). In particular, OPP are based on ‘the normative intuition that for each individual, opportunities to choose between alternative options have value, irrespective of whether or not those choices can be rationalized in terms of [consistent and] stable preferences’ (Sugden 2004: 1014, italics added; also Van Hees 2010: 687). The idea is that an individual ‘may act on different preferences at different times [but] it is good for her that, at each moment, she is free to satisfy whatever preferences she then has’ (Sugden 2004: 1016, italics added; also Buchanan 1986: part I).

There are at least two reasons to doubt that OPP’s claims about the putative intrinsic value of opportunities provide the basis to ground reliable and informative assessments of the value of opportunities. First, descriptively, individuals do not generally ascribe intrinsic value to the opportunity to act on their own preferences (e.g. Bernheim and Rangel 2004; Dold and Schubert 2018, on several cases where individuals who know that their own preferences are inconsistent and/or unstable do not ascribe intrinsic value to the opportunity to act on such preferences). This does not per se undermine OPP’s claim that opportunities have intrinsic value. Still, it challenges the proponents of OPP to provide more convincing reasons and/or evidence to explicate why many individuals fail to ascribe intrinsic value to opportunities and why exactly such individuals are mistaken. And second, the proponents of OPP have hitherto failed to substantiate the normative claim that opportunities generally have intrinsic value (e.g. Arneson 1999; Fleurbaey 2012 a, on several cases where giving individuals opportunities that do not satisfy any of their preferences fails to enhance what many theories of welfare regard as individuals’ welfare). In fact, it is hard to see how the proponents of OPP may substantiate such normative claim unless they relinquish OPP’s declared aim to ‘maintain neutrality between different conceptions of the good’ (Sugden 2010: 48; also Sugden 2006: 218). ⁷

A proponent of OPP may object that welfare economists can assess the value of individuals’ opportunities by examining individuals’ potential preferences – i.e. the preferences that individuals ‘might have … in the relevant circumstances’ – rather than individuals’ actual preferences (Sugden 1998 a: 323; also Sugden 2003: 792). However, appeals to the preferences that individuals ‘might have’ in ‘the relevant circumstances’ are unlikely to ground reliable and informative assessments of the value of individuals’ opportunities unless the proponents of OPP supplement these appeals with precise and plausible specifications of what ‘the relevant circumstances’ are and what preferences individuals ‘might have’ in such circumstances. In this respect, attempts to reconstruct individuals’ potential preferences seem vulnerable to the same difficulties that OPP take to affect informed/ideal PREF’s attempts to reconstruct informed/ideal preferences (e.g. Sugden 1998 a: 324, conceding that ‘there seems to be no way of avoiding appeal to contestable ideas of “normal”, “reasonable” or “natural” preferences’ when attempting to reconstruct individuals’ potential preferences). Moreover, even if reliable and informative assessments of the value of opportunities could be grounded on the notion of potential preferences, these assessments would seemingly assume that the value of opportunities is conditional on the value of preferences. And this assumption, in turn, would contradict OPP’s presupposition that opportunities have ‘normative significance in their own right … without explicit reference to individuals’ preferences’ (Sugden 2018: 115, italics added; also Sugden 2003: 797, conceding that the proponents of OPP ‘have to find a measure [of opportunity] that does not refer to preferences, actual or potential’).

A proponent of OPP may further object that although opportunities do not always have intrinsic value, OPP can provide a reliable and informative foundation for NWE because opportunities typically have instrumental value in virtue of their capacity to enhance what many theories of welfare regard as individuals’ welfare (e.g. Sugden 2018: 106, claiming that ‘the more opportunities [one] has, the more [one] is free to do [irrespective of whether she has] coherent preferences’; also Van Hees 2010: 702, holding that ‘we may value having the option to choose a non-valuable alternative because the alternative’s availability creates responsibility’). This objection correctly notes that expanding individuals’ opportunity sets often has a favourable impact on what many theories of welfare regard as individuals’ welfare (e.g. Vallentyne 2002; Bavetta et al. 2014, on various opportunities’ contribution to individuals’ freedom of choice). However, not all opportunities enhance what many theories of welfare regard as individuals’ welfare (e.g. Dold and Rizzo 2021, on opportunities to make choices that will predictably hinder one’s freedom of choice and one’s sense of agency; also Sen 2002: 605, holding that ‘the opportunity to choose is not always an unmitigated advantage’). In fact, expanding individuals’ opportunity sets may hamper (rather than enhance) what many theories of welfare regard as individuals’ welfare both in individual choice settings (e.g. Fleurbaey 2012 a; Hausman 2022, on several cases where expanding individuals’ opportunity sets leads to welfare losses due to greater cognitive/emotional decision costs and individuals’ vulnerability to harmful forms of addiction) and in strategic choice settings (e.g. Bicchieri 2017: Chs 3–4; Harless and Pancs 2022, on situations of strategic interaction where giving each player the option to defect turns the relevant cooperative games into prisoner’s dilemmas to the detriment of individuals’ payoffs). These concerns are exacerbated when one considers collective (rather than individual) opportunities. For showing that some collective opportunities are welfare-enhancing requires welfare economists ‘to convince each person separately that she benefits from being free to do as she likes’ (Sugden 2010: 61, italics added; also Sugden 2006: 209). This, in turn, is prohibitively hard to achieve in all policy settings where the examined policies only have statistical targets or target vast population segments rather than a limited number of well-identified individuals (e.g. Sunstein 2021: 406, holding that ‘it is hard to identify any social institution [or public policy] that it is in the interest of each individual to accept’).

3.3. Agency

OPP rest on a specific account of individuals as responsible agents. On this account, an individual is a responsible agent ‘to the extent that, at each moment, she identifies with her own actions, past, present, and future’ (Sugden 2018: 106, italics added; also Sugden 2010: 47–49). The idea is that a responsible agent ‘endorses the decisions made by all her selves, whether or not those decisions can be rationalised by a single system of coherent preferences’ (McQuillin and Sugden 2012: 563, italics added) and ‘thinks of the preferences on which [she] will act in the future as [her] preferences, even if [she] does not know what they will be and even if they will not be what [she] would currently like them to be’ (Sugden 2015: 299, italics added; also Sugden 2004: 1018). This account of individuals as responsible agents is not a dispensable tenet of OPP, but ‘provides philosophical underpinning’ to OPP (Sugden 2018: 106). For the claim that ‘any increase in [an] individual’s opportunity is good for her’ irrespective of whether she has consistent and stable preferences avowedly rests on the assumption that the individual is ‘a continuing locus of responsibility’ (Sugden 2018: 106) in the sense presupposed by OPP.

There are at least two reasons to doubt that OPP’s account of individuals as responsible agents provides the basis to ground a reliable and informative foundation for NWE. First, descriptively, individuals are not generally responsible in the sense indicated by OPP. In fact, many individuals do not seem responsible in such a sense. To illustrate this, consider individuals’ widespread and considered reliance on self-constraint, whereby individuals restrict their own short-term opportunities in order to expand their own long-term opportunities (e.g. Bernheim and Rangel 2004; Schubert and Cordes 2013). The proponents of OPP acknowledge that in some cases, self-constraint might be a sign of responsibility (e.g. think of instances of self-constraint intended to counter harmful forms of addiction). Still, they object that ‘these cases are exceptional’ (Sugden 2015: 302). However, such cases seem far from exceptional. For individuals frequently rely on self-constraint (e.g. Dowding and Oprea 2022; Harless and Pancs 2022). And OPP lack the resources to account for these decisions as the decisions of responsible agents (e.g. Sugden 2015: 300, holding that according to OPP, ‘present opportunities to restrict future opportunities [have] zero value’). And second, it remains unclear what exactly the normative import of individuals’ being responsible in the sense presupposed by OPP is. In particular, it is hard to see why welfare economists or policymakers should endorse OPP’s privileging of ‘the preferences of the acting self … rather than the self as the maker of plans or as the source of reflective judgements’ (Sugden 2004: 243). For in many cases, the choices of the planning self are more conducive to enhance what many theories of welfare regard as individuals’ welfare than the choices of the acting self (e.g. Rizzo and Whitman 2009 a; Schubert 2015). In this respect, what OPP depict as a responsible agent appears to resemble more ‘a faceless being that stumbles through life, driven around by whichever tastes she happens to acquire’ (Dold and Schubert 2018: 230) than an agent who qualifies as ‘responsible’ on any plausible common-sense definition of the term.

A proponent of OPP may object that pointing to common-sense definitions of responsible agents does not bear against OPP’s account of individuals as responsible agents since one of the major tenets of this account is precisely that welfare economists and policymakers ‘need a radically different conception of the continuing person’ (Sugden 2018: 105, italics added). However, this objection does not provide independent reasons to endorse OPP’s account of individuals as responsible agents since it presupposes (rather than shows) that such an account provides the basis to ground a reliable and informative foundation for NWE. In this respect, it would be of limited import to object that several preference-based models developed to represent inconsistent and/or unstable preferences ‘equat[e] agency with coherent preferences [and] have no way of recognizing the continuing identity of a person whose preferences are incoherent’ (Sugden 2010: 54; also Sugden 2018: 104–106, criticizing multiple selves models). For in primis, not all preference-based models face this limitation (e.g. Dietrich and List 2011; Dietrich 2018, for various models of preference change). And second, pointing to the limitations faced by some preference-based models falls short of indicating that the account of responsible agency presupposed by OPP provides the basis to ground a reliable and informative foundation for NWE. In particular, it remains hard to see why exactly welfare economists and policymakers should endorse OPP’s presupposition that responsible agency consists in one’s valuing ‘opportunities for herself as a continuing entity, whether or not her preferences remain constant’ rather than ‘in a coherent system of preferences or in a rational plan of life’ (Sugden 2010: 55).

4. A Qualified Defence of Preference-Based Approaches

In this section, I argue that although informed/ideal PREF face significant conceptual and practical challenges, OPP do not provide a more reliable and informative foundation for NWE than informed/ideal PREF and therefore the hitherto proffered calls to ground NWE on opportunities rather than preferences do not withstand scrutiny. Below I support this qualified defence of informed/ideal PREF against three influential objections that leading proponents of OPP put forward against informed/ideal PREF, namely: (4.1) the objection from unfeasible preference reconstruction; (4.2) the objection from paternalism; and (4.3) the objection from contractarianism. ⁸

4.1. Objection from unfeasible preference reconstruction

The objection from unfeasible preference reconstruction proceeds as follows. Informed/ideal PREF fit the available empirical findings documenting the inconsistency and the instability of individuals’ actual preferences better than actual PREF (e.g. Salant and Rubinstein 2008; Bernheim and Rangel 2009). However, ‘there is no known psychological foundation’ for informed/ideal PREF’s assumption that individuals have consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare (Infante et al. 2016 a: 22, italics added; also Infante et al. 2016 b: 34, holding that ‘there is no general reason’ to think that consistent and stable informed/ideal preferences ‘exist at all’). Moreover, even if individuals had consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare, the idea that welfare economists ‘can reconstruct [these] preferences is a mirage’ (Sugden 2018: 14, italics added; also Dold 2018). For welfare economists lack the information and the evidence required to reconstruct such preferences (e.g. Rizzo and Whitman 2019: Chs 6–7, on cases where different methods of elicitation elicit dissimilar informed/ideal preferences; also Sugden 2018: Chs 4–5, on putative cases where individuals reveal context-dependent informed/ideal preferences).

This objection correctly notes that welfare economists’ attempts to reconstruct informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare face significant conceptual and practical challenges. Still, there are at least two reasons to doubt that the objection hampers informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE. First, informed/ideal PREF do not rest on any psychological assumption as to whether individuals have inner rational agents with informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare. In particular, welfare economists may provide detailed specifications of the conditions under which individuals can be plausibly assumed to have informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare without having to posit any inner agent having such preferences (e.g. Hausman 2016; Beck 2022). To be sure, welfare economists often do posit that individuals have ‘some mode of latent reasoning that generates [consistent and stable] preferences’ (Sugden 2018: 62) in their attempts to reconstruct informed/ideal preferences. Yet, welfare economists may consistently posit that individuals have some mode of latent reasoning that generates consistent and stable preferences without having to make any assumption as to what mode of latent reasoning generates such preferences (e.g. Bernheim 2021). In this respect, it would be overly restrictive to require that welfare economists provide general cross-contextual specifications of what neuro-psychological processes generate consistent and stable preferences. For dissimilar sets of neuro-psychological processes contribute to generate consistent and stable (as opposed to inconsistent and/or unstable) preferences across contexts (e.g. Fumagalli 2013; Ross 2014, for empirically informed criticisms of proposed identifications of the putative neuro-psychological substrates of rational decision making). And informed/ideal PREF frequently enable welfare economists to ground reliable and informative evaluations of policies’ welfare implications without having to draw on specific assumptions concerning what neuro-psychological processes generate consistent and stable preferences (e.g. footnote no. 4 on the welfare evaluations yielded by various model-less approaches).

And second, welfare economists may succeed in reconstructing informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare in a wide range of cases where the targeted individuals fail to exhibit consistent and stable actual preferences (e.g. Chambers and Hayashi 2012, for reconstructions of informed/ideal preferences from non-rationalizable choice data; Nishimura 2018, for reconstructions of informed/ideal preferences from choice data that reveal a nontransitive preference relation). To be sure, welfare economists may find it unfeasible to reconstruct informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare in presence of widespread choice inconsistencies (e.g. Manzini and Mariotti 2014, on cases where model-less approaches fail to yield welfare rankings because the targeted individuals’ choices vary with changes in ancillary conditions; also Bernheim 2016, on cases where the reliability of the welfare evaluations grounded on model-based approaches is constrained by the limited reliability of the models posited by such approaches). Still, individuals’ choice inconsistencies are rarely so widespread that they prevent welfare economists from being able to reconstruct informed/ideal preferences (e.g. Ross 2011; Bernheim 2021, on how non-choice data may help welfare economists reconstruct individuals’ informed/ideal preferences in situations where individuals’ observed choices are inconsistent or unstable). That is to say, welfare economists’ attempts to reconstruct informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare typically depend on normative presuppositions (e.g. Fumagalli 2021, on normative presuppositions about the merits of different theories of welfare). However, this dependence does not reflect limitations inherent in informed/ideal PREF, but rather reflects the thickness of the notion of welfare, i.e. the fact that the notion of welfare involves both positive and normative dimensions and that in ascribing welfare to individuals, one implies that these individuals ‘have something worth having’ (Tiberius 2013: 217; also footnote no. 1). Moreover, such dependence would not be avoided by grounding NWE on opportunities rather than preferences (e.g. section 3 on the normative presuppositions required to measure and assess the value of individuals’ opportunities).

A proponent of OPP may object that multiple conceptions of preferences have been advocated in the NWE literature and that informed/ideal PREF cannot provide a reliable and informative foundation for NWE unless the proponents of informed/ideal PREF specify which conceptions of preferences they endorse (e.g. Lecouteux 2022, arguing that the merits of different justifications of preference satisfaction as a normative criterion depend on whether one endorses a mentalistic or behaviouristic conception of preferences). This objection correctly notes that the proponents of informed/ideal PREF have to specify which conceptions of preferences they endorse if they are to demonstrate that the satisfaction of the informed/ideal preferences they reconstruct in particular choice settings can be plausibly taken to reliably track or constitute welfare. For multiple conceptions of preference have been advocated in the NWE literature, which sharply disagree concerning the nature of preferences (e.g. Broome 1993; Hausman 2012: Chs 7–8; Moscati 2021, on the issue of whether preferences are more aptly characterized as judgments or feelings), the causal bases of preferences (e.g. Dietrich and List 2016 a; Ross 2011; Guala 2019, on the issue of whether preferences are more aptly characterized as mental states, dispositions or behavioural patterns) and the relationship between preferences and choices (e.g. Hands 2013; Vredenburgh 2020; Thoma 2021, on the relationship between preferences and actual or hypothetical choices). Even so, it would be misguided to subordinate the justifiability of informed/ideal PREF to the provision of general necessary and sufficient conditions for the notion of preference. For what preferences are plausibly taken to be depends on the theoretical and pragmatic presuppositions of the models and the policy applications where such preferences figure (e.g. Angner 2018). And different conceptions of preferences may be suitable for distinct modelling and policy purposes (e.g. Dietrich and List 2016 b). This by no means indicates that all conceptions of preferences are equally plausible or that any such conception can provide a reliable and informative foundation for NWE (e.g. Beck 2022). Still, neither the availability of multiple conceptions of preferences nor the lack of general necessary and sufficient conditions for the notion of preference per se hampers informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE.

4.2. Objection from paternalism

The objection from paternalism proceeds as follows. Suppose, for the sake of argument, that the proponents of informed/ideal PREF are able to reconstruct sets of consistent and stable informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare. On informed/ideal PREF, the inconsistency and the instability of individuals’ actual preferences indicate that ‘individuals [often] make inferior decisions in terms of their own welfare’ (Sunstein and Thaler 2003: 1162; also Thaler and Sunstein 2003; Bubb and Pildes 2014). Moreover, according to many proponents of informed/ideal PREF, individuals’ failures to make welfare-enhancing decisions license a wide range of paternalistic interventions, i.e. interventions that ‘violate (or interfere with) the autonomy of their target agents … are implemented without the explicit consent of these agents [and] are designed with the primary aim to enhance the well-being of those agents’ (Fumagalli 2016: 461; also New 1999: 65; Hausman and Welch 2010: 126–130). Regrettably, the objection goes, many of these paternalistic interventions are unjustified or otherwise objectionable (e.g. Sugden 2008; Rizzo and Whitman 2009 b). This, in turn, casts doubt on informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE. ⁹

This objection correctly notes that leading proponents of paternalistic interventions advocate informed/ideal PREF (e.g. Sunstein and Thaler 2003: 1162, assessing policies’ welfare implications by reference to what preferences individuals would have ‘if they had complete information, unlimited cognitive abilities, and no lack of self-control’; also Sunstein 2014: 154, advocating interventions designed to ‘move people in the directions that they would go if they were fully rational’). Still, there are at least two reasons to doubt that the objection hampers informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE. First, the proponents of the objection have hitherto failed to show that adopting informed/ideal PREF generally leads welfare economists and policymakers to advocate unjustified (or otherwise objectionable) paternalistic interventions. And second, it is dubious that adopting informed/ideal PREF generally leads welfare economists and policymakers to advocate unjustified (or otherwise objectionable) paternalistic interventions. In fact, the proponents of informed/ideal PREF have both welfare-related and non-welfare-related reasons to oppose these interventions. For those interventions often fail to enhance individuals’ welfare compared with a situation where such interventions are not implemented (e.g. Bernheim 2016; Hausman 2022, on cases where the individuals targeted by paternalistic interventions are better informed about their own welfare than third parties). Moreover, in those cases where paternalistic interventions enhance individuals’ welfare, these interventions frequently involve morally problematic violations of individuals’ autonomy or consent (e.g. Hausman and Welch 2010; Qizilbash 2012). And many proponents of informed/ideal PREF oppose those interventions because of their commitment to respect individuals’ autonomy and consent (e.g. Bernheim 2009: 291–293; Fumagalli 2016: 468–471).

A proponent of OPP may object that the commitment of many proponents of informed/ideal PREF to respect individuals’ autonomy and consent frequently conflicts with their aim to enhance individuals’ welfare (e.g. Cowen 1993; Rizzo and Dold 2020). The idea is that the proponents of informed/ideal PREF have to relinquish either their commitment to respect individuals’ autonomy and consent (e.g. Sunstein and Thaler 2003: 1167, claiming that ‘it would be fanatical [to treat autonomy] as a kind of trump, not to be overridden’) or their aim to enhance individuals’ welfare (e.g. Le Grand and New 2015: Chs 5–6). However, the proponents of informed/ideal PREF may consistently retain both their commitment to respect individuals’ autonomy and consent and their aim to enhance individuals’ welfare in a wide range of cases, including many cases where individuals’ choices fail to enhance individuals’ own welfare. For the proponents of informed/ideal PREF are not committed to the extreme welfarist position that welfare enhancement is the sole goal of public policy (e.g. footnote no. 2 on the relative importance of welfare enhancement and other policy goals). And respecting individuals’ autonomy and consent does not require the proponents of informed/ideal PREF to relinquish their aim to enhance individuals’ welfare in all cases where their attempts to enhance individuals’ welfare would involve any minor violation of autonomy or consent (e.g. short-term mandatory cooling-off periods that aim to benefit individuals by inducing them to critically reconsider their own decisions), but only requires the proponents of informed/ideal PREF to relinquish such aim in those cases where their attempts to enhance individuals’ welfare would involve morally problematic violations of autonomy or consent (e.g. manipulative subliminal advertising campaigns). That is to say, informed/ideal PREF are compatible with a plurality of positions about the relative importance of welfare enhancement and respect for autonomy and consent (e.g. Rubinstein and Arad 2018; Ferreira 2022). And pointing to cases where the commitment of many proponents of informed/ideal PREF to respect individuals’ autonomy and consent conflicts with their aim to enhance individuals’ welfare does not per se hamper informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE.

A proponent of OPP may further object that the difficulties facing attempts to reconstruct informed/ideal preferences whose satisfaction reliably tracks or constitutes welfare hamper the ability of the proponents of informed/ideal PREF to provide effective safeguards against unjustified (or otherwise objectionable) paternalistic interventions (e.g. Sugden 2018: Ch. 3). The idea is that in concrete policy contexts it is often unfeasible for policymakers to reconstruct individuals’ informed/ideal preferences and that, as a result, policymakers’ attempts to ‘make choosers better off, as judged by themselves’ (Thaler and Sunstein 2008: 5) frequently end up promoting what policymakers regard as individuals’ welfare. This objection correctly notes that in concrete policy contexts, policymakers frequently end up promoting what they regard as individuals’ welfare rather than what individuals themselves regard as their own welfare (e.g. Hausman and Welch 2010; Whitman and Rizzo 2015, for illustrations). However, as noted above, the proponents of informed/ideal PREF can provide effective safeguards against unjustified (or otherwise objectionable) paternalistic interventions (e.g. Bernheim 2009: 291–293; Fumagalli 2016: 468–471, opposing paternalistic interventions that involve morally problematic violations of individuals’ autonomy or consent; also Bernheim 2021, proposing to respect individuals’ consistent and stable choices while accepting normative ambiguity when individuals’ inconsistent or unstable choices do not result from demonstrable mistakes). Moreover, the proponents of OPP do not generally provide more effective safeguards against unjustified (or otherwise objectionable) paternalistic interventions than the proponents of informed/ideal PREF (e.g. think of cases where increasing the range of individuals’ opportunities involves trade-offs with other policy goals valued by those individuals). In fact, OPP’s presupposition that ‘it is good that people are free to do what they want, when they want, even if their preferences are unstable’ (Sugden 2007: 678–679) may hamper (rather than enhance) the ability of many proponents of OPP to resist unjustified (or otherwise objectionable) paternalistic interventions that exploit the inconsistency and the instability of people’s preferences (e.g. Qizilbash 2012; Sunstein 2021, for illustrations).

4.3. Objection from contractarianism

The objection from contractarianism proceeds as follows. OPP presuppose a contractarian approach which assesses policies’ welfare implications ‘from the several viewpoints of individual members of society, considered as potential parties to an agreement’ (Sugden 2018: 14, italics added). On this approach, for some agreement to be justified, ‘each party to [the] agreement must recognize that, for all the parties severally, agreement must be more beneficial than non-agreement’ (Sugden 2018: 38, italics added; also Sugden 1998 b: 60). In particular, to justify her policy recommendations, a contractarian ‘has to show that her recommendations are in the separate interests of each individual, as that individual perceives those interests’ (Sugden 2018: 29, italics added; also Buchanan 1986: part III). Conversely, informed/ideal PREF assess policies’ welfare implications from the putatively impartial viewpoint of a social planner or choice architect who aims to enhance individuals’ welfare while respecting individuals’ judgements about their own welfare (e.g. Sunstein and Thaler 2003: 1164, holding that a choice architect should ‘give consumers what she thinks they would choose on their own’; also Bernheim and Rangel 2007: 464, holding that a social planner should ‘respect the choices an individual would make for himself’). Regrettably, the objection goes, there is no plausible and informative impartial viewpoint from which policies’ welfare implications may be reliably assessed (e.g. Sugden 2013, 2018: Chs 2–3). This, in turn, casts doubt on informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE. ¹⁰

This objection nicely elucidates the contractarian approach presupposed by OPP. However, there are at least two reasons to doubt that the objection hampers informed/ideal PREF’s potential to provide a reliable and informative foundation for NWE. First, as illustrated in sections 3.1–3.2, in concrete policy contexts it is typically unfeasible to show that the proffered policy recommendations are ‘in the separate interest of each individual’ targeted by such recommendations. In this respect, it would be of limited import to object that informed/ideal PREF face similar difficulties since in concrete policy contexts it is often unfeasible to design and implement policies that satisfy the preferences of all the individuals targeted by such policies (e.g. Wilson 2011). For on informed/ideal PREF, public policies that satisfy the preferences of most (but not all) individuals may be justifiable if they do not significantly harm others (e.g. Adler 2019: Chs 4–5). And this justification is not available to the proponents of OPP unless they relinquish their requirement to show that policy recommendations are ‘in the separate interest of each individual’ targeted by such recommendations.

And second, OPP do not ‘defend any specific normative proposal about the distribution of [opportunities]’ and do not allow for ‘the analysis of distributional issues’ (Sugden 2010: 47 and 67, italics added; also Sugden 2004: 1020). This concern is exacerbated if one considers the relevant baseline from which OPP measure policies’ welfare implications for each individual targeted by these policies. For this baseline corresponds to individuals’ actual distributions of resources rather than ‘abstract principles of fairness’ (Sugden 2018: 174; also Buchanan 1972: 127, claiming that ‘we start always from here, not from an original position’). And such baseline is ill-suited to address distributional issues (e.g. if individuals’ actual distributions of resources result from previous injustice, then taking these distributions as the relevant baseline will likely tend to perpetuate injustice). In fact, as acknowledged by leading proponents of OPP (e.g. Sugden 2018: Ch. 8), the baseline presupposed by OPP sharply differs from the baselines presupposed by other leading contractarian theories (e.g. Qizilbash 2021, on Rawls’s theory; also Gauthier 1986: Chs 5–7, requiring absence of coercion and approximate equality in the relevant baseline). To be sure, my point is not that OPP yield implausible recommendations to a putatively impartial social planner or choice architect. For OPP are not addressed to such planner or choice architect (e.g. Sugden 2018: 24). Rather, my claim is that OPP yield recommendations that would often be regarded as unjustified (or otherwise objectionable) by many of the individuals targeted by such recommendations (e.g. Arneson 1989; Becker 2005, on the tendency of many agreements based on individuals’ actual distributions of resources to unfairly disadvantage the poor and the disabled). And this, in turn, is rather problematic for OPP given OPP’s requirement that a contractarian ‘has to show that [her] recommendations are in the separate interests of each individual, as that individual perceives those interests’ (Sugden 2018: 29, italics added).

A proponent of OPP may object that appeals to distributional issues do not undermine OPP’s potential to provide a reliable and informative foundation for NWE since ‘complete equality of opportunity conflicts with’ individuals’ freedom of choice (Sugden 2020: 72), which is ‘what ultimately matters’ according to OPP (McQuillin and Sugden 2012: 556). The idea would be that even if one started with a putatively equal distribution of opportunities, inequalities in individuals’ abilities and talents would tend to generate major inequalities in the distribution of resources unless policymakers systematically interfere with individuals’ choices (e.g. Nozick 1974: Ch. 7; also Sugden 2010: 67). For this reason, major inequalities in the distribution of resources may well be regrettable, but do not call for correction (e.g. Sugden 2004; also Sugden 2018: 106, claiming that a responsible agent ‘accepts sole responsibility for the consequences [of his decisions]’). This objection correctly notes that there are significant trade-offs between equality of opportunity and individuals’ freedom of choice (e.g. Fleurbaey 2012 b). Still, the existence of these trade-offs by no means implies that any major inequalities in the distribution of resources are normatively acceptable (e.g. Fumagalli 2020). On the contrary, most prominent theories of fairness hold that fairness requires to correct (or at least alleviate) major inequalities in the distribution of resources even if such correction leads to some efficiency losses (e.g. Vallentyne 2000; Peragine 2004; Fleurbaey and Maniquet 2011: Chs 2–3). This, in turn, challenges the proponents of OPP to specify why exactly all the theories of fairness that take major inequalities in the distribution of resources to call for correction would be mistaken or otherwise inadequate. ¹¹

5. Conclusion

In recent years, several prominent authors have criticized welfare economists’ reliance on preference satisfaction as the normative standard for welfare and have advocated grounding NWE on opportunities rather than preferences. In this paper, I argued that although PREF face significant conceptual and practical challenges, OPP fail to provide a more reliable and informative foundation for NWE than PREF. I then identified and rebutted various influential calls to ground NWE on opportunities rather than preferences to support my qualified defence of PREF. This result does not per se exclude the possibility that some further arguments may be articulated which support grounding NWE on opportunities rather than preferences. Still, together with the major challenges that affect OPP, it makes it pressing for the proponents of OPP to articulate and support such arguments.