Can the Capability Approach Live Up to Its Promises? The Conceptualisation of Freedom Matters

In this section we shall explore whether all variants belonging to the class of specifications of the general capability framework as defined by two char­acterising features, identified by Robeyns (2017), can indeed live up to the promise to address and overcome the problems the capability approach was originally meant to address and to move beyond welfarism at the same time.

However, such a move from achievements to freedom and from means and resources to ends is not sufficient to overcome the highlight­ed problems. As we show in this section, whether the problems can be overcome will depend on the conceptualisation and valuation of cap­ability, reflecting the real freedom a person enjoys. In order to do so, we shall draw on the freedom ranking literature in welfare economics that can and is often interpreted as the ranking of capability sets (Sen 2002; Baujard 2007).

Despite its prominent roots in welfare economic theory, the welfare economics literature on the capability approach has remained surprisingly limited. A notable exception is the small literature on freedom rankings, that can be interpreted as the ranking of capability sets.^[146] In this literature an axiomatic approach is employed to rank sets of options in terms of the freedom they offer to a person.

In recent decades scholars increasingly employed the axiomatic ap­proach to clarify the limits of welfarism in general (Suzumura and Xu 2009) and the capability approach and its unresolved challenges in particular (Echavarri and Permanyer 2008; Basu and Lopez Calva 2011; Foster 2015; Pattanaik and Xu 2012, 2020; van Hees 2016). In this formal literature, the main focus is on the construction, conceptualisation and ranking of capability sets. It should be noted though that many of the scholars in the freedom ranking literature do not explicitly defend one informational basis for welfare assessments. Instead, the interest is usually directed towards gaining a deeper understanding of the core concepts of the approach and possible problems by formalising them.

The formal literature on the capability approach usually starts out from (capability) sets of (mutually exclusive) bundles of functionings a person can choose from. One of the main questions discussed is that of which condi­tions should be satisfied by rankings of capability sets. The axiomatic literature dealing with this issue initially focused on the general question how sets of alternatives can be ranked in terms of freedom of choice. A crucial point when interpreting the freedom ranking literature in terms of the freedom offered by a capability set, is in how far the respective conceptions of freedom are in line with the notion of real freedom employed in the capability approach. In the following we shall discuss this issue in greater detail and explore some problems related to it (van Hees 2016; Pattanaik and Xu 2020). One important issue is what are the relevant constraints that have to be absent for a person to indeed have a certain bundle of functionings at her reach.

Once the conditions under which certain functionings are within a person's reach and thus form part of her capability set are clear, the question then becomes how such capability sets can be ranked in terms of the freedom they offer to a person. The literature on freedom rankings is concerned precisely with this question.^[147] Sets of mutually exclusive alter­natives, that can be interpreted as capability sets containing the bundles of functionings within reach of a person, are ranked in terms of the freedom they offer to her. The question is which conditions such rankings should satisfy. In other words, what does it mean to say that one set of functioning bundles offers more freedom or capability to a person than another one?

One of the first answers provided was the simple cardinality ranking (Pattanaik and Xu 1990), in which sets are ranked in terms of the number of options they contain: the more options available, the more freedom of choice a set offers. This ranking has been criticised for various reasons, one being that it does not account for the value of the alternatives (or function­ing bundles) available to a person. As Sen (1991) influentially pointed out, a dreadful alternative such as ‘being beheaded at dawn' would increase a person's freedom as much as say ‘going to school', which is counterintui­tive, particularly if we are concerned with the role and value freedom has for human well-being and development.

Pattanaik and Xu (in press) distinguish four different ways in which the value of alternatives could be incorporated in freedom rankings. For this purpose, they introduce the notion of a ‘Value Based Ordering' (VBO) which can, but does not have to coincide with a person's actual (or possible) preference ordering(s). More specifically, they distinguish be­tween (a) all possible VBOs a person might have in the future, (b) a person's actual VBO over functioning bundles, (c) all VBOs held by a certain group or by members of a society, (d) all VBOs a person could reasonably have (according to some perfectionistic theory of the good life), (e) all VBOs a person could possibly think of choosing herself. Whether a newly available functioning (bundle) then increases the freedom offered by a capability set will depend on whether it is ranked at least as high as all other functioning bundles available by at least one of the VBOs. For instance, whether the functioning ‘being beheaded at dawn' will increase the freedom offered by a capability set will depend on whether it is ranked at least as high in at least one VBO, as all other options.

One crucial issue is whether all of these ways to account for the value of the options that are available in a person's capability set allows one to move beyond welfarism and to overcome the problems the capability approach was meant to address. I argue that this is indeed not always the case. The first four classes of freedom rankings ((a)-(d)) are, in one way or another, in conflict with either the ability of the capability approach to address the problems that motivated its development or the possibility of moving beyond welfarism. Only if (as in case (e) above) the value of options is based on the VBOs a person could possibly imagine choosing herself (that is if a person's agency is taken seriously), a possibility might open up to move beyond welfarism while overcoming the problems the capability approach was originally meant to address.

In case (a) freedom is valued instrumentally to achieve a higher prefer­ence satisfaction (if VBO is equated with preferences) in the future (Kreps 1979). Thus, what is (intrinsically valued) is preference satisfaction and freedom is a mere means to achieve that in case of uncertainty over one's future preferences. This, however, does not do justice to the original aim to move beyond a welfarist informational basis, since freedom is only valued instrumentally to achieve a higher level of preference satisfaction. In our example this would mean that the availability of the option to wear a headscarf increases freedom only if the woman might one day prefer wearing a headscarf to not wearing one, say because of religious conver­sion, so that being able to choose this functioning bundle would lead to a higher satisfaction of her future preferences.

In cases (b) and (c) it can be shown that the adaptation problem, one of the criticisms of certain utilitarian variants that motivated the capability approach, can reoccur. To see this, it is important to see how the value of functioning bundles identified by VBO influences freedom rankings. In many contributions (see, e.g., Pattanaik and Xu 1998) an additional option increases the freedom offered by a set only if it is at least as good as all other options available (in terms of at least one of the relevant VBOs). If, however, adaptation can influence the valuation of functioning bundles itself, the adaptation problem can reoccur on the level of freedom. Say if a person adapts her valuation in the light of her (perceived) circumstances and starts to value ‘becoming a shoe cleaner' more than ‘becoming a doctor', then the addition of ‘going to university and becoming a doctor' to her capability set (in which ‘becoming a shoe cleaner' is also available) would not increase the freedom or capability offered by it.

In case of (d) one considers all VBOs a person can reasonably have (according to some perfectionistic theory of the good life). In this case there is a risk of running into a similar problem as the one that was the main criticism of resource-based approaches, namely to neglect the diver­sity of human beings and the plurality of their conceptions of the good life.¹² The problem that might occur is that it becomes conceptually impossible to identify acts of paternalism (Binder 2019). To see this, consider the following definition of paternalistic acts based on Dworkin (1972): paternalism refers to (i) an act (of interference) of an agent X with the freedom of another person Y, which is (ii) without the consent of Y, and (iii) promotes the good of Y.

Now, suppose we take freedom in clause (i) in the definition of paternal­ism to be measured along a ranking that incorporates the valuation of alternatives/functionings as in (d). This means that a particular freedom or bundle of functionings increases a person's real freedom (or capability) only if it is considered to be valuable according to some standard, inde­pendent of a person's own views. The standard could, for instance, be an objective list theory of well-being. Say the objective list theory at hand does

Note that case (d) would thus lead to a violation of the core characteristic of value pluralism in Robeyns's (2017) categorization.

not consider the option of wearing a headscarf in the office or at school to have any value for a woman. Next, assume person X removes some particular freedom of wearing a headscarf (x) from the opportunity set of person Y, motivated by the promotion of Ys good or interests, where Ys good is defined along the lines of the objective theory under consideration, without the consent of Y. Thus clauses (ii) and (iii) of the definition of paternalism are satisfied. To see whether this is an instance of paternalism, consider clause (i) in the above definition again. If interference is taken to mean that a person interferes with another person's freedom if, and only if, he diminishes it, we do not then speak of interference if the freedom of a person is increased or is left untouched. Given that x, that is, the option to wear a headscarf, is not considered to be valuable along the objective theory employed, the removal of it does not diminish Ys real freedom. Hence X does not interfere with Ys overall freedom, and we do not face an act of paternalism since clause (i) in the definition above is not satisfied. In a sense these acts of paternalism are ‘defined away'. However, this would make it per definitionem impossible to identify cases of paternalism, there­by making it impossible to justify such acts, opening the doors to excluding all conceptions of the good life that deviate from the employed objective list theory. To illustrate, consider the debate about prohibiting the wearing of headscarves in schools that became the topic of considerable public debate in many European countries in recent years. Such prohibitions to wear headscarves of hijabs are often justified on paternalistic grounds, in other words that they are for the women's own good. No matter whether one agrees or not, it is usually considered important for such arguments to be open to public scrutiny, debate and justification. The problem of the above highlighted case of ‘hidden paternalism', however, is that it is no longer in need of justification, as prohibiting the wearing of headscarves would not count as an interference in a woman's freedom and therefore not as an act of paternalism. As a result, the diversity in valuation and conceptions of well-being risk being severely violated (if the approach is for instance employed in policy evaluation or design).

This last part and the problem of paternalism deserves special attention if one is concerned with welfarism and its alternatives. Problems of pater­nalism are considered particularly problematic in welfare economics. As discussed in the introduction to this volume, avoiding paternalism or any value judgements that do not stem from the preferences of the individuals concerned, was considered crucial to safeguard consumer sovereignty and autonomy in welfare economics throughout the years. Including additional information beside a person's preferences, which is a characterising feature

of non-welfarist approaches, is thus often considered to raise suspicions of paternalism. In the literature in political philosophy, convincing argu­ments have been made for paternalistic policies to be justified in some cases (such as when agents are not being well equipped to take decisions on their own). To discuss whether acts of paternalism are justified, the possi­bility of identifying paternalistic policies is required. Note however that the problem highlighted here is even more severe, as it leads to a situation of ‘hidden paternalism', that is cases of paternalism can no longer be identi­fied (due to them being defined away) and are thus no longer even in need of justification.

This discussion thus illustrates that if freedom or capability is conceptu­alised along the lines (a)-(d), a number of different problems that the capability approach was originally meant to address risk re-emerging. To put it differently, the ways to conceptualise it along (a)-(d) all fail to satisfy at least one of the discussed desiderata, namely to move beyond welfarism and address the two problems at the capability approaches' origin respect­ively. If, for instance, sets are ranked in terms of their value to maximise uncertain future preferences (or VBOs), then freedom is valued instru­mentally to achieve higher utility or preference satisfaction in the future (but not for its own intrinsic value). The information that ultimately counts is a person's utility/preference satisfaction, not freedom. If, alterna­tively, the valuable functioning bundles in a set are identified on the basis of a person's or group's actual valuation function (b, c) or by using a list of objectively valuation functions that are independent of the person in question (d), then we saw that a version of the adaptation problem or an impossibility of paternalism, a problem of diversity neglect, can (re)occur. The remaining avenue still open that holds the promise to satisfy all of the three criteria is the one in which functionings are valued in terms of their contribution to a person's agency, that is to the process in which a person reflects upon and chooses her preferences.¹³

These points highlight that the importance ascribed to freedom in the capability approach is not enough to move beyond welfarism and over­come the problems at its origin. Whether the approach can move beyond welfarism in the assessment of societal states and succeeds in addressing the problems at its origin crucially depends on the way freedom is concep­tualised and capability sets are ranked. To move beyond achievement by including information about the freedom a person enjoys is thus neither

For one way as to how such a process of preference formation can be formalised, please see Binder (2013).

enough to ensure that the capability approach can overcome the problems at its origins, nor does the assignment of importance to freedom in itself guarantee one to move beyond welfarism.

12.5