The Quality of Work (QoW): Towards a Capability Theory

ABSTRACT This paper introduces a comprehensive conceptual framework for measuring the Quality of Work (QoW) using the Capability Approach (CA). Drawing from [Robeyns,, Ingrid. 2017. Wellbeing, Freedom and Social Justice: The Capability Approach Re-Examined. Cambridge, UK: Open Book Publishers.] modular framework for developing Capability Theories, it proposes we conceive of work as a body of resources existing in a “space” of work. Dimensions of QoW can be identified based on how work resources enhance, or impede, the achievement of important “beings and doings” (Functionings) both inside (intrinsic importance) or outside (instrumental) this space – such as intrinsic Functionings like meaningful work; or instrumental Functionings like family- and life-fulfilment. However, it further argues that many approaches to QoW are under-specified, since they neglect the crucial ways that peoples’ wider circumstances, outside this space of work, determine peoples’ overall work-related wellbeing. This calls for indices of multi-dimensional QoW to also measure (a) the range of wider Functionings people could achieve outside their current work activity (the Capability Set); and (b) personal, social, and environmental factors which affect how work resources are converted into Functionings (Conversion Factors). It is only by taking these circumstances into account that indices can capture the true impact of the worst forms of work, by understanding who is forced to engage in this work.


Introduction
The Capability Approach (CA) has transformed our understanding of human wellbeing across philosophy and the social sciences. It rejects approaches which equate wellbeing with subjective wellbeing (e.g. happiness), gross national income, or indeed any set of resources (e.g. Rawlsian philosophy). Instead, the starting point should be to ask what people are able to do or be: what valued Functionings they are able to achieve. Resources are therefore instrumentally rather than intrinsically important: the "means" to achieving Functionings, rather than "ends" in themselves. This contribution in itself has given rise to a considerable range of applications of the CA. However, the CA also makes two crucial additional contributions which have proved more elusive to conceptualisation and measurement. First, it defines our wellbeing not as Functioning achievement in our current state (described in the CA as our vector of achieved Functionings), but our freedom to achieve different states of wellbeing: it is the range of combinations of achievable Functionings available to an individual (the Capability Set), whether chosen or not. Second, it argues that the same set of resources do not create Functionings equally for all people: personal, social, and environmental factors (Conversion Factors) affect how these resources create Functionings. This means that because of their circumstances, some people will require more resources than others to attain the same achieved and achievable Functionings. In short, the CA calls for a holistic and person-centred assessment of wellbeing. The true impact that any set of resources have on peoples' wellbeing can only be understood if we consider peoples' wider circumstances: what they can do and be (Functionings); the range of combinations of achievable beings and doings (Capability Set); and the way that personal, social, and environmental factors affect their ability to convert resources into wellbeing (Conversion Factors).
Work is ubiquitous in our lives, and encompasses all forms of paid and unpaid productive activity (Budd 2011;Cooke, Donaghey, and Zeytinoglu 2013). It is therefore unsurprising that a considerable number of applications of the CA to job quality have been developed. 1 These have emerged in tandem with a growing number of attempts to measure job quality by academics, public policymakers, and international institutions (e.g. see Hovhannishan et al. 2022;ILO 1999;Leschke and Watt 2014;Muñoz de Bustillo 2011;OECD 2003), many of which already engage with the CA to at least some degree. However, there is still a lack of consensus about how to conceptualise and measure job quality, and this has impeded progress in improving job quality, not least in European public policymaking (Piasna, Burchell, and Sehnbruch 2019). I also submit that there is a tendency across job quality research, including within CA-based approaches, to look at job quality through peoples' Functioning achievement in their vector of achieved Functionings, neglecting the role of the aforementioned wider circumstancesthe Capability Set and Conversion Factorsin determining peoples' ultimate wellbeing. Such research is, I suggest, under-specified: it implies that a given set of work resources has the same fixed and time-static effect on the wellbeing of all individuals. To the contrary, work has a special status within the CA: it has all-pervasive effect on Functionings across our lives, and its impact on our wellbeing can only be fully captured once we consider the wider circumstances under which people negotiate access to, and carry out, work.
To redress this, this paper utilises Robeyns' (2017) modular framework on the development of applied approaches to the CA. She draws an important distinction between Capability Theories and applications of the Capability Approach. The CA is a deliberately "open-ended" and "under-specified" concept (Robeyns 2017). A Capability Theory requires the CA to be "closed" and "specified": used for a specific purpose and aim, with a set of normative principles external to the CA introduced. Building on Robeyns' framework, this paper develops five requirements for a Capability Theory of Quality of Work (QoW). It argues that previous approaches lack elements necessary to fulfil these requirements. I then take the first steps towards building a Capability Theory of QoW which meets these five requirements. I address questions about exactly how work relates to Functionings in an attempt to build consensus around a clearer framework for measuring and conceptualising QoW. Crucially, I then propose a way in which the Capability Set and Conversion Factors could be incorporated into indices of QoWoutlining how they relate to wellbeing, and proposing potential indicators of them.
Without incorporating the Capability Set and Conversion Factors, I suggest it is not possible to correctly distinguish between advantaged people (ie people with high work-related wellbeing) and disadvantaged people (low workrelated wellbeing). This is because the most disadvantaged people do not merely achieve few Functionings from work: they also, first, have few other achievable Functionings across their wider lives (a narrow Capability Set); and second, their personal, social, and environmental Conversion Factors prevent work from achieving the wellbeing which more advantaged people can achieve from even the same (or similar) work. Put another way, they lack the power to force employers to build work around their lives, rather than around the needs of employers. Advantaged people in society, by contrast, negotiate access to forms of work which maximise their achievement of Functionings at various time intervals, because they have a wide range of achievable Functionings and can therefore build work around their own Conversion Factors. They may use poor jobs for their instrumental value earlier in life whilst undergoing training or educationfor example by using the money from them to fund their studies, or taking on unpaid internships to help them progress in their careers. This enables them to access better, more secure forms of work later in life.
It follows that at its most extreme, the exact same work could lead to high work-related wellbeing for some individuals, but not for others. An identical part-time unpaid volunteering role for a charity, for example, might enhance the wellbeing of an older worker with another paid job, a stable career and a good pension who regards it as a supplemental meaningful work activity, but not for a younger worker with no paid work prospects for whom it is the only slim route they have to a stable, paid job. Platform labour in the gig economy is characterised by some as a potentially wellbeing-enhancing job, but I suggest this rests on implicit ideal-typical assumptions about the circumstances of the people accessing these jobs: it may work for a single person in training for a permanent job later in life, but not for someone with a family, no other opportunities for life fulfilment, and thus no prospect of long-term security for them or their dependents. We risk under-stating the impact of the worst forms of work by making such ideal-typical assumptions about the circumstances of the worker. The reason these and other forms of work might be so damaging is not simply that the people doing them generally achieve few Functionings from work, but because of who is forced to it. To measure multidimensional QoW, we therefore need to introduce indicators of these circumstances into existing indices. This paper is split into four sections. In the first, I review how foundational literature in the CA has addressed the issue of work. Second, I set out the requirements for a Capability Theory of QoW, drawing on Robeyns' (2017) modular approach. Third, I review subsequent literature against these requirements. In the fourth, I set out my proposed Capability Theory. The theory proposed in this paper is necessarily incomplete due to some unresolved debates in the literature; I discuss these unresolved debates and chart potential answers to them in the succeeding sections.

Work in Foundational Literature in the Capability Approach
Work features throughout foundational literature in the CA. Sen wrote a dedicated study on employment and development earlier in his career (Sen 1975). Poverty and Famines (Sen 1983) was prepared for the International Labour Organisation's (ILO's) World Employment Programme, and contains frequent references to the role of work in causing and preventing famines. Both Sen and Nussbaum appear to endorse a broad view how work is important to human wellbeing, going beyond its mere pecuniary benefits. In Poverty and Famines, for example, Sen highlights that despite carrying out the same work activities for the same income, two groups of agricultural labourerslandless waged labourers; and share-croppers who received their income from the crop they owned and soldfared very differently in the Great Bengal Famine (Sen 1981, 5): work can thus have a markedly different impact on the "exchange entitlement" of individuals, even for two jobs which have the exact same earnings. Both Sen and Nussbaum's writings also contain frequent references to the empowering role that the non-pecuniary aspects of work can play, particularly in liberating women from cycles of domestic abuse and oppression (Nussbaum 2000;Nussbaum and Sen 1993;Sen 1987).
Since formulating the CA, Sen has argued that unemployment needs to feature in our assessment of the "spaces of inequality", suggesting that the lower wage inequality in continental Europe versus the United States may simply be due to higher unemployment in the former, with otherwise unemployed workers in the US being captured in the labour force and driving up wage inequality (Sen 1997, 159). Later, in the wake of newfound interest in job quality in the ILO in the late-1990s, he argued for the inclusion of informal and unpaid work in our assessment of peoples' work-related wellbeing (Sen 2000). In his earlier writings, he also endorsed multi-dimensional approaches to understanding work (Sen 1975) and unemployment (Sen 1973).
Despite this, it is not possible to identify a clear way of measuring job quality from either Sen or Nussbaum's writings, since their research interests have lay elsewhere. Sen's later articles do not attempt to rigorously apply the CA to job quality so much as make a series of observations about work from a CA-based perspective. Nussbaum's research has focussed more on specifying important Capabilities at a higher level of abstraction and on dedicated applications of the CA to other areas of research, notably women's empowerment and animal welfare. Nussbaum criticised human capital approaches (Nussbaum 2019, 67) and as noted earlier, her writings on women's empowerment give a clear idea of the importance she attaches to the non-pecuniary aspects of work, but despite deriving her central capabilities from the early work of Marx (following Aristotle), she has been criticised for not including meaningful work as a Central Capability (Weidel 2018). 2 It has been left to subsequent scholars to explore how the CA can be applied to job quality.

Five Requirements for a Capability Theory of QoW
Although a wide range of papers on processes for operationalising the CA and identifying Functionings were written in the decades since the CA was formulated (see in particular Clark 2013;Jaggar 2006;Robeyns 2003;Sen 2002b), it was not until Robeyns' (2017) modular framework that scholars had a dedicated book on how to apply the CA. Robeyns argued that on its own, the CA is an open-ended and under-specified concept: it is, by its own definition, an approach. To properly apply the CA, scholars need to, amongst other things, be explicit about the purpose and aim of the application; and introduce external normative decisions to identify important Functionings. In doing this, we develop Capability Theories: these are closed and specified applications of the CA which contain all necessary modular elements. Whilst Robeyns allows for a diverse range of uses of the CA, she argues that "some ideas in this broad 'capabilities literature' do not survive careful analysis, and should be rejected" (Robeyns 2017, 21). Building on Robeyns' modular framework, I propose five requirements for a Capability Theory of QoW. These requirements draw heavily from Robeyns (2017), but given the specific focus of this paper is the measurement of QoW using the CA, they also suggest ways to re-frame and add to her requirements where necessary for the specific purpose of measuring QoW.
Firstly, there is the "specification" requirement: we need to specify the space of interest and the research focus. This requires a clear definition of the set of resources within this space of interest which we are measuring, and the purpose for which the Capability Theory is being built. At this initial stage of building a theory, this body of resources will be extremely broad. The space of interest for CA-based research needs no more philosophical justification than for any other area of research, but it is important to be very clear which resources are and are not in scope.
Secondly, we must make normative decisions to "identify Functionings". On their own, neither resources in our space of interest nor Functionings themselves are "important". Work resources, for example, comprise every single measurable and even non-measurable aspect of work. Since Functionings are simply beings and doings, they encompass all trivial and non-trivial things we can do or be. In the space of work, for example, they would range from the Functioning to sit in a red office chair, to have a vending machine at the workplace, to work from home if desired, to have maternity and paternity leave, or to have impartial mechanisms for redress. Any Capability Theory needs a process to identify a list of important Functionings. This is an inherently normative exercise: a value judgment, external to the CA, needs to be introduced to identify them. There are two broad schools of thought as to how lists of such Functionings can be identified: philosophical deliberation; or democratic / participatory engagement. Advancing the latter approach, Alkire (2005, 5-6), following Sen (2004), proposes that achieved Functionings must satisfy two criteria: (a) they must be "valued as being of special importance … to a significant proportion of the relevant population" to which a given individual belongs; and (b) they must be "socially influenceable" Functionings "that social and economic policies have the possibility to influence directly". I return to Alkire's requirements later.
Thirdly, having identified the list, we need to "relate resources to Functionings". Resources in our space of interest are only important to the extent that they affect important Functionings within the space of interest (intrinsic importance), or are a means to the achievement of important Functionings which are not in our space of interest (instrumental importance). The extent to which they are one or the other varies with the space of interest ("specification") and any external normative theories applied ("identify Functionings"). For example, if we identify work as our space of interest and identify important Functioningslike Nussbaum (2011)based on consideration of the what is necessary to live a life worthy of human dignity, then we need to establish how work resources relate to her list of Functionings.
Fourthly, we need to "introduce Conversion Factors and the Capability Set". In the first three steps, we will have established an approach for assessing the impact that resources in our space of interest have on peoples' Functionings in their current circumstances (their vector of achieved Functionings). However, it is an entirely distinct and separate exercise to conceptualise how Conversion Factors and the Capability Set affect peoples' wellbeing. Conversion Factors are defined as any individual, social and environmental factors which affect how resources create Functionings and, ultimately, Capabilities. Existing literature would suggest that Conversion Factors can be positively-or negatively-framed. In the former case, an individual with the Conversion Factor requires less resources to achieve a given level of Functioning achievement. This appears to be the case for many social or "institutional" Conversion Factors discussed in the literature, such as the way institutions affect peoples' sense of entitlement, and thus opportunities (Hobson 2018;Sen 1999, 142). The availability of paid parental leave in a society has also been framed in these terms (Browne, Deakin, and Wilkenson 2004). In the latter, an individual requires more resources to achieve Functionings, such as a pregnant woman for the achievement of the Functioning of being well-nourished. Which Conversion Factors there are, and whether they are positively or negatively framed, will depend on the purpose of the Capability Theory. The Capability Set, as discussed in the introduction, consists of the range of achievable combinations of Functionings outside of one's vector of achieved Functionings. They thus have an inherent "counterfactual nature" (Comim 2008, 173). There is detailed scholarly discussion about how they could be more directly measured using ranked opportunity sets (e.g. see Klemisch-Ahlert 1993; Pattanaik and Xu 1990), but as I will elaborate on later for the purposes of this Capability Theory there is also a case for more indirectly measuring them using proxies which would suggest an individual has a wide range of achievable Functionings. For both Conversion Factors and the Capability Set, a different set of factors need to be brought in at this stage. For the measurement of QoW, this necessarily requires the consideration of factors outside the space of interest.
Finally, there is the "operationalisation" requirement: we need to set out a process for measuring the wellbeing, in terms of Functionings and (ultimately) Capabilities, which people achieve from the resources in our space of interest. This can be either quantitative or qualitative, and will depend on the purpose of the Capability Theory. For the specific purpose of the quantitative measurement of QoW, the researcher will have ended this process with a set of indicators and dimensions of QoW. An account needs to be given of how these relate to achieved Functionings and, ultimately, Conversion Factors and the Capability Set. A process needs to be set out for which indicators to choose, how they form dimensions, and how dimensions determine any aggregation scoresas has received considerable attention in quantitative job quality literature (e.g. see Anand et al. 2009;Anker et al. 2003;Bescond, Chataignier, and Mehran 2003;Leschke and Watt 2014).

Literature on Work and the Capability Approach: A Critical Review
Applications of the CA to work span across all areas of the social sciences. They demonstrate varying degrees of engagement with the CA depending on the aims of the research, ranging from focussed applications by scholars dedicated to the CA to looser CA-informed research by the broader body of social scientists. Amongst other things, they encompass attempts to integrate the CA into a broader case for a heterodox economics of work (Muñoz de Bustillo 2011; Spencer 2015); a CA-based critique of employment-focussed Active Labour Market Policy and flexicurity (Fernandez-Urbano and Orton 2021; Lambert and Vero 2013;Lambert, Vero, and Zimmermann 2012;Laruffa 2020;Lehwess-Litzmann 2012;Orton 2011;Vero et al. 2012); the Capability Approach to Labour Law (Bueno 2017;2021;Deakin 2019;Langille 2019;Supiot and Meadows 2001); research on agency and work-life balance (Hakim 2000;Hobson 2011;Pandolfini 2012); literature operationalising the CA to identify measures of multi-dimensional job quality (Abma et al. 2016;Green 2001;2007;Green et al. 2022;Gürbüz et al. 2022;Ruiz-Tagle and Sehnbruch 2015;Sehnbruch, González, and Apablaza 2020) or lack of work (Bartelheimer and Verd 2012;Schokkaert and Van Ootegem 1990); and theoretical literature, which engages particularly on which aspects of work are intrinsic Functionings in themselves (Bartelheimer, Leßmann, and Matiaske 2012;Suppa 2019)such as meaningful work (Veltman 2016;Weidel 2018;Yeoman 2014;Yeoman et al. 2019), the "Capability to Aspire" (Hobson and Zimmermann 2022;Lambert and Vero 2013), the "Capability for Voice" (Bonvin 2012;De Munck and Ferreras 2013;Deakin and Koukiadaki 2011), or the instrumental role of job satisfaction (Leßmann and Bonvin 2011). 3 This body of literature provides elements of a Capability Theory, but I suggest it contains no single unifying Capability Theory for measuring job quality which satisfies the five requirements. I will now review this literature against these five requirements.
With respect to requirement one ("specification"), much literature is not specific about the resources of interest. Any restriction to paid work activity is increasingly unsatisfactory in the context of significant developments in the research of unpaid and informal care work, particularly within feminist literature (Barker and Kuiper 2003;Land 1980;Lewis and Giullari 2005), though not exclusively so (Budd 2011). It is also at odds with foundational literature in the CA, since as noted earlier Sen (2000) himself argued that unpaid care work and informal work is a core element of job quality. Broader views of paid and unpaid work are specified in some conceptual literature, as in Weidel's (2018) case for a Functioning for meaningful work, and they are operationalised in some qualitative literature on work (Cooke, Donaghey, and Zeytinoglu 2013), but these do not make it into quantitative literature on job quality due to a considerable lack of data on the working conditions of unpaid workers. I will return to how future surveys could overcome this in the "Operationalisation" subsection.
Much of the existing literature is not clear on any external normative theories they use to derive Functionings (the "identify Functionings" requirement) and on how work-related resources relate to these Functionings ("relate resources to Functionings"). Some literature has argued for a number of important work-related Functionings (which they variously term "capabilities" or "capacities"), with frequent references in this literature to a "capability / capacity" to aspire (Hobson and Zimmermann 2022;Lambert and Vero 2013), a "capability for voice", and a "capability for work" (De Munck and Ferreras 2013). But as highlighted by Goerne (2010, 10-11), they are not explicit on any external normative theories they use to identify these "capabilities" as important. This literature also focuses on a range of Functionings external to work which Active Labour Market Policy should foster, notably "active citizenship" (Laruffa 2020, 6-7), but it is unclear what role all other important Functionings outside the space of work should also play, such as any Functionings related to being healthy, having a family, etc. In the Capability Approach to Labour Law literature, Bueno (2021) develops a more useful and comprehensive framework distinguishing between Capabilities through work, Capabilities in work and Capability for work. Through this, he captures work's instrumental role as "a means to achieve an income in order to have capabilities" (Bueno 2022, 5) whilst also arguing for "freedom to choose work as a capability in itself" (Bueno 2022, 6-7). Given the debate over how to derive important Functionings, it is therefore unsurprising that there continue to be significant unresolved conceptual debates about whether there are any intrinsic Functionings within the space of work itself. We therefore have a full spectrum of approaches, ranging from scholars who advance one or numerous intrinsic Functionings within the space of work (Bartelheimer, Leßmann, and Matiaske 2012;Bueno 2022;Weidel 2018); to those who view work as an instrumentally-important "characteristic-providing activity", its quality assessed in terms of the impact it has on the achievement of "high-level" Functionings outside of this space (Suppa 2019).
It is in respect of requirement four ("introduce Conversion Factors and the Capability Set"), however, that there is a particular gap in existing literature. Most CA research into job quality focuses solely on wellbeing in the vector of achieved Functionings and neglects the role of circumstances outside the space of work in determining work-related wellbeing. Some welcome exceptions are Bueno (2021) and Sayer (2012), who both highlight how the existence of a range of achievable work Functionings in-itself can significantly affect peoples' wellbeing. The most dedicated conceptual attention to both Conversion Factors and the Capability Set is given by Suppa (2019). Conversion Factors, he highlights, could critically alter the wellbeing created by work, turning a good job into a bad job simply because of the way it interacts with a set of other characteristics. In the same vein, the Capability Set is crucial for understanding the devastating effect which the worst forms of work have on wellbeing, such as the extreme deprivation of liberties associated with slavery (Suppa 2019, 15).
Turning finally to requirement five ("operationalisation"), existing literature has varying degrees of engagement with this question. The CA-based critique of Active Labour Market Policy can be credited with spanning conceptualisation and operationalisation (e.g. see Bonvin and Orton 2009;Fernandez-Urbano and Orton 2021;Orton 2011), but the general picture is one of a gulf between operational and conceptual research, with conceptual approaches tending not to operationalise job quality (e.g. see Bueno 2021;Sayer 2012;Suppa 2019), and operational approaches tending to prioritise measurement over theory (Abma et al. 2016;González et al. 2021;Green 2007;Sehnbruch, González, and Apablaza 2020). This paper is an attempt to bridge both conceptual and operational debates.

Proposed Capability Theory for the Measurement of Multi-dimensional QoW
This section bridges these gaps by taking the first steps towards a Capability Theory of QoW, bringing together elements of existing research on work and the CA. Figure 1 outlines the various parts of the theory. I go through each element of the proposed theory in the succeeding pages, covering each of the five requirements.

Specification -Work Resources for the Purpose of Measuring QoW
Work is the space of interest in this Capability Theory. In line with Cooke, Donaghey, and Zeytinoglu (2013, 504), drawing from Budd (2011), I define work as much broader than paid employment alone: a purposeful human activity involving physical or mental exertion that is not undertaken solely for pleasure and has economic value[.] … [I]t includes paid and unpaid tasks inside and outside the home, volunteering, and seeking employment.
Work "resources" exist in this space of interest. In line with Robeyns (2005, 98-100), these resources can be conceptualised as conditions for the achievement of Functionings and Capabilities (see also Suppa's (2019) definition of work as a "characteristic-providing activity"). 4 Whereas general approaches to human wellbeing have a broad space of interest encompassing all resources, I suggest more applied approaches, such as job quality, can be distinguished by their focus on a narrower body of resources within their space of interest, and how the resources in this space affect wellbeing. Because no normative decision has yet been made on which aspects of work are important, at this stage of the theory work "resources" consist of all measurable and non-measurable aspects of work. A later, normative decision needs to be made to identify how these resources relate to Functionings; it is then that we identify indicators and dimensions of QoW.
All of these work resources exist at the individual level, since the CA is "individualistic" in the sense that it sees human welfare as individual-level (Robeyns 2005;Stewart 2005). However, individual-level work resources (and, ultimately, Functionings and Capabilities) can be collectively achieved. I agree that the role of groups and collective action has been neglected in previous literature on the CA (Ibrahim 2006), and since so many characteristics of work are achieved through collective action (e.g. union activism) and held by collective groups, it will undoubtedly be critical to any Capability Theory of work. But I suggest Sen (2002b, 85) is correct in arguing that this should not detract from the fact that resources, and thus Functionings and Capabilities, can only be individually-held.
Having identified the resources, I then specify the purpose of the Capability Theory. We need to consider whether we are interested in the worst forms of poor work, or a broader qualitative assessment of work. In line with most approaches to job quality, and in accordance with the Capability Approach to Labour Law, this theory proposes the latter approach: when we talk about work quality, we are interested not just in which aspects of work enable the fulfilment of minimum standards, but in how work enables (or inhibits) the ability to live a life of a certain quality by following one's personal values. This is an important distinction, since a Capability Theory for the purpose of measuring employment deprivation (e.g. see González et al. 2021) or poverty would give rise to different Functionings, indicators and dimensions at later stages.

Identify Functionings
At this second stage, I introduce normative decisions to identify Functionings. I start by introducing "wellbeing" as the over-arching "good" outcome of interest, in line with other applications of the CA (e.g. see Robeyns 2017; Suppa 2019). Wellbeing is sufficiently broad to encompass both subjective and objective forms of wellbeing and is adaptive enough to allow for the distinction between achieved wellbeing (Functionings) and the freedom to achieve wellbeing (Capabilities). However, the choice to identify wellbeing vs. other over-arching principles, such as a "thick" conception of human need (Dean 2009; Yeoman 2014), needs further discussion and debate in future research.
Agreeing with Sen (2004), I suggest that lists of important Functionings should be identified through a democratic and participatory process of public engagement. Through this process, people could themselves be asked to agree a list for the purpose of measuring the wellbeing people get from work resources (i.e. QoW). Although research to develop such a list through engagement with platform labour workers is being carried out (Ghirlanda 2022), no such list exists at present. We are therefore left with lists developed for different purposes, and based on different normative considerationssuch as philosophical deliberation on the Functionings or Capabilities necessary to live a life worthy of human dignity (Nussbaum 2011, 125-131); or participatory engagement to derive a list for the purposes of equality and human rights monitoring (Burchardt and Vizard 2011). Because of the different purposes of these lists, they inevitably specify Functionings at a higher level of abstraction: these lists thus contain general Functionings outside the space of work. They do not contain intrinsic work Functionings within the space of work which have received considerable discussion in the literature, such as meaningful work (Weidel 2018) or the capability for work (Bueno 2022)an issue I will return to in the next subsection.
Nonetheless, due to a high degree of overlap between the elements of these lists (Qizilbash 1996), pre-existing lists will suffice in identifying a range of important work resources for the measurement of QoW. For example, if a Capability Theory adopts Nussbaum's approach, important work resources could be identified based upon the extent to which they instrumentally affect Central Capabilities. A low-paid job with long and unsociable hours, low levels of autonomy and no task discretion would prevent an individual from building a family (impeding the Central Capability of life), engaging in civic and political life (affiliation) and socialising with their peers (play). Based on this, these would be identified as important indicators of QoW, based on their instrumental effect on these Functionings. This process is along the lines proposed by Suppa (2019, 10) to identify important work resources based on the effect on Functionings specified at what he terms "a higher level of abstraction". However, I suggest this is an unsatisfactory compromise: future research needs to develop a list of Functionings for the specific purpose of measuring QoW. This list would have the advantage of containing at least some Functionings within the space of work itself, in addition to Functionings outside the space of work. I develop this further in the below subsection.

Relate Resources to Functionings: Intrinsic and Instrumental Importance
There are two particular challenges with relating resources to Functionings for QoW. First, as discussed earlier there is debate over whether there are any Functionings within the space of work itselfwith different scholars advancing various intrinsic work Functionings (e.g. for work, for meaningful work, for voice, to aspire) or none at all (e.g. Suppa 2019). This gives work a distinct status in the CA, contrasting with some other areas of applied CA research where there appears to be greater agreement that many Functionings exist within the space of interest itself: in the space of education, for example, it can be agreed that many resources are intrinsic Functionings in themselves (e.g. see Robeyns 2006), in addition to enabling non-educational Functionings in peoples' wider lives. Second, it is hard to see how the work Functionings which have been identified in some literature in any way relate to existing lists of Functionings, because the normative process used to identify the former is often not made clear.
How should we address this challenge? I disagree with approaches which view work exclusively as an instrumental "characteristic-providing activity" based on its effect on high-level Functionings (Suppa 2019). This would mean that no intrinsic Functioning could exist within the space of work itself. In turn, there would therefore be no Capability for work, since the Capability Set can only comprise the range of combinations of Functionings. A number of significant conceptualisations of work have come from viewing work as a Functioning in itself, as part of a Capability Setnotably Bueno's (2022) discussion of a Capability for work. However, it is conversely the case that an approach to job quality which saw work as purely providing intrinsic work Functionings would not capture the considerable (instrumental) effect work resources have on Functionings outside the space of work. This too is an unsustainable position, since work self-evidently has an all-pervasive impact on all areas of our wider lives: as highlighted by Sayer (2012), work impacts every aspect of our lives, including our cognitive development in our earliest years.
It follows that the effect of work on peoples' Functionings can only be appreciated if we consider the impact work has on a wide range of Functioningsboth inside (intrinsic) and outside (instrumental) the space of work. I suggest the instrumental impact it has on Functionings outside the space of work is greater than any impact it has on intrinsic work Functionings, but it does not follow that there are no intrinsic work Functionings. Yet as argued the above subsection no specific list of Functionings for the measurement of QoW, which includes Functionings within the space of work, exists.
In the absence of such a list, I turn to the democratic conditions set by Alkire (2005), following Sen (2004): that important Functionings should (a) be of special importance to a population and (b) socially influenceable Functionings.
Many Functionings within the space of work satisfy Alkire's second condition, since public policies self-evidently have a significant influence on many work characteristics. However, I suggest many fail to satisfy the first condition. This is because people appear to have widely divergent views of what "good work" is, with even those in the same jobs disagreeing on whether they are good or bad. This was highlighted in a recent Government-commissioned review into good work in the UK, where people reported opposing views of the quality of the exact same work activity (Taylor 2017, 11): We were … taken by some of the diametrically opposed views of the same job presented to the Work and Pensions Select Committee[.] … Hearing one person describe a job as the best they have had followed by another person describing the same job as highly stressful or exploitative highlights the challenge for policymakers in seeking to promote better work for all. This heterogeneity of peoples' views about work is also reflected in statistics on subjective job satisfaction, which often show that people in jobs with objectively bad characteristics report high subjective job satisfaction (e.g. see Brown, Charlwood, and Spencer 2012;Léné 2019). This poses a challenge for any Capability Theory seeking to identify intrinsic Functionings within the space of work.
Bringing this together, I tentatively suggest that agreement could be reached that (a) carrying out work in itself (Bueno 2022), and within this (b) meaningful work (Veltman 2016;Weidel 2018;Yeoman et al. 2019) are Functionings within the space of work. I suggest that one core part of peoples' wellbeing is our opportunity to carry out paid or unpaid productive activity, and, distinct from this, meaningful productive activity. It follows that our freedom to achieve this activitythe range of productive activities and meaningful productive activities which we can devote ourselves tois part of our wellbeing, alongside our freedom to achieve other Functionings outside the space of work. This in turn would mean that an individual's Capability Set would comprise not just Functionings outside the space of work, but work Functionings: the range of work opportunities, and within this the range of meaningful work opportunities, available to people. I further suggest that the Capabilities to Aspire and for Voice have a case to be viewed as intrinsic work Functionings, but they may best understood as process freedoms necessary for the achievement of any Functioning. 5 Beyond this, I would challenge whether many other Functionings within the space of work could be agreedeither due to public disagreement over their value due to how their effect varies across individuals within societies (job characteristics such as hours, flexibility, voice, etc.); or because they do not fit within a CA-based framework.
In this section and the previous section, I have arrived at a means through which the effect of work on Functionings both inside and outside the space of work can be captured using the CA. I have however done this based on two separate and very different sets of normative principles: one based on the instrumental effect of work on pre-existing lists of Functionings (e.g. Central Capabilities); and another considering what Functionings within the space of work people would democratically agree existed, if they were asked to deliberate. Future research needs to investigate the potential for bringing these together into a unifying normative theory. I suggest that this could be done through a two-stage democratic and deliberative process: first, engaging with workers to develop lists of Functionings for the specific purpose of measuring QoW ("identify Functionings"); and second, agreeing with them how work resources relate to these lists of Functionings ("relate resources to Functionings").

Introduce the Capability Set and Conversion Factors
For the Capability Set, I suggest that there are broad or narrow ways of conceptualising it for QoW. A narrow conception would only consider the range of combinations of Functionings within the space of work which an individual can achieve. If we accept work and meaningful work as Functionings, this would consist of the range of combinations of work and meaningful work Functionings available to people, and would be broadly consistent with Bueno's (2021) articulation of a Capability for Work. A broad conception would go further, looking at all the Functionings inside and outside the space of work achievable for each individual worker. This would in effect consist of the overall wellbeing of all workers: their achieved work-related wellbeing, but also their freedom to achieve all forms of wellbeing, whether workrelated or not.
For the purposes of this Capability Theory, I advance a broad approach. Adopting a broad approach necessarily requires the measurement of circumstances outside the space of work: incorporating indicators of the wider wellbeing of people, outside their vector of achieved Functionings in the space of work. I suggest this is useful because it gives us an assessment of the power of these workers to build work around their own lives. Since this power is determined by the freedom to achieve all combinations of Functionings, and not merely work-related Functionings, there is a strong case for adopting a broad approach. I suggest Hirschman's (1970) conceptualisation of Exit, Voice and Loyalty is a useful way of understanding this: someone with a wide range of Functionings inside and outside the space of work can refuse work when it is not satisfactory for them, since they have other opportunities (Exit). This, in turn, gives them greater power within the workplace (Voice) and greater returns to engaging with their employer (Loyalty), since an employer knows they have genuine alternatives.
In the absence of any direct measure of the Capability Set, I suggest the use of less direct indicators which capture the broader skills, work opportunities, other earnings and assets of both the individual and other household memberssince these suggest that someone has a wider range of achievable Functionings outside of their chosen work activity. This would potentially include: . Any income they receive other than earnings, including welfare support, capital gains, pension, etc.; . The work history, work opportunities, skills and qualifications of the worker, as a signifier of the range of alternative work and meaningful work Functionings available to them; . The value of physical and financial assets they hold; . The wealth, earnings, and other income of all other household members; For measuring QoW, I define Conversion Factors as personal, social and environmental characteristics which, if possessed, mean that an individual needs a different amount of work resources to achieve a given level of QoW. As discussed earlier, these can be negatively-framed or positively-framed. For example an individual with dependents would require work to be more flexible, and would need more earnings from work, in order to achieve the same Functionings through work as someone who did not have these Conversion Factors (negatively-framed). Conversely, living in a welfare state with a strong social security safety net would lessen the impact of a precarious job on earnings, family-and life-fulfilment (positively-framed). By definition, Conversion Factors include circumstances outside the space of work. For the measurement of QoW, they could comprise a mix of positively-and negatively-framed Conversion Factors, including: . Personal Conversion Factors such as the number of dependents an individual has, whether they have disabilities, or any caring responsibilities; . Social Conversion Factors such as the social context in which they work, including attitudes to women's place in the labour market, their legal rights in work, the level of unemployment in their country, the institutional level of worker power to shape their own working environment, the sufficiency of welfare safety nets, and the nature of a country's Active Labour Market Policy; . Environmental Conversion Factors such as the climate in which they work; It is only after this process that we can fully measure peoples' work-related wellbeing, extending our analysis beyond a narrow assessment of work characteristics in peoples' current vector of achieved Functionings. Figure 2 overleaf gives an illustration of how this Capability Theory could be used to conceptualise the differing work-related wellbeing of two peopleone with high work-related wellbeing (an advantaged person) and the other with low work-related wellbeing (a disadvantaged person)throughout their life course. The point to emphasise is that at certain time intervals in their life course, the vector of achieved Functionings from work resources for these two people might be the same: early on in their lives, they might carry out work activity which is very similar, such as insecure and under-paid work. An index of QoW which only measured peoples' vector of achieved Functionings would therefore identify both workers as engaging in similar work activity: to refer back to Figure 1, their QoW Index score would be similar. It is only after adjusting for their work-related wellbeingby considering the Capability Set and Conversion Factorsthat the true differences between these two individuals could be identified. This is because one person has a wider range of combinations of Functionings outside their own vector of achieved Functionings in their current work activity (a wide Capability Set). This gives them the power to negotiate access to forms of work which build around their Conversion Factors at various life intervals. By contrast, the other person has a limited range of alternative combinations of Functionings outside their vector of achieved Functionings. Thus the first person can be described as advantaged and the second can be described as disadvantaged only once we consider the interaction of circumstances outside the space of work.

Operationalisation
To make up these indicators and dimensions, important work resourcesand thus dimensions and indicators of a QoW indexcan be identified based on the extent to which they affect, or impede, the achievement of an identified list of Functionings inside and outside the space of work. In principle this is an empirical rather than a normative exercise. Previous research has already identified potential indicators of Central Capabilities using existing survey data (e.g. see Anand et al. 2009), and with good data it would be possible to empirically assess which work resources are important to the achievement of these. In the absence of such data, a review of literature across the social sciences on this effect would suffice (e.g. see Muñoz de Bustillo 2011). It will be useful through this process to create dimensions of job quality, but as highlighted by Suppa (2019, 13), it is important to "distinguish the multidimensionality of labour activities carefully from the multidimensionality of human wellbeing". It follows that dimensions of QoW should instead be regarded as work resources, grouped according to the similar way in which they help achieve, or impede the achievement of, Functionings. I further suggest that in line with most applications of the CA, indicators in this index would strongly emphasise the objective characteristics of work rather than subjective aspects such as job satisfaction (see Felstead et al. 2019): by definition, such subjective characteristics satisfaction are an effect of a work resource (after the interaction of other circumstances outside the space of work), and not work resources in themselves. Whilst nonetheless crucial to the study of job quality (see Brown, Charlwood, and Spencer 2012), for the purposes of this Capability Theory subjective indicators are only useful to the extent that they suggest the existence or non-existence of a given important work resource, in the absence of a more direct measure of the existence of this resource. For example, to establish the use of temporary contracts in a country, one may have to rely on a survey question asking workers whether they think their job is permanent or temporary. Beyond this, this Capability Theory does not propose a specific way of aggregating indicator and dimension scores to determine index scores, and is compatible with a range of aggregation and weighting techniques.
Due to limits on data availability this proposed index would necessarily provide an incomplete picture. Despite including unpaid work in the definition of work resources at the Specification stage, it is likely that compromises will later need to be made due to a tendency for national surveys to only ask those in paid work about job quality. However, I suggest that it would be possible for future quantitative surveys of working conditions to incorporate insights from qualitative research which asks participants questions about both paid and unpaid work activity (Cooke, Donaghey, and Zeytinoglu 2013). All survey participants, in both paid and unpaid work, could be asked a range of questions about the conditions of unpaid work, such as the worth they attach to their unpaid activity, the hours they devote to it, the nature of this activity (e.g. whether it is unpaid caring, or volunteering), and whether they regard it as a supplement to or replacement for paid work. In time, this would allow quantitative research to start to integrate unpaid work activity into indices of QoW.
I suggest the key added value of a Capability Theory of QoW is in the introduction of the Capability Set and Conversion Factors, towards the right of Figure 1. A set of indicators for each of these could be used to adjust the QoW index score, creating a Work-Related Wellbeing index score. To measure these indicators, I suggest existing multi-dimensional indices of QoW make use of surveys which combine longitudinal and household data alongside data on individual working conditions. Social or environmental Conversion Factors could also be inferred based on the features of specific welfare states and labour markets. The proposed index would be a step forward on existing job quality indices by allowing us to explore how a range of individual, social and environmental factors interact with conventional measures of multidimensional job quality. This would allow us to understand the wider circumstances of people in "good" vs "bad" jobs.

Conclusion
Bringing together literature on the conceptualisation and measurement of job quality, this paper has charted the first steps towards a Capability Theory of QoW, designed to meet five requirements based on Robeyns' modular framework. It has been developed with operationalisation in mind, and has ended by outlining the core elements of an index of QoW. In advancing this theory, I have attempted to resolve some of the conceptual debates in the CA in particular, and in job quality literature more broadlybridging the gulf between conceptual and operational approaches to address some of the issues causing a lack of consensus about how to measure good work.
The theory particularly draws from Bueno (2021), Sayer (2012), Suppa (2019), and Weidel (2018). Consistent with Sayer, it recognises the all-pervasive effect that work has across every aspect of peoples' lives, through the (instrumental) impact that work resources have on Functionings. However, following Bueno and Weidel, it argues that intrinsic work Functionings exist, and should thus feature in the Capability Set. In line with Suppa, it conceptualises work resources in a way which is consistent with his idea of work as a characteristic-providing activity, and argues that Conversion Factors and the Capability Set play a crucial role in determining peoples' ultimate work-related wellbeing.
A key emphasis throughout this paper is that the true impact that work has on people can only be understood if we consider circumstances outside the space of work. As outlined in the introduction, the CA's central argument is not merely that we must regard Functionings rather than resources as the intrinsic aspects of wellbeing: it is that we must ground our understanding of the interaction of these resources and wellbeing in the social context people live in. Any set of resources do not create wellbeing in a fixed, time-static way. Rather, these resources interact with other factors to create wellbeing; and peoples' ultimate wellbeing is determined by their freedom to achieve other states of wellbeing outside of their vector of achieved Functionings. It is only by conceptualising work in this way that we will be able to fully understand the negative impact that the most damaging forms of work have on peoples' lives, and the stark inequities in the experience of work both within and between societies.
Notes 1. In this paper, I use the term "work" to refer to paid and unpaid productive activity (see the "Specification" subsection for a fuller definition). "Job quality" is used in reference to literature which measures, using multi-dimensional indices and other approaches, the effect that at least some of this work (usually paid work) has on peoples' wellbeing: to avoid confusion over multiple terms, it should be read as encompassing all the various synonyms for job quality which this research useswhether it be "employment deprivation", "quality of employment", or otherwise. The term Quality of Work (QoW) is exclusively used to refer to the specific approach to conceptualising and measuring the wellbeing which people achieve from work which is being proposed in this paper. "Wellbeing" is discussed in the "Specification" subsection. 2. Weidel's paper refers to a "Central Capability" for "meaningful labour", but to limit the use of terms in this paper I equate this with a Functioning for meaningful work. 3. The literature uses various terms to refer to Functionings, including "Capabilities", "Central Capabilities", "Capacities", "External Capabilities", etc. Again to avoid use of multiple terms, in this paper I exclusively use the term Functionings to refer to these important "beings and doings". "Capability Lists", "Central Capabilities" or otherwise are here defined as lists of important Functionings, identified through a normative processbe it democratic or participatory engagement, or philosophical deliberation. The Capability Set is strictly used to refer to the range of combinations of important Functionings which are achievable for a person, outside of their current vector of achieved Functionings. 4. To elaborate, work resources exist in a space of work. These resources affect the achievement of Functionings both inside this space of work (intrinsic effect) and outside this space of work (instrumental effect). This should not be confused with how resource-based philosophies, such as Rawlsian approaches, understand the term resources, since these philosophies see resources as having intrinsic value to human wellbeing: they do not draw a distinction between resources and wellbeing. 5. As Sen has highlighted (Sen 2002a, 10), the CA is interested not just in opportunity freedoms (the availability of genuine choices) but also in process freedoms (someone's agency/capacity to control their choices). There is not the space to do justice to this in this paper, but what are often termed "Capabilities" or "Capacities" for voice and to aspire appear to strongly relate to the notion of process freedoms. They also appear to have a distinct status as preconditions for the exercise of Functionings: for someone to have any given Functioning as part of their Capability set, they necessarily require the exercise of voice and the ability to aspire for (and thus knowledge of the availability of) the Functioning. Dimensions and indices of QoW could then be identified based on the work resources which enable, or impede the enablement of, these and other process freedoms, since these in turn determine the achievement of Functionings important to QoW. An alternative approach might be to treat them as what have recently been termed "agentic Capabilities" (Dold and Lewis 2023), but this may add confusionadding a further term to a field already heavily-laden with prefixes and suffixes to the word "Capabilities"and give the impression that they are less important than they in fact are. This is, however, an incomplete treatment of an issue which warrants considerable further dedicated attention.
Acknowledgements I am extremely thankful to Tania Burchardt, Kirsten Sehnbruch, Peter Bartelheimer, Raffaele Ciula, Pietro Ghirlanda, Francis Green, Stephen Overell, Cherise Regier, Nicolai Suppa and in particular the two anonymous reviewers and the Editorial Team for their helpful advice and comments. I also thank all participants in the panel on the measurement of multi-dimensional quality of employment at the 2022 HDCA conference in Antwerp and in an earlier presentation of this paper to the Work and Employment Thematic Group of the HDCA.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by the Economic and Social Research Council.

Notes on contributor
Thomas C. Stephens is a PhD candidate in Social Policy at the London School of Economics and Political Science, and an affiliate of the LSE's Centre for the Analysis of Social Exclusion. He is also a Teaching Fellow at the LSE's School of Public Policy and a Coordinator of the Work and Employment Thematic Group of the HDCA. He holds an MSc in Social Policy from the LSE (2015), where he was awarded two Titmuss Prizes for Outstanding Performance and Outstanding Dissertation. His research focus spans both the conceptualisation and the (quantitative) operationalisation of the Capability Approach to measure human wellbeingparticularly as it pertains to job quality, worklessness, and wider labour market disadvantage (including precarity, informality, and the impact of new forms of labour).

Data availability statement
Data sharing is not applicable to this article as no new data were created or analysed in this study.