The conceptualisation and measurement of poverty is a task of considerable sociological importance as well as policy relevance. Contemporary European poverty analysis continues to be heavily influenced by Peter Townsend’s relative deprivation approach, and while relative income measures of poverty—for example, measuring poverty at 60 per cent of national median income—are experiencing something of a challenge in recent years, they continue to represent the dominant approach to measuring poverty in Europe.
Nonetheless, poverty analysis is currently undergoing a multidimensional turn, due, in part, to a growing awareness of the limitations of such relative income measures. This turn towards multidimensionality has resulted in a growing number of multidimensional poverty applications in both European and non-European contexts (e.g. Coromaldi and Zoli, Reference Coromaldi and Zoli2012; Whelan et al., Reference Whelan and Maître2014; Waglé, Reference Waglé2008), as well as debates about and innovations in the measurement of multidimensional poverty itself (e.g. Alkire and Foster, Reference Alkire and Foster2011; Alkire and Santos, Reference Alkire and Santos2010; Ravallion, Reference Ravallion2011; Ferreira and Lugo, Reference Ferreira and Lugo2013). However, a central argument presented in this paper is that the problems of income-centric analysis are only in part measurement problems––they are also, very substantially, problems of conceptualisation, and the conceptualisation of poverty remains a neglected aspect of this multidimensional turn to date. In this paper, we discuss a number of problems associated with the dominant Townsendian framework for conceptualising and measuring poverty, and articulate an alternative framework, based on the capability approach, which, we argue, can overcome these difficulties.
The capability approach was developed initially by the economist and philosopher Amartya Sen. Its central concepts are functionings and capabilities. “Functionings” refer to the various things a person succeeds in “doing or being”, such as participating in the life of society, being healthy, and so forth, while “capabilities” refer to a person’s real or substantive freedom to achieve such functionings: for example, the ability to take part in the life of society, etc. (Sen, Reference Sen1999: 75). Of crucial importance is the emphasis on real or substantive—as opposed to formal—freedom, since capabilities are opportunities that one could exercise if so desired.
The central claim of the approach is that in interpersonal analysis, such as poverty analysis, our focus should be on what people can do and be and not just on what they have, or how they feel (e.g. Sen, Reference Sen1992). It thus focuses on ends—people’s capabilities—rather than means—their resources. Since what people can do and be is inherently plural, the capability approach offers a framework which is unambiguously multidimensional, focussing on the many ways in which human lives are impoverished.
Sen argues that resource-centric approaches to understanding poverty are deficient because (i) people have differing needs, which means that they require different amounts of resources to achieve the same capabilities (variations which he calls “conversion factors”), and because (ii) income is just one of the influences on what people can do and be (e.g. Sen, Reference Sen1999: 87; 2009). The approach insists on distinguishing between resources, which are considered to be only of instrumental importance, and capabilities, which, Sen argues, are of intrinsic importance (Sen, Reference Sen1999: 87). This, it turns out, provides the foundation for a critique not just of income-centric measurement approaches, but also of the Townsendian tradition of poverty analysis more broadly. It also provides the starting point for an alternative approach to conceptualising and measuring poverty. As we have previously argued (Hick, Reference Hick2012), however, the capability approach should not be understood as constituting a distinctive field of studies (“capability studies”) but as providing a lens with which to understand our existing concerns—in this case, the problems of poverty and deprivation.
For this reason, in this paper we restrict our attention to the contribution that the capability approach might make to the analysis of poverty in Europe, and do not discuss the many theoretical contributions and empirical applications in a range of other fields of study (for a recent discussion and critique in this Journal, see Kremakova, Reference Kremakova2013; see also Hick and Burchardt, forthcoming). Furthermore, we do not revisit the ill-tempered debate between Sen and Townsend themselves (Sen, Reference Sen1983; Reference Sen1985; Townsend, Reference Townsend1985), which both would in later years concede had been less than illuminating, preferring instead to draw on their primary ideas and works. While the paper discusses the conceptualisation and measurement of poverty with reference to the nations of Europe, the arguments presented herein are intended to be of wider significance to the analysis of poverty in rich nations and, indeed, in comparing the performance of richer and poorer nations more broadly.
The concept of poverty within the field of sociology remains firmly based on the path-breaking contribution of Peter Townsend (e.g. 1970; 1979). Townsend argued that poverty must be understood as a case of deprivation relative to wider society, refuting the idea that poverty could be understood without reference to the context in which it was found. Townsend’s work represented the culmination of a re-discovery of poverty in Britain (Coates and Silburn, Reference Coates and Silburn1973), overturning earlier optimism that poverty in Britain had, by and large, been eliminated in the post-war era.
In contrast with earlier “absolutist” approaches, which focused on subsistence needs, Townsend argued that the concept of poverty must reflect both material and social needs, and that these could only be considered relative to the societies in which people lived because people’s needs were socially determined. He noted that “[a] society which becomes more prosperous also becomes more complex and imposes new needs on its members” (Townsend, Reference Townsend1970: 18-9) and that, therefore, a poverty standard must evolve over time in line with changes in social customs and expectations.
As it turns out, the way in which relativity was specified by Townsend is not just the central contribution of his framework, but also its central problem. Townsend’s claim was that poverty should be understood as occurring when people’s resources fell below levels necessary to enable to them to participate in widely-accepted living standards and customs within society. Townsend intended his threshold of resources to refer to (and thus contain) a particular meaning—namely, the level below which participation was withdrawn (Townsend, Reference Townsend1979: 57). Nonetheless, the primary way in which this broad approach has been operationalised has been through the use of statistics of low income, such as a poverty threshold set at 60 per cent of median income. Such a threshold has often been criticised for being arbitrary (e.g. Gordon et al., Reference Gordon and Pantazis2000: 93; Bradshaw and Mayhew, Reference Bradshaw and Mayhew2011: 16, inter alia). However, if it lacks the specific meaning intended by the Townsendian framework at any point in time, we can at least expect that, over time, it rises as the framework predicts; that is, the threshold rises in line with median incomes in society. It is “relative”.
Using a threshold of 60 per cent of median income, Eurostat (n.d.) data suggest that, in 2005, poverty in United Kingdom was greater than in Bulgaria (19 versus 14 per cent), despite that fact that, once we adjust for differences in purchasing power parity, the poverty threshold in UK was more than four times greater than that of Bulgaria. Indeed, at 19 per cent, income poverty in the UK was as high as the average rate of the 12 “New” EU member states taken together.
The potential for relative income measures, such as the 60 per cent median income measure, to deliver seemingly counter-intuitive results when drawing comparisons between European nations has recently been the subject of significant scrutiny, especially in the context of EU enlargement since 2004 (see e.g. Kangas and Ritakallio, Reference Kangas and Ritakallio2004; Fahey, Reference Fahey2007; Whelan and Maitre, Reference Whelan and Maître2009 a, b, inter alia). Fahey (2007: 35) summarises the problem thus:
Following the recent eastern enlargement to the EU, the gap in living standards between the richest and poorest Member States has greatly widened, so much so that what is defined as the poverty threshold in the richest Member States would count as an above-average income in the poorest Member States, and the “poor” in some states have higher living standards than the well-off in other states.
Traditional relative income measures of poverty have typically been seen as the source of such counter-intuitive comparisons. It has been argued that, in response, analysts might supplement these with other measures, such as an income poverty measure set as a proportion of median European income (e.g. Fahey, Reference Fahey2007); a material deprivation measure, invariant across Europe, which captures respondents who are unable to afford a series of items such as three meals a day, new and not second hand clothes, and regular leisure activities (e.g. unicef, 2012); or a joint approach which focuses on the experience of both national relative income poverty and an EU-wide measure of material deprivation (e.g. Nolan and Whelan, Reference Nolan and Whelan2011).
This approach to measuring poverty using deprivation indicators—either on their own (Mack and Lansley, Reference Mack and Lansley1985) or, as the most recent UK Poverty and Social Exclusion survey intends, in combination with a low income indicator (e.g. PSE website; see also Nolan and Whelan, Reference Nolan and Whelan1996)—has become an important strand of analysis in the Townsendian tradition. It has also gained traction politically, with a deprivation measure comprising one of the three measures used to frame the European Union’s official poverty target to remove 20 million people from poverty and exclusion by 2020, as well as in the UK, where a joint low income and material deprivation measure was one of the four measures of child poverty enshrined in legislation in the Child Poverty Act 2010 (Kennedy, Reference Kennedy2014).
However, while comparisons of the kind outlined above—that is, between Britain and Bulgaria—might grate with our intuitions about the extent of poverty in these countries, they are not a by-product of a flawed measurement approach, but are, instead, entirely compatible with the Townsendian conceptualisation of poverty. As we argue below, from a Townsendian perspective, there is no reason why poverty, understood, to employ Nolan and Whelan’s (1996: 2) re-worded definition, as “exclusion from the life of the society owing to a lack of resources” would not be greater in Britain than Bulgaria—or even than in Burkina Faso!
The counter-intuitive international comparisons which have been highlighted by a number of authors are, thus, not measurement problems. The problem lies in the conceptualisation of poverty within the Townsendian framework: such examples may be counter-intuitive, but—importantly—they are not counter-conceptual. However, the conceptualisation of poverty remains a neglected aspect of the multidimensional turn currently being experienced within poverty analysis, and is at risk of being overlooked. Given the growing awareness of the counter-intuitive results which flow from relative income poverty analyses, it seems to be an opportune time to reconsider the conceptualisation of poverty itself.
This paper is comprised of four sections. First, we present an outline and critique of the Townsendian concept of poverty, demonstrating how this conceptualisation is prone to producing surprising and counter-intuitive results. In the second section, we outline an alternative, capability-inspired framework for conceptualising poverty and deprivation which we argue is more persuasive in terms of understanding poverty and deprivation in different nations and over time. Third, we discuss some implications of this framework in terms of the measurement of poverty with reference to a number of recent European studies. The conclusion summarises the preceding arguments.
Poverty as Relative Deprivation
In Poverty in the United Kingdom, Peter Townsend pioneered the use of deprivation indicators as a means of tapping into the prevailing style of living in the UK, with respondents’ deprivation scores being used to calibrate an income poverty line. Townsend’s influential definition of poverty was:
Individuals, families and groups in the population can be said to be in poverty when they lack the resources to obtain the types of diet, participate in the activities and have the living conditions and amenities which are customary, or at least widely encouraged or approved, in the societies to which they belong. Their resources are so seriously below those commanded by the average individual or family that they are, in effect, excluded from ordinary living patterns, customs and activities (Townsend, Reference Townsend1979: 31).
Townsend juxtaposed his own “relative deprivation” approach to poverty with previous absolutist, subsistence approaches—and, in particular, with the work of Seebohm Rowntree (1901; 1941; Rowntree and Lavers, Reference Rowntree and Lavers1951).
Rowntree attempted to set a primary poverty line at the level of resources required to achieve “merely physical efficiency”. Thus, the focus of his first study was on “bare subsistence rather than living… [t]he dietary I selected was more economical and less attractive than was given to paupers in workhouses” (Rowntree, Reference Rowntree1941: 102). This “absolute” conception of poverty related to wholly asocial living: “not a farthing was allowed in the course of the whole year for anything beyond mere physical needs” (1941: 451). However, while Rowntree uprated his primary poverty standard between his first and final surveys, he had, Townsend claimed, failed to provide a convincing account for why this uprating had occurred and how it could be consistent with a subsistence conception of poverty (see Townsend, Reference Townsend1970: 13; 1962: 215-218).
Townsend argued that people’s needs were both physical and social; that there could be no successful attempt to divide physical and social needs, since seemingly “basic” needs, for example for nutrition, were met in ways which served social functions. He provided the example of a cup of tea which, while nutritionally worthless, performed an important social role; offering tea to visitors made “a small contribution […] towards maintaining the threads of social relationships” (1979: 50). The concept of poverty should therefore, Townsend argued, focus on the full range of needs, both physical and social, insisting that the style of living to which the concept of poverty would relate should be participation (1979: 88) and not merely subsistence (1979: 33).
This was the relativist conception and, on the point of relativity, Townsend made two distinct, but inter-related arguments: first, that absolute needs, determined without reference to prevailing norms, did not exist and thus could not serve as the basis for an absolute conception of poverty (1962: 218-9; 1970: 2); and second, that the norms to which a poverty standard should relate ought to be determined with reference to the “societies in which they belong” (1979: 31, see also 1970: 2; 1954: 134). The distinction between these two positions is, as we shall see, of some importance.
Townsend’s conceptualisation of poverty was centred around the key concepts of “style of living”, on the one hand, and “resources”, on the other (1979: 33). A person’s style of living was understood as “the set of customs and activities which they are expected to share or in which they are expected to join” (1979: 54), and thus explicitly included aspects of social living. Furthermore, the customs which individuals and families were expected to participate in would evolve over time as a society became richer. For example, changes in social patterns whereby families would live in two or three bedroom, instead of one bedroom, houses, or the emergence in the twentieth century of a tradition of taking an annual holiday away from one’s home, raised the level of expectations which individuals and families were required to meet (1979: 52). This imposed new resource demands on households. And, for Townsend, the concept of poverty should relate to circumstances where individuals’ and families’ total resources were insufficient to allow such participation in established customs and consumption patterns in society (1979: 88; 1962: 225).
Rather than assigning a cost to each item on his deprivation index (i.e. Rowntree’s method), Townsend explored the relationship between resources and deprivation. Townsend’s hypothesis was that a threshold could be identified: “as resources for any individual or family are diminished, there is a point at which there occurs a sudden withdrawal from participation in the customs and activities sanctioned by the culture” (Townsend, Reference Townsend1979: 57). He claimed that the evidence for such a threshold was suggestive rather than clear-cut, but could be identified at approximately 150 per cent of the Supplementary Benefit scale (1979: 261).
One consequence of seeking to set a poverty line at a meaningful level, reflecting participation in the customary style of living of society, is that the poverty standard may vary as a percentage of relative income over time. So, while Townsend noted that the “threshold” he identified was set at 150 per cent of the Supplementary Benefit scale, there is nothing in his theory to suggest that this percentage would not rise, or fall, if he had repeated the survey ten years later. This is, of course, rather demanding on the social scientist, and particularly on analysts reliant on secondary data, who might not have the wealth of deprivation information available in dedicated poverty surveys such as Townsend’s. To that extent, it is perhaps not surprising that many analysts adopt arbitrary income thresholds, such as 60 per cent of median income, as a non-ideal solution to the problem of a variable ideal threshold. Indeed, this approach can be justified by reference to Townsend’s own writings:
Certainly no standard of sufficiency could be revised [over time] only to take account of changes in prices, for that would be to ignore changes in the goods and services consumed as well as new obligations and expectations placed on members of the community. Lacking an alternative criterion, the best assumption would be to relate sufficiency to the average rise in real incomes (1979: 52-53, emphasis added).
At least in terms of the trend over time, then, the 60 per cent median income threshold provides a rough proxy of the Townsendian concept of poverty.
The central aspects of the Townsendian conception of poverty are, therefore, the following: (i) the concept of poverty is itself based on the concepts of resources and styles of living; (ii) the style of living on which the concept of poverty is based should encompass all needs, including social needs, and should adopt a participation and not a subsistence standard; (iii) the relationship between resources and style of living will need to be examined empirically in order to identify a threshold; but (iv) in absence of detailed information about styles of living, we may expect the poverty line, expressed as a threshold of resources, to rise in line with average or median income over time. Relativity in terms of style of living was thus taken to imply relativity in terms of resources.
Problems with the Townsendian framework
One important problem with the Townsendian framework is that it treats the rising tide of “style of living” as being of precisely zero importance in itself, representing merely the “compliance costs” of participation, as Niemietz has recently argued (2011: 148). One important contribution of Townsend’s framework was in demonstrating the social nature of needs, and in arguing that the concept of poverty cannot be divorced from changes in those needs over time and place. However, in emphasising the rising tide of styles of living as being of no intrinsic importance, contributing only to the resource demands on households, the consequence is that an absolute increase in wealth, where this occurs without any change in the distribution of resources, will make no difference whatsoever in terms of the level of poverty in the Townsendian schema, and poverty rates are, on the contrary, driven almost entirely by the distribution of resources within a society, irrespective of the prevailing level of wealth.
On this point, the Townsendian schema seems to get things wrong, twice. First, since poverty is, in effect, determined by the distribution of resources alone, there may be more poverty in rich countries or cities than poorer ones—more, for example, in Britain than in Burkina Faso; more in Manchester than in Manila. This helps to explain why analyses operating within a Townsendian schema can produce some rather counter-intuitive examples, such as the comparison between Bulgaria and Britain discussed above.
Secondly, however, even when the Townsendian framework offers an answer that intuition suggests is correct (i.e. that there is more poverty in Burkina Faso than Britain; more in Manila than Manchester), it appears to offer the wrong reason for this result. If poverty in Burkina Faso is greater than in Britain, then under the Townsendian conception this is not because the standard or style of living is lower (since this would be, in effect, discounted) but because resources are more unequally distributed.
Direct approaches to conceptualising and measuring poverty
While Townsend himself presented a resource-centric conceptualisation of poverty, others who extended and developed his ideas attempted to conceptualise and measure poverty directly, using indicators of material deprivation (see especially Mack and Lansley, Reference Mack and Lansley1985), as we discuss below. Townsend’s classic study contained two sets of deprivation indicators: a full list of sixty indicators measuring “styles of living” across numerous dimensions, and a summary index of twelve items drawn from this longer list, with the latter used to calibrate his poverty line. As we have argued elsewhere (Hick, Reference Hick2012), Townsend’s summary deprivation index was in no way representative of the breadth of the full list, omitting many non-material dimensions of deprivation such as conditions at work, health, education, environmental deprivation, fuel, lighting, and clothing. The result was a unidimensional measure of material deprivation, focusing on marketable commodities and activities, which has become extremely influential in the literature (see Mack and Lansley, Reference Mack and Lansley1985; Nolan and Whelan, Reference Nolan and Whelan1996, inter alia).
At times, and in order to reflect the widespread belief that poverty is about “more than money”, this direct approach to measuring poverty is accompanied by a wider concern for social exclusion or multiple deprivation. However, these concepts remain underspecified within the literature on poverty and, most importantly, are typically treated as being a non-essential part of analysis—too often overlooked in favour of the narrower, unidimensional concept of material deprivation (Hick, Reference Hick2012). This narrower, unidimensional focus is problematic because it fails to capture the many ways in which people’s lives can be impoverished.
We have argued in this section that there are a number of problems with the conceptualisation of poverty in the Townsendian schema. The question is, then, whether there is a framework which can overcome some of these problems.
Poverty as Capability Deprivation
The purpose of this section is to outline an alternative, capability-based framework built around the two concepts of material poverty and multiple deprivation. In Poverty as Capability Deprivation, the framework (i) provides for a multidimensional assessment, (ii) focussing on both monetary and non-monetary dimensions and constraints, and (iii) the relationship between living standards and resources is subject to empirical scrutiny, and is not prescribed by theory.
Poverty as Capability Deprivation prioritises a focus on ends over means, and since the ways in which human lives can be impoverished are inherently plural, analysis will need to take on a multidimensional form, focussing on the deprivation of people’s core capabilities (the specification of which we discuss below). Indeed, from a capability perspective, the understanding of impoverishment itself is distinctive, for the approach suggests that it is the impoverishment of our capabilities which should be of concern, and these may extend beyond dimensions which are immediately related to material resources.
For Lister (Reference Lister2004), this shows that the capability approach is doing something quite different to a poverty approach since not all capability deprivation arises because of a lack of resources and, thus, would typically not be considered to be a case of poverty. One reason why the capability and poverty literatures have often arrived at a somewhat different answer to the question of how to understand poverty is because the question they ask is distinct: a poverty perspective typically asks “what is poverty?”, while the question the capability approach asks is closer to “what matters?”. The first question is conceptual, while the latter is ethical, and the distinction between these questions can, we suggest, shed light on the relationship between the capability and sociological approaches to understanding poverty.
This reflects, we suggest, a permanent tension between the conceptual and ethical aspects of poverty analysis. This tension arises because, while a lack of resources is typically considered to be at the core of the concept of poverty, the ethical importance of any particular dimension is not derived from its responsiveness to resources, but by an appeal to the living standard directly. This may take a number of forms. For Townsend, there was an ethical appeal to human needs (e.g. 1970; 18); for Nolan and Whelan (2011: 33), more recently, to “problems that people would avoid if they could”. But human needs and problems which all would wish to avoid inevitably extend to both monetary and non-monetary dimensions, which imposes a requirement to adopt a framework that can consider both.
Furthermore, if a particular dimension, such as the ability to participate in the life of society, matters, then a lack of resources cannot be the only constraint of interest, for there exists an ethical distinction between choice and constraint (Le Grand, Reference Le Grand1991; Piachaud, Reference Piachaud1981; Sen, Reference Sen2009: 237) that does not exist between one constraint and another (see also Hick, Reference Hick2012). There is, by now, reasonable agreement that non-participation in the life of society as a result of choice should not be considered to be as evidence of poverty or deprivation (see, inter alia, Nolan and Whelan, Reference Nolan and Whelan2011: 32; Piachaud, Reference Piachaud1981). But the “remainder” after subtracting choice is not a lack of resources. There may be other constraints which impede participation such as disability or discrimination and, while these are conceptually distinct from a lack of resources, they cannot be seen as being of no importance.
This tension is problematic because, while the delineation of the concept of poverty is often based on conceptual considerations (e.g. focussing only on resource-based dimensions and constraints), there is a desire to use the concept to make ethical statements. But the centrality of resources—which influences both the dimensions of interest and the constraints considered—is justified in conceptual and not ethical terms, and, from an ethical perspective, too much is ignored by the focus on material poverty which results.
Our focus, according to the capability approach, should be on what people can do or be; on impoverished lives and not on depleted wallets (Sen, Reference Sen2000). This approach to assessment reflects the primarily ethical orientation of the capability approach, which is more concerned with establishing “what matters” than with deciding whether and how this ethical terrain is to be divided up into constituent concepts. This latter point is important, for it implies that a narrower concept of poverty, centred around a lack of resources, can be retained if the broader additional terrain is taken up by another concept—for example, that of deprivation. In this case the “what we study” is therefore delineated by ethical concerns (what matters?) while the division of this terrain into concepts such as poverty and deprivation can be based on conceptual distinctions (i.e. poverty is about a lack of resources; deprivation considers a wider range of constraints). According to this view, two concepts are required in order to respond to the conceptual and ethical concerns which typically motivate poverty analysis, and a full assessment, capable of supporting ethical claims, would require a focus on both poverty and deprivation. In this sense, the division of this terrain between the two concepts of poverty and deprivation implies that an approach adopting a “narrow” focus on material poverty requires consideration of the broader concept of multiple deprivation (see Hick, Reference Hick2012, for a discussion).
The third area in which Poverty as Capability Deprivation is distinctive to the Townsendian framework is on the question of the relationship between resources and capabilities (or resources and styles of living in Townsendian terminology). In Poverty as Capability Deprivation, the relationship between resources and capabilities is one for empirical analysis, and not one to be specified by theory. Furthermore, since the relevant capabilities are multidimensional, the relationship between resources and capabilities is likely to vary from dimension to dimension. This implies that the uses of income are likely to be quite different—for example, it may be that reductions in severe malnutrition and stunting are largely determined by rising levels of absolute wealth, while participation in the life of society, for example, might be entirely relative in the space of resources.
Material poverty and multiple deprivation
The framework proposed here thus employs the two concepts of material poverty and multiple deprivation. It is necessary at this point to say something about the division between poverty and deprivation. Although the claim that poverty is about material dimensions while deprivation considers non-monetary dimensions and constraints appears to have a certain intuitive appeal, it is, we suggest, something of an oversimplification. Existing research has shown that the relationship between resources and dimensions which we expect to be related to resources (e.g. material deprivation) is often much weaker than typically assumed; conversely, other dimensions we define as multiple deprivation (e.g. subjective well-being, ill-health) are related to resources, albeit in a weaker fashion (Hick, forthcoming a). So attempting to distinguish between the resource-based and non-resource-based dimensions is likely to be far from straightforward in practice. With this caveat in mind, then, what do we mean by poverty and multiple deprivation?
Poverty can be defined as inadequate material living standards arising from a lack of resources. These living standards are understood as particular capabilities—the ability to afford participation in the life of society, the ability to avoid going into debt for ordinary living expenses, the ability to afford a basket of essential goods, the ability to afford to be adequately nourished, and so forth. This definition of poverty follows the “narrow” approach to conceptualising poverty, with a lack of resources at its core. However, while it bears a deliberate resemblance to Nolan and Whelan’s redefinition of the Townsendian concept, it shifts the emphasis from “exclusion from the life of the society” to “inadequate material living standards”, recognising that the former might be only one of the capabilities which comprise the latter. This is because, even amongst the “material” dimensions, there is no reason to assume that there is one dimension which, to the exclusion of all others, deserves priority in our understanding of material poverty. In this way, the capability approach would be consistent with multidimensionality even within the concept of material poverty.
The measurement of material poverty: fixed and moving parts
Moving beyond the issue of dimensionality and the broader assessment demanded by the capability approach, we may ask whether, in operational terms, the measurement of material poverty would, under Poverty as Capability Deprivation, proceed in the same fashion as the existing, Townsendian tradition. Here, we focus explicitly on one dimension—the ability to take part in the life of society, which Sen has suggested is likely to be one of the important dimensions for analysis (Sen, Reference Sen, Dominic and Nead1995: 15).
In understanding the relationship between measurement indicators and an underlying concept, we introduce a distinction between the fixed parts and moving parts of poverty measurement. According to this view, the fixed part is the relevant capability—e.g. the ability to participate in the life of society, and the moving parts are the indicator(s) that are required for measurement to be aligned with this conceptualisation. As the relationship between these moving parts and the fixed part (i.e. the meaning of the underlying capability) changes, it may be necessary to re-specify the moving parts—for example, by amending the income poverty threshold or selecting alternative deprivation items.
In practice, while it is straightforward to automatically up-rate an income poverty threshold with median income over time, amending the moving parts proves more difficult for deprivation indicator approaches. Authors working within the Townsendian tradition have, in dedicated surveys of poverty in the UK, sought to update and augment the list of deprivation items over time in order to reflect changes in socially perceived necessities over time (e.g. Mack and Lansley, Reference Mack and Lansley1985; Gordon and Pantazis, Reference Gordon and Pantazis1997; Gordon et al., Reference Gordon, Mack, Lansley, Main, Nandy, Patsios and Pomati2000; Reference Gordon, Mack, Lansley, Main, Nandy, Patsios and Pomati2013). However, this is typically not possible where analysts are reliant on secondary data since most social surveys collect information about a fixed set of deprivation items over time or in different countries.
This is problematic because, while in theory the fixed part should be the meaning of the underlying concept and the indicators should be the moving part, in practice it is often the indicators which are fixed, though they are the moving part, while the meaning will thus be variable, whereas it should be fixed. There is a parallel here with the concern expressed in the comparative social policy literature about the need to ensure measurement equivalence when conducting comparative analyses. Drawing on Marsh’s distinction between formal and functional equivalence, Kennett (2006: 293) notes that “using identical formal procedures when comparing different societies may produce functionally non-equivalent meanings”. If deprivation indicators are to be used to capture an underlying concept of poverty with a certain meaning, then we must be sensitive to circumstances in which different indicators (as moving parts) may be required in order to ensure that the meaning of the indicators (the fixed part) is the same across time and place.
A second critique of the use of deprivation indicators in measuring poverty comes from Berthoud and Bryan (2011: 137), who argue that while deprivation indicators may be of use in calibrating an income poverty threshold (i.e. Townsend’s method), “no index can support the strong set of assumptions required to treat it as a direct measure of poverty”. In essence, Berthoud and Bryan’s argument is that a short, summary index is not sufficient to act as a measure of poverty. At the very least, this critique poses a significant challenge to those who work with such indicators to ensure that the moving parts (the indicators) are both necessary and sufficient to capture the meaning of the underlying concept that they are intended to represent.
Distinguishing between the fixed and moving parts of poverty measurement, then, requires two tasks: first, deciding on the fixed part and, second, aligning the fixed and moving parts. The first task is thus to specify the meaning of the capability itself. Having established this, and taking exclusion from the life of society as being a relevant example here, the challenge is to identify an approach to measurement in terms of the moving parts (i.e. the indicators) which, at any point in time, is both necessary and sufficient to capture this underlying concept, and which will be amended over time and space as the relationship between the indicators and the relevant capability varies.
Conceptualising multiple deprivation
The concept of multiple deprivation is defined here as the enforced experience of low living standards. This multidimensional approach to conceptualising deprivation as being more than just material deprivation recognises that “the lives of human beings can be blighted and impoverished in quite different ways” (Anand and Sen, Reference Anand and Sen1997: 5). This presents an ethical requirement to consider both monetary and non-monetary dimensions and constraints, even if we wish to accommodate these within two concepts of poverty and deprivation. How, then, should we decide on the dimensionality of multiple deprivation? To answer this question, we turn to John Rawls’ (Reference Rawls1971; Reference Rawls1988) argument in favour of a resource-centric perspective of advantage and disadvantage.
Primary goods and primary goals
In articulating the claims of the capability approach vis-à-vis income-centric analysis, Sen often refers explicitly to the work of John Rawls’ (e.g. Reference Rawls1971). Rawls maintained that direct measures of living standards such as functionings and capabilities could not serve as the basis for understanding advantage and disadvantage because of the variability in the conceptions of the goods which people hold (1988: 255-256). The state is unable to prioritise one or other of these conceptions because such plurality of ends implies that any attempt to do so will not be “affirmed by citizens generally” and thus their pursuit “gives the state a sectarian character” (p. 256). Rawls’ response to this problem was to limit his own focus to what he called primary goods, defined as “rights and liberties, opportunities and powers, income and wealth” (1971: 92). Such primary goods are “things which it is supposed a rational man wants whatever else he wants” (1971: 92), because “whatever one’s system of ends, primary goods are necessary means” (1971: 93). Assessing disadvantage in the space of primary goods allows the analyst to be neutral towards the competing conceptions of the goods that people hold. In terms of the analysis of poverty, this would be consistent with a resource-centric measurement approach.
However, there is another way to respond to the “fact of pluralism” other than to retreat to primary goods (or from “ends” to “means”). To focus on ends themselves will only violate neutrality between the ends which people hold (the “fact of pluralism”, Rawls, Reference Rawls1988: 259) if there do not exist “primary goals”—ends which each person shares, whatever their conception of the good and whatever else they value.
We suggest that—at least at a sufficient level of generality—it is possible to find certain shared goals that people value, whatever else they value (see also Drydyk, Reference Drydyk2011; Brandolini and d’Alessio, Reference Brandolini and D’Alessio1998: 7, inter alia). On the question of the dimensionality of the quality of life, for example, the Stiglitz-Sen-Fitoussi Commission (2009: 58) noted that “[w]hile the precise list of these features inevitably rests on value judgements, there is a consensus that quality of life depends on people’s health and education, their everyday activities (which include the right to a decent job and housing), their participation in the political process, the social and natural environment in which they live, and the factors shaping their personal and economic security”.
If such primary goals exist, then the fact of pluralism does not necessitate a retreat from the ends to the means of living and it is these primary goals, we suggest, which provide the appropriate basis for a conception of multiple deprivation. This enables us to focus on capabilities for which each person is assumed to prefer more rather than less—at least over a certain range of achievement (Atkinson et al., Reference Atkinson and Cantillon2002: 21).
It is worth considering the various aspects of this definition. By capabilities, we mean ends which a person is unable to achieve, rather than simply those they do not achieve. In practical terms, the study of poverty and deprivation typically starts with a focus on an actual lack of achievement in a particular domain. In order to give information about lack of achievement a “clear and accepted normative interpretation” (Atkinson et al., Reference Atkinson and Cantillon2002: 21) it is necessary that either (i) there are explicit questions about whether the outcome arose because of choice or constraint (see also Nolan and Whelan, Reference Nolan and Whelan2011: 20-30), or (ii) the threshold is set at a sufficiently low level so that the level would be widely interpreted as representing constraint and not choice (i.e. it is implausible to suggest that the lack of achievement arose by choice).
There are a number of ways by which universality might be ascertained, one being common “necessities” surveys, which ask respondents whether a set of items are considered to be necessities or not. Most necessities surveys—at least within the poverty literature—focus on the components or indicators of particular dimensions, but there is, of course, no particular reason why such an approach could not be used for identifying the extent of agreement about the importance of particular dimensions or capabilities themselves.
Indeed, existing research demonstrates that, when given the opportunity, the public endorse a conception of necessity which is not limited to resource-based items or dimensions. For example, an Australian survey found that being “treated with respect by other people” and having “streets that are safe to walk in at night” were considered to be necessities by 98.5 and 97.7 per cent of the population, respectively (Saunders and Abe, Reference Saunders and Abe2010: 84). Similarly, the most recent Poverty and Social Exclusion survey in the UK found that “personal time for adults away from family responsibilities” (Fahmy et al., Reference Fahmy and Pemberton2011: 8) was one of the necessities of life in twenty-first century Britain. The resource-centric view of what is deemed to be a necessity is thus not held by the public themselves, but is imposed by analysts seeking to limit the analysis of poverty to unmet “material” needs.
Claims about the universality of particular dimensions or capabilities do not themselves imply that each person prefers to achieve these goals using the same means. For example, it seems obvious to suggest that there is more variation at the level of type of diet (i.e. whether one prefers apples or oranges, eats red meat or is a vegetarian) than in the goals to which such satisfiers are directed; for example, being adequately nourished. Doyal and Gough (1991: 155) note in this vein that “while the basic individual needs for physical health and autonomy are universal, many goods and services required to satisfy these needs are culturally variable”.
If a person or group does not value a particular capability—the fixed part—then the framework is to that extent non-ideal because it assumes individuals or groups value what they, in fact, do not. If a person does not value one of the indicators used to tap into the particular dimension then it may be that the wrong “moving part” has been selected for that group, and that another can be found. For example, research has suggested that typical deprivation items are, for many older people, “simply not consistent with lifestyle changes that occurred in old age” (Legard et al., Reference Legard, Gray and Margaret2008: 18; see also McKay, Reference McKay2004; Hick, Reference Hick2013, on the challenges of using deprivation indicators to measure material poverty across the life course). It may be expected that identifying the valued capabilities is more straightforward than identifying valued “moving” parts, or indicators, used to tap into these capabilities.
Taken together, what these conditions ultimately require is for the dimensions selected to be able to receive widespread acceptance—for the lack of achievement in particular dimensions to be understood in an unambiguous, normative fashion (Atkinson et al., Reference Atkinson and Cantillon2002) as outcomes which “we could reasonably expect most people to wish to avoid if possible” (Nolan and Whelan, Reference Nolan and Whelan2011: 32). These primary goals, or “core capabilities”, which all people value can, it is argued, meet this requirement.
Towards identifying dimensionality
If the focus of the concepts of poverty and deprivation is just this shared core of capabilities, these concepts must, as far as possible encompass all of this core. It is widely recognised that any multidimensional approach should be as comprehensive as possible in terms of the dimensions considered (e.g. Atkinson et al., Reference Atkinson and Cantillon2002: 24; Townsend, Reference Townsend1987: 140); in practice, that assessment considers “the darker side of the entire lifestyle of a people” (Townsend, Reference Townsend1987: 129).
Based on the aforementioned discussion and principles, which dimensions should analysis focus on in the rich-nation context considered here? This list in inevitably partial since further analysis might be undertaken in making such selection—for example, survey data could be used to further validate the dimensions selected here, or empirical analysis could be employed in determining dimensionality to avoid double-counting and so forth.
Nonetheless, it seems reasonable to suggest that assessment would include a focus on the following capabilities: the ability to participate in the life of society, to live in adequate housing, to avoid premature mortality and live a normal life span, to be well nourished, to avoid involuntary migration, to avoid ill-health and poor mental health, to avoid alcoholism and drug dependence, to avail of educational opportunities (at least at certain levels), to avoid discrimination and humiliation, to have autonomy, to be satisfied with life, to avoid crime and the fear of crime, to not experience unemployment. In each case, we assume that––at least up to a certain point––everybody prefers more rather than less of these things and thus that they can be interpreted in normative terms.
However, this list is intended to be illustrative: debate about the relevant dimensions is central to the literature on poverty analysis. It is more significant for our purposes here is to that we specify the importance of focusing on a comprehensive range of dimensions which match the definition of core capabilities we have outlined above.
On perfection and the burden of measurement
One final point on the question of consensus and universality relates to perfection. How far is the framework presented here weakened by departures from consensus and universality? Rather than considering attempts to reach consensus as either successful or not (i.e. as binary), it is worth considering all frameworks as specifying an ideal, to which actual empirical implementation approximates with some degree of distance. Thus, while the ideal aim is universality, it is to be expected that approaching universality will, in practice, be the relevant goal, as a non-ideal departure from universality itself. It is important, as Bradshaw et al. (Reference Bradshaw and Hoelscher2007) counsel, that perfection does not become the enemy of the good. In practice, the requirement of assessment is not, we suggest, perfection, but the ability to produce reliable and valid answers to a finite set of questions (such as “how many people are in poverty?”, “who is most at risk of poverty?”, “is poverty rising or falling?”, and so forth). The significance of departures from the ideal depends on their impact on the answers to such questions rather than intrinsically limiting the approach itself.
It may be legitimately asked whether the framework presented here imposes a severe burden on measurement and, of course, this is one of the central charges made against the capability approach (e.g. Sugden, Reference Sugden1993). It is certainly the case that as a framework becomes more complex, in particular by moving beyond a unidimensional approach, this entails a trade-off in terms of ease of measurement (Wolff and de-Shalit, Reference Whelan and Nolan2007: 21; see also Robeyns, Reference Robeyns2005: 32). A number of points may be made on the extent of the measurement burden imposed by the proposed framework. First, any multidimensional approach to measuring poverty involves a measurement burden—these are not, in the main, special problems for the capability approach.Footnote 1 Second, one particular burden imposed by any multidimensional poverty framework is that it may require the inclusion of dimensions not currently captured by social surveys. However, identifying the dimensions that we cannot currently satisfactorily measure can itself act as the first step in rectifying the situation (Robeyns, Reference Robeyns2003: 71). This is usefully illustrated by the work of Alkire and colleagues in identifying, and subsequently capturing, “missing dimensions” in multidimensional poverty analysis (Alkire, Reference Alkire2007).
Finally, Nolan and Whelan (2007: 147) claim that even if poverty is understood to be a multidimensional concept, this does not in itself mean that one cannot adopt a unidimensional measurement approach. Given the well-known weak association between different dimensions of poverty (e.g. Bradshaw and Finch, Reference Bradshaw and Finch2003; Hick, forthcoming b), we should perhaps not be too optimistic that a unidimensional measurement approach can ultimately capture the multidimensionality of poverty. Nonetheless, it may be possible to identify non-ideal proxies for forms of deprivation which are difficult to measure or which are not captured in many surveys. However, these would need to be identified empirically rather than chosen on the basis of assumed relationships and face validity. Any retreat to such proxies would, of course, need to be treated as a non-ideal deviation from the framework presented here.
Implications
In this penultimate section, we explore some implications of the framework outlined here in terms of conceptualising and measuring poverty over space and time, noting how Poverty as Capability Deprivation is likely to provide a somewhat different approach to analysis than Poverty as Relative Deprivation.
A number of recent studies have pointed to certain counter-intuitive examples that emerge when drawing pan-European comparisons using relative income approaches. Relying on a relative income measure of, say, 60 per cent of median income, may suggest that, for example, Ireland has a higher rate of poverty than Estonia (Whelan and Maître, Reference Whelan and Maître2009), or that the rate of poverty in Italy is twice that of the Czech Republic (Nolan and Whelan, Reference Nolan and Whelan2011: 57), or that child poverty in the UK is greater than in Hungary (unicef, 2012: 12). As we have noted, one important feature of most existing critiques is that they typically view these counter-intuitive results as problems of measurement—as an “artefact of the measure of poverty employed” (Whelan and Maître’s, Reference Whelan and Maître2009a: 76)—arising from the relative income poverty measure.
Poverty as Capability Deprivation takes a different approach. If the conclusion that Ireland has more poverty than Estonia or that Britain is has more poor than Bulgaria appears perverse, it is surely because “exclusion from the life of the society” is but one of the dimensions that we are be interested in. It is not that participation is trivial, just that it is not all that is important (see also Ravallion and Chen, Reference Ravallion and Chen2009). Other intrinsically important dimensions––such as avoiding housing deprivation and homelessness, being in reasonable health and achieving a full life span, etc.,––matter too, and deserve greater recognition in our analysis. Thus, and in line with one of the central themes of the capability approach, the informational base of much analysis conducted within the Townsendian tradition is simply too narrow. So the multidimensional focus provided by the capability approach is one reason why it may be a preferable approach when analysing poverty in different nations and, indeed, in one nation over time.
However, the approach may also recommend a distinctive approach, within any particular dimension, to the measurement of poverty. Contrary to Poverty as Relative Deprivation, there is no insistence that what is understood as an adequate level of capability should be determined solely by societal or national norms. Indeed, the reference point for the adequate level of capability is likely to vary from dimension to dimension, and should be determined by the meaning intended by the dimensions themselves.
This is not to suggest that poverty is an “absolute” concept—absolute in the sense of being drawn without reference to prevailing customs and norms. It is, rather, to note that Townsend’s critique of absolute approaches contained two claims: first, that an absolute concept of poverty, which did not rely on social customs and norms, was impossible; and, secondly, that the norms which should form the basis of the concept of poverty were those of one’s society or nation-state. The two claims are, of course, distinct and acceptance of the former does not imply the validity of the latter.
In response to the perceived limitations of state-bound income-centric measures of poverty, Fahey (Reference Fahey2007) has recommended that these be supplemented by an EU-wide income poverty measure, such as 60 per cent of median EU income. Such an approach produces patterns of poverty which are more intuitive, displays a closer association with a range of forms of multiple deprivation than standard approaches, and can, Fahey suggests, represent a “convergence indicator”, reflecting the EU’s competence in promoting convergence between EU regions. Nonetheless, this approach also has its shortcomings: unless calculated over a fixed set of Member States, in the event of further EU enlargement it may register a rise in poverty simply because median EU income would fall (Atkinson et al., Reference Atkinson, Cantillon, Marlier and Nolan2005: 114). More fundamentally, perhaps, it is not immediately clear what capability this might be said to represent in terms of what people, in a tangible sense, are able to do and be. In contrast, an anchored poverty line, uprated in line with prices, has a fairly straightforward interpretation in terms of the ability to afford a basket of goods over time.
An alternative response to the perceived deficiencies of relative income measures of poverty has been to rely to a greater extent on material deprivation measures. In terms of the pan-European analysis, one solution has been to employ a fixed deprivation index in order to measure poverty across Europe––either on its own (e.g. Guio et al., Reference Guio, Gordon and Marlier2012), supplementing a relative income measure (e.g. unicef, 2012), or as part of a combined low income-material deprivation measure (Nolan and Whelan, Reference Nolan and Whelan2011). Again, this may have the desirable characteristic of producing more plausible results (considerably more poverty found in Eastern Europe than in Western Europe), but it makes conceptual concessions that are not always acknowledged. As Berthoud’s (2012: 22) recent empirical analysis of poverty in twenty-six European countries finds: since the deprivation indicators are fixed, “objective deprivation can be interpreted as an indicator of absolute poverty”—i.e. the approach that Townsend’s work had sought to banish.
In terms of both of these strategies, and as argued above, adopting a capability perspective would require us to be clear about what capability—what being or doing—these measures are intended to capture, and to ensure that this falls within the concept of core capabilities outlined above. Clarifying the nature of the capability which one is seeking to capture is likely to determine, in turn, the decision about whether a state-bounded conception or European reference group is to be preferred (on the question of reference points and poverty analysis, see also Kangas and Ritakallio, Reference Kangas and Ritakallio2004). It is difficult to circumscribe, as a matter of theory, one particular reference point for each of the important dimensions. However, it seems reasonable to suggest that for dimensions which themselves relate explicitly to one’s society or to participation—for example, the ability to participate in the life of society; the ability to appear in public without shame, etc.—national reference groups are preferable. This is required in order to maintain the within-society or within-nation meaning of these capabilities.
For other dimensions where there is no explicit within-nation focus—for example, being able to achieve a full life span, being healthy, being adequately housed, and so forth—the thresholds might be selected with reference to the best achieving countries within an analytic comparison group. Here, a regional comparison, such as Europeanisation of reference groups, might be more appropriate, since there is little to suggest that understandings of a “normal” life span or “adequate” housing and so forth refer solely, or indeed primarily, to the societal or national norms. Thus, the inevitability of one’s own nation-state or society being the reference point would seem only to hold for dimensions which themselves explicitly relate to within-society comparisons. However, the crucial point is that the choice of reference points is based on the meaning intended by the capability itself.
Having clarified what capability a dimension was intended to capture, the aim of measurement, as we have discussed above, would be to ensure that the indicator(s) reflected this capability at each time and place (in the parlance adopted here, that it would be “fixed”). The existing literature on poverty often emphasises that deprivation indicators represent necessities, and evidence to this effect is often marshalled to overcome the charge of arbitrariness. However, as MacCárthaigh (Reference MacCàrthaigh2014) has recently argued, all uses of the language of necessity point —explicitly or implicitly—to some end state. The relevant question is: necessary for what?
Thus, adopting a capability perspective is distinctive in (at least) two ways. First, while there is a trend towards multidimensional analysis, the capability approach views such analysis not as desirable, but as positively necessary. Second, within any particular dimension, it requires us to be explicit about what capability—what being and doing—we are attempting to capture, and to give consideration to how the measurement approach may need to be amended over time or space in order to preserve its alignment with the intended meaning.
In this paper, we have provided a critique of the dominant framework for understanding poverty, Poverty as Relative Deprivation, and have outlined an alternative, Poverty as Capability Deprivation, which can overcome some of the limitations of the Townsendian framework. Poverty analysis is undergoing a multidimensional turn, due, in part, to a growing awareness of the limitations of relative income approaches. However, we have argued in this paper that the counter-intuitive findings which flow from cross-national relative income poverty analysis are not, in fact, artefacts of the measurement approach but are instead entirely consistent with the Townsendian conception of poverty.
The response to such counter-intuitive comparisons, we suggest, should be to adopt a multidimensional conception of poverty and deprivation, understood in this paper as being the deprivation of people’s core capabilities. Despite there being a plurality of ends which people hold, there are nonetheless, we have argued, certain ends which each person values, and it is these which form the basis for the framework presented here. Poverty as Capability Deprivation focuses on ends (people’s capabilities) rather than means (their resources) and its dimensionality is circumscribed by the core capabilities which all people are assumed to value. In this framework, multidimensionality is not merely desirable––it is necessary––and analyses which seek to be comprehensive will therefore need to consider both material poverty and multiple deprivation.
In this framework, poverty is defined as inadequate living standards arising from a lack of resources. Deprivation is defined as the enforced experience of low living standards. The division between “material” and “non-material” dimensions in this way allows the framework to preserve continuity with existing approaches and the widespread intuitive acceptance of a lack of resources as being central to the concept of poverty. At the same time, it emphasises the necessity of focusing on a broader terrain, taken up by the concept of multiple deprivation. However, we have suggested that, even with the concept of (material) poverty, there is no reason to assume there is one capability which, to the exclusion of all others, demands our attention, and have cautioned that distinguishing between material poverty and multiple deprivation may not be straightforward in practical terms.
When it comes to the measurement of particular dimensions, we have introduced a distinction between the fixed and moving parts of poverty measurement as a means of thinking about the relationship between measurement indicators and the underlying conception of any given dimension. This distinction implies two tasks: first, clarifying the capability which each dimension is seeking to capture (specifying the fixed part) and, second, selecting indicators which are necessary and sufficient to capture these capabilities at any particular place and time.
Taken together, then, it may be that Britain has a greater proportion of its population excluded from the life of society than Burkina Faso, but it is also a nation with substantially higher life expectancy, literacy, and so forth. It is not that participation is trivial, just that it is not all that is important. It is multidimensionality which is the key to understanding both why poor nations have considerably more poverty than rich nations and why poverty remains a problem even in the richest nations.
Ultimately, the challenge for any framework is to provide valid answers to a finite series of questions, such as: “is poverty rising or falling?”, “which groups experience the most poverty?”, “which nations have the most poverty?”, and so forth. In adopting a multidimensional perspective that focuses on people’s capabilities, or ends, rather than their resources, or means, it is argued that Poverty as Capability Deprivation can offer the basis for such an assessment, and that it can provide a conceptual underpinning to the multidimensional turn currently being experienced in poverty analysis.
Acknowledgements
The author thanks Tania Burchardt, David Piachaud, Tony Atkinson and Björn Halleröd for their comments and suggestions. Any remaining errors are my own. This work was supported by the Economic and Social Research Council [grant number ES/G01808X/1].