Measures of Deprivation in Europe and their Influence
Townsend (1979) conducted an early seminal study using a counting approach to poverty in the United Kingdom, analysing a 1968-9 survey covering about 2000 households in Britain. To assess the magnitude of ‘relative deprivation', Townsend defined sixty indicators covering twelve dimensions: diet, clothing, fuel and light, home amenities, housing conditions and facilities, the immediate environment of the home, conditions at work, family support, recreation, education, health, and social relations.
Each indicator was equally weighted, although the number of indicators within each dimension varied greatly. For ‘illustrative purposes', he then focused on a shorter list of twelve items covering major aspects of dietary, household, familial, recreational, and social deprivation. Townsend used a minimum score of five (out of the twelve) ‘as suggestive of deprivation' (p. 252). In other words, a poverty cutoff of five out of twelve was chosen to identify the poor.14 He did not use a union criterion because he recognized the potential problems: ‘No single item by itself, or pair of items by themselves, can be regarded as symptomatic of general deprivation. People are idiosyncratic and will indulge in certain luxuries and apply certain prohibitions for religious, moral, educational or other reasons, whether they are rich or poor' (p. 252).15 However, he actually did not use this counting approach to analyse poverty. Rather, he explored the correlation between deprivation scores and household income (adjusted for household size) in order to derive an income threshold below which people are ‘disproportionately deprived' (p. 255). In other words, he used a direct approach to ‘validate' the poverty line to be used in the indirect income poverty measure.Townsend's study inspired much subsequent work on poverty and social exclusion in Europe and, in particular, another benchmark study on poverty: Mack and Lansley's Poor Britain (1985).
This study was also influenced by Sen's writings on the direct approach to poverty measurement (Sen 1981). A novelty of this study was that the list of items considered as necessities was, for the first time, constructed using a survey of the public's perceptions of minimum needs (PSE 1983) known as Breadline Britain survey. That is why their method has been called the ‘consensual or perceived deprivation approach to measuring poverty'. Of the original thirty-five items, they retained the twenty-six that were considered necessities by strictly more than 50% of the population. The surveyintensity into one summary measure, which, like the Adjusted Headcount Ratio (M0), can be broken down by dimension.
14 Townsend used the terms ‘deprived' and ‘poor' interchangeably, whereas we define these terms differently.
15 Note however that he did not use the term ‘union' to refer to this criterion. usefully distinguished people who lacked an item because they could not afford it from those for whom it was a voluntary choice.[128] The authors identified as poor those who could not afford three or more of the equally weighted items (p. 178).[129] This poverty cutoff was selected after analysing the association between the number of deprivations, income levels, and spending patterns. Mack and Lansley proposed that a lack of three or more necessities was a matter of force rather than choice.[130] ‘Very few of the better-off lack this level of necessities. And nearly all those who lack this level of necessities cut back on non-necessities, a majority cutting back substantially' (p. 176). In addition to their benchmark cutoff, the authors reported degrees of deprivation using two additional poverty cutoffs: ‘Broadly speaking, those who cannot afford five or more necessities are sinking deeper into poverty; and those who cannot afford seven or more necessities are in intense poverty' (p.
184). British authors continue working along these lines, with new surveys in 1990, 1999, and 2012. Breadline-Britain-type surveys were also replicated elsewhere in Europe.[131] It also inspired the structure of two much-used datasets: the European Community Household Panel survey (ECHP) and the European Union Statistics on Income and Living Conditions (EU-SILC).[132]Gordon et al. (2000) compare the 1983, 1990, and 1999 Breadline Britain surveys in terms of the items considered as necessities and assess the evolution in poverty levels. Using an updated list of thirty-five items to evaluate poverty, they identified a household as poor if they could not afford two or more items and, additionally, had relatively low incomes.[133] The report also constructed a measure of child poverty using a list of twenty-seven socially perceived necessities for children. They used a poverty cutoff of one or more and another cutoff of two or more. In both cases the poverty cutoff was set using discriminant function analysis.[134] Note that the poverty cutoff selected using discriminant analysis, which is a data-driven approach, may provide different conclusions when applied to different datasets, making comparison across time difficult (section 3.4.4). Because the poverty cutoff is not normatively considered or justified, in contrast to other measures reviewed, there is no link to ethical assessments of poverty.
Building upon the work of Mack and Lansley (1985) and Ringen (1987, 1988), Callan, Nolan, and Whelan (1993) also proposed to identify the poor by combining both resource and deprivation measures.[135] They used data from a household survey conducted in Ireland by the Economic and Social Research Institute (ESRI) in 1987, which used Mack and Lansley's (1985) format. Starting from a list of twenty-four items, the authors used factor analysis to observe possible indicator clusterings and accordingly used three dimensions: (1) basic lifestyle (eight items such as food and clothes), (2) housing and durables (seven items related to housing quality and facilities), and (3) ‘other' aspects of lifestyle (nine items such as social participation, leisure activities, and having a car or telephone).
People's perceptions regarding the necessity of indicators restricted their material deprivation index to the eight-item basic lifestyle dimension.[136] They identified as poor anyone who both lacked one or more of the eight items and fell below the relative income poverty line, set at 60% of the average equivalent disposable income in the sample.[137] This work sparked a series of surveys and studies to monitor poverty in Ireland using variations on this combined method of resources and material deprivation. These were used to build a ‘consistent measure of poverty'[138] which identifies a person as poor if she is both income poor and deprived in some minimum set of deprivations.Muffels et al. (1992) built upon Muffels and Vriens (1991) and designed an index of relative and subjective deprivation using a Dutch socioeconomic panel survey inspired by Mack and Lansley (1985). Their innovation was to use household weights and poverty cutoffs. They first constructed an (objective) deprivation score for every head of household as the weighted sum of deprivation in each of a large set of items related to living conditions. The weight for each item varied across households and represented the respondent's perceived importance of the item, compared to the perceived importance of the item by the reference group of the household head (p. 195). They selected a subjective poverty cutoff, termed the ‘subjective deprivation poverty line', using an econometric model. In the model, the dependent variable was the respondent's subjective assessment of whether he/she was poor or not on a scale of one to ten. The explanatory variables were their deprivation score plus control variables such as income, age of the household head, family status, and financial stress factors. Using the estimated coefficients, the ‘subjective deprivation poverty line’ was calculated for each household separately as the deprivation score that would produce a subjective assessment score of 5.5.[139] Then they explored the degree of overlap between those identified as poor using the subjective deprivation poverty line and those identified as ‘insecure’ using three alternative income poverty lines (a subjective line, equal to the minimum income reported by households as ‘the minimum income they need to acquire a certain minimum standard of welfare’; a ‘national social minimum income standard’; and a ‘European statistical minimum income standard’).
They examined the relationship between both measures in terms of bivariate distributions using contingency tables and regression analysis and found significant mismatches. They concluded that a multi-method approach combining income and (direct) deprivation measures was needed to assess poverty.Hallerod (1994, 1995) used data from the Swedish standard of living survey in 1992, which also followed Mack and Lansley (1985). A key difference was that they retained all thirty-six originally included items but weighted them by the proportion of the population that regarded each as necessary. Weights were adjusted by certain groups to reflect significant differences in preferences.[140] The index was labelled the Proportional Deprivation Index (PDI). The author selected a poverty cutoff that produced the same headcount ratio as the Consensual Poverty Line (CPL). The CPL was an interpersonally comparable income level at which, on average, respondents in different circumstances would subjectively indicate that their current income was just sufficient for them to make ends meet. While both methods identified nearly 21.3% of the population as poor, only 8.8% of the population were identified as poor by both. Acknowledging that both the income and direct methods may be subject to substantial measurement problems, the author advocated the use of a combination of both methods and defined those 8.8% of the population who were poor by both CPL and PDI as the ‘truly poor’. Subsequently, Hallerod et al. (2006) used a variant of the PDI to compare poverty levels in Britain, Finland, and Sweden.
Using ECHP data, Layte et al. (2001) constructed a material deprivation index from thirteen items to assess the relationship between (relative) income poverty and material deprivation. For each country, they weighted each item by the proportion of households possessing that item, and they defined the poverty cutoff of the deprivation index endogenously as the threshold which generated a headcount ratio equal to that of the (relative) income poverty line of the country.
They performed this exercise for different relative poverty lines: at 40%, 50%, and 60% of the median income in each country. Their results showed that the overlap between the two poverty measures was very limited and thus supported a method that combines both measures. Whelan, Layte, and Maitre (2004), using the ECHP to identify persistently poor persons, found a similar mismatch, as mentioned in section 1.2.1.Eurostat (2002) constructed an index of non-monetary poverty (pauvrete d’existence) for European countries. Following the analysis of Whelan et al. (2001) of the first ECHP survey, a list of twenty-four dichotomous items (‘having'/‘not having') available in that survey were grouped into five dimensions using factor analysis.[141] For each individual, a deprivation score per dimension was obtained as the weighted sum of deprivations in the indicators of that dimension, where the weight attached to an indicator was inversely related to the deprivation rate in that indicator in the corresponding country. Then, the dimensional deprivation scores were also aggregated by taking a weighted sum, where the dimensional weight attached to a dimension was proportional to the weighted average of the coefficients of variation among that dimension's indicators (pp. 155-6). People with a deprivation score of 60% or more were considered poor.[142]
Additional implementations of the counting approach to identifying the poor in Europe included studies of poverty in Sweden (Erikson 1993), the reports on poverty in Belgium by Vranken and other authors (Vranken 2002), and recent work on the search for a relative deprivation index for Europe (Guio 2005, 2009; Guio and Maquet 2006; Decanq et al. 2013). In 2011, the European Commission implemented an ‘EU-2020' multidimensional poverty measure using union identification across three indicators: relative income poverty, severe material deprivation, and quasi-joblessness. This landmark measure identified those ‘at risk of poverty and social exclusion' in order to set and monitor a poverty reduction target for 2020. It represents the most high-profile policy application to date—hence, perhaps, the most closely scrutinized.
Nolan and Whelan's book Poverty and Deprivation in Europe offers a systematic conceptual and empirical study of ‘why and how non-monetary indicators of deprivation can play a significant role in complementing (not replacing) income in order to capture the reality of poverty in Europe' (2011: 1). It is thus relevant to this book at many points, as they too survey research on mismatches in identification between different indicators of poverty, by social group in one period and across time; scrutinize indicator design; apply robustness tests; consider the poverty cutoff; and propose ways of strengthening the EU-2020 Poverty Target. Maitre, Nolan, and Whelan (2013) offer a critical evaluation of the EU-2020 Target, and Whelan, Nolan, and Maitre (2014) explore the use of the AF method for the case of the European Union using EU-SILC data; they advocate the replacement of the current approach by the AF approach as it is more structured, less ad hoc, and more transparent, as well as being flexible in terms of the poverty cutoff and the axiomatic properties of its measures (see section 2.5 and Chapter 5). Alkire, Apablaza, and Jung (2014a) also apply an AF measure to EU-SILC data 2006-12 and explore the inclusion of social indicators.
Townsend (1979) and Mack and Lansley (1985) also influenced work outside Europe. For example, Mayer and Jencks (1989) severely criticized the income approach to poverty measurement in the United States based on a survey in Chicago on material hardship. They collected information on ten indicators covering dimensions of food, housing, and medical care. The number of hardships (equally weighted) were analysed alongside income and subjective satisfaction with living standard. They found that the family's income-to-needs ratio explained less than a quarter of the variation in the total number of hardships that families report.[143]
The consensual approach or socially perceived necessities to poverty measurement initiated by Mack and Lansley (1985) and its survey structure were replicated elsewhere. In particular, it served as a model for a Basic Necessities Survey (BNS) (Davies 1997; Davies and Smith 1998). The BNS method weights each item by the proportion of people who said it is a basic necessity. It suggests defining a poverty cutoff across the BSN score such that it identifies as poor the same proportion of people as those who have subjectively identified themselves as poor. Davies' BNS method was implemented in Vietnam and Mali (Nteziyaremye and MkNelly 2001), Bangladesh (Ahmed 2007), Ireland (Nolan and Whelan 1996), Japan (Abe 2006), Europe (Eurobarometer 2007), and South Africa (Wright 2008), among other countries.
4.3