Measures of Unsatisfied Basic Needs in Latin America and Beyond

Latin America is the other region where a counting approach to identifying the poor has been widely implemented. Rather than a focus on social exclusion or ‘relative deprivation' as in Europe, in Latin America it was operationalized under the unsatisfied basic needs (UBN) approach.

The selection of census indicators was first performed by ECLAC with an empirical study using data from the 1980 census of Argentina. The study acknowledges that the census did not provide data on income or consumption nor on key health variables such as nutrition. However, the census provided data from all areas in the country and, importantly, with a useful level of disaggregation at smaller geographical entities. Within these constraints, three criteria guided the selection of indicators (INDEC 1984: 11):

1. the indicators represented the degree of failure to satisfy some specific group of basic needs;

2. these indicators were significantly associated with [income] poverty;

3. they were comparable across regions of the country so that poverty maps could be constructed.

In order to fulfil the second criterion, as part of the project, ECLAC undertook an empirical study using data from a 1980 survey of two urban areas of Argentina: the Greater Buenos Aires area and Goya (taken as representative of other urban areas).^[144] The aim was ‘to select the characteristics that not only represented some intrinsically important deprivation but were also sufficiently associated with situations of [income] poverty so as to represent the other [unmeasured] deprivations that constitute such situations' (INDEC 1984: 500).

1. households with more than three people per room (overcrowding);

2. households with precarious housing;

3. households with no kind of toilet;

4. households with children of school age (6-12 years old) not attending school;

5. households with four or more dependents per occupied member (high dependency ratio) and whose household head's education was at most second grade of primary education.

The union criterion was used: all members of any household with at least one unsatisfied basic need were considered poor. The intuition was that because very low deprivation cutoffs were used for each indicator, one sole deprivation seemed sufficient to signal poverty (Rio Group 2006: 110). However, the information was reported in different ways and using different cutoffs: (1) the proportion of households and people experiencing each UBN; (2) the proportion of households and people with one or more UBN; and also (3) the proportion of people with two or more and three or more UBNs.

The set of census indicators outlined by INDEC and CEPAL for Argentina was replicated by official statistical institutes in many Latin American countries: Bolivia, Chile, Colombia, Ecuador, Guatemala, Honduras, Nicaragua, Paraguay, Peru, Uruguay, and Venezuela. While there was some variation in indicators, the dimensions considered remained essentially the same, as they were limited by the information contained in the countries' censuses.

1. access to housing that met minimum housing standards;

2. access to basic services that guarantee minimum sanitary conditions;

3. access to basic education;

4. economic capacity to achieve minimum consumption levels.

In all these countries, the UBN methodology was used to construct detailed and disaggregated poverty maps using census data. Poverty maps became a valuable tool for policy design (Kaztman 1996: 24). Coady, Grosh, and Hoddinott (2004) observe that poverty mapping has been widely used for geographical targeting purposes—and not only in Latin America. ‘Much of the history of poverty mapping has used a “basic needs” approach with poverty defined in terms of access to basic services. The simplest form of geographic targeting involves the use of a single variable such as nutritional status.... the choice of variables is largely guided by a combination of philosophy and data availability' (Coady, Grosh, and Hoddinott 2004: 63). In other cases, such as in Argentina or Chile, maps were constructed using the proportion of people with different numbers of UBNs. Poverty maps have guided investments in infrastructure, implementation of public works programmes and social funds, subsidized services, and the allocation of conditional cash transfer (CCT) programmes (usually alongside a complementary targeting method).^[145] While some countries have built consumption-based poverty maps (Elbers, Lanjouw, and Lanjouw 2002), these have been less common than the basic needs maps, and their policy interpretation is more challenging.^[146]

Beyond the policy impact that basic needs poverty maps have, it is worth noting that, while in Europe the direct method of measuring poverty was implemented alongside an effort to collect new data that would (a) reveal socially perceived necessities and (b) distinguish whether the lack of items was enforced or by choice, in Latin America the direct method was restricted to the data available at the time (census data).

Table 4.1 The UBN poor and the income poor

range of indicators that could be included was severely constrained. The direct method in Latin America did not seek to reflect people's views of their own necessities, and it did not deliberately permit ‘choice'. Relatedly, it must also be noted that in Latin America the definition of the UBN indicators was done from an absolute poverty perspective, whereas in Europe it was justified with respect to a relative or perceptual concept of poverty. Despite these differences, a strong common feature emerged: the interest in crossing the direct method with the indirect one. This gave rise in Latin America to the ‘integrated method' to measure poverty proposed by Beccaria and Minujin (1985) and Kaztman (1989). The indirect or income method was applied using data from household surveys, which started to be progressively implemented in the late 1980s and 1990s in the region. An absolute income poverty line approach was applied using the cost of basic needs method (Altimir 1979). The idea of the integrated method was to identify four sets of people: (1) the income and UBN poor, (2) the UBN poor but income non-poor, (3) the income poor but UBN non-poor, and (4) the non-poor by any method, as expressed in Table 4.1. This could be done using data from household surveys, which collected information on the UBN indicators as well as on income.

Kaztman (1989) terms the first group ‘chronically poor', not because of information on poverty over time but because he assumes that insufficient income coupled with at least one UBN (most had more than one) would reproduce poverty over time. This group would be equivalent to the ‘consistently poor' in the European literature.

As Boltvinik (1991) argued, these studies showed that the two methods, income poverty and UBN poverty, were (unintentionally) complementary, identifying to a great extent different slices of the population. Evidence from Montevideo (Uruguay) in 1984 and from Greater Buenos Aires (Argentina) in 1976 suggested that 10-15% of the households were in the ‘recently poor' category, 4-9% of the households were in the ‘with inertial deprivations' category, and only 7% were poor by both methods, i.e. in the chronically poor category. In Peru nearly 40% of the population were identified as both income and UBN poor (chronically poor). Yet 30% of the population were identified as either income poor or UBN poor but not both, which shows the ‘mismatches' between the two methods, covering nearly the entire population.

Boltvinik (1992) proposed an ‘Improved Integrated Method to Measure Poverty', which involved changes in each method separately, as well as in their combination. His method was applied in Mexico (Boltvinik 1995, 1996). He proposed first that UBN indicators be those associated with public investment, which could not be purchased individually, and dimensions that could be purchased using private resources should be considered in the income component.^[148] Second, a higher poverty line based on a more comprehensive basket of goods and services was generated. Third, he incorporated gaps in the measure­ment of income poverty and also of UBN. Thus, rather than dichotomizing achievements in each of the UBN indicators, he proposed (controversially) computing normalized deprivation gaps for each indicator as if they had cardinal data. Fourth, he allowed deprivation gaps to take negative values (reflecting achievements above the deprivation cutoff) thus permitting substitutability across deprived and non-deprived items. Finally he normalized the gaps to vary between minus one and one.

To identify the poor, Boltvinik suggested using three alternative poverty cutoffs. Boltvinik also discussed alternative methods for weighting the UBN indicators: (a) equal weights, (b) the complement of indicators' deprivation rates (Desai and Shah 1988), and (c) a combination of monetary and time valuations^[150] of each need. As in the case of the UBN index, negative gaps were allowed for income. These were normalized to range between minus one and one by dividing them by (the absolute value of) a normative maximum negative gap and replacing them by minus one whenever the absolute value of the negative gap was higher than the maximum normative gap. On the other hand, each person would have an individual UBN score which would be the weighted sum of the normalized deprivation gaps (ranging from minus one to one). In the combined method, each UBN indicator would be weighted by the proportion of the total cost required to fulfil each set of needs, and an individual's UBN score would be added to her income poverty score.

Boltviniks revised integrated method to measure poverty was altogether different from the integrated method outlined in Table 4.1 and is no longer a counting measure. It ceased to consider mismatches between the UBN poor and the income poor. His identification method is not a counting approach but relied on a score obtained as the weighted sum of the aggregated gaps of cardinalized ordinal data (permitting substitutability). The reasons for the ‘Improved Integrated Method to Measure Poverty' not acquiring popularity seem to be (a) that it required a number of controversial estimations, such as those related to time use and monetary valuations of UBN indicators, (b) that it attached a cardinal meaning to categories of response in ordinal variables and thus the intensity in the UBN index was dependent on the particular cardinalization used (which again could be contested), and (c) some steps such as the cardinalization of ordinal data and the consideration of negative gaps prevented the resulting measure from satisfying many desirable axiomatic properties outlined in section 2.5.^[151] Overall, in trying to accomplish too much, the method lost the public intuition and policy relevance that characterizes the counting approach and direct method of poverty measurement. That being said, many important distinctions were considered in this development, including the importance of time poverty.

The UBN approach has also been used in other parts of the world. In the Arab Region, Lebanon pioneered the UBN approach in 1997, using eleven indicators in four dimensions from the 1994-5 population and household survey (a mini census covering 10% of the population). The document MappingLiving Conditions in Lebanon was published in 1998 by the Lebanese Ministry of Social Affairs (MoSA) and UNDP, and was used to define poverty after the civil war in the absence of other data. This report, which mapped Lebanon's six governorates and twenty-six districts, was updated using 2004 survey data (UNDP and MoSA 2007), and an expanded index was published together with monetary poverty measures from the same survey in 2009. Iraq's Ministry of Planning together with UNDP also completed a significant three-volume study Mapping of Deprivation and Living Conditions in Iraq using 2003 data (UNDP and MPDC 2006). The study was used for budget allocation and policy priorities. In 2011 the same partners published a second study using 2007 data. A seven-country study was also produced using data from the Pan Arab Family and Health (PAPFAM) Surveys, which covered seventeen indicators grouped into five dimensions: education, health, housing, home necessities, and economic conditions (League of Arab States et al. 2009). Jordan published a similar two-volume study using 2010 data (first volume) and 2002-10 data (second volume). Other studies (ESCWA-AUDI) have covered particular topics such as an urban deprivation index in Tripoli, Lebanon (Nehmeh 2013).

The counting approach to identifying the poor has also been used in a new generation of poverty measures with renewed interest being shown in the direct method that uses solid aggregation methodologies based on axiomatic frameworks analogous to those which gave rise to the advances in income poverty measurement in the 1970s and 1980s (Alkire and Santos 2014). The following sections review some of these.

4.3.1 COUNTING APPROACHES IN MEASURES OF CHILD POVERTY

A counting approach to identifying the poor is a natural approach for various policy-oriented measures of child poverty. Understanding child poverty is widely agreed to require a multidimensional approach in both European and developing contexts (Trani, Biggeri, and Mauro 2013; Boyden and Bourdillon 2012; Minujin and Nandy 2012; Gardiner and Evans 2011). A pioneering and internationally comparable measure of child poverty in developing countries was computed in 2003 (Gordon et al. 2001,2003; UNICEF 2004), whose indicators and cutoffs reflect the Convention of the Rights of the Child. A number of studies have more recently measured and analysed child poverty using the AF method, including Alkire and Roche (2012), Apablaza and Yalonetzky (2011), Roche (2013), Trani et al. (2013), de Neubourg et al. (2012), and Dickerson and Popli (2013). In particular, it is worth highlighting that de Neubourg et al. (2012) is a step-by-step guide to implementing the Multiple Overlapping Deprivation Analysis (MODA) tool, developed at UNICEF's Research Office for global child comparisons adapting the M₀ measure of the AF method.^[152]

4.4