In this seminar, a comparative analysis of one and multidimensional poverty (lack of achievement) metrics for planning purposes is presented. Work focuses on Rural Water Supply and Sanitation (RWSS) services.

Alkire and Foster Method (AF) (www.ophi.org.uk, special issue of Social Indicators Research 112(2), 2013, on ‘A Multidimensional Approach: Poverty Measurement and Beyond’) has emerged as an alternative for poverty measurement in the international context, thanks to the recently disseminated UNDP Multidimensional Poverty Index. In welfare economics, poverty metrics are characterized by pre identifying the units contributing to the lack of achievement of the group, i.e. those individuals whose utility values fall below a certain threshold. The AF measures consider a multi-attribute identification function that includes the comparisons of the partial utilities with the corresponding thresholds. The dimension of the AF index corresponds to the number of lack of achievements considered for identification of poor units. New inference estimators and models have been developed for these measures (special issue of Econometric Reviews 32(1), 2012, on ‘Robustness in Multidimensional Well-Being Analysis’).

Investments in RWSS sector are mainly driven by measures as coverage, time to fetch water from improved water sources (or distance to them) and the amount of water provided per person by imposing the maximum number of served households per water point. Thus, the most basic decisions involve one or two related criteria.

Aggregated indicators as the Water Poverty Index are alternative frameworks that explicitly include multi-attribute nature of water - poverty relationships, beyond time to fetch water and consumption level. It is based on the fulfillment of five criteria, summarized by the keywords: Resources, Access, Use, Capacity and Environment. This framework is fed at different scales, with different indicators but sharing structure and problem understanding, and for different purposes (Gine, & Pérez-Foguet, 2010). Monitoring WASH sector is moving, within the post-2015 wave, from mapping basic infrastructure to wider water, sanitation and hygiene characterization (Gine & Perez-Foguet, 2013a, b). Even further, international monitoring mechanisms are looking for including components of the human right to water and sanitation, approach that has been used for an index construction elsewhere (Flores et al. 2013).

In this context, the usual measures and small area estimates for local administrative levels need  to be complemented with information that directly reflects households (and even people) characteristics. It is necessary to include the null valorization of their binary or categorized characteristics within the muti attribute index, instead of just considering utilities in terms of small area central estimators (as with proportions).

The AF approach offers an alternative to specifically work on those cases where there is an interest in characterizing low or even null utility situations. Other options can be devised, within the umbrella of unidimensional measurement, as modifying and adjusting the utility functions/poverty line thresholds, or even by using the multi-attribute utility theory instead of simple arithmetic or geometric means for index composition. However, as it is shown in this work, the analysis of the sensibilities of unidimensional thresholds with respect to multidimensional ones (i.e. between one and multi dimensional poverty lines) illustrate the limits of the former and justify looking for the later.

In this work, a Quality of Service Index at Water Points is used as a simple example for water provision measure to compare standard Poverty estimates on small area with Multidimensional Poverty metrics based on household data. The key points of Index modeling are briefly described, as well as data characteristics of two case studies (Kenya and Mozambique). Quality of service is characterized by time to fetch water, water consumption per capita, number of household’s members, secure/acceptable path to water point and usual provision carried out by children or adults.

Results show that AF approach is a good alternative for measuring muti-attribute utilities with null values. Even more, AF measures are essential when household data for service characterization must be included. Moreover, multidimensionality measuring offers new perspectives in terms of understanding the incidence and intensity of the lack of achievement.

Hopefully, the example can illustrate applications in other contexts; or even increase awareness of the audience on improving infrastructure’s services for the most deprived.