Measuring multidimensional poverty in latin america.
Fernandez, Ana Laura (1); Beccaria, Luis (2) (2018). 'Measuring multidimensional poverty in Latin America.' Paper presented at the annual conference of the HDCA, Buenos Aires, Argentina 2018.
In Latin American countries the income (or poverty line) approach has been widely used to identify poor households. However, growing efforts are being made to measure poverty from a multidimensional point of view, directly evaluating a wide range of needs of households and their members.
The multidimensional perspective is conceptually associated to the Human Development paradigm and the poverty definition in terms of Sen’s approach. This point of view broadens the income-centered definitions of poverty (mainly based on a utilitarian perspective). These refer to the combination of functionings and results associated to welfare and agency, focused on the possibilities available to each individual for promoting her own welfare.
Beyond the theoretical developments, different experiences have taken place so far, aiming at the construction of a poverty measure based on multiple dimensions related to Sen’s approach. One among these is the proposal of the United Nations Development Programme (UNDP) to complement measures of monetary poverty with a Multidimensional Poverty Index (MPI) that considers deprivation in different dimensions such as health, education and standard of living.
The Oxford Poverty & Human Development Initiative (OPHI) has also been involved in the estimation of the Global MPI and developed indicators associated to multiple dimensions of poverty (as well as promoted their use in different countries, including Latin America). Additionally, their research proposes the inclusion of other dimensions usually not included in the assessment of deprivation, such as social inclusion and psychological wellbeing.
The use of this kind of poverty measures –which are generally recognized as a step towards a broader analysis of deprivation beyond income standards- in Latin America still rises some questions, especially those related to the statistical information necessary for the construction of a multidimensional poverty measures.
Notwithstanding that, there is a long tradition of a sort of proto - multidimensional poverty indicators in the region as the Unmet Basic Need measures mainly estimated from Population Censuses.
More recently, official MDM measure appeared, as in Colombia or Mexico. The latter also includes the availability of economic resources as one of the dimensions. Some international agencies have made also their contributions, as the recent ECLAC measure.
In view of these previous studies, this article presents the first results of a research project aimed at contributing to the search of alternative measures of the process of human development, by proposing a multidimensional poverty measure that could be applied in different Latin American countries, based on public information drawn from official household surveys.
The hypotheses that guide this research are the following: i) the multidimensional approach for poverty measures allow a more comprehensive assessment of wellbeing, in comparison with income (or expenditure)-centered approaches; ii) it is possible to develop a multidimensional poverty measure that throws comparable data for different Latin American countries; iii) this measure can be estimated based on information that is already available and is produced by national statistics systems through household surveys (such as income and expenditure surveys, standard of living surveys, etc.).
The article starts with a review and a comparative analysis of the different attempts to provide multidimensional poverty measurements for Latin American countries. To do so, the specific objectives of each research will be accounted for, the different methodological approaches will be compared, and the variables and datasets used in for each country will be identified.
Then, considering recent developments for the region, as well as more general methodological advances in the field, a general methodology for a multidimensional poverty measure for Argentina will be presented. It will be based on public access information provided by official surveys. Alternative estimations reached with that methodology will be presented and an analysis to evaluate the sensitivity of results to changes in the selection of dimensions and variables, as well as in the definition of the variables, will be performed.
Since we are aiming at future application of this measure to other countries in the region, the selection of variables will focus on those which are comparable across different surveys and countries.
Methodologically, several issues should be tackled. The choice of dimensions and variables, the system of weights and the thresholds used for the identification of deprivation are complex matters. The resulting measures should prove to: i) consistently identify the poor households; ii) measure poverty and not another concept; iii) have an adequate number and choice of dimensions. In order to assess the quality of the methodology, previous literature suggests five criteria that should be able to establish the robustness of poverty measures: i) Validity of criterion (the selected indicators are correlated with results or causes of poverty) and of construction (the index measures poverty and not something else); ii) Reliability (consistency and repeatability of the index); iii) Additivity (the result is worse for households with deprivation in more dimensions); iv) Measurement invariance (the measure has equivalent outcomes for different groups). Additionally, the number of dimensions in which the different indicators are clustered, as well as the effects of the weights’ structure on the final estimation, should be analyzed. To assess the quality of the proposed methodology, it will be subjected to a range of tests that help to determine whether the resulting measure satisfies the above characteristics.