Category Archives: In Progress

On Track or Not? Projecting the Global Multidimensional Poverty Index

In this paper we compute projections of global multidimensional poverty. We use recently published estimates of changes over time in multidimensional poverty for 75 countries, which are based on time-consistent indicators. We consider and evaluate different approaches to model the trajectories of countries’ achieved and future poverty reduction. Our preferred model respects theoretical bounds, is supported by empirical evidence, and ensures consistency of our main measure with its sub-indices. We apply this approach to examine whether countries will halve their poverty between 2015 and 2030 if observed trends continue. Our results suggest that if observed trends continue, 47 countries will have halved their poverty by 2030—irrespective of the underlying model. As the current COVID-19 pandemic may severely disrupt progress in poverty reduction, we also assess its potential impact using simulation techniques and evaluate the resulting setback. Our analyses suggest a setback to multidimensional poverty reduction of about 3–10 years.

Citation: Alkire, S., Nogales, R., Quinn, N. N., and Suppa, N. (2020). ‘On Track or Not? Projecting the Global Multidimensional Poverty Index,’ OPHI Research in Progress 58a, University of Oxford.

Multidimensional Poverty Reduction in India 2005/6–2015/16: Still a Long Way to Go but the Poorest are Catching Up

Following Amartya Sen’s pioneering ideas on poverty and inequality measurement, the development economics literature proposes diverse classes of measures as well as poverty orderings. Yet in the Sustainable Development Goals (SDGs), the headcount ratio is the primary statistic for measuring monetary and multidimensional poverty. Rigorously analysing the trends of multidimensional poverty for India between 2005/6 and 2015/16, we illustrate how the headcount ratio is not able to observe certain centrally important requirements of the SDGs – such as whether anyone is being left behind, or how deprivations are interlinked. We propose using the adjusted headcount ratio or Multidimensional Poverty Index (MPI) as the primary poverty measure for policy assessment, supplemented by the headcount ratio, intensity, number of poor, and composition of poverty, to provide more accurate analyses. Exploiting cross-sectional data comprising of more than three million individuals and a panel of 29 states and several socio-economic subgroups, we show empirically how the reduction of multidimensional poverty by 271 million unfolded within a decade. In contrast to earlier periods in time, we find that the poorest of the poor saw the largest reductions in multidimensional poverty due to falling levels of intensity – a feature the headcount ratio alone cannot portray. Despite the importance of the MPI we also recognise the inherent and enduring need to probe the headcount ratio and number of poor statistics. Hence we corroborate these stark findings with an assessment of the dominance of the distribution of attainment scores which establishes the relationship between MPI and H in both periods. To assess the robustness, 19 additional MPIs are constructed, having different indicator definitions and combinations, and it is found that in nearly all of these a greater number of persons left poverty.

Citation: Alkire, S., Oldiges, C. and Kanagaratnam, U. (2020). ‘Multidimensional poverty reduction in India 2005/6–2015/16: Still a long way to go but the poorest are catching up’, OPHI Research in Progress 54b, Oxford Poverty and Human Development Initiative, University of Oxford.

Changes over Time in the Global Multidimensional Poverty Index and Other Measures: Towards National Poverty Reports

This paper compares trends in multidimensional and monetary poverty systematically across developing regions. The trends in multidimensional poverty draw on the global Multidimensional Poverty Index (MPI) and related sub- and partial-indices in 80 countries and 647 subnational regions, covering roughly 5 billion people, for which there is a recent MPI estimation and comparable datasets for two time periods. This paper uses two main techniques to assess the pro-poorness of multidimensional poverty reduction and triangulate monetary and nonmonetary poverty measures. First, utilizing the properties of subgroup decomposability and dimensional breakdown, it examines changes in the MPIT and its consistent sub-indices over time across sub-national regions and urban–rural regions. The decomposition analysis identifies relevant national patterns, including those in which the pace of poverty reduction is higher for the poorest subgroups. Next, it assesses overall annualized changes in the incidence of multidimensional poverty, compares this with changes in $1.90 poverty trends, and evaluates the pace and direction of various international poverty lines for monetary poverty, with national monetary and multidimensional measures, and for the family of global MPIT measures. This extensive empirical analysis illustrates how to assess the extent and patterns of reduction of multidimensional poverty, as well as whether it is inclusive or whether some people or groups are left behind, and triangulates various poverty measures to evaluate the reliability and credibility of their purposes. Naturally, some further research questions emerge.

Online Appendix E: Eighty national poverty reports, triangulating monetary measures and the global MPI family of multidimensional poverty measures.

Citation: Alkire, S., F. Kovesdi, M. Pinilla-Roncancio and S. Scharlin-Pettee. (2020). ‘Changes over time in the global Multidimensional Poverty Index and other measures: Towards national poverty reports’, OPHI Research in Progress 57a, Oxford Poverty and Human Development Initiative, University of Oxford.

Revising the Global Multidimensional Poverty Index: Empirical Insights and Robustness

The global Multidimensional Poverty Index, published annually since 2010, captures acute multidimensional poverty in the developing regions of the world. In 2018, five of its ten indicators were revised with the purpose of aligning the index to the SDGs insofar as current data permit. This paper provides comprehensive analyses of the consequences of this revision from three perspectives. First, we offer new empirical insights available from the revised specification. Second, we analyse its robustness to changes in some key parameters, including the poverty cutoff and dimensional weights. Third, we compare the revised and the original specifications by implementing both on the same 105 national datasets. The country orderings in the revised specification are found to be robust to plausible parametric alternatives. Largely, these country orderings are at least as robust as the original one. Additional research on robustness standards is suggested.

Citation: Alkire, S., Kanagaratnam, U., Nogales, R. and Suppa, N. (2020). ‘Revising the global Multidimensional Poverty Index: Empirical insight and robustness’, OPHI Research in Progress 56a, Oxford Poverty and Human Development Initiative, University of Oxford.

Evaluation of Programs with Multiple Objectives: Multidimensional Methods and Empirical Application to Progresa in Mexico

Development programs and policy interventions frequently have multiple simultaneous objectives. Existing quantitative evaluation approaches fail to fully accommodate this multiplicity of objectives. In this paper we adapt the multidimensional poverty measurement approach developed by Alkire and Foster (2011) to the estimation of treatment effects for programs with multiple objectives. We use the potential outcomes framework to show that differences in Alkire-Foster indices between treated and control samples correspond to average treatment effects estimates of outcomes of interest under experimental conditions, and develop further methods of analysis to explore these multidimensional treatment effects. We discuss issues of index design encountered in practice and provide an illustrative example. We apply the methods developed to evaluate the conditional cash transfer program Progresa in Mexico, finding significant multidimensional effects of the program. Further analysis shows that these treatment effects are driven mainly by impacts on school attendance and health visits, objectives that correspond directly to the conditions of the program. There is no evidence for heterogeneity of the treatment effects by the extent to which the beneficiary failed to achieve the objectives at baseline. This study complements the extensive literature on the evaluation of Progresa and other development programs, comprising studies that focus on particular objectives or outcomes of the program. We hope that the methods developed here will find wide application to the evaluation of programs with multiple objectives.

Citation: Vaz, A., Malaeb, B. and Quinn, N.N. (2019). ‘Evaluation of programs with multiple objectives: Multidimensional methods and empirical application to Progresa in Mexico’, OPHI Research in Progress 55a, University of Oxford.

Multidimensional Poverty Reduction in India 2005/6–2015/16: Still a Long Way to Go but the Poorest Are Catching Up

This paper assesses the change in multidimensional poverty in India from 2005/6 to 2015/16 using data from the NFHS-3 and NFHS-4 surveys. Estimates of changes are disaggregated by age cohort, state and by socio-economic group-level, and broken down by indicator; sampling errors are considered throughout. Multidimensional poverty is defined using the global Multidimensional Poverty Index 2018 (Alkire and Jahan 2018).  The paper finds a very strong reduction, indeed a halving of the MPI during that decade. Furthermore, subnational patterns of poverty reduction are strongly pro-poor, whereas from 1998/9 to 2005/6 they had been regressive. The reductions of MPI are hardly correlated with state level growth in GDP, making this a rich terrain for future research. District level analyses in 2015/16 only document extensive ongoing intra and interstate variation. These explorations confirm that at the end of the decade under study, at least 271 million fewer persons were living in multidimensional poverty – a magnitude of change rivalling the numbers exiting monetary poverty in China.

Citation: Alkire, S., Oldiges, C. and Kanagaratnam, U. (2018). ‘Multidimensional poverty reduction in India 2005/6–2015/16: still a long way to go but the poorest are catching up’, OPHI Research in Progress 54a, University of Oxford.

Towards a Global Assets Indicator: Re-assessing the Assets Indicator in the Global Multidimensional Poverty Index

This paper explains the revision of the assets indicator of the updated global Multidimensional Poverty Index (global MPI), which was launched just before the 73rd Session of the United Nations General Assembly in September 2018. The joint decision of the United Nations Development Programme (UNDP) and the Oxford Poverty and Human Development Initiative (OPHI) to revise the global MPI in 2018 to align it with the Sustainable Development Goals and to best monitor progress towards “leaving no one behind” provided the opportunity to assess the statistical validities of the assets indicator contained in the Original MPI, jointly designed by OPHI and UNDP Human Development Report Office (HDRO) in 2010, and an assets indicator included in an Innovative MPI, which was developed by UNDP HDRO in 2014. Further, considering the improvements in many Demographic and Health Surveys, Multiple Indicators Cluster Surveys and selected national surveys in recent years, from which the global MPI is constructed, the revision also offered an occasion to assess whether the inclusion of additional assets would add value to a revised asset index for the updated global MPI 2018. Taking into account a blend of inputs, including statistical test results, public consultations, normative reasoning and substantive trial measures of possible asset indices as outlined in detail in this paper, the revised assets indicator maintained the structure of the Original MPI, but added computer and animal cart as additional items. Here we explain the reasons and delineate the many decisions that were taken along the way.

Citation: Vollmer, F. and Alkire, S. (2018). ‘Towards a global asset indicator: re-assessing the asset indicator in the global Multidimensional Poverty Index’, OPHI Research in Progress 53a, Oxford Poverty and Human Development Initiative, University of Oxford.

Download supplementary data (xlsx)

On Data Availability for Assessing Monetary and Multidimensional Poverty

Data availability plays a crucial role in the fight against poverty. Yet, it lags behind the data available on most other economic phenomena. We catalogue and review existing data availability aiming to break the cycle of outdated poverty data. We identify countries that generate and analyse frequent and accurate poverty data to highlight potential improvements. Results show, data for both monetary and multidimensional poverty dramatically increased since 1980. Sixty countries now produce annual datasets, while internationally comparable short surveys and regional harmonised variable definitions are being implemented. These existing resources and experiences can inform much-needed efforts to expand data availability.

Citation: Alkire, S. and Robson, M. (2018). ‘On data availability for assessing monetary and multidimensional poverty’, OPHI Research in Progress 52a, University of Oxford.

Statistical Note: Disaggregating Bhutan’s MPI 2017 by Disability Status

Since 2010, Bhutan has used a Multidimensional Poverty Index (MPI) alongside consumption poverty to measure and fight poverty in all its forms and dimensions. Bhutan’s National MPI was updated on 2012 and 2017 using the Bhutan Living Standards Survey (BLSS). In 2017, the BLSS questionnaire included questions on disability status. This statistical note shows different ways by which the MPI can be disaggregated using the available information. Each way is implemented, and the results analysed. Thus, by presenting worked out empirical examples, we hope to contribute to the evolving methodological discussions of how best to disaggregate poverty measures including the MPI by disability status. In addition, we hope to contribute to robust and detailed understanding in Bhutan of the relationship between poverty and disability status, hence to inform policies that seek to address both. However, survey data are limited, and so, very importantly, we also advise re-running these results with the 2017 census data for a more precise picture. It is hoped this note will provide some structure for a census-based analysis.

Citation: Pinilla-Roncancio, M. and Alkire, S. (2018). ‘Statistical note: disaggregating Bhutan’s MPI 2017 by disability status’, OPHI Research in Progress 51a, University of Oxford.

Incorporating Environmental and Natural Resources within Analyses of Multidimensional Poverty

How can multidimensional poverty measures – that currently encompass social and economic dimensions – be extended to include environmental deprivations that strike the poor simultaneously? And can such extended measures better inform effective and integrated policy responses? Research on joint Environmental and Natural Resources (ENR) and poverty issues is rich, and has contributed to bringing the poverty-environment nexus to the fore. Yet, no widely used multidimensional poverty measure identifies who and how the socio-economically poor people are affected by ENR-issues, at a large enough scale, and in ways that can respond to and inform public policies over the medium term. This paper sets out how such a measure could be built. In particular, it sets out how to include indicators of ENR deprivations into the profile of the joint deprivations people experience. These deprivation profiles could then be used to compute multidimensional measures using the Alkire Foster (AF) methodology, with the difference that these would now encompass a subset of pertinent ENR deprivations. The paper clarifies the ENR data requirements for developing and analysing such a measure empirically.

Citation: Thiry, G., Alkire, S. and Schleicher, J. (2018). ‘Incorporating environmental and natural resources within analyses of multidimensional poverty’. OPHI Research in Progress 50a, University of Oxford.