Page 327 - Contributed Paper Session (CPS) - Volume 6

CPS1949 Khalid S.

Measure of multidimensional poverty Robustness
of indices to weighting schemes
Khalid Soudi
High Commission of Planning, Rabat, Morocco

Abstract
This paper analyzes the robustness of multidimensional poverty indexes by
considering various weighting schemes. In particular, it establishes
comparison between three approaches for measuring multidimensional
poverty, namely: fuzzy logic method, Alkire and Foster (AF) method and
Bourguignon & Chakravarty method. By combining two approaches: A
quantitative approach of factor analysis and a qualitative one taking into
account the perception of deprivation by the poor, this paper develops an
empirical approach to determine dimensions of multidimensional poverty. By
referring to the analytical frame of capabilities as developed by A. Sen, a list
of central and basic functioning was chosen. The results show that all poverty
indices estimated by using the AF weighting scheme are significantly higher
than those estimated using the other pre-determined normative weights. The
stochastic dominance of the curves of poverty allowed us to confirm the
robustness of these results. Moreover, if the items of deprivation are not
structured by dimension, the results show, whatever selected weighting
schemes, an overestimation of the indices of multidimensional poverty, in
particular those obtained through the AF weighting scheme. This pattern was
consistent for all weighting schemes and impacts both the headcount of
poverty as well as the multidimensional poverty index (MPI). The results of
Bourguinon and Chakravarty method show that, whatever is the adopted
weighting scheme, the indices of multidimensional poverty are lower than
those obtained from the AF approach. The differences become more
important when adopting the weighting scheme suggested by AF. Besides, the
differences observed become more pronounced if we don’t structure the items
of deprivation by dimension. Also, the multidimensional index of poverty
obtained according to fuzzy logic method not only shows an important
sensibility to the weighting design but also it remains widely lower than those
obtained according to the approach A.F.

Keywords
weighting schemes; multidimensional poverty; robustness of indices.

1. Introduction
In recent years, despite the many methods available for measuring
multidimensional poverty, some important methodological problems remain
316 | I S I W S C 2 0 1 9