Measuring Children’s Multidimensional Poverty under Constraints: An Empirical Exploration in Punjab, Pakistan

There is widespread concern that household measures of poverty, even if disaggregated by age cohorts, are not able to depict child conditions, and should be supplemented by individual child measures. There are normative reasons to prefer the most precise measure possible and practicable methods to do so. In parallel, because of the cost of data collection, and the implicit requirement that policy actors confidently master multiple indices, there is an empirical question of how results vary between these measures – e.g. for assessing which districts are the poorest. Using 2017–18 data for the Province of Punjab in Pakistan, this paper compares a household multidimensional poverty index (Proxy MPIP) that proxies the official MPI of Pakistan, with two others: 1) the child disaggregation of that same Proxy MPIP, and 2) an individual-level Child MPI, which is constructed at the level of the individual child and adds a fourth dimension of childhood conditions, with age-appropriate indicators such as nutrition, schooling, child labour, and child marriage. The analysis compares multidimensional poverty levels and indicator composition for all three measures across the districts of Punjab. Results show that although naturally the indicator information in the individual child measure is far richer, in this dataset, district-wise rankings are highly robust across the three multidimensional poverty measurement approaches.  Further empirical assessments from different datasets and using different specifications are required to assess the generality of this finding.

Citation: Ul Haq, R. and Alkire, S. (2025). ‘Measuring children’s multidimensional poverty under constraints: An empirical exploration in Punjab, Pakistan’, OPHI Research in Progress 69a, Oxford Poverty and Human Development Initiative (OPHI), University of Oxford.