How serious is the neglect of intra-household inequality in multi-dimensional poverty indices?

Income-based as well as most existing multidimensional poverty indices (MPI) assume equal distribution within the household and thus are likely to lead to yield a biased assessment of individual poverty, and poverty by age or gender. In this paper we first show that the direction of the bias depends on how these measures use individual data to determine the poverty status of households, while the impact of these assumptions on inequality between individual cannot be determined a priori. We then use data from the 2012 Indian Human Development Survey to create a standard household-based MPIs closely related to the MPI proposed by Alkire and Santos (2014) as well as UNDP (2014), and compare that to an individual level MPI that individualizes education and nutrition and some aspects of the living standards dimensions. We find that the poverty rate of females is 14 percentage points higher than that of men in our individual MPI measure but only 2 percentage points higher when using the household-based measure. Similarly, the age differentials in poverty are much larger using the individual-based measure. Using a decomposable inequality measure, we find the contribution of intrahousehold inequality to the total inequality in the individual deprivation score inequality to be 30% and total inequality is also some 30% higher using the individual-based measure, while inequality among the poor is found to be 5% smaller using the individual measure.