Welfare Comparisons with Multidimensional Well-Being Indicators: An Indian Illustration

Abstract

Problem of making welfare comparisons between populations with multidimensional discrete well-being indicators are well known. Application of a weighting scheme remains a convenient alternative for aggregating across dimensions but not without the limitation of subjectivity in the principle of weighting. Further dichotomous well-being indicators pose another complexity in comparison as regard counting ‘how many’ and ‘which ones’ at the same time. This paper attempts a welfare comparison of population where only ordinal information is available at the micro level in terms of multi-dimensional discrete well-being indicators. This does not involve any assumption either regarding strength of preference for each dimension or regarding the desirability of changes between levels within or across dimensions or the complementarities/substitutability between the dimensions. To carry out such a comparison, we adopt the concept of multidimensional first order dominance that enables us to make comparison across time and between populations based on a series of binary ordinal welfare indicators. This concept is applied to the data on Household basic amenities obtained in the National Sample Survey rounds and comparison is made across Indian states. Such a comparison offers a contrast to the welfare comparison made in terms of the deprivation prevalence across dimensions as well as its temporal changes.