Multidimensional and Persistent Poverty: Methodological Approaches to Measurement Issues

Multidimensional deprivation and persistent poverty are important research areas within the poverty measurement literature. Still, both encompass measurement issues for which methodological solutions are yet to be analysed. The thesis that I present here analyses three specific measurement issues, identified as relevant within these research areas, and proposes methodological approaches to tackle each of them. First, it evaluates the effect of different demographic population structures on societal multidimensional deprivation incidence comparisons. The results of this evaluation demonstrate that societal multidimensional comparisons reflect not only differences in relative deprivation but also differences in the demographic composition of the societies to be compared. These differences in the demographic structure of the population, thus, confound societal multidimensional deprivation comparisons. To tackle this comparability problem, the application of direct and indirect standardisation methods is proposed and analysed in this context. Second, it studies the effect of differences in need, exhibited across individuals from different demographic population subgroups or households of different sizes and compositions, on multidimensional deprivation incidence profiles. To address differences in needs and enhance individual or household comparability, I propose a family of multidimensional deprivation indices that describes how much deprivation two demographically heterogeneous units with different needs must exhibit to be catalogued as equivalently deprived. The obtained empirical results demonstrate that neglecting differences in needs yields biased multidimensional deprivation incidence profiles. The results also shed light on the ability of my proposed family of measures to capture these differences in need effectively. Third, this thesis analyses the reliability of persistent poverty measures in the presence of survey non-response. The obtained empirical results indicate that persistent poverty measures based on balanced panel estimates that do not account for the relationship between survey non-response and the socioeconomic status of the household provide a significantly biased picture of the intertemporal phenomenon. The methodologies that I present in this thesis are meant foremost to be easy to implement and understand by policymakers. As such, they are proposed as methodological tools to improve the measurement and analysis of poverty in the policy context.