Multidimensional indices are widely used to measure socioeconomic characteristics that are not directly observable. The choice of dimensions, indicators and the aggregation function determine the meaning and interpretation of a multidimensional index. Where the index appropriately reflects a population's values or preferences, I argue that the relevant ordering should be elicited to inform the choice of dimensions, indicators and aggregation function. Standard approaches to preference elicitation are not well suited to implementation, in the context of a field survey in a developing country, for the large number of indicators typically identified for multidimensional indices. An innovative experimental design that addresses these issues was implemented to elicit perceptions of empowerment among women in Tunisia. Estimation of a parametric model generates weights for an index of Women's Empowerment that reflects the perceptions and values of the survey respondents. The conventional assumption of equal weights is decisively rejected. The empirical approach resolves, in this context, an issue that emerges in the methodological literature on multidimensional indices: how to weight correlated indicators without 'double counting' latent factors.
About the speaker
Natalie is Senior Research Officer at Oxford Poverty and Human Development Initiative (OPHI) and Domus Fellow and Tutor in Economics at Lady Margaret Hall, both in the University of Oxford. Her research interests centre on applied welfare economics in developing countries, with a particular focus on issues of measurement. She has worked on the measurement of poverty over time, and is currently conducting research on methods to inform multidimensional index weights through elicitation of values and on the distributional impact of cash transfer programmes.
Date: 25 October 2018
Time: 12:00 - 13:00