Research

The Participatory Index of Women's Empowerment: development and an application in Tunisia [open access] (with Simone Lombardini)

• Oxford Development Studies, 2024 | (online appendices) (working paper version) (featured in an Oxfam blog post)

On track or not? Projecting the global Multidimensional Poverty Index [open access] (with Sabina Alkire, Ricardo Nogales and Nicolai Suppa) 

• Journal of Development Economics, 2023 | (working paper version)

Global multidimensional poverty and COVID-19: A decade of progress at risk?  [open access] (with Sabina Alkire, Ricardo Nogales and Nicolai Suppa)  

• Social Science & Medicine, 2021 | (working paper version)

• Measuring consumption in 12 minutes: Optimal survey design for impact evaluation.  With Geetika Nagpal and Kate Orkin.

Detailed measurement of consumption expenditure is the gold standard approach to welfare analysis in a low-income mixed-livelihoods setting, but is highly demanding of respondent time and survey resources. When resources are scarce, there is a trade-off between survey sample size and measurement precision for each unit of observation. It may not, therefore, be efficient to maximise measurement precision. In this paper we establish stylised facts about the relationship between survey resource demand and measurement precision for consumption expenditure in a low-income context. We use these to develop a theory of optimal measurement precision for consumption expenditure as the outcome of interest in an experimental impact evaluation. We illustrate the practical value of this approach through its application to the measurement of consumption expenditure for an impact evaluation in rural Western Kenya. Our (approximately) optimised consumption module required just 12 minutes of survey time on average across the sample.

• Early version titled Measuring consumption in 12 minutes: lessons from a field study in Rural Kenya presented at IARIW-World Bank Conference New Approaches to Defining and Measuring Poverty in a Growing World in November 2019.

• Chronic or acute?  The measurement of poverty over time.  With Catherine Porter.  Presented at ECINEQ 2019.

We consider the aggregation of information over people and over time to measure poverty. Mild ethical principles are shown to entail a form for the poverty measure that embodies a well-defined ordering of trajectories. Further principles are motivated and applied to restrict the class of possible trajectory orderings and the class of functions that represent them. In particular, we argue for a principle of non-increasing fluctuation aversion: less-poor people should not suffer more from the same absolute or even relative fluctuations in the wellbeing indicator. This principle is shown to be consistent with revealed consumption-smoothing behaviour in longitudinal data from rural Ethiopia as well as evidence in the empirical literature from both developed and developing countries. Measures proposed in the earlier literature on chronic poverty are not consistent with this principle. A parametric class of poverty measures is proposed that captures the principle, parameters are calibrated empirically, and the resulting measure is applied to evaluate poverty in rural Ethiopia between 1994 and 2009.

• Representation of a symmetric separable preorder.  Presented at ECINEQ 2015 and OECD 2014.

For theorists: Solvability (often in the form of connectedness) is central to theorems establishing additive representation of a separable preorder on a Cartesian product domain. I show that when the domain and the preorder are symmetric this requirement may be relaxed. The result has natural applications to anonymous welfare and poverty measures.

For welfare economists: Characterisation theorems enable the ethical and measurement principles that are embodied by particular socioeconomic measures to be precisely determined. The scope of the existing literature is limited as the underlying mathematical results do not encompass many applications of interest, in particular the case of a disconnected domain or discontinuities in the ordering represented. I show that under certain conditions, relevant for poverty and welfare measurement, the underlying mathematical results may be extended and thus broad classes of poverty and welfare measures may be characterised on the basis of intuitive axioms.