A preference-based theory unifying monetary and non-monetary poverty measurement

Multidimensional poverty measures are increasingly used in practice even though they face criticism and generate long-lasting debates. These contentions primarily find their origin in the divergence between standard poverty identification practices and a preference-based definition of the poor. This paper fills this gap by proposing a poverty measurement theory that (i) adopts a preference-based definition of the poor, (ii) acknowledges that the relevant preference is only partially known and (iii) encompasses both market and non-market dimensions of well-being. A preference-based definition satisfying two mild substitutability assumptions implies that two types of individuals are poor: individuals who experience an extremely low outcome in at least one dimension (“extremors”) and individuals who cumulate moderately low outcomes in several dimensions (“cumulators”). The paper shows that standard identification practices are not flexible enough to properly identify both types of poor individuals and proposes a simple refinement that improves on this issue. The theory provides a conceptual foundation from which practitioners may derive guidance for the many choices they face.