Poverty orderings with symmetric and asymmetric attributes

The motivation of this paper is that those who research on poverty, until recently anyway, have found it di ¢ cult to convince the policy community about poverty outcomes while using one-dimensional poverty measures. The idea that poverty is a multidimensional phenomenon and income is just an indicator is generally agreed. Multidimensional poverty measures have been the starting point in the lit-erature: Tsui (1997), Chakravarty et al. (1998, 2005), and Bourguignon and Chakravarty (2003) have suggested several functional forms to aggregate dif-ferent attributes in a multidimensional poverty measure. Creating an index requires aggregation across attributes and across individuals, a task which re-quires speci\u85c aggregation rules for which there is not consensus. This is the reason why a poverty index can be arbitrary, and hence the conclusions of the analysis. As in the one-dimensional approach, it is possible to reduce the degree of arbitrariness by considering a class of multidimensional poverty measures satis