August 2024
Abstract
We introduce a novel approach for eliciting relative poverty rankings that aggregates partial orderings reported independently by multiple neighbors. We first identify the conditions under which the method recovers more accurate rankings than the commonly used Borda count method. We then apply the method to secondary data from rural Indonesia and to original data from urban Cˆote d’Ivoire. We find that the aggregation method works as well as Borda count in the rural setting but, in the urban setting, reconstructed rankings from both the pairwise and Borda count methods are often incomplete and sometimes contain ties. This disparity suggests that eliciting poverty rankings by aggregating rankings from neighbors may be more difficult in ur-ban settings. We also confirm earlier research showing that poverty rankings elicited from neighbors are correlated with measures of poverty obtained from survey data, albeit not strongly. Our original methodology can be applied to many situations in which individuals with incomplete information can only produce a partial ranking of alternatives.