Abstract
<title>Abstract</title> <p>AbstractThe design of means-testing instruments requires not only choosing what information to collect about households, but deciding how that information should be aggregated into eligibility criteria. This paper addresses the second, comparatively neglected question by comparing the targeting efficiency of two multidimensional approaches to means testing, a money-metric index and a counting-based poverty measure, applied to an identical informational base comprising monetary resources, asset holdings, and household demographic characteristics. Because both frameworks draw on the same dimensions, the same deprivation thresholds, and the same equivalence scales, any difference in targeting performance can be attributed exclusively to aggregation structure. Using household-level data, we assess the relative ability of each approach to correctly identify eligible beneficiaries against an externally defined benchmark. The results show that the money-metric approach, which aggregates dimensions cardinally in a welfare-theoretically consistent manner that preserves interpersonal comparability, delivers substantially lower targeting errors than the counting approach, which treats dimensions as binary and non-compensable. These findings establish that aggregation structure, independently of informational content, is a first-order determinant of means-testing performance, with direct implications for the design and harmonization of social protection eligibility criteria. JEL Classification: D13, H31, I32, O15.</p>