STATISTICAL-INFERENCE AND THE SEN INDEX OF POVERTY

Statistical inference procedures are developed for Sen's distribution- sensitive index of poverty and each of its components-the headcount ra tio, income gap ratio, and the Gini index of the poor. Using results f rom U-statistics we show that estimates of the index and its component s all have a jointly asymptotically normal distribution and the varian ce-covariance structure can be consistently estimated. The inference t ests are illustrated by applying them to the same microeconomic data s et used in estimating official U.S. poverty statistics. The applicatio n reveals that the Sen index increased significantly in each of the th ree periods considered.