Nonparametric estimation of income distribution and poverty index (Q943652)

A class of estimators for the Foster, Greer and Thorbecke poverty index that applies to a large set of values of the poverty aversion index is proposed. The new class is buit employing the Parzen-Rozenblat kernel to estimate the density of the income distribution. Uniform almost sure consistency and mean-square consistency are established. A simulation study, for the special cases of the headcount ratio, the poverty gap index and the severity of power index, indicates that the new estimators perform well with finite samples too.