The regression approach to the estimation and additive decomposition of the Foster-Greer-Thorbecke poverty measures

We propose a new regression-based approach to the estimation and additive decomposition of the Foster-Greer-Thorbecke (FGT) poverty measures. Unlike the original additive decomposition for which subgroup population shares are used as weights, the new additive decomposition uses subgroup poor population shares as weights. Our simulation results reveal low biases of the regression-based estimators of the FGT poverty measures even for moderate sample sizes.