I've got it! Don't mind! Sorry ! :( Artur

For the Lorenz curve of income distribution, gini index derives from P and L where: P: cumulated % of individuals L: cumulated % of income and the step-by-step calculation I was talking about looks like: scalar area = 0 scalar gini = 0 loop i = 2..$nobs --quiet area = (L[i] + L[i-1])*(P[i] - P[i-1]) + area # the area under the Lorenz curve endloop area = (area + P[1]*L[1])/2 gini = (1 - 2 * area) The very small difference came from the additional last (and pretty small) term I missed : P[1]*L[1] best, artur