Hi everyone,
I am estimating the monetary value of in-kind benefits (like education) to be added to the household disposable income and observe the effect of the in-kind service on inequality measures.
Starting from the Gini coefficient, I expect that, ceteris paribus, adding up this lump-sum reduce the inequality because its relative effect will be stronger on the first quantile rather than the last. On the other hand, the household composition matters i.e. the number of children per household that are attending primary and secondary school. Given a positive correlation between income and household size, then the household composition effect may dominate the first one and education contributes positively on the overall Gini coefficient.
Is there a way to check which effect dominates? Or better, can I simply decompose the Gini coefficient with
Code:
ineqdec0 extended_income [w=rg002], by(quantiles)
(where the extended income is the equivalized household income including education benefits) and compare it with the same decomposition of the original non-extended income? I am aware of the necessary assumption of non-overlapping distributions for decomposition by subgroups of population, but I guess that grouping over quintiles this assumption should be met, right?


Thank you for the support