Decomposition of the Gini coefficient by income components

This paper aims at clarifying the notion "overall distributive effect" of an income component or a policy proposal and moreover discusses various approaches for assessing the distributional impact of the components of total income. We pay particular attention to the problem of evaluating the distributional consequences of including a new income component in the statistical income base. Our example is the value of unpaid household work, which statistically is new to the income base, although conceptually it is included in extended income or full income, so that individual time allocations are already reflected in data. In contrast, introducing a genuinely new income component (e.g. a new transfer payment) will lead to behavioral responses that should be accounted for in the distributional analysis. However, it is standard practice to ignore behavioral responses in official analyses of tax and benefit reforms (e.g.a new transfer payment) and to compare the Gini coefficients with and without the new income component given unchanged behavior. Rather than solely comparing the levels of the Gini coefficients we suggest that one should compare the decompositions of the Gini coefficients with and without the new income component. This result gives a clarification of the difference between contribution to inequality and (marginal) effect on inequality.