Stochastic dominance amongst swedish income distributions

Abstract

:Sweden's income distribution for the whole population and for subgroups, including its immigants, has been extensively studied. The interest in this area has grown with increasing availability of data, including panels. The previous studies are based on indices of inequality or mobility. While indices are useful for complete ordering and have an air of "decisiveness" about them, they lack universal acceptance of the value judgements inherent to the welfare functions that underlay any index. In contrast, uniformpartial order relations are studied in this paper which rank welfare situations over very wide classes of welfare functions. We conduct bootstrap tests for the existence of first and second order stochastic dominance amongst Sweden's income distributions over time and for several subgroups of immigrants. Analysis of immigrant's income is motivated by the fact that the development of income for immigrants has been different and strongly affected by their length of residence and countries of origin. We consider several non-consecutive waves of a panel of incomes in Sweden. Two income definitions are developed. One is pre-transfers and taxes, gross income, the other is a post-transfers and taxes, disposable income. The comparison of the distribution of these two variables affords a partial view of Sweden's welfare system. We have focused on the incomes of Swede's and immigrant groups of single individuals identified by country of origin, length of residence, age, education, gender, marital status and other relevant characteristics. We find that first order dominance is rare, but second order relations hold in several cases, especially amongst disposable income distributions. Sweden's incomes and welfare policies favor the elderly, females, larger families, and longer periods of residency. We find, in general, the higher the educational credentials, the higher is the burden of this equalization policy.