Distributions of Intergenerational Earnings: A Nonparametric Analysis of Racial Differences

Nonparametric methods are used to analyze distributions of intergenerational earnings. A stochastic kernel generates stationary distributions for African-Americans and whites, indicating a long-run outcome not in favor of racial convergence. Tests indicate that the distribution of white earnings stochastically dominate the earnings distributions of African-Americans from similar economic backgrounds. Within-race decile comparisons demonstrate that the distributions for whites improve as fathers' earnings improve; each increase in fathers' deciles progressively improves the entire distribution for sons. The same results do not exist for African-Americans who appear to have extensive within-race mobility.