Evaluating the Distributive Incidence of Growth Using Cross-sections and Panels

We study decennial anonymous and non-anonymous growth incidence curves in the United States during the past 50 years. The former show income growth by quantile independent of initial incomes and are typically upward sloping, reflecting increasing inequality. The latter are conditional on initial ranks and are flat or downward sloping. This suggests distributional neutrality of growth when accounting for initial income positions. We explain this difference by decomposing the non-anonymous curves into mobility and shape components. The former is always downward sloping, whereas the latter is upward sloping in periods of increasing inequality. Thus, flat non-anonymous curves can be observed even with increasing inequality. We exploit the decomposition to show that the slope of non-anonymous curves in the United States is determined by the evolution of cross-sectional income distributions. This enables inferring the shape of non-anonymous curves from cross-sectional data and test them for a generalized pro-poorness social welfare criterion.