What About the Rest of the Pie? A Dynamic Compositional Approach to Modeling Inequality

Objective To demonstrate how a novel method enhances our understanding of determinants of inequality. Methods We take advantage of recent advances in dynamic models of compositional dependent variables to simultaneously study tradeoffs across multiple slices of the composition of income in the United States between 1947 and 2014. Results Our analyses demonstrate the utility of dynamic compositional models of income shares. Factors that increase the income share of the top income group often also increase the income share of the second income group and decrease the shares of lower groups at different rates. Polarization and marginal tax rates have large effects on relative income shares, while returns to labor, returns to capital, and partisan control of Congress have smaller, but statistically significant effects. Conclusion Our suggested approach allows researchers to effectively explore interesting variation across multiple income groups in response to changes in determinants of inequality.