Inequality and the distribution of human capital

New cross-country data on income inequality has spurred research investigating the macroeconomic sources of income inequality. However, this research suffers three shortcomings. First, the panel of aggregate inequality data is incomplete complete and fraught with problems of measurement error and comparability. Second, with no clear theory-driven relationship between aggregate inequality measures and other macroeconomic data (analogous to the role of the Solow model in growth regressions) no consensus exists regarding an appropriate regression specification, complicating both inference and comparisons among published results. Finally, the number of potential variables of interest are nearly as many in number as the observations that exist to test them.; These issues are addressed in three chapters. The first chapter reviews the results from previous work, and then employs data-driven approaches to search for a “consensus” specification. A data set is constructed from the current universe of available data with careful attention to coverage and comparability. Nesting all of the variables found in previous specifications into a single model, hypothesis testing is used to find an appropriate and parsimonious framework. To better propagate specification uncertainty into final parameter estimates, Bayesian model averaging is employed to construct estimates of coefficients' posterior means and standard deviations in the search for specification-robust covariates.; The second chapter addresses the link between data on inequality and neoclassical factor-based economic theory to develop a theory-based specification for cross-country Gini coefficient regressions. Labor market shocks are shown often to have ambiguous effects on the Gini coefficient in equilibrium, complicating interpretation of standard reduced form regressions. An accounting formula is developed in which the income Gini coefficient is decomposed into factor endowment and price effects, and various forms of the equation are estimated using data on the distribution of educational attainment.; The third paper addresses a related question, the determinants of educational inequality. A general equilibrium framework of educational investment and returns is tested using cross-country data at the primary, secondary and higher levels of education. The system of equations is estimated using three stage least squares, with particular attention given to the impact of credit markets, educational expenditures, and expected lifespan.