Differences in Wealth, Evidence from Structural Regression Decomposition, 1850-1870

Recent studies have used regression decomposition to analyze recent data and found that over seventy percent of the black-white wealth differences remained unexplained (See, e.g., Gittleman and Wolff 2000; Altonji, Doraszelski and Segal 2000; and Blau and Graham 1990). Their results are limited to the variation in modern data. This study contributes improved methodology and historical empirical results to the literature on economic discrimination. In this paper, (i) James Curtis Jr presents structural regression decompositions, which are modifications to methods developed by Becker (1957) and Oaxaca (1973); (ii) James Curtis Jr presents a basic empirical test when analyzing structural regression decompositions; (iii) James Curtis Jr reports the estimated sources of black-white differences in wealth directly before and after emancipation; and (iv) James Curtis Jr links these findings to recent studies. Empirical estimates confirm that the size and persistence of modern black-white wealth differences have historical roots