Date of Award
Doctor of Philosophy
Scott Drewianka, John S. Heywood, James H. Peoples
This dissertation consists of two essays which are concerned with the measurement and description of Income Inequality. Chapter 1 studies the measurement of inequality when data is presented in binned form. While various methods have been proposed for this purpose, I reveal an issue which has not yet been addressed in the literature. I demonstrate that differences in mean earnings can cause differences in the measures of inequality, even if underlying income shares remain constant. The consequence of this issue is that the usage of many of the available methods will create estimates of inequality not suitable for descriptive purposes and especially problematic for regression analysis. In this chapter I describe the nature of binned data, the methods available for estimation of inequality from this data, and a scaling issue detrimental to many estimation techniques. I then perform a series of simulations to demonstrate when certain methodologies perform better than others. I also establish if there are qualities of a specific dataset that would make a given method more appropriate in certain situations.
In chapter 2, using long run and geographically precise income data from the Internal Revenue Service, I estimate annual, state-level distributions of household income of the United States population from 1946 to 2015. Additionally and critically, by studying the central range of the U.S. distribution of income over this time period, I focus on an area of investigation that has received little or no attention in the literature which either focuses on a shorter time period or on top earners. I also use the findings from chapter 1 to select the
most appropriate estimation technique for the data at hand. Various measures of inequality derived from the estimated distributional parameters are presented. National inequality represented by the Theil index shows the rise of inequality is concentrated in periods beginning in the late 1940s and the early 1960s. The movement of relative earnings among the 10th, 50th, and 90th percentile of earners supports the well-known hollowing out effect. Owing to the rich historical aspect of the data collected, I extend the understanding of the presence of this effect to the mid-20th century, i.e., the late 1940s and 1950s. State-level inequality follows a similar pattern to the national trend, but with some variation. While within-state inequality is the most substantial part of national inequality, an inverted W shaped pattern of between-state inequality is also found, with peaks in the early 1950s and early 2010s. The findings of this chapter both describe the history of United States inequality more fully and lay the foundation for better analysis of its causes and outcomes.
Schwendel, John Albert, "Inequality in the United States 1946-2015" (2020). Theses and Dissertations. 2420.