Date of Degree


Document Type


Degree Name





Wim Vijverberg

Committee Members

Timothy Goodspeed

Chu-Ping Vijverberg

Subject Categories

Economics | Income Distribution | Labor Economics


Income inequality, Wage inequality, Inequality of opportunity, Financial crisis, Decomposition


This dissertation is about inequality of income in the United States. The first essay examines how inequality in total personal income relates to the inequality in each income component, such as wage income, interest income, transfer income, and so forth. The second essay analyzes the percentile shares of wages to find the factors that contribute to U.S. wage inequality. Furthermore, wage inequality is decomposed into the parts that are explained and unexplained by these contributing factors. The third essay distinguishes the proportion of the overall inequality that is due to pre-determined conditions of an individual.

Chapter 1: The Source of Income Inequality in the United States

The measurement of income inequality has been a focus of considerable study. However, the determinants of income inequality are still unclear. This study explores which income sources are primarily responsible for the observed U.S. income inequality using a decomposition method. Moreover, by estimating the marginal effect of a specific source income, I analyze its impact on overall income inequality. By doing so, this study identifies which source income has been alleviating overall inequality the most – or has been increasing it the most – before and after the financial crisis.

Among eight different sources, wage income has the largest as share in total income, yet it is most equally distributed income source. However, the increase in the Gini of wage income after the financial crisis suggests that the intensifying unequal distribution of this source income contributed to rising inequality.

The role of government transfers is important in reducing inequality, especially after the financial crisis. The equalizing effects of a ten percent increase in these transfers is a 0.84 percent reduction in the Gini of total income after the financial crisis. Among different government transfers, Social Security is the greatest equalizer. Besides the government transfers, retirement income reduces total inequality.

Interest and self-employment income show unequalizing effects on total inequality. The importance of interest income rose during the financial crisis and increased again more recently. The shares of self-employment income decreased since the financial crisis, and thus its contribution to overall inequality became smaller.

Chapter 2: Unpacking Wage Inequality in the United States

Wage earnings are the single most important income source in the U.S. They account for over 70 percent of total personal income. Therefore, wage may be the key determinant of income inequality. Since the 1980s, slower economic growth, higher unemployment and reduced wage shares have been observed in the U.S. Along with these economic trends, the U.S. income inequality has been rising and many believe that these dispersions are due to the widening of the U.S. wage structure since 1980 (Karoly and Burtless 1995). Severe wage stagnation caused by the deep recession from the most recent financial crisis (2007-2009) makes an analysis of wage inequality a challenging task. This study attempts to untangle the trend of the U.S. wage inequality before and after the financial crisis and identify the source of the differences in wage distributions.

The Gini coefficient is the most popular inequality measure but is not sufficiently practical for our purposes. While it does provide a global level of inequality, it does not reveal the detailed process underlying the changes of this measures. Furthermore, this inequality measure has an inherent disadvantage: it varies whenever the distribution of wage income changes without regard to the location of the changes, such as whether they happen at the top or at the bottom of the distribution (Litchfield 1999). Therefore, this essay analyzes the percentile shares of wages, as in the well-known work of Piketty (2014). By means of this measure, this study investigates potential contributors to U.S. wage inequality during 2000-2014, including gender, race, ethnicity, age, concentration of the top earners, occupation, economic sector, location, difference in the level of educational attainment, and other conditions such as employment status and a worker’s English language proficiency. Utilizing two leading decomposition methods, I differentiate between components that can be explained by individuals’ core characteristics and components that cannot be explained by these potential contributors and thus, account for wage structural differences within our society. In addition, by further dividing the decomposition results, this study examines which factors are most likely to contribute to the difference in the wage distribution.

During 2000-2014, while wage income is more unevenly distributed among men, there exist the glass ceiling for women in the form of a significant wage gap at the top decile. Among five racial groups, within-group wage inequality among Asians is highest, while it is lowest among the Other Races category. At every quintile, White workers are paid more than non-White workers, and their relative gains tend to increase at higher quintile of the distribution. This result suggests the existence of sticky floors for non-Whites that prevents them from achieving an equivalent wage due to the higher returns to characteristics in favor to Whites, in particular for men. The estimated wage gaps between non-Hispanic White and others – either Hispanics or non-White non-Hispanics – is evidence of the presence of the U.S. labor market polarization.

The college wage premium increases non-monotonically and the recent increase in wage inequality is not only due to the higher returns to higher education but also due to larger within-group wage dispersion among more educated workers. Pre-determined conditions to workers such as working status or English language proficiency widen wage gaps, particularly at the top-end, up to 33.59 dollars per hour.

From a regional point of view, most of the wage dispersion in the U.S. is attributed to the within-state wage inequality. The between-state inequality accounts only for less than 20 percent during the sample period.

The wage shares held by top one or top ten percent highest earners are stable even during the financial crisis. They preserve about ten percent and thirty two percent of total annual wages, respectively. The within-age cohort wage inequalities tend to increase with age except for the teens as their potential experience increases.

The between-occupation inequality is smaller than the within-occupation inequality. There exists large relative gain of top one percent in managerial occupation relative to those in other occupations, which is 149.37 dollars per hour. Workers are in the financial sector are paid up to 116.97 dollars per hour more, depending on their relative ranks in the wage distribution.

The U.S. wage dispersion increases after the financial crisis and it is mostly due to the widening difference in the upper tail. In terms of composition or wage structure effects, wage structure effects account more for wage inequality. The composition effects linked to the industry has shown as increasing overall inequality. The most important contributor to overall inequality is different at the top and at the bottom of the wage distribution. Wage structure effects linked to education are contributed to increasing wage inequality in the upper tail, but these are attributed to reducing wage inequality in the lower tail of the wage distribution.

Chapter 3: Inequality and Unequal Opportunity in the United Sates

Although many studies focus on inequality of outcomes (e.g., household income or personal income or educational attainment), the outcome inequality does not adequately reflect the inequality in society since this inequality is also the result of different levels of efforts. Thus, inequality itself can be neither all good nor all bad (Ferreira and Gignoux 2011). The primary concern for inequality of opportunity (IOP, see Roemer, 2000) emphasizes the aspects of pre-determined factors that are beyond an individual’s control such as family background or individual core characteristics. Measuring IOP starts by sorting the determinants of outcomes into two exclusive classes: the factors that are beyond individual responsibility (circumstances) and factors are under the control of the individual (efforts). Utilizing an ex-ante approach, this study measures how much of the observed U.S. income inequality is attributable to IOP. In addition, the relative importance of each circumstance is examined using the Shapley decomposition.

During 2000-2014, about 35 percent of U.S. inequality should not be considered due to the differences in the individual efforts but rather as due to circumstances. This share of inequality that is considered as due to circumstances increases after the financial crisis. These results may imply that the effort characteristics, which are under the individual’s control, became less important in the U.S. labor market due to the shrinkage of the middle quality jobs after the financial crisis. Among five racial groups, IOP among Blacks is the highest while that of Asians is the lowest. After the financial crisis, the IOP increased for Blacks and Other Races but decreased for Whites, Asians, and Mixed-Race members of society. IOP of non-Hispanics is higher than that of Hispanics.

As for the decomposition results, while own circumstance contributes more to total IOP in the group of Whites and Mixed-Race individuals, family background was more responsible for total IOP among of Blacks, Asians and members of the Other Races group. For non-Hispanics, the larger contribution of the own circumstance was especially evident after the financial crisis. Meanwhile, for Hispanics, family background became increasingly more important after the financial crisis.

This work is embargoed and will be available for download on Sunday, June 02, 2019

Graduate Center users:
To read this work, log in to your GC ILL account and place a thesis request.

Non-GC Users:
See the GC’s lending policies to learn more.