Black-White Differences in Intergenerational Economic Mobility in the United States
Mazumder, Bhashkar, Economic Perspectives
Introduction and summary
The large and persistent gap in economic status between blacks and whites in the United States has been a topic of considerable interest among social scientists and policymakers for many decades. The historical legacy of slavery and segregation raises the question of how long black Americans are likely to remain a disadvantaged minority. Despite the enormous literature on black-white inequality and its historical trends, few studies have directly measured black-white differences in rates of intergenerational mobility, that is, the ability of families to improve their position in the income distribution from one generation to the next. Estimates of rates of intergenerational mobility by race can provide insight on whether racial differences in the United States are likely to be eliminated and, if so, how long it might take. Furthermore, they might also help inform policymakers as to whether there are lingering racial differences in equality of opportunity and, if so, what the underlying sources for these differences are.
More generally, the relatively low rate of intergenerational mobility in the United States compared with other industrialized countries has been a growing concern to policymakers across the political spectrum. (1) Understanding the sources of racial differences in intergenerational mobility might also shed light on the mechanisms behind the relatively high degree of intergenerational persistence of inequality in the United States.
In this study, I attempt to advance our understanding along several dimensions. First, I use two data sets containing larger intergenerational samples than have been used in the previous literature. One of the data sets matches individuals in the U.S. Census Bureau's Survey of Income and Program Participation (SIPP) to administrative earnings records from the Social Security Administration (SSA). This matched data set provides many more years of data on parents' earnings than most surveys and is likely to be less prone to measurement error, since it is derived from tax records. In addition, the SIPP contains data on key characteristics of the parents, such as wealth levels and marital history. The other data source I use is the U.S. Bureau of Labor Statistics' National Longitudinal Survey of Youth (NLSY). In addition to containing a rich array of information as children transition from adolescence to adulthood, such as test scores and personality traits, the NLSY also measures total family income in both generations, giving it an advantage over the SIPP. Using a measure of economic status that includes the income of the spouse avoids selecting only women who participate in the labor market.
Second, I use two types of measures of intergenerational mobility. The first is a set of transition probabilities of relative income status across generations. An example is the probability of moving out of the bottom quintile of the income distribution from one generation to the next. Hertz (2005) was the first to use transition probabilities to examine black-white differences in intergenerational mobility. Using the Panel Study of Income Dynamics (PSID), Hertz found that blacks were less upwardly mobile and more downwardly mobile over generations than whites. Since then, a few other studies mostly using the PSID have found similar results. (2)
The second set of measures, called directional rank mobility, compares whether the rank of a child in the income distribution is higher or lower than their parents' rank in the previous generation. (3) Both types of measures are able to distinguish upward movements from downward ones and can be measured at different points in the income distribution. The directional rank mobility measure is a useful complement to the transition probability because instead of using an arbitrarily chosen cutoff, it uses a natural yardstick, one's own parents' rank. As I discuss later, the directional rank mobility measures also appear to be very robust to many measurement issues. …