NLSY79 Data Data and Empirical Strategy

local costs of living. 15 As shown in Table 1, Panel B, we fi nd that the average black woman lives in an area with higher wages among low- mobility occupations. The gap between local wages where black and white women live is highest in CZs at the lower end of the local wage distribution that is, the 10 th and 25 th percentiles of respondents’ locations. 16,17 In addition to cost of living, our preferred specifi cations control for years of educa- tion. If, conditional on AFQT score, black women acquire more years of education than white women, then omitting years of education would result in an upward bias on the coeffi cient estimate for the black indicator variable. Lang and Manove 2011 show that black women in the NLSY79 had acquired more years of education by 2000 than white women with the same AFQT score. We confi rm that this is true in 2006 as well. Incorporating these methods, our preferred estimate of the black- white wage gap among women is the estimate for β in: 2 Ln wage i = ␣ + ␤BLACK i + ␥ 1 AFQT i + ␥ 2 AFQT i 2 + ␦ 1 age i + ␦ 2 age i 2 + ␾COL i + ␭EDUC i + ␧ i . This equation includes local cost of living COL and years of education EDUC. The model of employer discrimination in the Appendix implies that local costs of living and individual productivity traits like education are important controls to in- clude; otherwise, the regression is unlikely to identify wage differences due to racial discrimination. We estimate Equation 2 using OLS and median regression. Observed log hourly wage among workers is the dependent variable for our OLS estimates. Median regression estimates also include nonworkers with imputed potential wages, described above.

B. NLSY79 Data

In Table 2, Columns 1 and 2, we present descriptive statistics for the black and white women in the sample we use for our OLS analysis. This sample includes women who worked and have valid wage information for either the 2006 or the 2004 survey. We fi rst collect wage data from the 2006 survey but, for those missing wages in 2006, we use the 2004 wage if it is available. We convert wages to 2008 dollars using the 15. We identify noninstitutionalized workers ages 16-64 and employed in the private sector using the pooled 2005-2007 ACS at IPUMS Ruggles et al. 2010. We collect residuals from a regression of log hourly wages on a quadratic in age and indicators for educational attainment, sex, and raceethnicity. The average of those residuals among CZ residents employed in “low-mobility occupations” is our local wage index. An occupa- tion is considered “low-mobility” if more than half of the occupation’s 3-digit OCC1990 workers were still living in their state of birth. An occupation is considered “high-mobility” if less than half of the occupation’s three-digit OCC1990 workers were still living in their state of birth. 16. To address further concerns that the most productive workers choose to locate in the highest cost-of- living areas, we fi nd that although both of our cost-of-living measures are positively correlated with years of education and AFQT score, the magnitude is quite small. For example, increasing a respondent’s AFQT score from the 25 th to the 75 th percentile results in a 1.7 percent increase in the low-mobility occupation wage measure. 17. Table 1, Panel C shows differences between average wages in “high-mobility” occupations in locations where black and white NLSY79 respondents live. Wages in this group are higher than among the low- mobility group because more productive workers are more geographically mobile on average. Wages in these high-mobility occupations tend to be higher where blacks live as well. Table 2 Descriptive Statistics from the NLSY79 2006 Sample: OLS Sample Individuals with Low Imputed Wages Individuals with High Imputed Wages Spouse’s Earnings at or Above 90th Percentile Individuals with High Imputed Wages Spouse’s Earnings Fall Between 75th–89th Percentile Black Women White Women Black Women White Women Black Women White Women Black Women White Women 1 2 3 4 5 6 7 8 Mean hourly wage Columns 3–8: imputed wage 15.66 11.06 19.31 14.68 1 1 45 45 45 45 Median hourly wage Columns 3–8: imputed wage 12.87 15.40 1 1 45 45 45 45 Age at time of interview 44.67 2.18 44.71 2.22 44.65 2.24 45.03 2.01 45.67 2.65 45.15 2.18 45.17 2.48 44.32 2.13 AFQT score standardized –0.60 0.69 0.46 0.92 –1.15 0.37 –0.65 0.71 –0.19 0.75 1.10 0.75 –0.001 0.86 1.01 0.78 Years of education 13.32 2.09 13.73 2.33 11.20 1.36 11.03 1.77 16.33 2.06 16.38 1.76 15.67 2.42 16.15 1.48 Years of work experience at time of interview 18.65 6.44 21.07 5.50 5.04 6.08 7.28 5.13 16.64 5.08 13.28 6.21 17.35 8.12 16.07 5.85 N 968 1,505 84 36 9 40 6 34 Notes: Standard deviations in parentheses. We impute a low potential wage of 1 for women who: 1 received any benefi ts from the Temporary Assistance for Needy Families TANF, Supplemental Security Income SSI, or Food Stamp programs between 2002 and 2006; 2 have a high school degree or less education; and 3 report no spousal income in the previous fi ve years. We impute a high potential wage of 45 for women who meet the following two criteria: 1 earned at least some college education and 2 married to a high- earning spouse. A spouse is considered “high earning” if spousal average annual earnings over the past fi ve years place him at or above the 90 th percentile for men of his race in the 2006 NLSY79 in Columns 5 and 6 or above the 75 th percentile for men of his race in Columns 7 and 8. CPI- U. When we compare unconditional mean and median hourly wages for the two groups, we see that white women earn more. For example, the mean hourly wage for white women is 19.31 and only 15.66 for black women: Blacks experience an unconditional wage penalty of 18.9 percentage points. Of course, comparing un- conditional means or medians does not take labor market skills into account. The bottom three rows of the table show that white women in the sample score higher on the AFQT, acquire 0.4 more years of education, and obtain nearly 2.5 more years of work experience. 18 In Columns 3 and 4, we present descriptive statistics for the women who get a low imputed potential wage of 1, a group that includes twice as many black women as white women 84 versus 36. In Columns 5–8, we present summary statistics for the women who get a high imputed potential wage of 45. Both spousal earnings cutoffs imply high potential wage imputation for over four times as many white women as black women.

IV. Results