Immigrants and Ethnic Differences

Source neal and johnson 1996 table 2 testing racial

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: not only because they look at younger workers but also because they include numerous controls for work history and family background. Source: Neal and Johnson (1996) TABLE 2 TESTING RACIAL FOR DIFFERENCES THE RETURNO A FQT: MEN IN T All Races ( N = 1 ,593) (1) White (N = 8 25) (2) Black (N = 4 66) (3) Hispanic (N = 3 02) (4) Black Hispanic Age AFQT A FQT~ Black x A FQT Black x AFQT* R i To~E.-The "all races" sample includes all men from the sample described in table 1 . All respondents were born after 1961. Standard errors are in parentheses. T ABLE 3 TESTING RACIAL FOR DIFFERENCES THE RETURNO A FQT: WOMEN IN T All Races ( N = 1 ,446) (1) White (N = 7 2 6 ) (2) Black ( N = 4 28) (3) Hispanic (N = 2 92) (4) Black Hispanic Age AFQT AFQT~ Black x A FQT Black x AFQT' R NOTE -The "all races" sample includes all women from the sample described in table 1. All respondents were born after 1961. Standard errors are in parentheses. Source: Neal and Jonhson (1996) BLACK-WHITE WAGE DIFFERENCES TABLE 4 Black Hispanic Age AFQT - ,352 ( .029) -.I80 (.034) ,067 (.015) less is explained when selection is accounted for . .. N o~€.-The dependent variable is log hourly wages. The sample is the sample described in table I plus the sample of nonparticipants. Nonparticipants include workers who report not working between their 1989 and 1991 interviews. Nonparticipants also Include workers who did not work between their 1989 and 1990 interviews and were not interviewed in 1991. Some respondents are excluded from the previous regression analyses solely because their wage observations are invalid. These respondents are also excluded from this analysis. All respondents were born after 1961. Standard errors are in parentheses. construct our sample of log wage offers by assigning log wages of zero (hourly wages of one cent) to all male nonparticipants. This strategy ensures that our imputed offers for nonparticipants always fall below the relevant conditional medians. Table 4 presents median regression results based on this sample. The racial wage gap at the median moves from - .352 to - . 134 when AFQT is added to the regression. Whether we condition on AFQT or not, these median regressions show a larger negative effect of being black than the regressions on participants in table 1, where the adjusted gap for men was - .072. The contrast between the results at the mean and at the median supports the view that looking only at participants masks some discrimination. Nonetheless, over 60 percent of the difference in medians is explained by our one measure of skill. Smith and Welch (1986) use a different method to estimate the racial difference in the conditional means of the wage offer distributions. They observe that the mean of the wage offer distribution, E ( w ) , is a weighted average of the mean wage offers for participants and nonparticipants: E (w) = LFPR E (w ( participate) + (1 - LFPR)E (w1 don't participate). (2) Fortin – Econ 560 Lecture 4B Because AFQT scores are themselves influenced by years of schooling, more recent rese...
View Full Document

This document was uploaded on 02/26/2014 for the course ECON 560 at The University of British Columbia.

Ask a homework question - tutors are online