Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
0Activity
0 of .
Results for:
No results containing your search query
P. 1
Skin Tone Effects Among African Americans

Skin Tone Effects Among African Americans

Ratings: (0)|Views: 8|Likes:
Published by Marina Bessel

More info:

Published by: Marina Bessel on Dec 07, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

12/07/2012

pdf

text

original

 
Skin-Tone Effects among African Americans:Perceptions and Reality
 By
J
ONI
H
ERSCH
*
There are considerable racial disparities ineconomic outcomes and health, as well as evi-dence that these effects of race differ by skintone, with darker skin tone being associatedwith inferior economic outcomes and higherblood pressure. Using data from three sources, Ifind consistent evidence that African Americanswith lighter skin tone have higher educationalattainment than those with darker skin tone,with some, but limited, evidence that the racialdifference in wages is attenuated by lighter skintone. I explore explanations for these findings,considering the influence of possible measure-ment error, perceived attractiveness, access tointegrated schools or work groups, perceiveddiscrimination, and genetic differences. Theperception that there is differential treatment onthe basis of skin tone is more pronounced thanthe observed disparities.
I. Datasets and Background Statistics
I examine data from the National Survey of Black Americans (NSBA), 1979–1980; theMulti-City Study of Urban Inequality (MCSUI),1992–1994; and the 1995 Detroit Area Study(DAS).
1
In each dataset, interviewers reportedrespondents’ skin tone in five or three catego-ries. Both the NSBA and MCSUI provide in-formation on education, childhood background,and labor market characteristics. The DAS doesnot include information on earnings but in-cludes a series of questions about perceptions of treatment not available in other datasets. In ad-dition, respondents to the DAS self-reportedtheir skin tone, uniquely allowing corroborationwith interviewer reports.Table 1 provides average values of education,hourly wage, and employment status using theNSBA and the MCSUI. Lighter skin tone isclearly associated with higher employment ratesfor women and higher educational attainmentfor both women and men. The employment ratefor women with very dark skin tone in theNSBA is strikingly lower than for women withlighter skin tone.In contrast, evidence that skin tone affectswages is limited. For both sexes, in both data-sets, those in the light category have the highestaverage hourly wage, but this value is signifi-cantly different from those with darker skinonly for men in the NSBA. Furthermore, thepattern for women based on the MCSUI doesnot show an increasing wage from darker tolighter skin tone, but instead shows that womenin the medium-skin-tone category have the low-est average wage.
II. Education and Wage Regressions
To control for factors other than skin tonewhich may affect education and wages, I esti-mate education and wage regressions by gender,using the NSBA and MCSUI. The NSBA iscomprised of black respondents only, so black/ white comparisons cannot be made. I include inthe MCSUI analyses only those respondentswho were non-Hispanic black, or white, thusremoving from the sample those reported as anyother category (mainly Hispanic and Asian).Both datasets have information on parents’ edu-cation as well as other characteristics thatinfluence educational attainment, such as residen-tial characteristics when growing up. All equa-tions control for cohorts in 10-year intervals.ThosereportedlivingmainlyoutsideoftheUnitedStates are excluded from the years-of-educationanalyses, as are those who are under the age of 18,over the age of 65, or missing information onvariables included in the equations.
* Harvard Law School, 1545 Massachusetts Avenue,Cambridge, MA 02138 (e-mail: jhersch@law.harvard.edu).
1
Arthur H. Goldsmith et al. (2005) have used both theNSBA and the MCSUI to examine the effect of skin tone onwages among males.
251
 
The wage equations are estimated using aconventional log-wage specification augmentedby measures of skin tone. I exclude the self-employed from the wage analyses and controlfor education, tenure, an approximation of ac-tual work experience, whether the job is cov-ered by union contract, whether the worker is agovernment employee, and marital status. In theequations using the NSBA, I include indicatorsfor residence in the South or an urban area;using the MCSUI, I include indicators for city.Table 2 reports the coefficients of skin tonefor the education and wage regressions, esti-mated separately by gender and dataset.
2
Thepattern of effects is similar to the unadjustedaverages reported in Table 1. Those with darkerskin color attain significantly less educationthan those with lighter skin color. Adjustmentfor characteristics such as parents’ educationcuts the magnitude of the disparity by up to half.There is a clear pattern, with the educationalattainment penalty being smaller as skin tonelightens, although the coefficients are not al-ways significantly different from the coefficientin adjacent skin tone groups.The wage equations, likewise, show a patternsimilar to those indicated by the unadjustedaverages of Table 1. Using NSBA data, there isan earnings penalty of about 20 percent for allmen with darker skin tone relative to men withlight or very light skin tone, although there areno significant differences in the magnitude of the disparity within the darker skin tone cate-gories. Data from the MCSUI show that womenwith medium skin tone face a wage penaltyrelative to women with light skin tone. Thewage regressions do not, however, show thepattern of advantage as skin tone lightens ob-served in the education regressions.
III. Possible Mechanisms
The preceding results indicate that there is aconsistent effect of skin tone on educational at-tainment, with suggestive, but less consistent, sup-port for an effect of skin tone on wages. I explorepossible mechanisms by which skin tone differ-ences may influence economic outcomes.
2
Complete results available from the author uponrequest.T
ABLE
1—E
DUCATION
, W
AGE
,
AND
E
MPLOYMENT
S
TATUS BY
S
KIN
T
ONE
C
ATEGORY
Very dark Dark Medium Light Notes*Panel A: National Survey of Black Americans (NSBA)FemaleEducation (years) 10.44 10.95 11.52 12.17 b, c, d, e, Hourly wage ($1994) 8.96 8.55 9.29 9.69Employed if age 1865 43.75 58.92 61.71 64.56 a, b, cMaleEducation (years) 11.14 11.18 11.75 12.41 c, eHourly wage ($1994) 11.29 11.62 11.55 13.97 e, Employed if age 1865 77.61 77.57 79.32 78.89Panel B: Multi-City Study of Urban Inequality (MCSUI)FemaleEducation (years) 12.53 12.79 13.24 d, e, Hourly wage ($1994) 10.16 9.71 10.55Employed if age 1865 50.29 56.82 58.44 d, eMaleEducation (years) 12.45 13.13 13.23 d, eHourly wage ($1994) 11.11 11.72 12.44Employed if age 1865 64.49 69.57 69.89* Significant differences in means of skin tone categories at the 10-percent level based on Bonferroni multiple comparisontest, where “a” compares very dark to dark; “b” compares very dark to medium; “c” compares very dark to light; “d” comparesdark to medium; “e” compares dark to light; and “f” compares medium to light.
252 AEA PAPERS AND PROCEEDINGS MAY 200
 
Note that measurement error could bias anyestimated effects of skin color. In contrast tostudies in the medical literature that measureskin color by a reflectance meter, skin tone isreported by interviewer observation in theNSBA and MCSUI. Despite skin tone beingreported by trained interviewers, both randomand systematic measurement error may be con-siderable and may bias estimated skin tone ef-fects. Calculations from the DAS showsubstantial disagreement between self-reportsand interviewer reports of skin tone, with amatch for only 65 percent of the sample of black respondents. While disagreement is not neces-sarily identical to measurement error, the dis-parity does have a large random component, asmismatched respondents were about as likely tobe reported by interviewers as darker than aslighter, relative to self-reports. Evidence fromthe MCSUI shows that nonblack interviewerssystematically reported skin tones of black re-spondents as darker than did black interviewers(Mark E. Hill, 2002). If lighter-skinned blackshave higher earnings, but are reported as darker,the negative effect of darker skin tone on wagesis muted. Regressions using the MCSUI, re-stricted to the sample in which black respon-dents were interviewed by black interviewers,suggest such bias is possible, with the estimatednegative effect on wages of dark skin 26 percentgreater for men than in estimates using the fullsample.A possible explanation for the disparate re-sults for education and wages may lie in differ-ential employment rates on the basis of skincolor. Schooling is mandatory up to a certain
T
ABLE
2—S
KIN
-T
ONE
E
FFECTS ON
E
DUCATIONAL
A
TTAINMENT AND
W
AGE
Dependent variable:Education Ln(wage)Female Male Female MalePanel A: National Survey of Black Americans (NSBA)Very dark 
Ϫ
1.156**
Ϫ
1.065* 0.059
Ϫ
0.214**(0.328) (0.441) (0.069) (0.079)Dark 
Ϫ
0.534*
Ϫ
0.793* 0.014
Ϫ
0.175**(0.228) (0.345) (0.042) (0.063)Medium
Ϫ
0.403
ϩ Ϫ
0.476 0.038
Ϫ
0.207**(0.206) (0.335) (0.038) (0.061)
 N 
1047 637 473 366
 p
-value
a
Very dark 
ϭ
dark 0.05 0.47 0.50 0.55Very dark 
ϭ
medium 0.01 0.10 0.74 0.91Dark 
ϭ
medium 0.47 0.19 0.49 0.45Panel B: Multi-City Study of Urban Inequality (MCSUI)Dark 
Ϫ
0.403*
Ϫ
0.968**
Ϫ
0.146**
Ϫ
0.206**(0.171) (0.204) (0.027) (0.032)Medium
Ϫ
0.267
Ϫ
0.507**
Ϫ
0.172**
Ϫ
0.185**(0.163) (0.178) (0.026) (0.035)Light 0.092
Ϫ
0.424
ϩ Ϫ
0.113**
Ϫ
0.146**(0.178) (0.224) (0.038) (0.056)
 N 
2682 1729 1724 1161
 p
-value
a
Dark 
ϭ
medium 0.07 0.00 0.39 0.55Medium
ϭ
light 0.00 0.67 0.05 0.54Dark 
ϭ
light 0.00 0.00 0.22 0.31
 Notes:
ϩ
, *, ** significantly different from 0 at the 10-percent, 5-percent, and 1-percent level,respectively. Standard errors in parentheses, with standard errors in regressions based onMCSUI adjusted for sample stratification design. See text for additional variables included inequations.
a
 p
-value for tests of hypothesis of equality of coefficients on indicated variables.
253VOL. 96 NO. 2 SKIN-TONE DISCRIMINATION AND ECONOMIC OUTCOMES 

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->