Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 49
Measuring the Eject of Occupational
Sex and Race Composition on Earnings
ELAINE SORENSEN
During the past 20 years, the federal
government has instituted laws and regu-
lations to combat economic discrimination
against women and minorities. Despite these
efforts, women and minorities still earn
consiclerably less than white men. Because
progress toward greater equality has been
slow, many policymakers are directing their
interest toward a strategy of equal pay for
comparable worth, or pay equity.
Proponents of comparable worth strate-
gies believe that such an approach addresses
a substantial portion of sex- and race-basec!
pay disparities (National Committee on Pay
Equity, 1987~. They argue that a principal
factor contributing to these pay differentials
is the persistence of occupational segrega-
tion in the U.S. labor market. According
to this view, occupational segregation allows
firms to pay lower wages to workers in jobs
with an overrepresentation of women or
minorities. The purpose of a comparable
worth policy is to eliminate the effect of
occupational segregation on earnings within
a firm once legitimate factors that influence
earnings have been taken into account.
Opponents of comparable worth argue
49
that occupational segregation within a firm
is not a major factor contributing to earnings
disparities. Instead, they argue, differences
in productivity-relatec! characteristics (e.g.,
men have, on average, more work experi-
ence than women) are the major factors
contributing to the persistence of the earn-
ings differentials (Polachek, 1987). Others
claim that a significant portion of the earn-
ings differentials is explained by industrial
(differences (e.g., men are more likely to
work in manufacturing, and women are more
likely to work in services, Johnson and
Solon, 1986~. Comparable worth policies
thus cannot effectively reduce the national
sex- or race-based earnings disparities.
The purpose ofthis research is to measure
the extent to which occupational segregation
by sex anti race contributes to the national
earnings disparities between different sex/
race groups. Other factors contributing to
the national earnings disparities are also
examined, such as industrial and produc-
tivity-related characteristics. These esti-
mates in(licate the magnitude of the phe-
nomenon that a national comparable worth
policy would attempt to reme(ly.
OCR for page 50
50
RESEARCH METHOD
This section develops an appropriate
method for analyzing the effect of occupa-
tional segregation on earnings. It builds on
the basic design applied in comparable worth
studies in the public sector and extends that
approach to the national economy.
Comparable worth studies in the public
sector have measured the extent to which
occupational segregation affects earnings in
two steps. First, they conduct a job eval-
uation of all occupations, typically employ-
ing an a priori factor-point job evaluation
plan. These plans consist of a set of factors
and weights that are expected to reflect the
requirements of a job. The factors usually
fall into four broad categories: skill, effort,
responsibility, and working conditions. The
weights are applied to each factor and in-
dicate its relative importance. An evaluation
team evaluates jobs in terms of each factor
and assigns points commensurate with the
amount of the factor required on the job.
These factor scores are summed for each
job to produce a total point score, or job
evaluation score. Employers often conduct
job evaluations without reference to the
issue of comparable worth. In such cases,
the results from the preexisting job eval-
uation plan are typically used in the com-
parable worth study.
Once the job evaluation plan is com-
pleted, an earnings equation is estimated
with the occupational salary as the depen-
dent variable and an occupation's job eval-
uation score and a variable indicating the
sex composition of the occupation as the
independent variables. The latter variable
can be measured by the proportion of wom-
en in an occupation. It varies from 0 to 1,
with 0 indicating that none of the employees
in the occupation is a woman and 1 indi-
cating that all are women.
The dependent variable, the occupational
salary, is defined as one of the salary levels
within an occupation to which workers may
be assigned. Thus, an occupational salary
PAY EQUITY: EMPIRICAL INQUIRIES
is not the same as an individual salary;
rather, it represents one of the salaries that
an individual could receive if employed in
that job. Comparable worth studies have
tended to examine occupational salaries and
job requirements rather than individual sal-
aries and individual productivity-related
characteristics because in the public sector,
where these studies have been conducted,
occupational salaries and job requirements
are well specified. Moreover, it is the con-
tention of most civil service systems that
occupational salaries reflect the require-
ments of the job and not the characteristics
of the individuals who hold the job.
The functional form of the occupational
earnings equation has varied among com-
parable worth studies in the public sector,
but the following equation typifies the ap-
proach:
W0 = aO + alIO + a2 PFo + uO (1)
where, the subscript 0 indicates the set of
occupations; w is the occupational salary; l
is the occupational job evaluation score; PF
is the proportion of women in an occupation;
and u is the random error term.
It is expected that the coefficient on the
variable measuring the proportion of women
in an occupation will be negative. In other
words, occupational earnings are expected
to be lower in female-dominated jobs even
after controlling for differences in the re-
quirements of the job. The goal of com-
parable worth initiatives is to eliminate this
negative effect on occupational salaries. One
approach for eliminating it is to subtract id
PFo from each occupational salary. Once a
comparable worth policy is implemented in
this manner, occupational salaries would
reflect the requirements of the job and not
the predominant sex of the workers em-
ployed in the occupation.
Occupational segregation by race has been
examined in some comparable worth studies
in the public sector, most notably in the
state of New York. These studies typically
add another independent variable to the
OCR for page 51
EFFECT OF SEX AND RACE ON EARNINGS
earnings equation to reflect the race com-
position of the occupation (National Com-
mittee on Pay Equity, 1987~. In this analysis,
an independent variable that measures the
proportion of minority workers in an oc-
cupation is added to the earnings equation.
Thus, the race composition of an occupation
is treated in the same manner as the sex
composition of an occupation. It is expected
that the coefficient on this variable will be
negative, indicating that earnings decline
as the proportion of minority workers in an
occupation increases. A comparable worth
policy might then eliminate the negative
effect of both variables from the earnings
equation.
This analysis can be extended to individ-
ual earnings if we make an assumption re-
garding the relationship between individual
and occupational earnings. Let us assume
that individual earnings within a firm are a
function of the entry-level occupational wage,
an individuaT's tenure on the job, the sex
and race of the employee, and any education
or previous experience that the employee
may have that is greater than the amount
required on the job. ~ Then, individual earn-
ings can be written in the following manner:
w; = wO + be ti + bisect; - ed,`)
7 /
_ ~ \ _ .
Cl/
+ D2(Xi - To) + b3 Si ~ b4 Rt
+ b5 SRi + Hi
where, wi is the inclividual's wage; tZ is the
individual's tenure on the job; Eli is the
iThe job evaluation score is separated into its con-
stituent factor scores in this model. This allows a
comparison between actual and required levels of
education and experience. Using factor scores rather
than the overall job evaluation score is a basic difference
between policy-capturing and a priori job evaluation
plans. Policy-capturing plans use regression analysis
to determine the appropriate weights for each factor
score. In contrast, a priori plans apply previously
determined weights to each factor and sum those scores
to produce a total factor score. An earnings regression
is then estimated using the total factor score as an
independent variable. Thus, we are assuming that a
policy-capturing approach is used to estimate the oc-
cupational earnings equation.
51
indiviclual's educational attainment; edO is
the level of education required for the job;
xi is the incliviclual's experience level; xO is
the level of experience required for the job;
Si is 1 if the worker is a woman and O if
not; Fi is 1 if the worker is a minority and
O if not; SRi is 1 if the worker is a minority
woman and O if not.
Substituting Equation (1) into Equation
(2), adcling the race composition of the oc-
cupation as an independent variable into
Equation (1), separating the job evaluation
score into its factor scores, and rearranging
terms yield the following equation:
Wi = aO + car Jo + a2 PFo+ a3 PMo
+ C41i + bits. + bier
+ b5 SRi + vi
(3)
where, .! equals a vector of job characteris-
tics (i. e., educational and experience
requirements as well as other job charac-
teristics, such as working con(litions); PF
equals the proportion of women in an oc-
cupation; PM equals the proportion of mi-
norities in an occupation; I equals a vector
of indiviclual characteristics (i.e., educa-
tional attainment, work experience, and ten-
ure); S. R. and SR are defined as above;
and v is the ran(lom error term.
Thus, individual earnings within a firm
are assumed to be a function of the char-
(2) acteristics of the job, the productivity char-
acteristics of the incliviclual, the sex and race
composition of the occupation, and the sex
and race of the incliviclual. It is expected
that the estimated coefficients on the job
characteristics that have individual coun-
terparts (i. e., education ancT experience) will
be (lifferent in the indiviclual earnings equa-
tion than in the occupational earnings equa-
tion. But the estimated coefficients on the
variables measuring the proportion of wom-
en and minorities are expected to be un-
changed. Similarly, it is anticipated that the
estimated coefficients on the sex and race
of the individual will be unchanged between
Equations (2) ant! (31.
It is expected that as and as are negative.
OCR for page 52
52
A comparable worth policy could eliminate
this negative effect by subtracting a2 PFo
and d3 PMo from each occupational salary.
If this is clone, a comparable worth policy
eliminates the effect of the variables mea-
surin~ the proportion of women and mi-
nor~t~es in an occupation from individual
earnings ant] controls for differences in pro-
cluctivity-related characteristics and the sex
and race of the individual workers involved.
Scholars have noted that different sex/
race groups have significantly different es-
timated earnings equations. Thus, it would
be more appropriate to estimate separate
earnings equations for each sex/race group.
In addition, I would like to estimate the
effect of occupational segregation on earn-
ings within firms, since comparable worth
policies ad~lress intrafirm effects of occu-
pational segregation. Unfortunately, nation-
al data samples do not identify an individ-
uaT's firm. Thus, the analysis includes
variables that approximate the firm, such
as region, Standard Metropolitan Statistical
Area (SMSA), union status, employer size,
and major industrial categories. Taking these
issues into account, Equation (3) can be
rewritten for each sex/race group. The fol-
lowing equations are for white males and
white females:
Wwm = do + d'~.Jwm + (~2 PFwm
+ 43 PMwm + d:4 Iwm
+ d'5 Zwm + Vwm
an(l
WWf = eO + en |wf + e2 PFWf + e3 PMWf
+ e4 lwf + eS Zwf + Vwf (5)
where, Z equals a vector of industrial/re-
gional characteristics.
Consequently, with a national (lata set,
it is assumed that the indiviclual earnings
of each sex/race group are a function of job-
related productivity characteristics, the sex
and race composition of the occupation,
individual-related productivity characteris-
tics, ant] industrial/regional characteristics
PAY EQUITY: EMPIRICAL INQUIRIES
(i.e., region, SMSA, major industrial cat-
egories, union status, and employer size).
These earnings equations are different
trom those in most previous economic re-
search on sex- and race-based earnings (lif-
ferentials. First, they include the sex and
race composition of an occupation as in-
dependent variables. These variables are
included to measure the extent to which
occupational segregation affects earnings.
Although most research has not addressed
this issue, a few studies have inclucled these
variables to measure the effect of occupa-
tional crowding on earnings (see Treiman
ant] Hartmann, 1981~.
The second atypical characteristic of these
earnings equations is that they include both
human capital variables and job character-
istics. Most previous models of wage de-
termination have focused on either human
capital or job characteristics, not both. This
analysis includes both types of variables
because of the assumptions made above
about the relationship between occupational
earnings and individual earnings. Those as-
sumptions imply that an adequate assess-
ment of the effect of a comparable worth
policy on individual earnings cannot be made
without both types of variables in the equa-
tion.
Finally, these equations include detailed
variables representing industrial categories
and employer size. These variables are in-
(4) eluded to control for the type of firm as
closely as possible given that the data are
national in scope. Since comparable worth
initiatives are intrafirm policies, such con-
tro} is important.
DISCUSSION OF THE DATA
Earnings equations are estimated using
the Current Population Survey (CPS) from
May and June 1983 (data tapes available
from Bureau of the Census). The sample
refers to all nonagricultural civilian wage
and salary workers who are at least 16 years
old. Four sex/race groups are distinguished:
OCR for page 53
EFFECT OF SEX AND RACE ON EARNINGS
white males, white females, minority males,
and minority females. White males and fe-
males consist of all whites who are not
Hispanic. Minority males and females in-
clude all blacks and Hispanics. Other mi-
nority groups are excluded from the anal-
ys~s.
The CPS is particularly well suited for
this analysis because it has cletailec] occu-
pational information on a large national sam-
ple (approximately 17,000 individuals). Oth-
er surveys have a much smaller sample and/
or no detailed occupational information. The
May and June 1983 matched sample has
particular advantages over other CPS data
because the survey included questions re-
garding job tenure and employer size, in
addition to the usual questions concerning
employment status. Thus, the data provicle
better estimates of a worker's human capital
and more detailed information on the em-
ployer than most CPS data. In addition,
the survey was conducted while the econ-
omy was experiencing relatively normal un-
employment.
The CPS provides the dependent variable
in the analysis, which is the logarithm of
the ratio of usual weekly earnings to usual
weekly hours, referre(l to as "earnings."
The data user! to construct the proportion
of women in an occupation and the pro-
portion of blacks and Hispanics in an oc-
cupation are from the 1980 census. The
three-digit census occupational categories
are used to define occupations in this anal-
ysis. There are 503 such occupations. lob
characteristics are proxied by the following
variables from the Dictionary of Occupa-
tional Titles, as reported by Miller et al.
(1980~: general educational development,
specific vocational preparation, strength,
physical demands, and undesirable envi-
ronmental conditions. The CPS provides
the variables for inclividual characteristics
and industrial setting. Complete definitions
of the variables are given in Appendix A.
Appendix B reports the means of the vari-
ables for each sex/race group.
53
EMPIRICAL RESULTS
First, simple earnings regressions are es-
timated for each sex/race group that include
the proportion of women and the proportion
of minorities in an occupation as the only
inclependent variables. These estimates cle-
scribe the gross relationship between earn-
ings and the sex ant] race composition of
occupations. The first line of Table 2-1 re-
ports the estimated coefficients from these
regressions. For white men, the estimates!
coefficients for the proportion of women and
the proportion of minorities in an occupation
are - .332 and - 2. 498, respectively. This
implies that white male earnings decline by
3.3 percent for each 10 percentage point
increase in the percentage of women in an
occupation. In a(l(lition, white male earn-
ings clecline by 25 percent for each 10
percentage point increase in the percentage
of minorities in an occupation. The esti-
matec] coefficients for the other sex/race
groups are smaller than for white men, but
the magnitudes are similar.
These coefficients (lo not take into account
differences in worker or in(lustrial charac-
teristics. The second row of Table 2-1 re-
ports the estimated coefficients for the sex
an(l race composition variables from the full
regression model. The most striking result
is the dramatic decline in the estimated
coefficients for the proportion of minorities
in an occupation after other independent
variables are accounted for in the model.
In(leed, these estimated coefficients (recline
by 80 to 113 percent after controlling for
other factors. The estimate(l coefficients for
this variable in the full mode} are -.490
for white men; .012 for white women; - .077
for minority men; and .166 for minority
women. In other words, white male earn-
ings decline by 4.9 percent for each 10
percentage point increase in the percentage
of minorities in an occupation. For the other
sex/race groups, however, earnings do not
significantly (decline as the Percentage of
. ~
minorities in an occupation increases in the
OCR for page 54
OCR for page 55
EFFECT OF SEX AND RACE ON EARNINGS
full model. Thus, the negative relationship
between earnings and the concentration of
minorities in an occupation is eliminated
for all sex/race groups except white men
when other factors are taken into account.
In contrast, controlling for other inde-
pendent factors does not dramatically re-
duce the estimated coefficients for the pro-
portion of women in an occupation, except
in the earnings equation for white women.
This estimated coefficient declines by 55
percent for white women, but it declines
by only 18 percent for white men and by
19 percent for minority women, and it ac-
tually increases by 31 percent for minority
men. Thus, the negative relationship be-
tween earnings and the concentration of
women in an occupation remains signifi-
cantly negative for all sex/race groups even
after accounting for other explanatory fac-
tors. The actual estimated coefficients for
this variable are the following: -.272 for
white men; - .151 for white women; - .269
for minority men; and .209 for minority
women. According to these results, white
and minority men earn an average 2.7 per-
cent less per 10 percentage point increase
in the percentage of women in an occu-
pation. For white women, wages decrease
by 1.5 percent per 10 percentage point
increase in the percentage of women in an
occupation. Minority women's wages de-
crease by 2.1 percent for a similar increase
in the percentage of women in an occu-
pation. 2
2Variables indicating the marital status and the pres-
ence of children in the home were also included as
explanatory factors in preliminary earnings regressions.
They were not included, however, in the final versions
of the equations because they are difficult to classify
as legitimate factors for differentiating salaries. None-
theless, the estimated coefficients for the proportion
of women in an occupation in the white and minority
female earnings equations were not affected by the
inclusion of these variables. White (minority) women
still earned an average 1.5 percent (2.1 percent) less
per 10 percentage point increase in the percentage of
women in their occupation. In the white female earn-
ings equation, the estimated coefficients (and their
55
The estimated coefficients for the other
variables in the analysis are reported in
Appendix C. These estimated coefficients
are similar to those found in other research
on wage determination. For example, earn-
ings increase as education and job tenure
increase for all sex/race groups. On the other
hand, only white male earnings significantly
increase with potential experience. In acI-
dition, the earnings of white and minority
men increase as the number of undesirable
working conditions increases (such as noise
and cold temperature), but the earnings of
white an(l minority women are not affected
by the existence of such conditions. Oddly
enough, increasing requirements of strength
on the job decrease male earnings and have
no significant effect on female earnings.
Living in a SMSA increases earnings for all
sex/race groups, except minority males.
Earnings tend to be higher for indivicluals
working in such industries as mining, con-
struction, certain clurable goods industries,
transportation, communications, public util-
ities, and the federal government than in
personal services. Finally, union member-
ship significantly increases earnings for all
sex/race groups.
Using the full regression moclel, the sex-
an(l race-based earnings differentials can be
divided into four components. The first com-
ponent measures the effect of sex- ant] race-
based occupational segregation. The second
component measures the effects of industrial
and regional differences (i. e., two-digit
standard errors) for the dummy variables married,
divorced, and the presence of children in the home-
were .062 (. 016), . 113 (. 019), and - .020 (. 014), re-
spectively. In other words, married and divorced white
women earned 6 and 11 percent more, respectively,
than single white women. White women with children
earned 2 percent less than white women without
children. Similarly, the estimated coefficients (and
their standard errors) for minority women were .061
(. 033), . 135 (. 036), - .052 (. 028~. Thus, married and
divorced minority women earned 6 and 14 percent
more, respectively, than single minority women. Fur-
ther, minority women with children earned 5 percent
less than minority women without children.
OCR for page 56
56
Standard Industrial Classification PSIC] cat-
egories, SMSA, region, union status, and
employer size). The third component con-
sists of differences in job characteristics and
individual productivity-related characteris-
tics (i.e., education, work experience, ten-
ure, and part-time status). The final com-
ponent measures the amount of the earnings
gaps that is unexplained by these factors.
There are two methods for estimating the
extent to which each component contributes
to the national earnings differentials. For
instance, when decomposing the earnings
differential between white women and white
men, the coefficients from either the white
male earnings equation or the white female
earnings equation can be used as weights
in the difference equation. These difference
equations are reported below, first using
white male coefficients as the weights and
then using white female coefficients as the
weights. Since there is no clear reason for
selecting one set of weights over another,
both are used to estimate the decomposition
of the earnings gaps. The average of these
two estimates is then calculated and re-
ported in Table 2-2.
In Clan In Waif = D(Xw~n Xwf)
+ XWf(DE)
and
In Ww7n
In Wwf = E(xwmXwf ~
+ XWm(D E)
where, Xw~n (Xwf) is a vector of mean values
for white males (white females) for the in-
dependent variables in the earnings equa-
tion and D (E) is a vector of coefficients in
the white male (female) earnings equation.
Allocating the national earnings differ-
entials for different sex/race groups into four
components is particularly useful because
it separates the effect of occupational seg-
regation from other factors that are fre-
quently cited as more important explana-
tions for these earnings differences. As stated
before, the goal of a comparable worth policy
is to eliminate the effect of occupational
segregation, which is measured by the first
PAY EQUITY: EMPIRICAL INQUIRIES
component described above. The second
and third components measure the extent
to which differences in industrial segrega-
tion and productivity contribute to the earn-
ings differentials. These are the variables
that are frequently cited as more important
than occupational segregation.
Table 2-2 reports the extent to which
each component contributes to the national
earnings differentials. A substantial amount
of the sex- and race-based earnings dispar-
ities is explained by the sex and race com-
position of an occupation. In particular, the
proportion of women and the proportion of
minorities in an occupation account for 21
percent of the earnings disparity between
white women and white men; 20 percent
is due to the sex composition, and 1 percent
is clue to the race composition of occupa-
tions. These two variables explain 23 per-
cent of the earnings differential between
minority women and white men (21 percent
is due to the sex composition; 2 percent is
due to the race composition). For minority
men, however, both the sex and race com-
position of an occupation play a less im-
portant role in maintaining low relative
earnings. They explain 9 percent of the
6 earnings gap between minority and white
men (3 percent is due to the sex composition;
6 percent is due to the race composition).
In contrast, Table 2-2 shows that differ-
ences in the variables measuring industrial
(7) and regional characteristics (i.e., union sta-
tus, size of employer, two-digit SIC code,
region, and SMSA) account for much less
of the earnings disparities between different
sex/race groups. These variables explain 16
percent of the earnings disparity between
white women and white men and 6 percent
of the earnings disparity between minority
women and white men. They actually in-
crease the earnings of minority men, re-
ducing their earnings gap by 4 percent.
Table 2-2 also shows that variables mea-
suring productivity characteristics (i. e.,
schooling, potential experience, tenure, job
characteristics, and part-time status) explain
a large portion of the national earnings dis-
OCR for page 57
EFFECT OF SEX AND RACE ON EARNINGS
TABLE 2-2 Percentage of Earnings Disparities Accounted for by Different Factors
57
White Female/ Minority Male/ Minority Female/
White Male White Male White Male
Factors Earnings Gap Earnings Gap Earnings Gap
Total $3. 32 $2. 11 $3. 53
Occupational segregation by sex (race) 20% (1%) JO (6%) 21% (2%)
Industrial and regional differences 16% -4% 6%
Differences in job and productivity characteristics 26% 49% 23%
Unexplained residual 39% 46% 48%
NOTES: Variables included as productivity characteristics are schooling, potential work experience, job tenure,
job characteristics, and part-time status. Variables included as industrial and regional characteristics are 42 two-
digit SIC industrial categories, union status, employer size, region, and SMSA. Percentages may not add to 100
because of rounding.
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983); Dictionary
of Occupational Titles as reported in Miller et al. (1980).
parities. These variables explain 26 percent
of the white female/white male earnings gap
and 23 percent of the minority female/white
male earnings gap. Forty-nine percent of the
earnings disparity between minority ant] white
men is clue to these variables.
Although these earnings equations in-
clude an extensive array of explanatory vari-
ables, Table 2-2 shows that differences in
these variables explain only 52 to 61 percent
of the earnings disparities between cli~erent
sex/race groups, which leaves large unex-
plained resicluals.
In summary, these findings suggest that
even though differences between women and
men in productivity and industrial charac-
teristics explain about 40 percent of the na-
tional earnings disparity between women and
men, another 20 percent is due to occupa-
tional segregation by sex, the portion of the
earnings disparity that a comparable worth
policy seeks to eliminate. Thus, this study
finds that a national comparable worth policy
would address a sizable component of the
national sex-based earnings cli~erential.
jobs with a Disproportionate
Number of Women and Minorities
In this section of the paper, alternative
measures of the independent variables mea-
suring the sex and race composition of an
occupation are examined. Many comparable
worth studies use dummy variables that
inclicate the predominant sex or race in an
occupation as explanatory factors in their
earnings equations, rather than the pro-
portion of women or minorities in an oc-
cupation. These dummy variables are clis-
continuous; they equal 1 if there is a
disproportionate number of women or mi-
norities in an occupation and O otherwise.
Most of these studies use the 70 percent
rule to define female (male)-clominated jobs,
which states that any occupation in which
70 percent or more of the employees are
female (male) is a female (male)-dominatecl
job. As others have pointed out, this rule
is arbitrary. Moreover, there is no custom-
ary rule for defining jobs with an overre-
presentation of minorities, or minority-(lom-
inated jobs. Despite these limitations, I
estimate earnings equations for each sex/
race group using the customary 70 percent
rule for defining female-dominatec! jobs and
a (lefinition clevelope(1 for minority-domi-
nated jobs (described below).
Histograms of the variables measuring
the percentage of women and the percent-
age of minorities in an occupation were
examined (see Figures 2-1 and 2-2~. They
are based on the CPS sample, which iclen-
tifies the three-digit census occupational
code for each individual. This code is then
matche(l to data from the 1980 census indi-
cating the percentage of women and minor-
OCR for page 58
58
FIGURE 2-l Percentage of work
force employed, by percentage
female in an occupation. Source:
Current Population Survey data
tapes, May and June 1983;
Bureau of the Census (1983).
FIGURE 2-2 Percentage of work
force employed, by percentage of
minorities in an occupation. Note:
"*" indicates values less than .05
percent. Source: Current
Population Survey data tapes,
May and June 1983; Bureau of the
Census (1983~.
PAY EQUITY: EMPIRICAL INQUIRIES
35
30
IIJ
o
~ 25
CC
O 20
30
C)
LO
2 25
LL
o 'V
1 1
15
10
Q
15
10
5
o
5
35 _
_
o
7
qHq~
5 10 15 20 25 30 35 40 45 50 55 6065 70 75 80 85 90 95 100
PERCENTAGE FEMALE IN AN OCCUPATION
1
5 10152025 035404550 556065707580859095100
PERCENTAGE MINORITY IN AN OCCUPATION
OCR for page 59
EFFECT OF SEX AND RACE ON EARNINGS
ities in that occupational category. Figures 2-
1 and 2-2 show the percentage of the work
force that is employed in each 5 percent
interval of the variables measuring the per-
centage of women or minorities in an occu-
pation. For example, Figure 2-1 reports that
13.5 percent of the work force is employed
in occupations in which, at most, 5 percent
of the employees are women. At the same
time, Figure 2-2 reports that 33 percent of
the work force is employed in occupations in
which between 5 and 10 percent of the em-
ployees are minorities. As Figures 2-1 and
2-2 show, the histograms of the percentage
of women and the percentage of minorities
in an occupation are quite different. The
histogram in Figure 2-1 is almost U-shaped,
with two peaks, one at 5 percent and one at
100 percent female. The histogram in Figure
2-2 has a skewed distribution with one peak
at 10 percent minority.
Since the histograms of these two vari-
ables are so different, the definitions de-
veloped for minority- and white-dominated
jobs are not patterned after the 70 percent
rule used in clefining female- ant] male-
dominatec] jobs. In the sample, minorities
make up 15 percent of the work force. Thus,
a minority (white)-dominated job is defined
as one in which the percentage of minorities
in that occupation is greater (less) than 20
percent (10 percent). Any job that employs
10 to 20 percent minority workers is con-
sidered a minority-integrated job.
Table 2-3 reports the hourly wage that the
average member of each sex/race group can
expect to earn given employment in different
types of occupations. It shows that hourly
earnings for all sex/race groups are substan-
tially affected by employment in jobs that
have a disproportionate number of women
or minorities. For instance, Table 2-3 shows
that an average white man can expect to earn
$9.52 if he is employed in a white male
dominated job. In contrast, if he is employed
in a job that has a disproportionate number
of minority female workers he can expect to
earn $7.57, 20 percent less than the salary
59
received in a white male dominated job.
Similarly, an average minority woman can
expect to earn $5.02 if she is employed in a
job that has a disproportionate number of
minority female workers, but she can expect
to earn $6.31 if she is employecl in a job held
predominantly by white men. (The full regres-
sion results for these earnings equations are
available from the author.)
Empirical Results by Industrial Sector
Finally, separate earnings equations are
estimated for white men and white women
within each of the following major inclustrial
sectors: manufacturing, public sector, and
nonmanufacturing industries. These equa-
tions use the proportion of women and
minorities in the worker's occupation as the
measure of occupational segregation. The
results are reported in Appendix D. From
these results, comparisons can be macle
across different sectors. Other researchers
have argued, for example, that the variable
measuring the proportion of women in an
occupation affects earnings in the public
sector more than in the private sector and
that this explains, in part, why the public
sector has taken the initiative in this area
(see Johnson and Solon, 1986~.
The results show that both the sex and
race composition of an occupation strongly
affect white mate and white female earnings
in the nonmanufacturing private sector. White
male (female) earnings decline 2.7 (2.1) per-
cent for each 10 percentage point increase
in the percentage of women in an occupation.
Moreover, if the percentage of minorities in
an occupation increases 10 percentage points,
white male (female) earnings (recline 9.7 (3.5)
percent. In this sector, the ratio of white
female to white male earnings expressed in
percentage terms is 63 percent, which leaves
an earnings disparity between white women
and white men of 37 percentage points. Ap-
proximately 26 percent of this earnings dis-
parity is attributable to the sex and race
composition of the occupation.
OCR for page 60
60
PAY EQ UIf"Y: EMPIRICAL INQ UIRIES
TABLE 2-3 Predicted Hourly Earnings for Each Sex/Race Group by Type of Occupation
White Male White Female Minority Male Minority Female
Group Dominated Job Dominated Job Dominated Job Dominated Job
White males $9.52 $8.19 $~.80 $7.57
White females 6.11 5.51 6.03 5.45
Minority males 7.60 6.44 6.92 5.86
Minority females 6.31 5.52 5.74 5.02
NOTES: Definitions for the different types of occupations are the following: a white male dominated job is any
job in which 90 percent or more of the employees are white and not Hispanic and 70 percent or more are male;
a white female dominated job is any job in which 90 percent or more of the employees are white and not Hispanic
and 70 percent or more are female; a minority male dominated job is any job in which 20 percent or more of the
employees are Hispanic or black and 70 percent or more are male; a minority female dominated job is any job in
which 20 percent or more of the employees are Hispanic or black and 70 percent or more are female.
Predicted hourly earnings are calculated from earnings regressions that are available from the author on request.
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983~; Dictionary
of Occupational Titles as reported in Miller et al. (1980~.
In the public sector, white male and white
female earnings are significantly influenced
by the proportion of women in an occupation,
but the proportion of minorities in an oc-
cupation has no significant effect. White male
(female) earnings decline 4.2 (1.8) percent for
each 10 percentage point increase in the
percentage of women in an occupation. The
female/male earnings ratio in this sector is
71 percent, a much higher figure than in the
nonmanufacturing and manufacturing sectors.
Although the earnings disparity is smaller in
the public sector, one-third of it is explained
by the sex composition of an occupation. As
a result, this variable has its largest impact
in the public sector.
In the manufacturing sector, only white
male earnings are significantly affected by the
sex composition of an occupation, declining
1.8 percent for each 10 percentage point
increase in the percentage of women in an
occupation. The race composition of an oc-
cupation does not significantly influence white
male or white female earnings. The white
female/male earnings ratio in this sector is
63 percent, the same as in the nonmanufac-
turing sector, which leaves an earnings dis-
parity of 37 percentage points. Only 6 per-
cent ofthis earnings disparity can be explained
by the sex composition of an occupation.
In brief, these results show that the sex
composition of an occupation is an important
factor in determining wages in the public
and nonmanufacturing sectorsit explains
one-third of the earnings gap between white
women and white men in the government
sector and one-quarter in the nonmanufac-
turing sector. It does not appear to be as
influential in the manufacturing sector it
explains only 6 percent of this earnings
disparity. For white women and white men,
the race composition of an occupation was
significant only in the nonmanufacturing
sector, reducing earnings by 9.7 and 3.5
percent, respectively, for each 10 percent-
age point increase in the proportion of mi-
... . .
norltles in an occupation.
SUMMARY AND CONCLUSION
This study argues that the purpose of a
comparable worth policy is to eliminate the
effect of occupational segregation by sex and
race from earnings within firms. It then es-
timates this effect as closely as possible using
a national data set, the Current Population
Survey. It finds that occupational segregation
by sex significantly affects the earnings of all
sex/race groups, even after controlling for
differences in productivity and industrial char-
acteristics. It accounts for 20 percent of the
earnings disparity between white women and
OCR for page 61
EFFECT OF SEX AND RACE ON EARNINGS
white men, 21 percent of the earnings dis-
parity between minority women and white
men, and 3 percent of the earnings disparity
between minority and white men. At the
same time, occupational segregation by race
is not a significant factor influencing earnings
once other factors are accounted for, except
for white men. It does, however, explain 6
percent of the earnings disparity between
minority and white men ant] 2 percent of the
earnings disparity between minority women
and white men.
Contrary to earlier studies Johnson and
Solon, 1986), this research finds that in-
dustrial and regional differences explain
considerably less of the national earnings
disparities between different sex/race groups
than do both the sex and race composition
of an occupation. These factors accounted]
for only 16 percent of the white female/
male earnings gap and 6 percent of the
minority female/white male earnings gap,
and they increased the earnings of minority
mates, reducing their earnings gap by 4
percent. In contrast, productivity-related
differences explained considerably more of
the earnings disparities than industrial and
regional differences; the former explained
26 percent of the earnings gap between
white women and white men, 23 percent
of the gap between minority women and
white men, and 49 percent of the gap be-
tween minority and white men.
Finally, the study fincis that the effect of
occupational segregation on earnings de-
pends on the industrial sector. In particular,
occupational segregation by sex explains one-
third of the earnings disparity between white
women and white men in the public sector
ancT one-fourth of this earnings gap in the
nonmanufacturing private sector. But, in
the manufacturing sector, it explains only
6 percent of this earnings differential. This
suggests that a national comparable worth
policy would address a larger phenomenon
in the public and nonmanufacturing private
sectors than in the manufacturing sector.
In conclusion, the study suggests that a
61
national comparable worth policy would seek
to eliminate approximately 20 percent of
the white female/male earnings disparity.
Moreover, such a policy would address more
of the earnings disparity between minority
women and white men than between any
other sex/race groups. On the other hand,
this policy approach would affect only about
10 percent of the earnings gap between
minority and white men. If a comparable
worth policy is implemental on a smaller
scale, of course, its ability to remedy these
national earnings disparities would! be re-
cluced. Further, the initial gains of a national
comparable worth policy could be atten-
uated by secondary effects caused by the
policy.
REFERENCES
Aldrich, Mark, and Robert Buchele
1985 The Economics of Comparable Worth. Cam-
bridge, Mass.: Ballinger.
Bureau of the Census
1983 Detailed Occupation and Years of School
Completed by Age, for the Civilian Labor
Force by Sex, Race, and Spanish Origin:
1980. PC80-S1-8. Washington, D.C.: U. S.
Government Printing Office.
Johnson, George, and Gary Solon
1986 Estimates of the direct effects of comparable
worth policies. American Economic Review
7645~:1117-1125.
Miller, Ann R., Donald J. Treiman, Pamela S. Cain,
and Patricia A. Roos, eds.
1980 Work, Jobs, and Occupations: A Critical Re-
view of the Dictionary of Occupational Titles.
National Research Council, Committee on
Occupational Classification and Analysis.
Washington, D.C.: National Academy Press.
National Committee on Pay Equity
1987 Pay Equity: An Issue of Race, Ethnicity, and
Sex. Washington, D.C.: National Committee
on Pay Equity.
Polachek, Solomon William
1987 Occupational segregation and the gender wage
gap. Population Research and Policy Review
6:47-67.
Treiman, Donald J., and Heidi I. Hartmann, eds.
1981 Women, Work, and Wages: Equal Pay for
Jobs of Equal Value. National Research Coun-
cil Committee on Occupational Classification
and Analysis. Washington, D. C.: National
Academy Press.
OCR for page 62
62
APPENDIX A Variable Definitions
PAY EQUITY: EMPIRICAL INQUIRIES
Variable Names
Definitions
Log wage Logarithm of usual weekly earnings divided by usual weekly hours
Occupational variables
Proportion female Proportion of women in a three-digit census occupational category
Proportion minority Proportion of blacks and Hispanics in a three-digit census occupational cat-
egory
Human capital variables
Schooling Number of years of schooling completed
Experience (potential) Potential years of experience, computed as age minus education minus 6
Experience squared Potential experience squared
Tenure Number of years worked for current employer
Tenure squared Tenure squared
Voluntary part time 1 if voluntarily employed part time, 0 otherwise
Involuntary part time l if involuntarily employed part time, 0 otherwise
Job characteristics
G.E. D. Formal and informal education required to perform the job (ranges from 1
to 6)
S.V.P. Specific vocational training required to perform the job (ranges from 1 to
9)
Strength Strength involved in performing the job (ranges from l [sedentary] to 5
Every heavy work])
Physical demands Number of physical demands required on the job (e. g., climbing, stooping,
kneeling, and reaching; ranges from 0 to 4)
Environment Number of environmental conditions existing on the job (e. g., extreme cold
or heat, noise, or hazards; ranges from 0 to 6)
Industrial/regional variables
Northest 1 if lives in Northeast region, 0 otherwise
North Central 1 if lives in North Central region, 0 otherwise
West 1 if lives in West region, 0 otherwise
Large SMSA 1 if lives in a Standard Metropolitan Statistical Area (SMSA) with a popula-
tion of at least 3 million people, 0 otherwise
Medium SMSA 1 if lives in an SMSA with a population of at least 1 million people, but less
than 3 million people, 0 otherwise
Small SMSA 1 if lives in an SMSA with a population of less than 1 million people, 0
otherwise
Mining through forestry Dummy variables that equal 1 if employed in the relevant two-digit SIC
industrial cate~orv. 0 otherwise
~ ~ .
Local government 1 if employed in local government, 0 otherwise
State government l if employed in state government, 0 otherwise
Federal government 1 if employed in federal government, 0 otherwise
Small firm 1 if employed in a location with 25 to 99 employees, 0 otherwise
Medium firm 1 if employed in a location with 100 to 499 employees, 0 otherwise
Large firm 1 if employed in a location with 500 to 999 employees, 0 otherwise
Very large firm 1 if employed in a location with 1,000 or more employees, 0 otherwise
Union member 1 if a union member, 0 otherwise
NOTES: G.E.D. = General Educational Development; S.V.P. = Specific Vocational Preparation.
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983); Dictionary
of Occupational Titles as reported in Miller et al. (1980).
OCR for page 63
EFFECT OF SEX AND RACE ON EARNINGS
APPENDIX B Means of Variables in Log Wage Equations
63
White White Minority Minority
Variable Names Males Females Males Females
Log wage 2.1955 1.7349 1.9284 1.6970
Constant 1.0000 1.0000 1.0000 1.0000
Proportion female 0.2415 0.6782 0.2681 0.6762
Proportion minority 0.1354 0.1486 0.1886 0.1944
Female-dominated job 0.0564 0.5696 0.0881 0.5626
Female-integrated job 0.2342 0.2834 0.2576 0.3078
Minority-dom. job 0.1990 0.2300 0.4320 0.4478
Min.-integ. job 0.4005 0.3757 0.3960 0.3452
Schooling 14.2052 14.0560 12.5319 13.1374
Experience (potential) 16.8770 17.7282 18.0763 17.6478
Experience squared 459.0151 500.2187 502.8012 478.4948
Tenure 8.1440 5.7788 6.7563 6.0026
Tenure squared 141.3462 77.2591 99.9474 73.7422
G.E.D. 3.8316 3.7799 3.2781 3.4656
S.V.P. 5.6754 5.1133 4.7283 4.6418
Strength 2.3372 1.8947 2.6721 2.0817
Physical demands 1.7245 1.4806 2.0371 1.6112
Environment 0.6049 0.1973 0.7700 0.3435
Northeast 0.2328 0.2336 0.1628 0.1365
North Central 0.2944 0.2935 0.1208 0.1365
West 0.1679 0.1753 0.2424 0.2070
Large SMSA 0.1251 0.1245 0.2609 0.2104
Medium SMSA 0.2456 0.2503 0.2710 0.2826
Small SMSA 0.3022 0.2839 0.2517 0.2896
Mining 0.0188 0.0043 0.0109 0.0026
Construction 0.0741 0.0097 0.0839 0.0043
Lumber 0.0072 0.0018 0.0193 0.0017
Furniture 0.0070 0.0042 0.0050 0.0026
Stone 0.0085 0.0039 0.0109 0.0078
Primary metals 0.0148 0.0028 0.0117 0.0035
Fabricated metals 0.0214 0.0091 0.0252 0.0070
Machinery, exe. elect. equip. 0.0438 0.0145 0.0294 0.0087
Electrical equipment 0.0254 0.0204 0.0134 0.0235
Automobiles 0.0171 0.0027 0.0185 0.0078
Aircraft 0.0091 0.0042 0.0059 0.0026
Other transport equip. 0.0093 0.0015 0.0042 0.0043
Professional equip. 0.0102 0.0078 0.0042 0.0043
Toys 0.0009 0.0012 0.0008 0.0035
Miscellaneous 0.0033 0.0036 0.0025 0.0035
Food 0.0209 0.0124 0.0268 0.0209
Tobacco 0.0013 0.0004 0.0008 0.0009
Textiles 0.0055 0.0070 0.0076 0.0165
Apparel 0.0039 0.0162 0.0067 0.0443
Paper 0.0097 0.0054 0.0059 0.0052
Printing 0.0242 0.0168 0.0151 0.0070
Chemicals 0.0182 0.0090 0.0117 0.0087
Petroleum 0.0040 0.0007 0.0034 0.0026
Rubber and plastics 0.0102 0.0057 0.0185 0.0052
Leather 0.0012 0.0036 0.0050 0.0070
Transportation 0.0379 0.0141 0.0436 0.0070
Communications 0.0153 0.0159 0.0134 0.0322
Utilities 0.0169 0.0048 0.0159 0.0017
Wholesale trade 0.0640 0.0300 0.0411 0.0157
Continued
OCR for page 64
64
APPENDIX B Continued
PAY EQUITY: EMPIRICAL INQUIRIES
White White Minority Minority
Variable Names Males Females Males Females
Retail trade 0.1418 0.2019 0.1409 0.1565
Banking 0.0191 0.0464 0.0159 0.0357
Insurance 0.0256 0.0387 0.0336 0.0348
Business 0.0289 0.0262 0.0210 0.0235
Repair 0.0179 0.0054 0.0227 0.0043
Entertainment 0.0111 0.0087 0.0101 0.0035
Hospitals 0.0130 0.0631 0.0243 0.0617
Medical, exe. hosp. 0.0057 0.0584 0.0042 0.0504
Educational 0.0119 0.0246 0.0126 0.0174
Social 0.0030 0.0145 0.0067 0.0296
Professional services 0.0338 0.0428 0.0151 0.0183
Forestry 0.0008 0.0003 0.0000 0.0000
Local government 0.0876 0.1260 0.1049 0.1565
State government 0.0411 0.0499 0.0302 0.0496
Federal government 0.0428 0.0306 0.0721 0.0530
Small firm 0.2376 0.2517 0.2584 0.2687
Medium firm 0.2059 0.2090 0.2030 0.2443
Large firm 0.0598 0.0640 - 0.0663 0.0800
Very large firm 0.1318 0.0969 0.1158 0.1209
Voluntary part time 0.0688 0.1996 0.0587 0.1078
Involuntary part time 0.0348 0.0604 0.0705 0.0843
Union member 0.2796 0.1894 0.3532 0.2765
NOTES: G.E.D. = General Educational Development; S.V.P. = Specific Vocational Preparation.
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983); Dictionary
of Occupational Titles as reported in Miller et al. (1980~.
APPENDIX C Estimated Coefficients (and Standard Errors) in Log Wage Equations
White White Minority Minority
Variable Names Males Females Males Females
Constant 1.3010 * 0.3426 * 1.0356 * 0.8256*
(0.0960) (0.0828) (0.2066) (0.1703)
Proportion female 0.2716 * - 0.1511 * - 0.2695 * - 0.2089*
(0.0342) (0.0291) (0.0751) (0.695)
Proportion minority 0.4902 * 0.0125 - 0.0771 0.1659
(0.1326) (0.1233) (0.2801) (0.2357)
Schooling 0.0415 * 0.0457 * 0.0266 * 0.0264*
(0.0029) (0.0034) (0.0054) (0.0059)
Experience (potential) 0.0076 * - 0.0030 * 0.0040 - 0.0086*
(0.0015) (0.0014) (0.0035) (0.0032)
Experience squared 0.0002 * 0.0001 * - 0.0001 0.0001
(0.0000) (0.0000) (0.0001) (0.0001)
Tenure 0.0292 * 0.0324 * 0.0396 * 0.0369*
(0.0021) (0.0023) (0.0052) (0.0054)
Tenure squared - 0.0006 * - 0.0007 * - 0.0009 * - 0.0009*
(0.0001) (0.0001) (0.0002) (0.0002)
OCR for page 65
EFFECT OF SEX AND RACE ON EARNINGS
APPENDIX C Continued
65
White White Minority Minority
Variable Names Males Females Males Females
G. E. D. - 0.0351 0.0729 * 0.0918 0.1086*
(0.0190) (0.0203) (0.0482) (0.0462)
S.V.P. 0.0452 * 0.0386 * - 0.0021 0.0011
(0.0084) (0.0098) (0.0201) (0.0228)
Strength - 0.0727 * - 0.0021 - 0.0777 * - 0.0104
(0.0174) (0.0131) (0.0391) (0.0311)
Physical demands - 0.0122 * 0.0497 * - 0.0027 - 0.0016
(0.0100) (0.0114) (0.0263) (0.0261)
Environment 0.0350 0.0029 0.0539 * - 0.0343
(0.0127) (0.0194) (0.0245) (0.0309)
Northeast 0.0148 - 0.0001 0.1023 * 0.0936*
(0.0160) (0.0164) (0.0461) (0.0442)
North Central 0.0208 - 0.0184 0.1244 * 0.1554*
(0.0150) (0.0153) (0.0480) (0.0421)
West 0.1366 * 0.1390 * 0.2293 * 0.1454*
(0.0175) (0.0178) (0.0382) (0.0369)
Large SMSA 0.0941* 0.1440* 0.0203 0.1343*
(0.0196) (0.0200) (0.0464) (0~0457)
Medium SMSA 0.0877* 0.1271* 0.0667 0.2004*
(0.0154) (0.0155) (0.0414) (0 0377)
Small SMSA 0.0374* 0.0479* - 0.0122 0.0899*
(0.0143) (0.0147) (0.0414) (0.0367)
Mining 0.3739 * 0.4655 * 0.5578 * 0.1428
(0.0658) (0.0928) (0.1580) (0.2507)
Construction 0.3331 * 0.2928 * 0.2234 * 0.0866
(0.0560) (0.0673) (0.1037) (0.1991)
Lumber 0.1193 0.0970 - 0.0561 - 0.1047
(0.0833) (0.1381) (0.1330) (0.3030)
Furniture 0.0730 0.3622 * 0.3884 - 0.2860
(0.0842) (0.0943) (0.2040) (0.2498)
Stone 0.0592 0.2203 * - 0.0719 0.1222
(0.0795) (0.0974) (0.1541) (0.1539)
Primary metals 0.2229* 0.1984 0.2686 0.2654
(0.0696) (0.1120) (0.1528) (0.2225)
Fabricated metals 0.1555* 0.3044* 0.1396 0.0984
(0.0642) (0.0691) (0.1208) (0.1650)
Machinery, exe. elect. equip. 0.2482* 0.2902* 0.0915 0.1311
(0.0582) (0.0590) (0.1174) (0.1488)
Electrical equipment 0.1877* 0.2268* 0.1971 0.1446
(0.0625) (0.0535) (0.1446) (0.1054)
Automobiles 0.2692 * 0.3413 * 0.0916 0.3880*
(0.0684) (0.1147) (0.1328) (0.1584)
Aircraft 0.2006 * 0.3358 * 0.3849 * - 0.2570
(0~0785) (0.0954) (0.1955) (0.2503)
Other transport equip. 0.1339 0.2262 0.2117 0.0048
(0.0782) (0.1509) (0.2242) (0.2008)
Professional equip. 0.1510 * 0.1676 * 0.0846 0.1433
(0.0757) (0.0733) (0.2203) (0.1985)
Toys 0.0544 0.3142 0.0232 - 0.0145
(0.1919) (0.1676) (0.4662) (0.2202)
Miscellaneous - 0.0710 0.0811 0.0241 - 0.0649
(0.1086) (0.1007) (0.2767) (0.2203)
Continued
OCR for page 66
66
APPENDIX C Continued
PAY EQUIP: EMPIRICAL INQUIRIES
White White Minority Minority
Variable Names Males Females Males Females
Food 0.1295 * 0.2303 * 0.2673 * 0.1870
(0.0645) (0.0618) (0.1198) (0.1100)
Tobacco 0.2635 0.2316 0.5054 - 0.1910
(0.1634) (0.2692) (0.4638) (0.4228)
Textiles 0.0427 0.1562 * 0.1676 0.2023
(0.0910) (0.0768) (0.1760) (0.1179)
Apparel 0.0709 0.0893 - 0.2527 - 0.0205
(0.1031) (0.0572) (0.1831) (0.0870)
Paper 0.1849 * 0.2561 * 0.1338 0.2202
(0.0770) (0.0847) (0.1927) (0.1820)
Printing 0.0592 0.1201 * 0.2550 - 0.1211
(0.0624) (0.0558) (0.1386) (0.1608)
Chemicals 0.3101 * 0.2902 * 0.3076 * 0.2541
(0.0665) (0.0698) (0.1500) (0.1502)
Petroleum 0.3341 * 0.4492 * 0.5412 * - 0.0892
(0.1020) (0.2103) (0.2443) (0.2507)
Rubber and plastics 0.1751 * 0.1557 0.0840 0.1528
(0.0757) (0.0831) (0 1299) (0.1828)
Leather 0.1311 0.1805 - 0.0913 0.1675
(0.1708) (0.1014) (0.2044) (0.1627)
Transportation 0.2523 * 0.3919 * 0.2839 * 0.1652
(0.0591) (0.0592) (0.1095) (0.1610)
Communications 0.2899 * 0.3522 * 0.3104 * 0.2742*
(0.0691) (0.0577) (0.1458) (0.0977)
Utilities 0.2641 * 0.2955 * 0.2608 0.2347
(0.0673) (0.0892) (0.1361) (0.3027)
Wholesale trade 0.1581* 0.1785* 0.0257 -0.0176
(0.0562) (0.0476) (0.1096) (0.1183)
Retail trade - 0.0336 - 0.0282 0.0251 - 0.0887
(0.0530) (0.0372) (0.0944) (0.0727)
Banking 0.2366 * 0.1691 * 0.1799 0.0280
(0.0652) (0.0436) (0 1358) (0.0944)
Insurance 0.2228 * 0.1761 * - 0.0491 0.0791
(0.0618) (0 0451) (0.1132) (0.0950)
Business 0.1589 * 0.2774 * 0.0317 - 0.0228
(0.0606) (0.0490) (0.1272) (0.1033)
Repair 0.0894 - 0.0403 0.2425 0.0126
(0.0663) (0.0844) (0.1257) (0.1978)
Entertainment 0.0767 - 0.1511 * 0.2121 - 0.0514
(0.0733) (0.0696) (0.1582) (0.2171)
Hospitals 0.0729 0.1567 * 0.1658 0.1041
(0.0718) (0.0435) (0.1233) (0.0850)
Medical, exe. hosp. 0.1241 0.1587* 0.0194 0.0011
(0.0900) (0.0416) (0.2242) (0.0848)
Educational - 0.1080 0.0120 0.1480 - 0.0569
(0.0730) (0.0504) (0.1471) (0.1162)
Social - 0.2371 * - 0.0952 0.3112 - 0.0715
(0.1142) (0.0585) (0.1836) (0.0967)
Professional services 0.0130 0.2067* 0.1365 0.1006
(0.0605 ~ (0.0442) (0.1397) (0.1129)
Forestry 0.1783 0.4155 0.0000 0.0000
(0.2063) (0 3285) (0.0000) (o.
OCR for page 67
EFFECT OF SEX AND RACE ON EARNINGS
APPENDIX C Continued
White White Minority Minority
Variable Names Males Females Males Females
Local government 0.0378 0.1082* 0.0424 - 0.0157
(0.0553) (0.0397) (0.0974) (0.0731)
State government 0.0287 0.1414* 0.1469 0.0259
(0.0590) (0.0438) (0.1177) (0.0855)
Federal government 0.2646 * 0.2431 * 0.2402 * 0.1471
(0.0589) (0.0485) (0.1023) (0.0857)
Small firm 0.0768* 0.0388* 0.0413 0.0876*
(0.0153) (0.0154) (0 0363) (0.0353)
Medium firm 0.1127* 0.0889* 0.1396* 0.0547
(0.0172) (0.0172) (0.0420) (0.0384)
Large firm 0.1699 * 0.1559 * 0.2458 * 0.1213*
(0.0262) (0.0268) (0.0624) (0 0557)
Very large firm 0.1991* 0.1607* 0.1599* 0.1001*
(0.0216) (0.0237) (0 0538) (0.0504)
Voluntary part time 0.4939 * - 0.3224 * - 0.4461 * - 0.3139*
(0.0247) (0.0155) (0.0604) (0.0432)
Involuntary part time 0.2747 * - 0.2939 * - 0.2154 * - 0.3051 *
(0.0314) (0.0250) (0.0540) (0.0485)
Union member 0.1183* 0.0971* 0.1078* 0.1017*
(0.0144) (0.0167) (0.0322) (0.0324)
Adj. R2 0.4131 0.4168 0.4096 0.4124
N = 7,762 6,675 1,150 1,192
NOTES: G.E.D. = General Educational Development; S.V.P. = Specific Vocational Preparation.
*p < .05.
67
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983), Dictionary
of Occupational Titles as reported in Miller et al. (1980~.
APPENDIX D Estimated Coefficients (and Standard Errors) in Log Wage Equations, by
Industrial Sector
Public Sector Nonmanufacturing Manufacturing
White White White White White White
Variable Names Males Females Males Females Males Females
Constant 1.5040 * 0.3224 * 1.5926 * 0.5880 * 0.7828 * 0.4926*
(0.1802 ~ (0.1537) (0.1137) (0.0982) (0.1979) (0.2406)
Proportion female - 0.4241 * - 0.1826 * - 0.2725 * - 0.2166 * - 0.1837 * 0.0235
(0.0746) (0.0564) (0.0451) (0.0372) (0.0780) (0.0882)
Proportion minority 0.0209 0.2009 - 0.3541 * - 0.3541 * 0.2095 - 0.2637
(0.2989) (0.2407) (0.1827) (0.1681) (0.2840) (0.2934)
Schooling 0.0376 * 0.0546 * 0.0411 * 0.0418 * 0.0467 * 0.0454*
(0.0068) (0.0071) (0.0041) (0.0045) (0.0050) (0.0074)
Experience 0.0018 0.0004 0.0112 * - 0.0024 0.0017 - 0.0078*
(potential) (0.0039) (0.0030) (0.0020) (0.0018) (0.0030) (0.0032)
Experience squared - 0.0001 0.0000 - 0.0002 * 0.00005 -0.00004 0.0002*
(0.0001) (0.0001) (0.00004) (0.00004) (0.0001) (0.0001)
Continued
OCR for page 68
68
APPENDIX D Continue~
PAY EQUITY: EMPIRICAL INQUIRIES
Public Sector Nonmanufacturing Manufacturing
.
White White White White White White
Variable Names Males Females Males Females Males Females
Tenure 0.0325 * 0.0320 * 0.0320 * 0.0337 * 0.0265 * 0.0258*
(0.0054) (0.0048) (0.0030) (0.0031) (0.0035) (0.0047)
Tenure squared 0.0007 * - 0.0007 * - 0.0007 * - 0.0007 * - 0.0005 * - 0.0006*
(0.0002) (0.0002) (0.0001) (0.0001) (0.0001) (0.0002)
G.E.D. 0.0297 0.0923* -0.0863* 0.0615* 0.0796 0.1025
(0.0388) (0.0425) (0.0259) (0.0267) (0.0435) (0.0540)
S.V. P. 0.0100 0.0321 0.0556 * 0.0433 * 0.0351 0.0089
(0.0204) (0.0228) (0.0110) (0.0123) (0.0181) (0.0248)
Strength - 0.0661 0.0187 - 0.1143 * - 0.0422 * 0.0832 * 0.0845
(0.0366) (0.0254) (0.0238) (0.0162) (0.0409) (0.0446)
Physical demands - 0.0719 * 0.0364 0.0131 0.1119 * - 0.0932 * - 0.0595
(0.0261) (0.0221) (0.0132) (0.0154) (0.0219) (0.0306)
Environment 0.0605 * 0.0096 0.0506 * 0.0382 - 0.0005 - 0.0098
(0.0257) (0.0408) (0.0183) (0.0267) (0.0258) (0.0369)
Northeast 0.0394 0.0404 0.0008 - 0.0249 0.0326 0.0768*
(0.0396) (0.0356) (0.0224) (0.0217) (0.0282) (0.0339)
North Central 0.0009 - 0.0563 0.0251 - 0.0220 0.0508 0.0577
(0.0347) (0.0315) (0.0211) (0.0202) (0.0264) (0.0322)
West 0.1691 * 0.1464 * 0.1282 * 0.1401 * 0.1401 * 0.1213*
(0.0388) (0.0354) (0.0236) (0.0233) (0.0365) (0.0420)
Large SMSA 0.1300* 0.0493 0.0970* 0.1824* 0.0872* 0.1485*
(0.0456) (0.0432) (0.0277) (0.0262) (0.0347) (0.0423)
Medium SMSA 0.1369* 0.0913* 0.0938* 0.1481* 0.0412 0.1265*
(0.0372) (0.0329) (0.0213) (0.0203) (0.0284) (0.0342)
Small SMSA 0.0378 0.0542 0.0413 * 0.0509 * 0.0341 0.0365
(0.0325) (0.0296) (0.0204) (0.0197) (0 0249) (0.0307)
Mining 0.2462 * 0.2972 *
(0.0468) (0.0924)
Construction 0.2056 * 0.1115
(0.0308) (0.0628)
Transportation 0.1195 * 0.2293 *
(0.0360) (0.0524)
Communications 0.1346 * 0.1880 *
(0.0514) (0.0516)
Utilities 0.1286 * 0.1222
(0.0488) (0.0882)
Wholesale trade 0.0642* 0.0256
(0.0293) (0.0369)
Retail trade 0.1304 * - 0.1424 *
(0.0232) (0.0210)
State government 0.0079 0.0590
(0.0349) (0.0319)
Federal 0.2514 * 0.1843 *
government (0.0363) (0.0403)
Durable 0.0260 0.0788*
manufacturing (0.0211) (0.0249)
Small f~rm 0.0202 - 0.0483 0.0876 * 0.0561 * 0.0484 0.0940*
(0.0372) (0.0327) (0.0197) (0 0194) (0.0349) (0.0437)
Medium firm 0.0326 - 0.0281 0.1440 * 0.1386 * 0.0695 * 0.1460*
(0.0397) (0.0364) (0.0245) (0.0224) (0 0327) (0.0407)
Large firm 0.0314 - 0.0209 0.2442* 0.2246* 0.1488* 0.2540*
(0.0580) (0.0563) (0.0436) (0.0353) (0.0417) (0.0513)
OCR for page 69
EFFECT OF SEX AND RACE ON EARNINGS
APPENDIX D Continued
Public Sector Nonmanufacturing Manufacturing
White White White White White White
Variable Names Males Females Males Females Males Females
69
Very large firm 0.0611 0.0051 0.2232 * 0.2149 * 0.2429 * 0.2907*
(0.0452) (0.0427) (0.0375) (0.0334) (0.0353) (0.0474)
Voluntary part time 0.5652 * - 0.3105 * - 0.4322 * - 0.3247 * - 0.8767 * - 0.3587*
(0.0640) (0.0356) (0.0300) (0.0187) (0.0759) (0.0566)
Involuntary part 0.4105 * - 0.3899 * - 0.2835 * - 0.2991 * - 0.1307 - 0.2565*
time (0.1206 ~ (0.0752 ~ (0.0369 ~ (0.0306) (0.0795 ~ (0.0534)
Union member 0.0366 0.0867* 0.1830* 0.0933* 0.0685* 0.0785*
(0.0292) (0.0266) (0.0228) (0.0277) (0.0243) (0.0310)
Adj. R2 0.3693 0.3723 0.3411 0.3711 0.4240 0.3896
N = 1,331 1,378 2,150 1,035 4,281 4,262
NOTES: G.E. D. = General Educational Development; S.V. P. = Specific Vocational Preparation.
*p < .05.
SOURCES: Current Population Survey data tapes, May and June 1983; Bureau of the Census (1983); Dictionary
of Occupational Titles as reported in Miller et al. (1980).
Representative terms from entire chapter:
comparable worth
