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 70
Erects of Excess Supply on the
Wage Rates of Young Women
ALICE NAKAMURA and MASAO NAKAMURA
In this paper we begin with the premise
that wages are determined over time by
the interaction of supply and demand forces
within labor markets. It is assumed that
excess labor supply conditions wit! tend to
Tower wage rates. Thus, we would expect
wages to be lower in a relative sense in
labor markets in which the supply of labor
is more abundant than the demand. Berg-
mann (1974, 1986) argues that excess supply
pressures of this sort have been particularly
severe in female labor markets ant] that this
crowding is one important reason why wom-
en s wages are as low as they are. In this
paper we address the related question of
whether the severity of crowding pressures
on wage rates diners among various groups
of working women. Reasons for expecting
this to be the case are advanced, and sup-
porting empirical evidence is presented an
discussed.
Pay equity programs can be thought of
as one type of measure for making female
wage rates in jobs covered by these pro-
grams less vulnerable to crow(ling effects
originating in female labor markets. We en(1
the paper with a (liscussion of pay equity
programs viewed from this perspective.
70
INVESTIGATING CROWDING
EFFECTS
In the language of economists, what is
being set by the interaction of supply and
demand forces in labor markets are rates of
remuneration for the years of schooling and
work experience of inclividuals with (liffer-
ent occupational specializations an(l other
productivity-related characteristics (Pola-
chek, 1976, 1981~. The expectation is that
the rates of return on human capital re-
sources will be lower, anti hence wage rates
will be lower after controlling for years of
schooling and work experience, in occu-
pational labor markets in which the supply
of labor is more abundant relative to the
(demand. This will be the case because mar-
ket forces permit, or even compel, em-
ployers to pay relatively less for labor in
slack occupational employment markets. It
will also be the case because would-be em-
ployees in these stack markets must often
settle for jobs that do not fully utilize their
accumulated human capital.
Everyone understancls these relative sup-
ply concepts of wage determination on an
intuitive level. When recruiting for aca-
OCR for page 71
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
demic staff, for instance, business schools
usually filch larger numbers of well-qualifiec!
applicants for the business economics and
quantitative methods positions than for the
accounting positions, ant] it is understood
by all concerned that that is why the ac-
countants receive higher salary offers.
Moreover, some of those who have trained
for academic jobs in business economics or
quantitative methods end up having to take
other sorts of jobs in business or government
that do not fully utilize their research train-
ing and teaching skills. Excess supply pres-
sures on the wages of aspiring artists (mu-
sicians, dancers, painters, and so forth) are
even more severe. When artistic work can
be found, the wages typically provide a
paltry return on the investment in artistic
training and experience. When artistic work
cannot be founcI, aspiring artists end up
supporting themselves waiting on tables or
cloing other jobs that make little or no use
of, and hence provicle little or no return
on, their accumulates] artistic human cap-
ital.
Nevertheless, crowding effects on wage
rates cannot be easily observed or measured
in a direct sense. For most occupations there
are no direct measures of labor supply or
demand; only those who were or are actually
working in an occupation can be identified.
Thus, researchers investigating possible ex-
cess supply effects on wage rates have typically
proceeded by attempting to demonstrate that
conditions that might be expected to cause
labor market crowding accompany observed
wage rates that are lower than would oth-
erwise be expected. Several researchers, for
example, have presente(l evidence that men
born in large cohorts generally have lower
earnings profiles than men born in relatively
small cohorts (see, for instance, Easterlin,
1980; Freeman, 1979; Welch, 1979~.
Differences in Female Labor Markets
Bergmann's assertions about the relative
severity of crowding effects in female labor
71
markets hinge on the following empirical
observations. First, female labor force par-
ticipation rates have risen (Iramatically since
World War II. Largely as a consequence of
this, the growth rate for the female labor
force has been much higher than for the
male labor force. Second, female employ-
ment is concentrates] in a narrower array
of occupations than is male employment
(Blau, 1977; Blau an(l Hendricks, 1979; Trei-
man ant] Hartmann, 1981~. Moreover, there
is little overlap between female and male
labor markets. Even within narrow occu-
pational classifications that appear from the
numbers of female anti male workers to be
sex integrated, women and men often have
clifferent job titles or work in separate es-
tablishments. Based on their study of 400
California business establishments employ-
ing nearly 47,000 men ancT over 14,000
women, Bielby and Baron (1984:50-51) con-
clude that
in most establishments, few job classifications are
staged by both men and women. Indeed, com-
plete segregation was the norm in establishments
studied . . . and segregation levels were virtually
constant in these organizations during the late
1960s and 1970s.
This occupational segregation by sex pre-
sumably limits the potential for a more
general diffusion of crowding effects origi-
nating from excess supply in female labor
markets. Corresponding to these conditions
that presumably have resulte(l in crow(ling
in female labor markets is the empirical
evidence that women have been and con-
tinue to be paid less for their market work
than men are (see, for instance, Nakamura
and Nakamura, 1985; Oi, 1982; and O'Neill,
19854.
In the empirical portion of this study, we
attempt to examine crowding effects in fe-
male labor markets by relating conditions
deeme(l likely to result in excess labor sup-
ply to observe(l female wage rates. A study
of this sort cannot yield precise estimates
of the clownward impact on wages resulting
OCR for page 72
72
from various degrees of excess labor supply.
Nor is it easy to see how circumstantial
evidence of this nature can illuminate the
question of hou: much more severe the
crowding effects are in female compared
with male labor markets. For one thing,
some of the conditions that can readily be
identified as likely causes of labor market
crowding probably affect female labor mar-
kets differently from the way they affect
male labor markets. For instance, prime-
aged men who are laid offdue to a downturn
in some sector of the economy are likely to
remain in the labor force as unemployed
workers until they locate new jobs. On the
other hand, a substantial proportion of prime-
aged women who lose their jobs are likely
to simply drop out of the labor force. Also
young men are more likely than their female
counterparts to migrate for job-relatecI rea-
sons. In this study we address the more
limited question of whether there is any
evidence that some segments of the female
work force are more vulnerable than others
to crowding-related wage erosion. To our
knowledge, there have been no previous
empirical investigations of this question.
Differences by Occupation
Our reasons for expecting wages to be
more vulnerable to crowding effects in some
occupations, ant] in some positions within
occupations, than in others include the fol-
lowing. First, barriers to entry, such as
training requirements, appear to stem the
flow of job seekers into some occupational
labor markets, making it less likely that
excess supply pressures will (levelop. The
potential magnitude of these effects can be
seen from the numbers presented in Table
3-1. The numbers in the first two columns
indicate a (lramatic surge in me(lical school
applications following the end of World War
II and the return to civilian life of large
numbers of young men whose educations
had been interrupted by the war and who
were now eligible for GI education benefits.
PAY EQ UI TY: EMPIRI CAL INQ UIRIES
Because of constraints on medical school
capacity, acceptance rates dipped for both
men and women as the numbers of appli-
cants rose (see columns 3 and 4~. Thus, as
can be seen from the last column of Table
3-1, the numbers actually admitted to U. S.
medical schools (and hence the numbers
graduating from those schools) rose only
modestly over the 16-year period from 1939
through 1956. In general, higher educa-
tional requirements probably act as a barrier
to entry into an occupation even in the
absence of binding limitations on the ca-
pacities of the relevant training programs.
This is a consequence of the time and mon-
etary costs required to obtain this training.
Second, more educated and also relatively
more scarce types of workers may be more
successful in securing concessions from em-
ployers that protect their wages against ex-
cess supply pressures when they (lo ~le-
velop. These concessions may include long-
term contractual agreements, job ladders
requiring employers to promote from within
to fill most positions, wage scales virtually
guaranteeing rising levels of pay with in-
creases in seniority, and the institutionali-
zation of powerful collective bargaining units
that can push the workers' point of view in
disagreements with employers. Thus, in
general, we expect to find that the returns
to education and work experience (that is,
the expected wage increases associated with
each a(l(litional year of education an(l each
additional year of work experience) are low-
er and crowding effects on wage rates are
more severe for women working in occu-
pations with lower educational require-
meets.
Within occupations, we would also expect
wage rates to be more vulnerable to crowd-
ing effects for some jobs than for others.
Moreover, even after controlling for general
measures of human capital, such as years
of schooling, women with certain charac-
teristics might be more likely than other
women to end up in those jobs within an
occupation for which wage rates are more
OCR for page 73
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
TABLE 3-1 Applicants and Admissions to U. S. Medical Schools, 1939-1956
73
Percentage of
Number of Applicants Applicants Admitted Total Number
Year Women Men Women Men Admitted
1939-1940 632 11,168 50 51 6,012
1940-1941 585 11,269 53 52 6,170
1941-1942 636 11,304 57 51 6,128
1942-1943 810 13,233 49 46 6,484
1949-1950 1,390 23,044 29 29 7,086
1950-1951 1,231 21,049 33 31 6,931
1951-1952 1,109 18,811 38 38 7,570
1952-1953 1,021 15,742 47 43 7,249
1953-1954 972 13,706 53 49 7,231
1955-1956 1,002 13,935 54 50 7,508
SOURCE: Cole (1986:Table 1~.
adversely affected by excess labor supply
pressures. This might be the case, for in-
stance, for women belonging to racial, eth-
nic, or religious groups that are discrimi-
nated against in the labor market; for women
having relatively little work experience due
to child-related withdrawals from the work
force; and for recent entrants or reentrants
to the work force.
METHODS
younger cohorts of working women are larg-
er in number, which leads to larger sample
sizes for empirical analysis. Second, the
careers of these younger women have been
less affected by any discriminatory training
or labor market practices that were, in fact,
reduced as a result of the equal opportunity
rulings that came into effect in the 1960s
and early 1970s.
General Labor Market Variables
We estimate log wage equations using Our choice of explanatory variables was
microdata from the 1980 census (5 percent motivated not only by the objectives of this
A public-use sample) for individual working study, but also by the limitations of our data
women 20 to 24 years of age. Separate
results are presented for eight broad oc-
cupational groups: personal service, other
clerical, secretarial, sales, managerial, health
professional/technical, and teaching (de-
fined in Appendix A). Results are also pre-
sented for women (working in all occupa-
tions) in various demographic groups. We
do not show equations for women sorted
by both occupation and demographic char-
acteristics because the sample sizes are too
small for many occupational-clemographic
categories.
We focus on the wage rates of women
ages 20 to 24 for two reasons. First, the
source.
Unemployment rates are sometimes used
as indices of general excess labor supply
conditions in macroeconometric models. We
include state-specific unemployment rate
variables for women and men ages 20 to 29
in our log wage equations. We have included
those ages 25 to 29 along with those ages
20 to 24 in the computation of these state-
specific unemployment rates because that
leads to more reliable estimates of the un-
employment rates (lue to the larger sample
sizes and also because employers may regarc]
those ages 25 to 29 as potential substitutes
for workers who are 20 to 24 years old.
OCR for page 74
74
The unemployment rate for women and
for men in each state was computed as the
number who were unemployed in the cen-
sus reference week divided by the number
who participated in the civilian labor force
in that week. Unfortunately, the interpre-
tation of the coefficients of these variables
is not straightforward. High male unem-
ployment rates, for instance, may be in-
dicative of more male competition for jobs
usually held by women, which results in
downward supply pressure on female wages.
But high male unemployment rates may
also cause some married women to be more
serious about finding and hoisting on to jobs
that pay relatively well because of the higher
degree of uncertainty associated with their
husbands earnings. High male unemploy-
ment rates could even be indicative of a
substitution of cheaper female labor for higher
priced male labor.
1 he state- specific enects on la nor markets
of general changes in the demand for goods
and services, as well as differences among
states in other relevant factors, such as the
cost of living should be reflected in the
ma. . . . a. fir . ~ .
~ .
average earnings of prime-aged men (who
still make up the largest share of the work
force). In our log wage equations, we include
a state-specific variable for the log of the
average earnings of men ages 25 to 45,
working in all occupations. The coefficient
of this variable is expected to be positive.
That is, it is expected that the effects of
general labor market conditions on the wage
rates of young women will be in the same
direction as the effects on the earnings of
prime-aged men.
After controlling for more general labor
market conditions, we wish to determine
whether there are any additional effects on
female wage rates due to crowcling in female
labor markets. In an attempt to identify
acIditional effects, we include in our log
wage equations the logarithm of the number
of women ages 20 to 24 in each state divided
by the corresponding number of women 25
PAY EQUITY: EMPIRICAL INQUIRIES
to 29 years of age. We refer to this state-
specific variable as the log population ratio.
When the female population is growing, the
number of women ages 20 to 24 will exceed
the number who are ages 25 to 29. In this
case, the ratio of these two numbers will
be greater than unity and the log of the
ratio will be positive. Our expectation is
that positive values of the log population
ratio will be associated with increases in
labor supply in those occupations in which
women seek employment.
In some occupations, of course, increases
in the log population ratio may also be
specifically associated with increases in the
(lemancl for female labor. Increases in the
number of women ages 20 to 24, for in-
stance, may lead to increases in the demand
for primary schoolteachers as well as day-
care and other service workers to care for
the children of these young women. In
general, however, there are no obvious rea-
sons why the number of jobs for young
women will expand as quickly as the number
of potential jobholders with increases in the
population of young women. Excess supply
conditions in female labor markets will re-
sult when the number of women seeking
jobs rises more rapidly than floes the num-
ber of jobs. If there are crowding elects
on the wage rates of young women after
more general, state-specific labor market
conditions have been controlled for by the
unemployment and male earnings variables,
we expect the coefficient of the log popu-
lation ratio in our log wage equations to be
negative.
For similar reasons, increases over time
in female employment rates might be ex-
pecte~d to lead to excess supply conditions
in female labor markets. Thus, we include
the logarithm of the state employment rate
for women ages 20 to 24 divided by the
state employment rate for women ages 25
to 29 as an explanatory variable. We refer
to this variable as the log employment ratio.
Our presumption is that the more positive
OCR for page 75
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
the value of this variable is, the more likely
that there is crowding pressure in female
labor markets. Values of the state-specific
variables (before taking the logarithms) are
shown in Appendix B.
Key Explanatory Variables
The explanatory variables of key interest
are the state-specific log population ratio
and log employment ratio variables, the log
of the state average for the earnings of men
ages 25 to 45, and individual-specific vari-
ables for the number of years of schooling
and potential labor market experience (age
minus years of schooling minus six). Our
expectation is that groups of women with
more negative values for the coefficients of
the log population ratio and log employment
ratio variables will also have smaller coef-
ficients for the years of schooling and po-
tential work experience variables and for
the state-specific average male earnings
variable.
Unfortunately, the relationship between
the potential work experience variable in
this study and work experience of the sort
that might be reflected in higher wages is
tenuous. For any given woman, suppose
we denote the value of the potential ex-
perience variable by PEXP and the actual
number of previous years in which the
woman worked by EXP. Then,
EXP = PEXP L
where, L denotes the number of years in
which the woman was not employed or in
school (plus any discrepancy between com-
pleted years of schooling and the number
of years required to reach that level of
educational attainment). Our data source
contains no information concerning the val-
ues of EXP or L. That is why we use the
potential experience variable.
To the extent that the average values of
L differ across the cli~erent occupational
and demographic groups of working women,
75
there will be systematic differences in the
estimated constant terms for the log wage
equations for the different groups of women.
Because the estimate<] constant terms play
no role in our subsequent analysis, this is
not a serious problem. Within the groups
of women, however, there may also be
correlations between the inclividual values
of the omitted variable L an(l the values of
the potential experience variable. Corre-
lations of this sort could contribute to es-
timates of the effects of an additional year
of work experience on a woman s wage rate
that are systematically too high or too low.
Moreover, these biases could differ system-
atically among the various groups of women.
As a consequence, results obtained in this
study concerning the return to years of work
experience await confirmation from further
research basec] on a data source containing
information about actual past work expe-
rience.
In adclition to the variables discussed, we
inclucle several other explanatory variables
in our log wage equations. To capture in-
dustry-specific demand effects on each
working woman s wages, we include an in-
dustry-specific deman(l index clefined as the
log of the earnings of female workers ages
20 to 24 in the woman s industrial group
clividecl by the average earnings of all wom-
en ages 20 to 24. (Values for this variable
are shown in Appendix C.) We include an
individual-specific variable for the number
of chiIclren ever born (except in the equation
tor women with no chilciren), as well as
dummy variables for race (black or nonblack,
except in the equations for black women
and for nonblack women), for whether a
woman has ever been marrie(l, and for
whether a woman was working 5 years ago
(except in the equations for women who
were working 5 years earlier and for those
who were not). Finally, in the log wage
equations for the separate demographic
groups, we also inclu(le a set of dummy
variables for the eight occupational cate-
OCR for page 76
76
gories developed from the census codes (see
Appendix A for the definitions of these
occupations).
EMPIRICAL RESULTS
The main estimation results of this stucly
are displayed in Tables 3-2 through 3-4. We
present descriptive statistics and regression
results for log wage equations for working
women ages 20 to 24 classified by occupation
and personal characteristics.
Occupation
The mean years of education and of po-
tential labor market experience for women
ages 20 to 24 in each of the eight occupa-
tional categories are shown in the first two
rows in Table 3-2, with the occupations
ordered from lowest to highest in terms of
mean years of schooling. From the mean
wage figures shown in the third row, women
in the top four occupations in terms of
eclucational qualifications (managerial; health,
excluding doctors and dentists; professional/
technical, including doctors, dentists, and
university teachers; and teaching, excluding
university teachers) are better paid, on av-
erage, than women working in the personal
service, other clerical, secretarial, and sales
occupations.
From the next two rows of the top panel,
women working in the four occupations with
the highest educational requirements are
less likely to have wage rates below $2.50
(approximately the mean wage rate for a
secretary) than women working in the four
other occupations, and they are less likely
to have wage rates below $4. 50 than women
working in the personal service, other cler-
ical, and sales occupations. On human cap-
ital grounds, one would expect the wage
rates to be higher for the occupations with
higher educational qualifications even if
workers in all occupations received the same
returns, on average, to additional e(luca-
tion. As expected, however, the top two
PAY EQUITY: EMPIRICAL INQUIRIES
rows of coefficient estimates in the middle
pane! of Table 3-2 show that returns to both
a(lditional years of schooling and adclitional
years of labor market exposure are also
generally higher for the four occupations in
which the average educational levels of the
work forces are higher.
Ignoring the teaching occupation, the
coefficient estimates for the log population
ratio and the log employment ratio are not
significantly different from zero for the oc-
cupations with higher educational require-
ments, but one or the other of these two
coefficient estimates is statistically signifi-
cant and negative for each of the four occu-
pations with lower educational require-
ments. These findings are in line with a
priori expectations that crowding effects on
wage rates will tend to be more severe in
occupations in which the educational level
of the workers is lower. The results for the
teaching occupation seem to confirm that
our method for identifying crowding effects
will not work for those few occupations for
which changes in the size of the specifie
demographic group, or in the labor force
participation rate of that group, are directly
linked to changes in the numbers of jobs
available in those occupations.
The coefficient estimates for the female
and male unemployment rate variables are
shown in the fifth ant] sixth rows of the
micIdie panel of Table 3-2. With none of
the coefficients of the male unemployment
rate variable being significantly different
from zero, there is no evidence of adverse
effects on female wage rates in any of the
eight clesignated occupations as a result of
men in slack mate labor markets competing
for jobs usually or sometimes filled by wom-
en. The coefficient estimates for the female
unemployment rate variable are only sig-
nificantly negative for the high-education
professional/technical an(l teaching occupa-
tions.
Looking at the last row of coefficient
estimates in the mi(ldle panel, female wage
rates do move together with male earnings,
OCR for page 77
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
over states, for all of the designated occu-
77
over, different sorts of jobs within the same
patrons except personal service, sales, and broad occupational group tend to be filled
health. The health occupation has relatively by different sorts of workers.
high average educational requirements, and To examine the possibility that, regardless
it is not one of the occupations for which of occupation, certain types of women are
we have found significant crowding effects more likely to end up in those jobs in which
on wage rates. The other clerical and see- accumulated human capital is less rewarded
retarial occupations are ones with relatively and in which wages are more subject to
low educational requirements, end they are crowding effects, we estimated log wage
also occupations for which we have found equations separately for women with less
evidence of crowding effects on wage rates. than 12 years of schooling and for those
In the bottom pane! of Table 3-2 we show with at least 12 years, for women with
the predicted wage impacts of changes in children and for those with no chiTclren ever
thelog population ratio,thelog employment born, for black women and for nonblack
ratio, and the male average earnings variable women, and for women who were working
associated with "moving" a woman who was 5 years earlier and for those who were not
earring the average wage for her occupation (Table 3-3~. From the first row of Table
from South Dakota to Oregon. According 3-3, women with children, black women,
to our measures of excess supply conditions anc] (obviously) women with less than 12
in female labor markets, there is less crowd- years of schooling have lower mean edu-
ing in Oregon than in South Dakota. Also, cational levels than the other groups of
average earnings for men ages 25 to 45 are women. From the wage statistics presented
almost $5,000 higher in Oregon. We chose in the top panel, it is clear that the women
these two states to demonstrate the mag- in these three groups, as well as those
nitude of the estimated wage gains associ- women who were not working 5 years ear-
ated with a move from a poorer into a more tier, are more poorly paid in general than
favorable labor market for female workers. the women in the other groups.
Due to the crude nature of our proxies for T ~ · . .1 a.
excess supply conditions, our rudimentary
understanding of how wages are cleter-
mined, and a variety of possible statistical
problems with our estimation results, how-
ever, these predictions may be very im-
precise.
Personal Characteristics
One problem with looking for crowding
and other related wage effects on an oc-
cupation-by-occupation basis is that each
one of our broad occupational groups ac-
tually contains a range of jobs with differing
educational requirements and other barriers
to entry; differing pay scales, including dif-
fering institutionalized practices governing
wage growth with increased seniority; and
differing balances of bargaining power be-
tween employers and employees. More-
Looking now at the first two rows of
coefficient estimates in the misfile pane} of
Table 3-3, women with less than 12 years
of schooling, women with children, black
women, and women who were not working
5 years earlier also earn lower rates of return
on additional years of schooling or for acI-
(litional years of potential labor market ex-
perience, or both. These are the groups of
women, too, for whom the estimated coef-
ficients of the log population ratio are sig-
nificantly negative. Moreover, the estimat-
e(l coefficients for the female unemployment
rate variable are consistently negative for
these four groups, and they are statistically
significant for both women with less than
12 years of education and women with chil-
lren.
Again, there is no evidence of adverse
effects on female wage rates from men in
states with higher unemployment rates com-
OCR for page 78
78
Cal
o
on
Cal
U)
be
Ē
o
bE
,_
o
,_
o
40
V'
o
._
Ct
bt
JO
,_
do.
U)
-
U)
o
._
Cal
U)
_
by
a'
V:
.o
V)
._
Cal
U)
~ cO
> ._
._ ~
o
Ct
-=
C)
~o
~ ~ ~ Cal
U) Cry ~
Ct ~
~ Cal
_
Cut>
ACE
.o
.
CĒ
~ .
-
C)
au
a_
~ o
O ~ ~ U: CD
. . . . .
U)
_
fir 0 ~ ~ An
. . . . .
~ CM U: ~ Cal
~ - ~
~ X
A) ~ O ~ G)
. . . . . .
Cal Cal ~ ~ ~ ~
_ ~ ~ ~
a_
o ~
~ CO Cal Cal
. . . .
~ - 'c
c~ ~
x c~ ~ ~ c~
. . . . .
- ~ ~ c~ -
- ~
r~ ~ d~ C~
. . . . . .
c9' c~ ~ ~ x ~
- ~ - 'o
'o
c5:
. . . .
~ ~ o ~
- c~ ~
=5
~ c~ ~ 3
0 ~ ~ _
0 ~ ~ ~
O ~, =~: c; 69 ~ 3 =0
e; e~ c c~ tC e 3 69
~ ~ _ ~ ~ ~ ~ _
c~ ct c~ c;
~ ~ ~ 5~
-
~ c) c~q ~ c~ o~ ~ ~ oo x ~r
c~ 0 0 0 0 0 _ ~ cc c~ (5)
, ~ _~ ~ ~ _ ~
c
._
b~ * * ^ *
c" ~ c~ ~ c~ ~ ~ c~ 0 ~ 0
c~ c~ ~ x ~ _
~ ~ ~ , ~ , ~ , _
b.C 1 1 1
o
3
~4
bt
c~
3 o
u:
'_ .
.5 6~
;> ~
~ ,.
c~
b~
._
- o
o
-
c~
v'
o
~n
ct
0 ~ 0
.o
c~
· ~ S.4
~ ~ ._ ~
0 ~ ~ 0
o ~ o ~
v, ~ ~ ~
ct ~0 bt
~ o o
-
4-
ct
1
~ o
']
- -
o ~
- o
OCR for page 79
79
~ * ~ u: o
c~ cM ~ ~ ~ ~ ~ ~ ~
~ ~ ~ ~ ~ c~ ~ ~ o
~ . ~ . ~ . . .
~ - - ~ c~ c~
1
*
O ~ o oo ~
c~ cc ~ c: o ~ - c~ o
~ ~ ~ ~ 1 ~
o
bt
-
o
o
~ ~ x ~ ~
X O oo CD O ~ ~ O U) O
c~ oo '~ ~ ~ x ~ o o 'o
o
1
c~
-
o
u:
^* ^o ~ o
C>] C'] C,~) ~ Xi_4 ~L CC~ ~ t_
M ~ cM
1- ~ · ~ ·
ct
o x ~ ~ ~ ~ ~ o ~
~ x c~ ~ o ~ ~ o c~ ~
~ ~ ~ .
~ ~ l ~ ~ ~
, ~ 1
~ l
o ~ - ~
_ 5 ~ _ _ 5
~ _ o ~ ~ ~ O
a~ bt bt bt bt
~ o C~ Z o ~o ~o
O ~ ~ _'
~k ~ c_ ~ c~
_ Ct ~ - _
_ ~ O ~ ·
~ ~ a~ ° V2
V, ~ ~ V2
C~ Ce :t ~ O
~ ~ ~ o o
·- o
· -
3 c, ,,,
° o ~ o .~-
C~ ~ =-_
~ ~ . ~ _
a~ c:: ~ ~ ~
bt bt 5 ~ ~
bt a' ~ 0 3°
W
.s °
,) ~ ._
a' C ~ ~
- a.; ~ O ~°
~ ~ o
.=
:> .= o '~
o ~ - C)
Ct -- ° (t C~
~: ~ C~ ~ _
a' =4 _ ~ ~
~ ~ ~ ~ .'
C~
E_, ~ ~ .o
o o
=,, ~ ~ ~
o ~ ~
P~2 b.0 Ce
_ =: `- C
e,: -< ~ ~ ·_
.C~ ~ ~ ~ .=
C;~ (~)
C~
~ ~ ~ ~ .
· V' ~ ~ 0 . ~
_ v~ (~: ~ CD '
) C) ~ ~ U)
~ z) 50 ~ O S
O ~ ~ ~ _
Ct U' O
b~
C~
~ 3
~ _
~ ~o,
o
_
V) o
~ ~ (,,
U) ·
a; ~ ~
bC ~ au
° - ~o ~ ~
_d
C] c; _
C~ _
~~
. °
;~b ~ V,
o ~ ~
3 ~ ~ °
o ~ o
~ ~ ._
~ ~ ~ o
~ 5 ~ ~
~ ~ c~
x ~ ~ ~ x
_
_ ~ ~,
c~ _
° c`~ ~ a
3°
c~ ,. 0 ~ ce
.= ~ 0
0
0 °
, ~ c c y u~
.' Ct ~ ~ ~ ~ 0
c; ~ ~ ~ ~ u: ~
~ cĒ ~ ~ ~ O bC
.o ~ ~ ~ ~ _
~ ~ ° ° ~ c~ ~
° s_ ~ ~ ~ ~
~ .~ 0 0 ;>
S~ ~ a; a.
_ ~ ~ ~ ~ c
~ 0 ~ ~ ~ 0
~ ~ ~ ~ C: o bt
~ ~ ~ ~ ~ ~ ct
OCR for page 80
80
C~
bt
bt
o
.~
40
U)
o
._
U)
bt
.'
o
~o
C~
_ C~
;~
.o
,,~
C)
._
.
c~ 5
.> _
·~ O
C ~ ;~
U:
,
._
40
._
`,~
~,,
-
C~
o ~ ~ X
. . . . . .
C~ ~ ~ o ~ ~
_ _ _ U:
* ~ * CD
X o ~ o
o o o o
. . . .
~ * ,,,
C~ ~ ~ o ~ o
c~ c~ ~ ~ ~r 0 0 0 0
. . . . . . . . . .
C~ ~ U: GO ~ CC
_ ~ _ ~
* ir: * ~
c~ ~ 0 ~r 0
C~ C~ =~ ~14 _ o o o o
. . . . . . . . . .
C~ ~ ~ X U)
_ ~ _ ~ U:
s~
o
C~ X C~ o C~ ~
. . . . . .
CO _ ~ X ~ ~
_ ~ _ U)
.~
._
V,
._
C~
~ ~ C~
._ ~ C~
· _
C~
V,
_ `=
~ ~, o
~o
C~ C)
U)
~ CD
. .
_ _
~ ~.
s~
e
5
Ct * ~ * ~
CC O ~ O
~7 o o O O
. . . .
U, ~ ~
o
._
C~
bt
C~
O0 ~
X ~ ~ O
. . . .
~ ~ X C~
_' _ CD
_ _
53 0 ~
. . .
~ C~ C:
- ~ C']
CO
O
. . . .
CO C~ C~
C~
-
* CO * CO
CC O C~ O
O O O O
. . . .
U,
Ct
.s * ~* ~
~ > C';l C~ _
U' o o O O
~1 . .
S~
·C)
o
* '_ ~
~ ~ C~ _ _
O O O O O
._ . . . .
U7 ~ _'
V'
~;
~ 3P ~ ~
CM O _
O ~ _ O O O O
. . . . . . . .
C o 0 5 5
c) C~3 c) ~ ~ ~ ·
~ ~ ~ 3 ~ ~ o
·~ bt ~ 3- ~ ~ ~ o
X ~ ~ ~ ~ ~ ~ o o
~ ~ ~ ct C~
OCR for page 81
81
1
C5> CO O X CC io CM X ~ X ~ O oo O CM
o o o - o c~ ~ - ~ o o ~ - o ~
. ~ . . . . . . . . . ~
1 ~ 1 ~ ~ ~ ~
o
bC
o
* ~ ~ - * ~ * ~ c~ ~ o
~ 'o ~ oo ~ ~ ~ x u: x 'o x ~ ~ cM
O O O ~ O ~ ~ ~ ~ O O C~ c;5 C~ O
. . . . . . . . . ~ ~
1 ~ ~ ~ _ °v 1
ct
o
u)
~ ~x ~ ~ x ~ ~ o o ~ oo
0 0 0 _ O c~ ~r ~ 0 0 0 ~ c~ 0 x
. . . . . . . . . . . ~ . . .
1 ~ 1 ~ ~ ~ ~'
o
x
ct
3
* ~ ~ ~ * ~ * ~ X X ~
~ ~ C~ C~ X ~ ~ ~ o o ~ ~ ·C' ~ _ _
_ o o CM C~ CO ~ CM ~ _ o ~ o U: o ~
· · · · · · ^ U~ · ·
1 ~ 1 1 ~ ~ ~
-
C~
* ~ ~ ~ * ~ * ~ ._ _ s~
x ~ ~ ~ c~ ~ ~ r~ ~ u: ~ c~ ~ ~ ~ cc
C~ ~ CC C~ ~ ~ ~ o C~ o X ~ X ~ o
1- 1 1 ~ 1 ~ ~ ·
C)
-
V)
40
o
X ~ C~ CS _ X C~ ~ ~ C~ ~ ~ ;,3 ~ ~ ~
C~ ~ ~ ~ ~ ~ o ~ o C~ o _ ~ C~ o
. . . . . . . . . . . ~ _ . . .
1 1 ~ ~ ~ c~ . _
C~
C)
C~
* ~ ~ * ~ * ~ ~ ~ ~ .C)
_ ~ ~ ~ ~ o ~ ~ ~ _ ~ C~ oo X
_ C~ ~ ~ X X CO C~ C~ o ~ ~ o o
1 ~ ~ ~ ~ c~ ~ _ 1
o
._
.° C~
~ _
- o ~
~ -
s~ ~
o ~
bt b~
o o
~ s~
40 ~
a; ~ _
c~ ~ ct
~ c~ ~ ~
~ o ~ ~ ~ ~
~ ~ ~ o ~ ~
0) ;~ c~ (D
o ~ o ~ ~ ~ ~
~ 3 ~ ~ ~o ~ ~
~ ~ bt
o
_ ~
-
! ~
-
:-
C:
.0
O ~ _
_ ~ ~ _
~ ~ C~ ~0,
. _ _ ,_ ~
C a~ ~ ~'
~ =, ~ ~ ~
O ~ ~ ~ U)
:) C ~ CM
bf bf
~i
1
a;
E~ ~
00 o
~ =d
O ~
C)
~ O
U) ~
C~ ~
~ O
~
L _
O ~
7 ~
OCR for page 82
82
CM
o
of
Cal
in
be
Ē
o
._
_
to
-
<~o,
U)
o
._
be
C~
bf
o
-
C--o.
U)
U.
o
._
U)
-
b~
U:
._
U:
U)
~ C)
~ ._
C~ U.
_
. _
. _ _
C~ 5
~ C)
_ ~
~ O
5 v:
_
_ c,
~9
C~ ~
~ V:
C~
~ C~
E~ C)
,'
C)
U)
C~
~C
a:
~
.'
5~
C~
_ ._
E~ - o
o
C~
U.
C~
._
U'
._
C~
U) V~
~
;>
·_
~ ._
~
._ ~
_ ~
C~
a~
~ ~ ~ ~ C~ ~ ~ X CS ~ o o
_ ° ~ - C~ ~ ~ o _ ~
Oo ~ ~ ~ _ ~ ~ ~
z
~ ~.o -- 8~= ~ ~oO
o o ~ C~ C~ C~ ~ U: CS X
z z ~ ~ ~ ~_
Zo ~
A _ _
O . - ~
Z ~ ~
~ C,) C~ X C~ C~ ~ CM
~ ._ . . . . . .
Z .' ~ ~ ~ X ~ U:
:Ē
U)
._
_
=i ;)
~ X
C~ ~ C~ ~ O
. . . . .
C~ ~ X
C~
O ~ X X
. . . . . .
C] ~ ~ U: ~ C~
~ ~ ~ C~
O
O O
. . . . .
C~ C~ ~ CD
~ ~ C~
C~
X CM ~ C~
-
S~
o
-
b.O _
_ ~
O
O ~
C) =.
O O
.'' '' ~ ~
_ s~ a;
a' 50
:^
_ ~ ~
~ C~
X
0 00 ~ C~
C`;i ~4
C~ CD
-
V,
~_
'_
o
O3 b.t
-
.>
-
P~
5 ~
-
O
C~
., . -
:>
c~
* CD
CD O
o o o
. . .
cO ~
r' * ~
CD O
~ o o
bC ·
c~
bt
o
-
40
v)
5
._
5
a;
·C)
*
o O O
C;
* ~
O O
. .
-
o
._
s~
bt
C~ CD
O O
. .
Ct
s~
3
o
-
~ .5
3 o
`_ o
~ _
o 3 o '';
~ ~ 40
~ . ~
C~ ~ ~ ~
V'
C~ ;-
s~ Ct
OCR for page 83
OCR for page 84
OCR for page 85
OCR for page 86
OCR for page 87
OCR for page 88
OCR for page 89
OCR for page 90
Representative terms from entire chapter:
labor markets
83
C~ o C~ C~ CM X o ~ CS) C~
o o _ o o C~ ~ C~ ~ _
. . . . . . . . . . . .
~ ~ 1 ~ ~ ~
*
o C~ CC ~ X CO
o o o o o _ C
. . . . . . .
* ~ * X O CD ~ ~ ~ CD O C~ CS U)
O O _ O _ CM _ ~ C~ CM U) _ O CD C~ O
~ 1~ ~ _
84
peting with women for jobs. (The coefficient
estimates for the male unemployment rate
are positive, not negative.) Also, as ex-
pected, the coefficients for the male earn-
ings variable are insignificantly different from
zero for those groups of women found to
be most subject to adverse crowding effects.
In the bottom pane! of Table 3-3 we show
predicted impacts for selected variables.
The results presented in Table 3-3 raise
a number of questions. For instance, would
the wages of women with less than 12 years
of schooling still be relatively unresponsive
to human capital differences and the level
of male earnings, but vulnerable to crowding
effects, if we limited this low-education sam-
ple to nonblack.women? From the results
presented in the second column of Table
3-4, the answer to this question is probably
yes. Based on the results in Table 3-3 we
might also wonder if the wage rates of
women with children would be found to be
more responsive to human capital variables
and the state level of male earnings, and
less responsive to crowding effects, if we
limite(l the sample of women with children
to nonblacks. Comparing the coefficient es-
timates in the fifth column of Table 3-4 with
those in the second column of Table 3-3,
the answer to this question is probably yes
with respect to the human capital effects
and probably no with respect to the other
effects.
In Table 3-3, the wages of black women
are adversely affected by crowding and un-
related over states to the average level of
mate earnings. From column 3 of Table 3-
4, these conclusions still hold even if we
limit the sample of black women to those
with at least 12 years of schooling. Finally,
we might wonder whether we wouIc] still
find evidence of crowding effects on the
wage rates of young women who were not
working 5 years earlier if we limitecl this
sample to those who are nonblack and chfl~-
less. From the coefficient estimates in col-
umn 7 of Table 3-4, the answer to this
question is yes. In general, the conclusions
PAY EQUITY: EMPIRICAL INQUIRIES
reached on the basis of Table 3-3 seem to
be borne out by the findings for more de-
tafled groupings of women presente(1 in
Table 3-4.
ALTERNATIVE CAUSAL
EXPLANATIONS
In the empirical literature on female work
behavior, log wage equations similar to those
estimated in this study are viewe(l as re-
flecting the wage offers macle to women
with specified productive attributes in com-
petitive labor markets in which the wage
distributions are determined by marketwide
supply and demand conditions not subject
to the control of in(liviclual employers or
labor force participants. Thus, these equa-
tions usually include both individual-
specific explanatory variables, such as years
of schooling, and certain marketwicle vari-
ables, such as the state or county unem-
ployment rate (see, for instance, Heckman,
1981; Nakamura and Nakamura, 1981~. The
estimate(1 coefficients of the marketwi(le
variables are presumed to represent the
responsiveness of the female wage (listri-
bution, and hence the responsiveness of the
wage offers received by indivi(lual women,
to marketwide supply and demand condi-
tions reflected in the values of the mar-
ketwide variables. The log wage equations
estimate(l in this study contain a number
of marketwide variables. It has been argued
that negative coefficient estimates for two
of these variables, the state-specific log pop-
ulation and log employment ratios, are in-
dicative of negative crowding effects on the
wage rates of working women.
Could it be, however, that the log em-
ployment ratio variable is serving as a proxy
for other sorts of impacts on the wage dis-
tributions of women ages 20 to 24? It is
often argued that the probability that a
woman Will work is positively related to the
wage offers she receives. If marketwide
supply-side effects of this sort were strong
enough, there would be a tendency for the
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
values of the log employment ratio to be
higher in states where wages are relatively
high for women ages 20 to 24 versus women
ages 25 to 29. As a result, the coefficient
estimates for this variable might turn out
to be insignificantly different from zero or
significantly positive, even though there are
also negative crowding effects on female
wage rates. Because of their positive effects
on wage rates, however, supply-side fee(l-
backs of this sort could not contribute to
an erroneous conclusion that there are
crowding effects.
On the other hancI, it has also been argued
in the literature that employers make min-
imal specific training investments in women
and, therefore, are generally willing to sub-
stitute younger for older female workers (or
vice versa) in order to minimize labor costs.
If marketwide demand-side effects of this
sort were sufficiently strong, there would
be a tendency for the values of the log
employment ratio to be lower in states
where wages are relatively high for women
20 to 24 versus 25 to 29 years old. As a
result, the coefficient estimates for the log
employment rate variable might be found
to be significantly negative even in the
absence of crowding effects on the wage
rates of female workers. However, when
women lose or cannot finct jobs due to the
substitution of lower priced female labor
from another age bracket, those women are
being "crowded out" in terms of employ-
ment opportunities. Thus, negative coeffi-
cient estimates for the log employment ratio
variable can still be considered to be in-
dicative of crowding effects in female labor
markets, although these crowding effects
may be employment rather than wage re-
lated.
CONCLUSIONS
The evidence presented in this study
suggests that increases in the female work
force, brought about by population increases
and increases in female employment rates,
85
have resulted in wage erosion that has been
more serious for working women in some
demographic and occupational groups than
in others. In particular, we find there have
been adverse crowding effects on the wage
rates of women employed in the personal
service, other clerical, secretarial, anti sales
occupations, and on the wage rates of wom-
en with less than 12 years of education,
those with children, black women, and
women who are relatively recent labor mar-
ket entrants or reentrants (that is, those
who were not working 5 years earlier). On
the other hancI, we find no evidence of
crowding effects for working women who
have at least 12 years of education, or who
have no children, or who are not black, or
who were aIrea(ly working 5 years earlier.
Nor do we find any evidence of crowding
effects for women working in managerial,
health, an(l professional/technical occupa-
tions. We conclude that either there is less
crowding in the segments of the female
labor market in which these women are
employed or the wage rates for the sorts of
jobs these women have are better protected
by institutional and other factors from ero-
sion due to excess supply pressures. To the
extent that these results reflect reality, it
is appropriate to ask what the policy im-
plications of these findings might be.
Bergmann (1986:128) made the following
observation:
Every woman now on the women's labor market
who would be allowed into a job in the men's
market would reduce the pay gap between the
sexes. Her move would push the wage scale in
the two markets toward equality by increasing
the supply of labor to the men's market and de-
creasing the supply to the women's.
Our empirical results inclirectly support this
position. They also lend abided credence to
Bergmann's assertion that"continue(l toc-
cupationall segregation would make the ef-
forts to close the pay gap between women
and men a continual uphill struggle." If this
is the case, the continued importance of
86
vigorous affirmative action and other pro-
grams promoting employment opportuni-
ties should not be lost sight of in the political
rush to institute pay equity programs.
The results of this study also suggest a
general need to examine practices and in-
stitutions that prevent the development of
excess supply conditions in some labor mar-
kets or prevent wage declines in the face
of excess supply conclitions. Women could
seek an expansion of such practices and
institutions in female labor markets. The
push to increase the representation and
power of women in labor unions falls under
this rubric. Pay equity programs can also
be viewed as an institutionalized concession
from employers that may serve to insulate
the wages of jobholders in some sectors of
the female labor market from excess supply
pressures.
Affirmative action and equal opportunity
programs act to break clown barriers reg-
ulating female entry into occupations and
cushioning the wage rates of those holding
jobs in these occupations against excess sup-
ply pressures that might otherwise clevelop.
Thus, affirmative action and equal oppor-
tunity programs contribute to the devel-
opment of a labor market in which excess
supply pressures on wage rates spread more
evenly and more quickly through all sectors
of the labor market and, hence, are borne
more equally by all groups of workers. On
the other hand, we have characterized pay
equity programs as yet another measure for
protecting the wages of those who have jobs
against potential excess supply pressures.
We do not see any incongruity in the fact
that groups pushing for improve(1 labor mar-
ket conditions for women are supporting
both of these types of measures. The goal
is clearly to reduce the sensitivity of wages
to excess supply pressures in the secretarial
and certain other female labor markets, ant]
at the same time open up to women a wider
range of what have traditionally been male
occupations.
Finally, we believe that the results of this
PAY EQUITY: EMPIRICAL INQUIRIES
study convey a cautionary message. We
have found that some groups of women are
more likely than others to have their wage
rates adversely affected by excess labor sup-
ply conditions. These results suggest that
those implementing pay equity programs
for the other clerical and secretarial occu-
pations, for instance, must be careful to
(resign those programs so that the employ-
ment opportunities of groups of women,
such as blacks, those with children, and
those with low educational levels, are not
improperly infringed upon. The other cler-
ical and secretarial occupations have offered
black women and women with low levels
of education some of the most attractive
jobs, in wage terms, available to them. If
wages rise in these occupations due to pay
equity adjustments, competition for these
jobs will presumably intensify. What needs
to be guarded against is the formal or in-
formal institutionalization of educational and
other requirements for these jobs that are
not dictated by productivity considerations
an(l that woul(1 exclude from consideration
for these jobs some of the sorts of women
who currently rely on them for employment.
Policy conclusions are the result of judg-
ment as well as logic. As such, they inspire
disagreement. In this particular case, more-
over, there is bound to be disagreement
not only about the particular policy conclu-
sions stated, but also about whether any
policy conclusions at all can be drawn from
a study such as ours.
Some will argue, for instance, that no
policy implications can be (Irawn from our
study because we have not established the
extent to which excess supply pressures in
female labor markets are the result of vol-
untary career decisions rather than a re-
flection of sex-relatecl employment discrim-
ination. A woman might have become a
secretary because of a preference for this
sort of work, because she thought this job
woul(l blend more easily with homemaking
responsibilities than other possible jobs,
because this was the best job she could find
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
when forced to look for work by economic
necessity with little or no career prepara-
tion, or because she was rejected for training
programs or jobs in her chosen career area
(perhaps because of discrimination). Our
data source only provides information on a
woman's occupation, not on her reasons for
being in that occupation.
Some economists would argue that the
erosion offemale wage rates due to crowding
pressures is an appropriate public policy
concern only if the crowding can be shown
to be largely due to employment discrim-
ination against women. Demonstrating this
point would probably require a fully artic-
ulated model of occupational choice. Thus,
the acceptance of this position would cer-
tainly forestall any timely (rebate of policy
measures intencled to reduce or counter-
balance crowding pressures on female wage
rates because occupational choice is one of
the most poorly developed topic areas in
labor economics. A number of other econ-
omists are also pushing forward with the
analysis of implications of pay equity pro-
grams, despite the paucity of empirical ev-
idence on key behavioral responses (see,
for instance, Beider et al., 1986~.
Others would argue that no policy im-
plications can be drawn from our empirical
results because we have not clemonstrated
that crowding effects on wage rates are more
severe in female than in male labor markets.
It is true that our results do not shed light
on the question of whether differences in
the severity of crowding pressures in female
versus male labor markets are an important
cause of the female-male wage gap. But we
do not agree that consideration of policies
to deal with crowding effects in female labor
markets must await evidence that this
crowding is due to discrimination or that
the wage effects of crowding are more severe
for female than for mate workers. Perhaps
our position on this question can be clarified
by pausing for a moment to consider some
of the issues in the health care area that
have come to be viewed as appropriate
87
topics for public policy debate and potential
government action. Virtually everyone agrees
that providing some minimal level of health
care for the nation's elderly is an appropriate
public concern. This is despite the fact that
many elclerly indivicluals are suffering from
ailments that are the result of their own
voluntary life-style choices (both present
and past).
There is also special public concern about
factors that appear to affect adversely the
health status of those in subgroups of the
elderly population that have particularly
severe overall health problems. There is
special concern, for instance, about the ef-
fects of high blood pressure and poor nu-
trition on the health status of el(lerly blacks.
This is the case even though these same
factors may have just as severe adverse
effects on the health of those belonging to
groups for whom the overall health picture
is more satisfactory. Demonstrated need
provides both the justification for public
expenditures on health care for the elderly
and the rationale for making extra funding
evade tor programs designed to provide
special assistance to certain subgroups of
the elderly population.
The case for public policies to improve
female labor market conditions rests on a
similar foundation. There is clear evidence
that the earnings situation for female work-
ers is poor in an overall sense, even though
employment and wage conditions for wom-
en have improved in recent years. More-
over, the poor earnings situation of large
numbers of working women causes suffer-
ing. There is widespread and growing pov-
erty, for instance, among widowed, cli-
vorcecI, and separated women and the
depenclent children of those women. To us
the question is not whether, but how, public
policy should be harnessed to improve the
economic status of women. We undertook
this study in the hopes of improving the
understanding of how wages are determined
in female labor markets, for it is this un-
derstanding that must form the basis for
88
policy discussions on how the wages of
working women can be enhanced.
ACKNOWLEDGMENTS
We are grateful to Francine Blau, Harriet
DuTeep, Ronald Ehrenberg, and particu-
larly, Arthur Goldberger for their helpful
comments on earlier versions of this paper.
REFERENCES
Beider, Perry C., B. Douglas Bernheim, Victor R. 1985
Fuchs, and John B. Shoven
1986 Comparable Worth in a General Equilibrium
Model of the U.S. Economy. Working Paper
Series, No. 2090. Cambridge, Mass.: National
Bureau of Economic Research.
Bergmann, Barbara
1974 Occupational segregation, wages and profits
when employers discriminate by race or sex. O'Neill,
Eastern Economic Journal 1 :103- 110. 1985
1986 The Economic Emergence of Women. New
York: Basic Books.
Bielby, William, and James Baron
1984 A woman's place is with other women: Sex
segregation within organizations. Pp. 27-55
in Barbara F. Reskin, ea., Sex Segregation
in The Workplace: Trends, Explanations,
Remedies. National Research Council, Com-
mittee on Women's Employment and Related
Social Issues. Washington, D. C.: National
Academy Press.
Blau, Francine
1977 Equal Pay in the Office. Lexington, Mass.:
D. C. Heath.
Blau, Francine, and Wallace Hendricks
1979 Occupational segregation by sex: Trends and
prospects. Journal of Human Resources
14~2~:197-210.
Cole, Stephen
1986 Sex discrimination and admission to medical
school, 1929-1984. American Journal of So-
ciology 92(3~:549-567.
Easterlin, Richard A.
1980 Birth and Fortune: The Impact of Numbers
on Personal Welfare. New York: Basic Books.
PAY EQUITY: EMPIRICAL INQUIRIES
Freeman, Richard B.
1979 The effects of demographic factors on age-
earnings profiles. Journal of Human Re-
sources 14:289-318.
Heckman, James J.
1981 Heterogeneity and state dependence. Pp. 91-
139 in Sherwin Rosen, ea., Studies in Labor
Markets. Chicago: University of Chicago Press.
Nakamura, Alice, and Masao Nakamura
1981 A comparison of the labor force behavior of
married women in the United States and
Canada, with special attention to the impact
of income taxes. Econometrica 49:451-489.
The Second Paycheck: A Socioeconomic Anal-
ysis of Earnings. Orlando, Fla.: Academic
Press.
Oi, Walter
1982 The fixed employment costs of specialized
labor. Pp. 63-116 in Jack Triplett, ea., The
Measurement of Labor Costs. Chicago: Uni-
versity of Chicago Press.
June
The trend in the male-female wage gap in
the United States. Journal of Labor Eco-
nomics 3:S91-116.
Polachek, Solomon
1976 Occupational segregation: An alternative hy-
pothesis. Journal of Contemporary Business
5:1-12.
1981 Occupational self-selection: A human capital
approach to sex differences in occupational
structure. Review of Economics and Statistics
58:60-69.
Treiman, Donald, and Heidi 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.
Welch, Finis
1979
Effects of cohort size on earnings: The baby
boom babies' financial bust. Journal of Po-
litical Economy 87:565-597.
White, Halbert
1980 A heteroscedasticity-consistent covariance
matrix estimator and a direct test for het-
eroscedasticity. Econometrica 48:817-838.
EFFECTS OF EXCESS SUPPLY ON WAGE RATES
APPENDIX A Definitions of Occupational Groups
89
Group Name
1980 Census Codes
Personal service
Other clerical
Secretarial
Sales
Managerial
Health
Professional/technical
Teaching
403-4()7, 449, 453, 457, 467, 468
275, 276, 319, 323, 325-329, 335-339, 343-349, 353, 354, 364, 365, 368, 374,
377-379, 383-387, 389
313-315
256, 259, 263-269, 274, 283
004-009, 013-019, 024-029, 033-037
095, 097-099, 103-105, 203-208, 445-447
023, 043-049, 053-059, 064-069, 073-079, 083-089, 113-119, 123-129, 133-
139, 143-149, 153, 154, 164-169, 173-179, 183-189, 193-199, 213-218, 223-
229, 233-235
155-159, 163
APPENDIX B State-Specific Variables Calculated Using Public-Use Sample (5 percent)
Data from the 1980 Census
Ratio for Women 20-24 Versus
Women 25-29 Unemployment Rate Mean Annual
Employment Women Men Earnings for
State Population Rate 2()-29 20-29 Men 25-25 ($s)
Alabama 1.13 .94 .15 .08 15,684
Alaska 1.00 .96 .06 .06 26,031
Arizona 1.22 1.00 .11 .05 15,992
Arkansas 1.21 .98 .10 .06 14,482
California 1.05 .97 .07 .09 18,192
Colorado 1.10 1.01 .08 .06 18,481
Connecticut 1.03 .96 .03 .06 19,442
Delaware 1.16 ~ 1.05 .11 .02 17,140
District of Columbia .92 1.02 .10 .10 14,726
Florida 1.17 .97 .07 .06 16,345
Georgia 1.11 .93 .11 .09 15,731
Hawaii 1.05 1.04 .10 .04 17,284
Idaho 1.64 .89 .07 .08 16,589
Illinois .98 .96 .08 .12 19,227
Indiana 1.03 .98 .10 .12 18,033
Iowa 1.19 .95 .05 .09 17,332
Kansas 1.22 1.00 .10 .09 17,511
Kentucky 1.18 .97 .08 .11 16,934
Louisiana 1.11 .90 .09 .07 17,726
Maine 1.17 .99 .09 .16 12,764
Maryland 1.14 .96 .09 .07 18,963
Massachusetts 1.10 .96 .04 .07 17,576
Michigan 1.01 .98 .12 .19 20,020
Minnesota .98 1.03 .04 .10 18,313
Mississippi 1.48 .92 .16 .12 14,101
Missouri .96 .99 .10 .11 17,080
Montana 1.06 1.00 .08 .16 17,355
Nebraska .81 .96 .02 .01 17,596
Nevada .89 1.03 .04 .04 19,081
New Hampshire 1.57 .97 .02 .08 18,610
New Jersey 1.09 .95 .09 .08 18,815
New Mexico 1.14 .99 .10 .09 16,153
New York 1.06 .91 .08 .12 17,428
North Carolina 1.18 .98 .10 .07 14,314
Continued
so
APPENDIX B Continued
PAY EQUITY: EMPIRICAL INQUIRIES
Ratio for Women 20-24 Versus
Women 25-29 Unemployment Rate Mean Annual
Employment Women Men Earnings for
State Population Rate 20-29 20-29 Men 25-25 ($s)
North Dakota .87 1.00 .08 .06 18,091
Ohio 1.06 .99 .11 .14 18,004
Oklahoma 1.01 .95 .05 . 06 16,487
Oregon . 74 .97 .10 .11 18,204
Pennsylvania 1.26 .94 .07 .11 17,506
Rhode Island 1. 12 .98 .02 .07 15,328
South Carolina 1.20 .99 .08 .06 14,243
South Dakota 1.29 1.05 .10 .11 13,565
Tennessee 1. 53 .99 .08 .12 16,166
Texas 1. 07 .96 .07 .04 17,960
Utah 1.16 .94 .07 .04 17,314
Vermont 1.05 1.06 .12 .15 14,394
Virginia 1. 13 .95 .07 .06 16,953
Washington 1. 13 .95 .10 .12 18,439
West Virginia 1.09 .99 .04 .11 15,172
Wisconsin 1. 07 .99 .09 .08 17,171
Wyoming .84 1.07 .03 .02 19,137
NOTE: Natural logarithms of these variables were used in the regressions for which results are reported in Tables
3-2, 3-3, and 3-4.
APPENDIX C Definitions of Industrial Groups and Values for an Industry-Specific
Relative Earnings Variable
1980 Census Industry-Specific
Industrial Group Codes Demand Indexa
Durable goods 230-391 1.29
Food and kindred products 100-122 1.02
Textile products 132-152 .92
Other nondurable goods 130, 160-222 1.11
Transportation 400-411, 420-432 1.37
Communications 440-442 1.47
Utilities and sanitary services 460-472 1.28
Wholesale trade 500-571 1.19
Retail trade 580-691 .76
Finance, insurance, and real 700-712 1.21
estate
Business and repair services 732-760 1.12
Personal services 761-791 .77
Entertainment and 800-802 .70
. .
recreation services
Health services 812-840 1.12
Legal services 841 1.31
Educational services 842-851, 860-861 .83
Other services 852, 862-892 .86
Other All remaining codes 1.01
aDefined as the average earnings of women 20-24 in the given industry divided by the average earnings of all
women 20-24. The natural logarithm of this variable was included in the log wage equations to account for demand
effects.
