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 222
10
Women's Pay in Australia,
Great Britain, and the United States. The Role
of Laws. Regulations, and Human Capital
R. G. GREGORY, R. ANSTIE, A. DALY, and V. HO
The 1970s were a remarkable time for
women. Labor force participation rates in-
creased markedly throughout the Organi-
zation for Economic Cooperation and De-
velopment (OECD) (Mincer, 1985), an(1 in
most OECD countries the pay of women
relative to that of men increased signifi-
cantly. Two of the largest changes in the
pay ratio occurred in Australia and Britain
(Figure 10-1~. In Australia women's pay
increased 30 percent relative to that of men.
In Britain the pay ratio increased about 20
percent. The United States seems to be an
exception to the general trend in that the
relative earnings of women did not in-
crease. ~
At the beginning of the 1970s, the ratio
of women's to men's earnings, which we
will call the female earnings ratio, was sim-
ilar in Australia and the United States.
Women in Britain fared relatively worse.
By the end of the decade Australia hacT
joined those countries that pay women well,
PA more detailed discussion of the U.S.-Australian
comparison can be found in Gregory and Ho (1985~.
The British-Australian analysis is developed further in
Gregory et al. (1986~.
222
relative to men; Britain had caught up to
the United States, but both ranke(l Tow
among countries groupe(1 by the female
earnings ratio. A considerable amount can
be learned from comparing the three labor
markets, which encompass the range of ex-
periences of the clecade. We focus on three
sets of questions.
First, why is the female earnings ratio so
(li~erent in the three countries? In 1981,
for example, adult women who worked full
time in Australia earned about 79 percent
of the fulI-time average earnings of men.
In Britain and the Uniter] States the female
earnings ratios were about 64 an(1 60 per-
cent, respectively. How much of these dif-
ferences can be explained by human capital
models?
Second, why was it that two of the coun-
tries—Britain and Australia exhibited
su(l(len and sharp changes in female pay?
What explains the speec] of the changes and
their extent? What has been the role of
equal pay laws and regulations? These ques-
tions are acl(lressed by assessing the im-
portance of institutions.
Third, what are the relationships between
changes in the female earnings ratio an
OCR for page 223
WOMEN'S PAY
80
75
70
o
i:
.
223
r
60
55
-
50 I I . ,
1966 1968 1970
If._. ',''-
~ ;=;~-
United States
Britain
1972 1974 1976
YEAR
1978 1980 1982 1984 1986
FIGURE 10-1 Ratio of female to male average weekly earnings. Notes: Australia: Average weekly earnings for
full-time (more than 35 hours) adult nonmanagerial employees in the private sector. Australian Bureau of Statistics,
warnings and Hours of Employees, Cat. No. 6304. United States: Median usual weekly earnings for full-time
(more than 35 hours) wage and salary workers, May 1967 to May 1978, and second quarter, 1979 to 1983. U.S.
Department of Labor, Bureau of Labor Statistics, The Female-Male Earnings Gap: A Review of Employment and
Earnings Issues. September 1982. Report 673, pp. 9-10. Washington, D.C. Great Britain: Average weekly earnings
for full-time (more than 30 hours) males age 21 and over and females 18 and over. Department of Employment,
The New Earnings Survey. London.
changes in the employment and unem-
ployment of women? What can we learn
about supply and demand elasticities for
female labor from a comparison of countries
with such different wage experiences? We
address these questions in a discussion of
employment and unemployment in the three
countries.
RELEVANCE OF THE HUMAN
CAPITAL MODEL
Method of Analysis
There are many reasons why men and
women are paid at different rates within a
country. These reasons may include differ-
ences in the quality of workers, the clistri-
bution of workers across industries and oc-
cupations, the (legree of pay (discrimination
against women, and the relative deman(l
an(l supplies of labor. The usual way to
measure the contribution of these factors is
to fit earnings equations to the data for each
sex. The most common earnings equations
adopt a human capital framework an(l hy-
pothesize that the differences in the earn-
ings of men and women can be explainer!
primarily in terms of differences in human
capital, as measure(1 by years of schooling,
work experience, marital status, and so on
(Mincer, 1974; Oaxaca, 19731.
For simplicity, we can add the male and
female earnings equations that are (derived
OCR for page 224
224
from human capital theory and form one
equation, which can be written as
n n
Ei = ~ Bj Xij +~. ~ ~jXtF + Ui
j- 1
(1)
where Ei is the log of the earnings of the
ith person and Xj are human capital and
experience variables. The superscript F re-
fers to female individuals. Consequently,
male workers earn Bj for each attribute and
female workers (Bj + Jo. Ui is an error
term.
Once Equation (1) is fitted to the `data
the earnings gap between males and females
is usually decomposed into two components;
one component is attributable to the dif-
ference in human capital endowments of
men and women, the X variables, and the
other component is attributable to the dif-
ference in the estimated coefficients for men
and women, OFF. With the information de-
rived from the estimation of (1) the female
earnings ratio can be written as
n
E F — EM = ~ B (XF — XM)
n
+ ~ X Tj
j=1
(2)
where the superscript M refers to males.
The first term on the right-hand side of (2)
captures the contribution of the difference
in endowments, an(l the second term cap-
tures the contribution of the difference in
coefficients.
As a general rule, both (differences in
endowments and differences in coefficients
are important contributors to the earnings
gap. The contribution of endowment dif-
ferences is explained in the context of the
human capital model. Human capital en-
clowmer~ts generate labor productivity in
the marketplace, and to a large extent,
workers choose the optimal amount of hu-
man capital that they wish to invest. Women
are paid less, according to this theory, be-
cause they invest less and are therefore less
productive.
PAY EQUITY: EMPIRICAL INQUIRIES
The contribution of coefficient differences
to the earnings gap is not explained by the
human capital mo(lel. This effect is usually
referred to either as pay discrimination, to
be explained largely by noneconomic fac-
tors, or within the human capital framework,
as a measure of our ignorance of the factors
that cause the earnings gap. This ignorance
is usually assumed to stem primarily from
two sources: mismeasurement, as a result
of data deficiencies, and omission of relevant
variables from the regression equation. .
Over the past decade, considerable work
has been ~lone, particularly in the United
States, to exten(1 the list of variables that
may be included in regression equations,
such as Equation (1), an(1 to refine the data
so that the contribution to the earnings gap
of the (differences in coefficients, the unex-
plaine(1 contribution, is reduced. At this
stage, however, progress has been slow in
the United States so that the contribution
of the differences in coefficients to the earn-
ings gap is still usually around 50 percent
(Blau and Ferber, 1986:233, 235; Daymont
and Andrisani, 1984~. In Britain and Aus-
tralia the (differences in coefficients also con-
tribute to at least half of the earnings gap,
although the size of the gap is very different
in each country (see Greenhalgh, 1980, for
Britain; Chapman and Mulvey, 1986, for
Australia). In Britain the earnings gap is
similar to that for the United States. In
Australia the earnings gap is very small.
The human capital approach for explain-
ing the female earnings ratio within a coun-
try may also be used to explain differences
across countries (Gregory and Ho, 1985~.
In each country, for example, the estimated
coefficients may be similar, but the earnings
gap may be generated by differences in the
relative endowments of human capital. A1-
ternatively, there may be the same relative
stock of human capital endowments of men
and women in each country, but the relative
rewards for the endowments of men and
women may be different. In the latter case
the human capital model would have noth-
OCR for page 225
WOMEN'S PAY
ing to contribute to explaining the difference
in the female earnings ratio across countries.
What would be needed is a theory to explain
why the coefficients are different in each
country.
To use the human capital mode! to explain
the different pay ratios across the countries,
our analysis would proceed in the following
steps:
1. Fit earnings equations for men and
women in each country; see Equation (1~.
2. Calculate the female earnings ratio
from Equation (1) and derive the differences
in the ratios between the countries. The
differences provide the cross country earn-
ings gaps to be explained. If the female
earnings ratio is 66 percent in Britain and
78 percent in Australia, for example, we
would attempt to explain the 12 percentage
points difference.
3. Calculate the contribution of differ-
ences in relative endowments of men and
women to the cross country earnings gaps.
This might be best explained by considering
two countries, say Britain and Australia, for
which we have identified in step 2 a cross
country earnings gap of 12 percentage points.
In step 3 we substitute the average endow-
ments of male and female workers in Britain
into the Australian earnings equation to
derive the earnings of British workers if
they were paid according to the Australian
pay structure. A comparison of this hypo-
thetical British earnings ratio with the Aus-
tralian earnings ratio will provide a measure
of the contribution to the cross country
earnings gap of the different relative en-
clowments of men and women (that is, the
calculation for Britain and Australia uses
identical coefficients taken from the Aus-
tralian equation).
For example, if the hypothetical British
earnings ratio and the Australian earnings
ratio were equal, then we would know, on
the basis of the Australian coefficients, that
the cross country earnings gap cannot be
explained by the different relative endow-
225
meets of men and women in each country.
All the earnings gap would be explainecl by
the difference in coefficients in each coun-
try. In this instance, the human capital
theory would make no contribution to an
explanation of the cross country earnings
gap. The calculations of step 3 can be re-
peated using the coefficients of Britain.
4. Finally, find the difference between
the original British earnings ratio, calculated
from the British equation, and the hypo-
thetical ratio, using the Australian coeffi-
cients. This residual might be interpreted
a number of ways. It might be interpreted
as the difference between the two countries
in the level of pay discrimination against
women. Alternatively, it might be inter-
pretecl as the outcome of different biases
in the coefficients of each country. These
biases may have arisen either by variables
being omitte(l from the regression equations
or from the mismeasurement ofthe incluclec]
variables.
Results from the Earnings Equations
The sample is restricted to full-time wage
and salary earners. The Australian data
(N = 17,100), for workers aged 15 to 54
years gill, are drawn from the 1981 House-
holds Sample File of the Australian Bureau
of Statistics' Census of Population and Hous-
ing (Australian census). The British data (N
= 7,018), for workers aged 16 to 54 years,
are drawn from the 1981 General Househoicl
Survey ((lata tape available from Her Maj-
esty's Stationery Office, London). The U. S.
(lata (N = 13,949) are from the March 1982
Current Population Survey (data tape avail-
able from Bureau of the Census) and refer
to the labor force status of individuals in
1981.
The regression equations are as in Equa-
tion (1), with the addition of a constant
term. The dependent variable is the natural
log of weekly earnings. We use weekly
earnings because the Australian data do not
provide good estimates of hourly earnings.
OCR for page 226
226
In each country a fulI-time worker is em-
ployed 35 hours or more per week. The
coefficients of Equation (1) are interpreted
as the percentage changes in earnings in
response to a one-unit increase in the value
of the inclependent variable.
The constant terms measure the average
Tog of weekly earnings of a male high school
graduate, of urban residence, never mar-
ried, working full time, and in his first year
in the labor market. The coefficients for the
variables estimate the additional payoff for
men over and above the constant term.
Thus, an estimate of the average earnings
of a male university graduate, with all the
other attributes included in the constant
term, is given by the acIdition ofthe constant
term to the estimated coefficient, B. at-
tached to the graduate dummy variable.
The estimated earnings of a female univer-
sity graduate, with all the other attributes
of the constant term, is given by the addition
of the constant term to the sum of the
university graduate coefficients B and By. By
presenting the data in this way the t statistics
for the It's indicate whether the female
coefficients are significantly different from
the male coefficients. (The definitions of the
variables are given in the appendix.)
The individual coefficients from the earn-
ings equations for each country are mostly
as expecter] and will not be discussed in
any detail (Table 10-1~. Among full-time
workers in each country those with more
education earn more. Women also earn less
than men in each educational category.2
Other coefficients indicate that married
women earn less than married men, and
women with children earn less than those
without children. For the workers of each
country more potential experience increases
earnings and, once again, there is a differ-
2The British sample is not as large as that for
Australia, and in some cells the sample size is quite
small. This may explain why the female coefficients
in the higher education categories are not significantly
different from those of men.
PAY EQUITY: EMPIRICAL INQUIRIES
ence between men and women. For women,
the earnings experience profile is less steep.
One difficulty that we share with many
other researchers (e.g., Johnson and Solon,
1986) is that the experience variable is not
correctly measured. We use potential ex-
perience (age minus years of school minus
six). The correct measure would be actual
experience but such data are not available
for cross country comparisons. The use of
potential experience limits our ability to
make comparisons between men and wom-
en within a country, because for women as
a group actual experience is so much less
than potential experience. Across countries,
however, the problem should be much less
serious in that relative to males the over-
statement of female experience in each
country should be similar.3 In other words,
there should be a close correlation across
countries in the ranking of the unobserved
ratios of actual experience of men ant! wom-
en and the ratios of potential experience
that are used in our calculations.
The fit of the earnings equations is similar
in each country; the R2 is .49 for Britain,
.46 for Australia, and .36 for the United
States. Our first conclusion, therefore, is
that within each country, and at a point of
time, the human capital mo(le} seems to
perform reasonably well, and to a similar
degree, as an explanation of the variation
of earnings among men and women. Within
each country, however, the coefficients at-
tache`d to the human capital variables are
different for each sex. As inclicated earlier,
other studies suggest that the human capital
model, on average, can explain only about
half of the pay gap within each country.
We now turn to an investigation of the
3There are some differences in labor force partici-
pation rates for women across the countries, and a
more extensive study would attempt to take those
differences into account, especially because they may
affect the gaps between potential and actual experience.
At this stage, however, we do not believe that our
inability to measure actual experience is a significant
source of error.
OCR for page 227
WOMEN'S PAY
cross country earnings gaps. The basic data
are presented in Table 10-2. Row 1 lists
the calculated earnings ratio in each country
using the coefficients from the British earn-
ings equations but the own endowments
from each country. Similarly, the data listed
in row 2 are derived from the U. S. earnings
equation, and those in row 3 from the
Australian equation. The cells along the
principal diagonal represent the earnings
ratio calculated for each country using its
own regression equation. A comparison of
these numbers produces the earnings ratios
to be explained. The earnings gaps relative
to Australia are listed in row 4.
Proceeding down the principal diagonal,
the first point to emerge from Table 10-2
is the similarity of the earnings ratios be-
tween the United States and Britain. On
the basis of these data, female workers in
Britain earn 64.2 percent of the average
weekly earnings of mates; in the United
States the ratio is 61.7 percent. This pro-
cluces a relatively small earnings gap of 2.5
percentage points between Britain ant! the
United States.
The second point to emerge is that Aus-
traTia is obviously different. Women are
relatively well paid compared with their
U. S. and British counterparts. The earnings
gap between Australia and Britain is lS.1
percentage points; between Australia and
the United States, 17.6 percentage points
(row 4~.
Finally, the human capital endowments
of women, relative to those of men, seem
to be much the same in each ofthe countries.
This is illustrated by the fact that there is
little difference in the earnings ratios across
the columns. Proceeding across row 1, for
example, if workers with the average level
of endowments of U.S. workers are paid
according to the British pay structure, then
the female to male ratio of average weekly
earnings would be 64.0 percent, an earnings
ratio very similar to that in Britain. Simi-
larly, male and female workers with the
same level of endowments as the average
227
of Australian workers, when paid according
to the British pay structure, wouIc] receive
an earnings ratio of 68.6 percent, again a
ratio quite close to that of Britain.
This fincling, that the human capital en-
clowments of women, relative to those of
men, are similar in each country, is a general
one. It holds whichever set of country coef-
ficients is used to calculate earnings. For
example, if British ant] American workers
were paid according to the Australian pay
structure (row 3), their earnings ratios would
be 76.3 and 77.6 percent, respectively. These
are earnings ratios that are quite close to
those prevailing in Australia.
It follows from the fact that the relative
endowments of human capital are so similar
across countries that most of the difference
in relative earnings, especially with regard
to the difference between Australia and the
others, flows from differences in coefficients
an(l not from (li~erences in enclowments.
Our second major conclusion is that the
simple human capital model, as usually
specified, cannot explain the difference in
earnings ratios across countries. Only a small
fraction of the earnings gap between Aus-
tralia, on the one han(l, and the Unite
States and Britain, on the other, can be
explained by different endowments of hu-
man capital.
The next major question, therefore, is
why is Australia different? Why does Aus-
tralia pay women so well?4 To answer these
questions it is necessary to provi(le some
institutional background to the Australian
4Some of the difference in the earnings ratios across
countries might be explained by differences in the
average hours worked by men and women in each
country. From other data sources (see Gregory and
Ho, 1985: footnote 10), it is apparent that the ratio of
full-time weekly hours worked by men and women is
much the same in Australia and the United States. In
Britain, women's hours relative to men's hours are
lower than in the other two countries, so some ad-
justment should be made for this fact. The adjustment
required, however, does not seem to change the results
to any significant degree (see Gregory et al., 1986:11~.
OCR for page 228
Measure
Australia
228
PAY EQUITY: EMPIRICAL INQUIRIES
TABLE 10-1 Earnings Equations for Australia, Great Britain, and the United States
(standard] errors in parentheses)
Great
Britain
United
States
Educationa
Dropout —0.1463 - 0.1433 - 0.3053
(.0086) (.0145) (.0107)
High school
Postsecondary 0.0663 0.0015 0.1228
qualifications (.0091) (.0167) (.0098)
0.2213 0.2106
(.0130) (.0178)
University degree 0.4664 0.4368 0.3035
(.0121) (.0248) (.0107)
Postgraduate degree 0.6011 0.6382 0.3791
(.0220) (.0504) (.0114)
Female x dropout —0.0717 - 0.1135 - 0.0705
(.0159) (.0252) (.0258)
Female x high school - 0.0700 - 0.1192 - 0.1375
(.0127) (.0247) (.0220)
Female x postsecondary —0.1883 - 0.1448 - 0.1128
qualifications (.0364) (.0440) (.0232)
- 0.1616 - 0.0727
(.0200) (.0296)
Female x university —0.0861 - 0.0381 - 0.1265
(.0201) (.0522) (.0237)
Female x postgraduate —0.0682 - 0.2443 - 0.0182
(.0404) ~ 1733) (.0268)
Experiences
Experience 0.0469 0.0484 0.0428
(.0013) (.0020) (.0015)
Experiences - 0.001 - 0.001 - 0.0008
(.00003) (.00005) (.00004)
Female x potential —0.0016 - 0.0109 - 0.0106
experience (.0022) (.0033) (.0024)
Female x potential —0.0001 0.0002 0.0001
experiences (.00004) (.0001) (.0001)
Area
Rural —0.1217 - 0.0122 - 0.110
(.0102) (.0098) (.0070)
Female x rural 0.0307 - 0.0239 - 0.0194
(.0199) (.0178) (.0114)
Urban
Marital Status
Spouse present 0.1577 0.2128 0.1734
(.0106) (.0159) (.0129)
Other marital status 0.0852 0.1145 0.1298
(.0158) (.0282) (.0161)
Female x spouse present - 0.1049 - 0.1588 - 0.1453
(.0165) (.0247) (.0187)
Female x other marital 0.0123 - 0.0576 -0.0724
status (.0241) (.0417) (.0226)
Single never married
Children - 0.0117 - 0.0165 - 0.0160
(.0087) (.0127) (.0088)
OCR for page 229
WOMEN'S PAY
TABLE 10-1 Continued
229
Great United
Measure Australia Britain States
Female x children - 0.1691 - 0.0946 - 0.1154
(.0154) (.0241) (.0135)
Constant 5.0126 4.1867 5.311
(.0079) (.0168) (.0145)
R2 .46 .49 .36
Dependent variable, in week-
ly earnings
aSee the appendix for definitions of the education variables.
bFor Australia and Britain, this group has been divided into two parts; the first coefficient relates to those who
completed trade qualifications, and the second to those with other postsecondary qualifications.
CSee the appendix for the definition of experience for each country.
SOURCES: Australia: Australian Bureau of Statistics, 1981 Census, Household Sample file, full-time wage and
salary earners, ages 15 to 54. Great Britain: 1981 General Household Survey (data tape available from Her Majesty's
Stationery Office), full-time wage and salary earners, ages 16 to 54. United States: Bureau of the Census, Current
Population Survey, March 1982 (data tape), full-time wage and salary earners, ages 15 to 54.
labor market. This will give rise to a new
and related question: Why is it that the pay
ratios have changed in some countries but
not in others?
THE IMPORTANCE OF
INSTITUTIONS
Australia
In Australia the female earnings ratio is
vitally affected by a complex network of
federal and state tribunals that evolved in
the early years of this century. The tribunals
set minimum rates of pay, referred to as
awards, for each occupation. The pay of
university professors is fixed along with that
of bus drivers, laborers, fitters and turners,
storemen, and so on. Before 1975 the tri-
bunals acIjusted downward the award rates
of pay for women, relative to the pay of
mere. Occupations were categorized as to
whether they were filled predominantly by
mates or females: blacksmith was a male
occupation, for example; milliner was a fe-
maTe occupation. When an occupation was
determined to be male, the tribunal cal-
culated the man's wage what he needed
to support himself, his wife, and his children
living in "a civilized community" and then
added a margin for the work value of the
occupation. When an occupation was de-
termined to be female, the tribunals macle
the calculations as though it were a male
occupation and then acljuste(l the notional
rate of pay downward.
From 1950 to 1969 the markdown in pay
for a female occupation was usually to 75
percent of the notional male wage for each
occupation. Because women work fewer hours
an(1 are disproportionately represented in
low-paying occupations and because men
often earn significantly more than the min-
imum rates of pay, the 75 percent rate
procluce(1 an aggregate weekly earnings ratio
of full-time workers of about 60 percent.
Before 1970, it was clear that the tribunals
believed that they were discriminating against
women, and so (lic1 the community at large,
although there seems to have been no ex-
plicit questioning as to what the market pay
relativity might be if the tribunals were not
an active participant in pay setting. Con-
sequently, although there was explicit dis-
crimination according to tribunal criteria,
the question was never posed as to whether
the prevailing pay ratio could be thought
of as discrimination relative to a national
free market for labor.
The 6 years between 1969 an(1 1975 wit-
OCR for page 230
230
TABLE 10-2 Explaining the Pay Gap
Among the United States, Great Britain,
and Australia: The Ratio of Female to
Male Full-Time Average Weekly Earnings
-
Great United
Britain States Australia
Measure
Aggregate
British pay structure
U.S. pay structure
Australian pay struc-
ture
Earnings gap to be ex-
plained (compared
with Australia)
Attributable to
Endowments
Coefficients
64.2 64.0
60.6 61.7
76.3
15.1
77.6
17.6
4.4 1.9
10.7 15.7
68.6
63.6
79.3
SOURCES: Australia: Australian Bureau of Statistics,
1981 Census, Household Sample file, full-time wage
and salary earners, ages 15 to 54. Great Britain: 1981
General Household Survey (data tape available from
Her Majesty's Stationery Office), full-time wage and
salary earners, ages 16 to 54. United States: Bureau of
the Census, Current Population Survey, March 1982
(data tape), full-time wage and salary earners, ages 15
to 54.
nesse(l a very great change. On June 19,
1969, the federal tribunal, following the lead
of four of the six state tribunals, ruled that
the sex of the worker was not to be used
as a wage criterion in those jobs that were
neither preclominantly male nor predomi-
nantly female. By 1972 there should be
"equal pay for equal work. " It would become
illegal, for example, to pay graduates in the
public service different starting salaries based
on their sex. Similarly, in the private sector,
beginning bank tellers would be paid the
same whether they were male or female.
Women working in female occupations, such
as nursing en c] secretarial work, were to be
excluded from "equal pay for equal work"
provisions.
Then, in 1972, the federal tribunal de-
ci(le(1 that the concept of equal pay for equal
work should be widened to "equal pay for
work of equal value" or approximately in
U. S. terms "comparable worth." This wider
concept was to be introduced in three uni-
PAY EQUITY: EMPIRICAL INQUIRIES
form steps over the perio(l to June 1975.
After 1975 award rates were to be deter-
mined without regard to the sex of workers
(Niland and Isaac, 1975~.
Finally, between 1950 and 1974 the fed-
eral tribunal ha(l maintaine(1 a "basic wage,"
which was the minimum wage that any fulI-
time worker couIc3 be paid within any award
determination. Over this period the basic
wage for females was 75 percent of that of
males. Because the basic wage is such a
large fraction of the average wage, this
formula ten(le(1 to produce average award
wage relativities between men and women
that were very close to 75 percent. In 1974,
the federal tribunal decided to extend the
male minimum wage to females, an(1 this
reinforced the substantial pay increase for
low-paid women on the bottom of the pay
scales in female occupations. As a result, a
substantial pay increase for all women was
ensured no matter what the outcome of the
"equal pay for work of equal value" decision
in each individual instance. (See Nilan(1 anti
Isaac, 1975, for a fuller discussion. ~
The changes in award rates of pay are
presented in column 1 of Table 10-3. After
almost two decades of constancy, the rel-
ative awards began to increase in about
1970, and by 1977 they were 29.6 percent
higher than 9 years earlier. Most of the
large increase in awards occurre(3 from 1972
onward, so it is evident that the "equal pay
for work of equal value" and the basic wage
decisions were the important factors.
The arbitration system (letermines min-
imum awards an(l (loes not directly cleter-
mine the earnings that are paid. Many work-
ers, particularly men, receive over-awarc3
payments. There is no reason, therefore,
why changes in relative award rates shoul(1
necessarily be reflected in changes in rel-
ative earnings. After the equal pay ~leci-
sions, male workers coul(1 systematically
seek over-awarcT payments to offset the change
in the award relativities.
The impact of the award rate decisions
on earnings is shown in columns 2 and 3
OCR for page 231
WOMEN'S PAY
TABLE 10-3 Female to Male Awards and Earnings Ratios
237
Australian Ratios
British Ratios
Earnings, Awards/ Awards, Earnings, Awards/
Private Earnings Manual Manual Earnings
Year Awards Sector 1976= loo Workers Workers <1976= loo'
964 72.0 59.2 98.6 83.1 s9.8 101.0
969 72.0 58.4 97.2 83.3 59.s 100.
970 73.2 59.1 96.8 82.6 60.1 102.
971 74.6 60.7 97.6 84.9 60.6 100.1
972 77.4 63.2 99.6 85.6 60.7 99.4
973 79.4 65.9 99.5 87.4 62.5 100.3
974 85.2 70.9 99.7 92.1 67.0 102.1
975 90.8 75.7 100.0 95.1 68.0 100.3
976 92.4 77.1 100.0 100.0 71.3 100.0
977 93.2 76.6 98.6 100.0 71.8 100.2
1979 92.1 74.1 96.5 100.0 70.7 99.2
NOTES: Australia: Awards = adult average minimum award rates for a full week's work, all industry groups,
average of four quarters to December 31 each year (Gregory et al., 1985~. Earnings = adult average weekly
earnings for full-time (more than 30 hours) nonmanagerial employees in the private sector (Gregory et al., 1985~.
Great Britain: Awards = weighted average of minimum rates laid down in collective agreements (Tzannatos and
Zabalza, 1984~. Earnings = relative hourly earnings of full-time manual workers (Tzannatos and Zabalza, 1984~.
Of Table 10-3. Earnings increased by 30
percent over the 1970 to 1977 period, so
there was remarkably little slippage from
the tribunal decisions. Award changes were
fully translated into earnings.
The data in Table 10-3 are interesting
because they illustrate the effectiveness of
the equal pay decisions, and as is eviclent
in Figure 10-1, they show that it is only
after discrimination has been removed from
the Australian pay structure that the earn-
ings ratio increases relative to the 1981
earnings ratios of Britain and the United
States. This raises two interrelated ques-
tions.
First, could it be argued that the current
British and U.S. pay structures reflect the
discrimination that used to prevail in Aus-
traTia? Could we go a little further and argue
that with respect to pay Australia is now a
discrimination-free country but the others
are not? This is a position that is at least
consistent with the results of the earlier
section, which indicate that the difference
between the pay relativities across countries
relates to differences in coefficients and not
differences in human capital endowments.
Second, given that labor market insti-
tutions have been so important in changing
the pay relativities in Australia, what has
been the role of labor market institutions
in Britain an(l the United States?
Britain
British labor market institutions appear
to be similar to those that prevail in Aus-
tralia. Trade union membership is large and
national agreements set minimum rates of
pay in a wide range of industries. A relatively
small number of collective agreements de-
termine the minimum rates of pay of a very
large number of workers. The four largest
agreements cover almost one-fifth of the
work force.5
The network of wage agreements in Brit-
ain, however, is not as extensive as the
coverage of fe(leral an(l state awards in
Australia. In Australia 90 percent of female
workers are directly coverer! by the aware]
This section draws heavily on Zabalza and Tzannatos
(1985~.
OCR for page 232
232
wage system. In Britain 41 percent offemale
workers are directly covered by agreements.
As in Australia, before 1975, the British
labor market institutions explicitly recog-
nized pay discrimination. Different rates of
pay for men and women who performed
the same job were written into wage agree-
ments, anti before the 1970s it was common
not to provide equal pay for equal work.
An attempt was made to remove this explicit
discrimination by the introduction of The
Equal Pay Act of 1970, which was to become
effective by December 1975.
The act included three important clauses.
First, it sought to institute "equal pay for
equal work" within establishments. This
clause might be thought of as the equivalent
of the main clause in the Equal Pay Act or
the Civil Rights Act in the Uniter] States.
Second, the act provicled for a pay change
if a female job had been given equal value
to a different male job by means of a job
evaluation. At the broadest level this would
provide for "comparable worth" or, in Aus-
tralian terms, "equal pay for work of equal
value." The act did not require job evalu-
ations, however. Third, and perhaps most
important, if a female pay rate with no male
equivalent was included in a wage agree-
ment, the act provicled that the female pay
rate must be at least equal to the lowest
level of the male pay provision in the agree-
ment. From the point of view of changing
the pay relativities, this clause could be
loosely thought of as the equivalent of the
1974 Australian basic wage judgment, al-
though the level of the minimum pay in
Britain would differ from one wage agree-
ment to another.6
6 An agreement may lay down a rate of pay for
women workers only in a particular category while
making no provisions for men in the same category,
because, for example, there are at the time, no men
doing that kind of work. In such a case, if a rate of
pay applying to women only is lower than the lowest
rate of pay applying to men in the agreement, the
Committee is required to raise the rate applying to
women to the level of the lowest rate applicable to
men" Zabalza and Tzannatos (1985:100).
PAY EQUITY: EMPIRICAL INQUIRIES
On the basis of the Australian evidence
it might be expected that the provisions of
the British Equal Pay Act would be fairly
easily carrier! into the labor market by the
extensive set of wage agreements and the
large degree of unionization. Some incli-
cation of the success of the legislation is
seen in column 4 of Table 10-3, which refers
to the covered sector in Britain. It lists the
average ratio of female to male minimum
rates of weekly pay for full-time manual
employees covered by wage agreements.
The story is similar to the Australian ex-
perience. After very little change over two
clecacles, the minimum rates of pay sullenly
began to increase in the early 1970s, anc!
by 1977 the index was 21.6 percent greater
than in 1969.
In column 5 we list the actual earnings
of British manual workers for the economy
as a whole, that is, the aggregation of the
covere(1 and uncovered sectors. It is evident
from column 6, which compares minimum
awards for the covered sector to average
earnings for the economy as a whole, that
there was no significant slippage between
changes in minimum rates of pay for the
covered sector an(l changes in average earn-
ings for the economy as a whole. Over the
period 1969 to 1977 the ratio of the average
earnings of female to male manual workers
increased by about 24 percent, marginally
more than the minimum rates of pay. The
parallels between Australia an(l Britain are
remarkable (see Figure 10-1~. The earnings
ratios increase in tandem.
In Australia the "equal pay for equal
work" (recisions of the federal tribunal prob-
ably increased female pay by less than 5
percent. In the Unitecl States, if the Equal
Pay Act hail any effect at all, it is likely that
it changed the female earnings ratio in ag-
gregate by considerably less than 5 percent.
Why then was the British Equal Pay Act,
which di(1 not require comparable worth or
equal pay for work of equal value, so ef-
fective? It seems to us that the institutional
structure trade unions and national wage
agreements provi(led the essential pre-
OCR for page 236
236
_
X
_
_
V,
Cal
Cal
.
U)
o
C)
-
o
Cal
C)
o
V)
o
Ct
U'
be
· ~
C~
o
C~
50
U:
o
. ~
C~
oo
-
·C~
.
50
:i ~
~ CD 00
. . .
co ~ GO
_ C~
CO
. . .
C~ ~ oo o
_ o
-
x
. . .
oo ~ o
C~ o
-
x o
. . .
C~ ~ o
_ o
-
oo _ _
o
CO o
-
o o o
. . .
~7 o
oo _ o
oo oo ~ ~ o o ~
. . . . . . .
O ~ C~ CD CO C~
CD ~ ~ 00 ~ O ~
~ ~ ~ ~ ~ ~ o
. . . . . . .
- . ~ CD — — 0
oo ~ ~ ~ o oo
U: C~ _ o C~
C~ o ~ C~ o C~ C~
_
° C -
o ~ ~ C C,,
C c,, te bc C o ~ C
C ~ =, ~ ~ o ~ ~ o O ~ ~ ~ ~ ~
g0 C 0 c 5 ~ ,~ O -c — ~ ~ C O ~ C
11
X
C~
. ~
C~
C~
s~
O
3
4-
o
V:
C~
s~
o
C~
s~
o
C:
11
. C~
o
:d oo O
_
b4 c4
e~
~o =, ~
C~ ~ ~ oo
O ~ P4 _
C~ =50
C) C
O ,o ~
oo ~ o
_ C) ~
a' O v, O
~ °
+ ~ oo
_
V'
~ ~ C) s~
~ ~ :~ ~
. ~ ._ Ct ~
C C
C) ~ U) ~ O
4~ C,:
C~ ~ ~
Ct
·= ~ =-
s~ ~ ~ ~
0 2 u:
3 ~ ~,
E -> e 'E e
~ o~ O ~ ~ ~o O
_ ~ ~ — C~3 S _
c ~ ~ ~ e ~
c ~ - o~ o
~ 3
O ~ O
:^
u,
. .
v,
ct
c~
._
3
o
, ~
0 °
~ u'
_ . _
o o
OCR for page 237
WOMEN'S PAY
reflects the rankings of the aggregate earn-
ings ratio. In the early 1980s, women in
male occupations received 89.0 percent of
the average male wage in Australia, 82.6
percent in Britain, ancI 72.7 percent in the
United States. This gap between the United
States and the other countries leads us to
suggest, somewhat tentatively, that perhaps
it is true that the Equal Pay Act in the
United States was ineffective, not because
equal pay already existed before the act,
but because the nature of the act and the
institutional framework of the labor market
in the Uniter! States prevented the equal
pay initiative from being effective.
EMPLOYMENT AND
UNEMPLOYMENT
The sudden and dramatic change in the
female earnings ratio in Britain and Australia
was not reflected in a sudden and dramatic
change in relative employment growth. The
ratio of total hours worked in the labor market
by women to the total hours worked by men
is given in Figure 10-2 for the three countries.
The vertical scale is plotted as a log of the
employment ratio so that the slope of these
lines is a measure of the growth rate of the
female to male hours worked.
In each country women have increased
their share of total hours worked. In Australia
the increase of female to male hours worker]
between 1970 and 1984 was 25 percent. The
increase was 27 percent in Britain and 31
percent in the United States.
The growth of female employment relative
to that of males has been greatest in the
United States, where the female pay relativ-
ities have not increased; but the surprising
feature of these data is that each of the
employment series is dominated by a trend.
There is no noticeable break in the trends
for Britain and Australia despite relative pay
increases on the order of 20 to 30 percent.
This suggests a very low elasticity of substi-
tution between men and women in the pro-
duction process for example, an elasticity
237
of one, that implied by a Cobb-Douglas pro-
cluction function, should have led to a 20 to
30 percent fall in female hours worke(1 relative
to male hours worked. Nothing like that
occurred. The slow upward trends persisted.
Alternatively, there could have been shifts
in the demand for female labor at the same
time as the pay increases so that the effects
of the relative cost changes are undetectable
in a simple diagram such as Figure 10-2. We
JO not believe that this is an important phe-
nomenon because it requires both that the
unusual shifts be confined to the period of
the pay changes and that the shifts just offset
the effects of the pay change. It all seems so
unlikely, especially because the trenc! in the
United States towar(l female employment is
also fairly smooth. At this stage, we prefer
to think that the elasticity of substitution
between men and women in production is
very low because the labor market is so
segregate(l into malt? and female inbc {T~hl~
10-5).
The insensitivity of the relative employ-
ment growth in Figure 10-2 need not mean
that there are no employment effects at all.
We have not investigated the macroconse-
quences of the pay changes, that is, the (legree
to which male and female employment to-
gether may have been reducecI. If, as a first
approximation, male and female labor can be
thought of as being used in fixed proportions,
then it wouIcl be total employment that is
affected. An increase in female wages would
increase the total wage bill, in real terms, if
the increase in female pay is not offset by a
reduction in male pay.
Finally, a cursory glance at relative un-
employment rates also suggests that the
impact of the large change in pay relativities
on female demand and supply must have
been marginal (Figure 10-3~. In Britain fe-
male unemployment is not well measurecl.
The statistics are based on recipients of
unemployment benefits an(1 many married
women are ineligible; consequently, we
compare unemployment only for Australia
and the United States. The data are col-
~ ~ ~ ~ \ ~ ~~.~
. . . .
OCR for page 238
238
80
70
60
50
40
O _ _
~ _ . _
CC
_
-
-
-
_
30 i 1 1 1 1 1 1 1 1 1 1 1 1
1964 1968 1972 1976
YEAR
-
_
-
PAY EQUIIT: EMPIRICAL INQUIRIES
!
r.
i \_
,!united States
I'
. · _
'_ __ _
Great Britain
Australia
1980 1984 1988
FIGURE 10-2 Ratio of female to male aggregate weekly hours worked. Sources: Australian Bureau of Statistics,
Labour Force, Cat. No. 6303, August. U.S. Department of Labor, Bureau of Labor Statistics, Employment and
Earnings Monthly Bulletin, annual averages. Great Britain, Annual Abstract of Statistics, mid-June.
lected from household surveys. During and
after the period of the equal pay initiatives
in Australia, the unemployment of women
continued to fall relative to that of men and
appears not to have been affected by the
pay changes to a great degree. In both
countries female unemployment has been
marginally above that of male unemploy-
ment for most of the period and subject to
a strong downward trend. It is remarkable
how highly correlated the relative unem-
ployment series is across the two countries.
Again, there is no evidence of Australian
women being seriously (lisadvantaged after
the equal pay judgments.
CONCLUSIONS
In the three countries female employ-
ment, relative to that of males, has grown
strongly over the past decade and a half.
The history of the pay gap between men
an(1 women in each country, however, has
been (different. There was a significant nar-
rowing ofthe pay gap in Britain and Australia
cluring the 1970s, but there was no change
in the Unitecl States. Recently, there has
been relative constancy in Britain and Aus-
tralia, but a more moderate narrowing of
the pay gap in the Unite(l States.
This analysis has focused on the pay gaps
of the 1970s and leacls to the following major
conclusions. First, an application of the usu-
al human capital mode] cannot explain pay
gaps across countries. In 1981 the relative
endowments of human capital of men and
women seem to be much the same in each
of the three countries. Within a human
capital framework the variation in the pay
gap among these countries stems from dif-
ferences in coefficients rather than from
differences in enclowments.
Second, between 1969 anti 1976 the ratio
of female to male pay increased about 30
OCR for page 239
WOMEN'S PAY
3.00
2.50
o
CC
239
2.00
1.50
1.00
-
\
0.501
1 964
\~W
\~stralia
\ ~
United States
~ 1 1 ~ ~ 1 ~ 1
1968 1972 1976
YEAR
._, tic
, , ,
1980 1984 1988
FIGURE 10-3 Ratio of female to male unemployment rates. Sources: Australian Bureau of Statistics, The Labour
Force, Historical Summary, Cat. No. 6204, August. U.S. Department of Labor, Bureau of Labor Statistics,
Employment and Earnings Monthly Bulletin, annual averages.
percent in Australia and by 20 percent in
Britain. The human capital mocle] cannot
explain these large pay changes, which flowed
from official intervention in the labor mar-
kets. The pay changes followed two clecades
of relative constancy. The Australian and
British experience, therefore, was different
from that of the United States, where, dur-
ing the early 1960s, official intervention in
the form of the Equal Pay Act and Title 7
of the Civil Rights Act prover] to be inef-
fective. These results suggest that research
in this area must place increaser] emphasis
on institutions and the impact of the law.
Third, we are not sure why official in-
tervention was effective in Australia and
Britain but ineffective in the United States.
The labor markets of Britain ancl Australia
possessed institutional features within which
change was easily effected. The key feature
seems to have been national wage agree-
ments, which, before the large pay changes,
explicitly discriminated between men en c!
women by giving them different rates of
pay for the same job. As a result, some part
of pay discrimination was easily identified.
This, however, seems to be only a small
part ofthe answer. The fraction ofthe female
work force covered by "equal pay for equal
work" is not sufficient to bring about directly
such large changes in the pay ratio. A larger
part of the explanation probably relates to
the fact that the interventions in Australia
ant] Britain went beyond equal pay for equal
work by adopting simple, across-the-board
rules. In Britain these simple rules related
to setting equal the male ant] female min-
imum rates of pay within each wage agree-
ment. In Australia, there was one minimum
wage the basic wage and that, too, was
set equal for men and women.
Fourth, currently, Australian women are
much better paid relative to men than their
British or U.S. counterparts. An important
OCR for page 240
240
PAY EQUITY: EMPIRICAL INQUIRIES
Chapman, B. J., and C. Mulvey
1986 An analysis of the origins of sex differences
in Australian earnings. Journal Of Industrial
Relations 28~4):504-520.
Chiplin, B., M. Curran, and C. Parsley
1980 Relative female earnings in Great Britain and
the impact of legislation. Pp. 57-126 in P. J.
Sloane, ea., Women and Low Pay. London:
Macmillan.
Daymont, T., and P. Andrisani
1984 Job preferences, college major, and the gen-
der gap in earnings. Journal of Human Re-
sources (3) (Summer):408-428.
Department of Employment
1975 Progress towards employment. Employment
Gazette (London).
Greenhalgh, C. A.
1980 Male-female wage differentials in Great Brit-
ain: Is marriage ah equal opportunity? Eco-
nomic Journal 90(December):751-775.
Gregory, R. G., and R. G. Duncan
1981 Segmented labor market theories and the Aus-
tralian experience of equal pay for women.
Journal of Post Keynesian Economics II(3):40
428.
reason for this is the pay ratio in female
occupations. In the three countries between
62.8 and 53.7 percent of women work in
occupations in which 70 percent or more of
the workers are female. In Australia, these
women, on average, are paid 71.4 percent
of the average male weekly wage. In Britain
and the United States the ratio is 60.3 and
56.0 percent, respectively.
Fifth, in Britain and Australia it appears
that all women shared in the large pay in-
creases. There is no evidence of a significant
development of uncovered sectors, nor of
groups of fi~-time workers whose working
conditions ~leter~orated as the working con-
ditions of those covered by equal pay im-
proved. It does not appear that the labor
market is evolving toward two classes of wom-
en, although there has been a faster rate of
growth of women working part time in Britain
and Australia.
Finally, in Britain and Australia the ex- Gregory, R. G., and V. Ho
perience of large pay changes seems to suggest 1985
that the relative employment response is not
large. To a significant degree, women (lo not
seem to lose jobs restive to men. In fact,
the relative employment of women increased
in all three countries during the 1970s. We
have not unclertaken an analysis of aggregate
employment growth, in which an important
consideration would be who pays for the
female wage increases. If there is an offset
in male wages, then the employment response
in aggregate is likely to be much less than if
men maintain their real wages and compa-
rable worth acts as a profit tax.
REFERENCES
Beller, A. H.
1979 The impact of equal employment opportunity
laws on the male/ female earnings differential.
Pp. 304-330 in C. B. Lloyd, E. Andrews,
and C. Gilroy, eds., Women in the Labor
Market. New York: Columbia University Press.
Blandy, R.
1963 Equal pay in Australia. Journal of Industrial
Relations 5:13-28.
Blau, F., and M. Ferber
1986 The Economics of Women, Men and Work.
Englewood Cliffs, N. J.: Prentice-Hall.
Equal Pay and Comparable Worth: What Can
the U.S. Learn from the Australian Experi-
ence? Discussion Paper No. 123. Centre for
Economic Policy Research, Australian Na-
tional University, July.
Gregory, R. G., A. Daly, and V. Ho
1986 A Tale of Two Countries: Equal Pay for Women
in Australia and Britain. Discussion Paper No.
147. Centre for Economic Policy Research,
Australian National University, August.
Johnson, G., and E. Solon
1986 Estimates of the direct effects of comparable
worth. American Economic Review (Decem-
ber):lll7-1125.
Mincer, J
1974
-
Schooling Experience and Earnings. National
Bureau of Economic Research. New York:
Columbia University Press.
1985 Intercountry comparisons of labor force trends
and of related developments: An overview.
Journal of Labor Economics 341) (Part 2~:S1-
S32.
Niland, J. R., and J. E. Isaac
1975 Australian Labour Economics Readings. Mel-
bourne: Sun Books.
Oaxaca, R.
1973 Male female wage differentials in urban labor
markets. International Economic Review 14
(October):693-709.
Zabalza, A., and T. Tzannatos
1985 Women and Equal Pay. New York: Cam-
bridge University Press.
OCR for page 241
WOMEN S PAY
APPENDIX: DEFINITION OF
VARIABLES USED IN THE
REGRESSION EQUATIONS
The results relate to full-time workers,
that is, those working more than 35 hours
per week. In Australia the sample was men
and women aged 15 to 54 years; in Britain,
16 to 54 years; ant] in the United States,
15 to 54 years.
Australia
Education Variables
Dropout Age on leaving school was less
than or equal to 15; no further qualifications.
High school- Age on leaving school was
greater than or equal to 16, but the person
hac] no postsecondary qualifications.
Postsecondary—Trade certificate or oth-
er postseconciary certificate.
University degree completion of a bach-
elor's degree.
Postgraduate higher degree level.
Experience
Age minus years of schooling minus six.
Chiil(lren
A crummy variable taking the value of 1
if children under the age of 18 were present
in the household and were the responsibility
of the head of the household or spouse.
Area
Rural Those living in a community of
less than 1,000 people.
Urban—Those living in a community of
more than 1,000 people.
Marital Status
Spouse present Currently married and
living with spouse.
241
Other marital status Widowed, sepa-
ratecI, and clivorced indivicluals.
Single Never married.
Great Britain
Education Variables
Dropout Those with no qualifications
or with ungraded] or grades 2 to 5 of a
Certificate of Secondary Education.
High school—The person had one of the
following: a Certificate of Secondary Eclu-
cation grade 1, school certificate, one or
more General Education Certificate "0" lev-
els or the Scottish equivalent (Scottish Leav-
ing Certificate of Education), or clerical and
commercial qualifications.
Postsecondary trade apprenticeship,
GCE "A" level or other postsecondary qual-
ifications.
University completion of a bachelor's
degree.
Postgraduate higher (legree level.
Experience
Age Age on leaving fulI-time education.
Children
A dummy variable taking the value of 1
if children under the age of 16 were present
in the household and were the responsibility
of the heat] of the household or spouse.
Area
Rural Those living in rural local au-
thority areas.
Urban Those living in urban local au-
thority areas or in the conurbations (e.g.,
Greater London).
Marital Status
Spouse present—Currently married and
living with spouse.
OCR for page 242
242
Other marital status Widowed, sepa- Children
rated, and divorced indivicluals.
Single Never married.
United States
Education Variables
Dropout Completer] less than 4 years
of high school.
High school Completed 4 years of high
school.
Postsecondary qualifications Complet-
ed 1 to 3 years of college.
University degree Completed 4 years
of college.
Postgraduate degree—Completed 5 or
more years of college.
E.
xper~ence
Age minus years of schooling minus six.
PAY EQUITY: EMPIRICAL INQUIRIES
A crummy variable taking the value of 1
if children uncler the age of 18 were present
in the househoicl and were the responsibility
of the head of the household or spouse.
Area
Rural—Those living in communities of
less than 1 million people.
Urban Those living in central cities or
other communities of more than 1 million
people.
Marital Status
Married spouse present Currently mar-
rie(1 and living with spouse.
Other marital status Widowed, sepa-
ratecI, and divorced individuals.
Single Never married.
OCR for page 243
Commentary
RONALD G. EHRENBERG
At the start of the 1970s, the average
weekly earnings of employed women rel-
ative to the average weekly earnings of
employed men (which I henceforth call the
female relative wage) was approximately equal
in the United States and Australia and some-
what lower in Great Britain. During the
decade, however, the female relative wage
rose substantially in Australia and Great
Britain, but remained roughly constant in
the United States. As a result, by the early
1980s, the female relative wage in Australia
exceeded that for the other two countries
and the relative wage in Great Britain had
reached roughly the same level as the rel-
ative wage in the United States.
With these facts as background, the Greg-
ory, Anstie, Daly, and Ho paper aciclresses
four questions. First, why do female relative
wages slider across countries in the early
1980s—do the differences reflect cliffer-
ences in relative (female/maTe) human cap-
ital endowments or differences in labor mar-
ket institutions across countries? Second,
why dicl women achieve such large relative
wage gains vis-a-vis men in Great Britain
and Australia during the 1970s? Third, what
is the implication of the experience in Aus-
243
tralia and Great Britain for the debate over
comparable worth in the United States and
elsewhere? Finally, what can we learn from
the Australian and British experiences about
the effects of such relative wage changes
on employment and unemployment of wom-
en?
To answer the first question, the authors
use the now standard Oaxaca decomposition
method to determine whether differences
in female relative wages across countries
are clue to differences in relative (female/
mate) human capital endowments or differ-
ences in wage equation coefficients across
countries. "Comparable" micro-level data
sets from househoIc] surveys are used for
each country, and log weekly earnings equa-
tions for men and women are specified to
be a function of education, potential labor
market experience (age minus years of
schooling minus six), marital status, the
presence of children in the home, and rural/
urban location.
The estimates (fount] in Table 10-1) are
used to compute the extent to which the
female relative wage in a country changes
when one substitutes the mean value of
male and female characteristics from either
OCR for page 244
244
of the other two countries into its wage
equation to arrive at a predicted female
relative wage. By substituting the mean
characteristics of British workers into the
Australian wage equations, for example, one
can compute what the female relative wage
in Australia wouIcT look like if Australian
workers had the same human capital en-
dowments as British workers. In fact, the
authors find (Table 10-2) that the predicted
female relative wage in each country appears
to be quite insensitive to which country's
characteristics are used. Thus, they con-
clude that most of the clifference in female
relative wages across the three countries is
due to differences in the wage structures
(coefficients of the wage equations); human
capital differences are relatively unimpor-
tant.
To answer the second question, why fe-
maTe relative wages rose in Australia and
Great Britain during the 1970s, the authors
focus on the unique labor market institutions
that exist in each country and the funda-
mental changes that took place in them
cluring the decade. Turning first to Aus-
tralia, minimum wage rates are set there
by occupation through a system of federal
and state tribunals. Prior to the early 1970s,
minimum wages in male-clominated occu-
pations were set by determining some min-
imal living standard for a family and then
acIding to that a premium for the "work
value" of the occupation. Minimum wages
in other occupations were similarly deter-
mined, but in femaTe-dominatec! occupa-
tions an explicit downward adjustment was
then maple (usually 25 percent cluring the
1950-1969 period). Discrimination against
women, then, was explicit in Australia.
Between 1969 and 1975, two very im-
portant changes occurred in the tribunal's
behavior. First, by 1972 the federal and
most state tribunals had ruler] that the sex
of a worker should not be used as a criterion
in setting wages in those jobs that were
neither predominantly male nor pre(lomi-
nantly female (equal pay for equal work).
PAY EQUITY: EMPIRICAL INQUIRIES
Second, between 1972 and 1975, the tri-
bunals introclucec] the concept of"equal pay
for work of equal value?' by eliminating in
three stages the (lownwarcl adjustment of
wages in femaTe-dominate(1 occupations.
Thus, something akin to comparable worth
was mandated by the fe~leral government
for all workers in Australia. As these changes
occurred, the female relative wage in Aus-
tralia rose from roughly 60 percent in 1970
to 75 percent in 1979 (Table 10-3~.
Turning next to the British experience,
the British labor force, as is well known, is
heavily unionized; the four largest national
agreements cover almost one-fifth of the
work force. Prior to 1975, explicit sex dis-
crimination in pay (different pay rates for
men and women doing the same job) was
built into the agreements. The Equal Pay
Act of 1970, which was to become effective
in 1975, required equal pay for equal work
within a firm, equal pay for jobs of equal
value within a firm if a job evaluation was
undertaken, and that women's wages be at
least equal to the lowest male wage rate in
the firm. The authors conclude that the act
appeared to be effective: The female relative
earnings of manual workers rose from rough-
ly 60 percent in 1970 to 71 percent in 1979.
The authors next address the third issue,
the implications of these fin(lings for the
comparable worth debate. They point out
that in both Great Britain and Australia pay
discrimination was explicit, and it was thus
easy to identify where (discrimination was
occurring. Given the unique labor market
institutions in each country, government
intervention could be direct; they observed
that "the marketplace" did not appear to
frustrate the equal pay efforts.
For the future, the authors see little room
for expansion of comparable worth in Aus-
traTia (effectively, a variant is aireacly in
place). An Equal Pay Act amendment in
Great Britain (effective January 1984) now
permits British women to bring claims
through the judicial process if they believe
they are not receiving equal pay for work
OCR for page 245
COMMENTARY
of equal value. The amenciment's ultimate
effects will depend on the judicial process.
The authors also claim that although the
introduction of comparable worth in the
United States in the state and local sectors
has led to considerable initial pay changes
(although some economists would dispute
this cIaim), the decentralized wage deter-
mination process in the Unitecl States is
likely to make the overall effects of any
larger comparable worth policy much small-
er (due to likely incomplete coverage and
employment of women in low-paying es-
tablishments).
Finally, the authors address the issue of
whether the female relative wage changes
that have occurred in Australia ant] Great
Britain have had any adverse employment
and unemployment effects. They note that
during the 1966-1984 period! women in-
creased their share of hours worked in all
three countries, and that the increase was
greatest in the United States. They claim
a growth trend was dominant in all cases,
and no sharp slowdown in growth was ob-
servec] in Great Britain or Australia after
those countries implementecl their antidis-
crimination policies. They assert (without
proof) that this is due to small elasticities
of substitution between male- and female-
clominated occupations. A cursory glance at
the pattern of female relative unemploy-
ment rates in each country over time sim-
ilarly leads them to the conclusion that there
is no evidence that the relative wage changes
affecter] these rates either. They note, though,
that they have not analyzed the effect of
the female wage adjustments on total em-
ployment, on male wages, or on corporate
profits.
My reaction to this paper is mixed. On
the one hand, it represents one of the few
serious efforts I know of to place discussions
about comparable worth in a comparative
perspective and to bring evidence from oth-
er countries' experiences into the (rebate
about policy in the United States. For this
the authors should be resoundingly applaud-
245
eel. On the other hand, I am left with the
feeling that they have not pusher] their
empirical analyses as hard as they might
have, and because of this, in places they
may have drawn some inappropriate con-
clusions. My discussion will elaborate on
this latter theme.
Consider, first, the analyses of the de-
terminants of intercountry differences in the
female relative wage differentials. Although
the authors, probably justifiably, conclude
that the differences are due to cliffering
coefficients of wage equations across coun-
tries, not to differences in human capital
endowments, they do not attempt to explain
why the coefficients found in Table 10-1
might (lifter across countries. The presence
of children in the home, for example, ap-
pears to have a much larger negative effect
on women's wages in Australia than it floes
in the United States. Is this because Aus-
tralian families have more children than
U. S. families, or because the lack of chil~l-
care facilities in the former makes it more
likely that women will leave the labor force
temporarily to care for children? To take
another example, the return to potential
experience for women appears to be greater
in Australia than in the United States (Table
10-1~. When one takes into account the
higher labor force participation rates of aclult
women in the United States, it is likely that
potential experience (age minus years of
schooling minus six) systematically over-
states actual experience by more in Australia
than in the Unite(l States and, thus, that
the actual returns to female labor market
experience are certainly greater in Australia.
I wish the authors tract provided an expla-
nation for this, as well as for other findings.
Consider, next, their analysis of how fe-
male relative wages changed after the pas-
sage of the anticliscrimination laws in Aus-
tralia and Great Britain. They base their
analysis here on casual analysis of obser-
vations on annual data from 1964 to 1979.
I am troubled by their conclusion ~com-
parable-worth-type policies caused relative
OCR for page 246
246
wage changes), for a number of reasons.
First, their data inexplicably end in 1979.
Given the availability (presumably) of data
for another 5 or 6 years, they couIc3 have
studied whether the trencis they observed
in the data continued. Second, one must
question their failure to estimate a multi-
variate mode} that would permit other forces,
such as aggregate (leman(l pressures, to
influence the female wage ratio; their con-
clusions are implicitly based on simple cor-
relations of policy changes and wage ratio
changes. Third, they implicitly treat the
policy changes as exogenous. No thought is
given to the possibility that social or eco-
nomic pressures that might leac] female rel-
ative wages to rise might also lead to the
policy changes ant] tribunals' decisions. Put
another way, they may have the direction
of causation backwards. Indeed, their Fig-
ure 10-1 suggests that the female relative
wage started to rise in both Australia (1965-
1969) and Great Britain (1970-1971) prior
to the implementation of the policies that
they describe.
The conclusion that female relative wage
changes have hack no ejects on female rel-
ative employment and unemployment lev-
els is similarly based solely on cursory ex-
aminations of trends in the data. Since
Gregory and Duncan's earlier paper (Iour-
PAY EQUITY: EMPIRICAL INQUIRIES
nal of Post-Keynesian Economics, 1981) es-
timated relative (male/femaTe) employment
equations for Australia for the 1938-1978
period and fount] some role for relative
wages tin the aggregate a -.3 elasticity of
substitution with respect to relative wages),
after controlling for trend terms and macro-
leve! conditions, one wonders why a similar
structures] analysis was not clone here. At
the very least, such an analysis wouIcl pro-
vide some comparative data on elasticities
of substitution between men and women.
To do this correctly, of course, would re-
quire a formal model of employment anc]
labor force behavior.
Ignoring my concerns about the nature
of their empirical evidence, I take away a
message from this paper that is a simple
but important one: It is likely to be much
easier to improve the female relative wage
rate by a comparable-worth-type policy in
a world in which wages are set centrally
and discrimination is overt than it is in a
(decentralized market economy in which we
still argue over whether labor market dis-
crimination occurs. Proponents of compa-
rable worth in the United States should take
heed. Widespread comparable worth ini-
tiatives here are unlikely to improve the
female relative wage by as much as they
did in Australia and Great Britain.
Representative terms from entire chapter:
human capital
