Unformatted Attachment Preview
Why Are Married Women Working More?
Some Macroeconomic Explanations
BY AUBHIK KHAN
F
or the past 60 years, the number of hours
worked per person in the U.S. has changed
very little. Nonetheless, the labor force has
undergone some pronounced shifts over that
same period. One prominent change is the sharp increase
in the number of hours worked by married women. In
this article, Aubhik Khan discusses how the composition
of the labor force has changed since 1945 and how
macroeconomists explain these changes.
Since the Second World War,
there has been little overall change
in the number of hours worked per
person in the United States. Hiding
under this apparent constancy lie some
pronounced shifts in the composition of the labor force. The share
of employment attributable to men
and women, to older workers, and to
married households has changed, in
some instances rather dramatically.
How have these shares changed, and
what does recent economic research
have to say about these compositional
changes?1
My discussion of changes in hours worked uses
the work of Ellen R. McGrattan and Richard
Rogerson closely. The advanced reader should
consult their 1998 and 2004 papers for a far
more thorough analysis.
1
Aubhik Khan
is a senior
economist in
the Research
Department of
the Philadelphia
Fed.
16 Q4 2004 Business Review
Perhaps the most prominent
change in the composition of the labor
force has been the sharp rise in the
hours worked by married women. Motivated by this rather striking phenomenon, macroeconomists have developed models to explain the asymmetric
rise over the past 40 years in weekly
hours worked by married women.
Three basic changes in the economy
likely have contributed to a rise in
hours worked by married women (in
no particular order of importance): (1)
technological progress that has made
durable consumer goods more productive; (2) a reduction in the gender
wage gap associated with lower pay for
women than for men; and (3) a change
in social attitudes toward married
women working outside the home.
CHANGES IN COMPOSITION
OF THE LABOR FORCE
If we ignore differences in the
sex, age, or marital status of workers
and look at aggregate average hours
worked, the number of weekly hours of
market work per person has remained
roughly constant over the postwar
period from 1950 to 2000 (Table 1).2
This is not to suggest that there have
not been short-term fluctuations. For
example, we know that hours worked
per person fall during recessions
(as firms’ demand for employment
decreases) and rise during expansions.
However, aside from such cyclical fluctuations, the long-run value
changed little between 1950 and 2000:
The data indicate that average weekly
hours worked per person were 22.34 in
1950 and rose slightly to 23.90 in 2000.
Average weekly hours are,
of course, considerably less than the
familiar 40-hour workweek, since not
all persons are employed. However,
the first indication that the aggregate
measure hides changes across different
groups of workers comes from an examination of the employment to population ratio. Over the same 50 years,
this ratio has risen from 0.53 to 0.59.
Thus, while five out of 10 people were
working in 1950, 50 years later nearly
six out of 10 people were employed in
the economy. This substantial rise in
the employment-to-population ratio
and the smaller increase in average
weekly hours per person together imply
that the hours worked by the typical employed individual have fallen.
Indeed, on average, workers worked
two fewer hours per week in 2000 then
they did in 1950.
Of course, the constancy of
average weekly hours per person does
I survey the postwar data using the decennial
U.S. census, which is taken in the final year of
every decade. All tables are based on data taken from McGrattan and Rogerson’s 2004 article.
I thank Ellen McGrattan for making these data
available to me.
2
www.PhiladelphiaFed.org
TABLE 1
Average Weekly Hours
Year
Average Weekly Hours Worked
Per Person
Per Worker
Employment-to-Population
Ratio %
1950
22.34
42.40
52.69
1960
21.55
40.24
53.55
1970
21.15
38.83
54.47
1980
22.07
39.01
56.59
1990
23.86
39.74
60.04
2000
23.90
40.39
59.17
6.98
-4.74
12.30
% Change
1950-2000
Source: Table based on data presented in McGrattan and Rogerson (2004); original source for the
data is the U.S. census.
not reflect a constancy of earnings.
Average real compensation per hour is
a common measure of real labor earnings, which includes workers’ benefits
and controls for the effects of inflation
on nominal earnings. Between 1950
and 2000, average real compensation
per hour rose more than 150 percent
(Figure 1). Thus, while workers are
earning much more, the population as
a whole is not working more.3
A SIMPLE ECONOMIC MODEL
OF THE LABOR-LEISURE
TRADEOFF
The constancy of hours
worked in light of changes in wages is
interesting to economists, as it offers
some insight into workers’ preferences.
To see this, consider the following
very primitive model of labor supply
sometimes used by macroeconomists.
Average real compensation per hour is a better
measure of earnings than wages. For example,
it would make little sense to focus on a measure
that ignored health insurance provided by an
employer. However, I will use the term wage
in what follows as shorthand for compensation
per hour.
Assume for simplicity that each worker
values two goods. We call the first
consumption, a single commodity that
represents all the different goods and
services we use. The second good is
leisure, which is not produced but is
granted to the worker as time. The
worker may devote his time to either
leisure, which he values, or to labor,
which earns him a wage.4 Since wages
pay for consumption goods, any worker
faces this fundamental tradeoff: The
time spent enjoying leisure could have
been spent working for wages. Given
any real wage — the amount of consumption goods that can be purchased
with a given money wage — the worker must choose how much of his time
to allocate to labor, and the remainder
is leisure.
A rise in his wage will induce
a wealth effect: With no change in
hours worked, the worker is now
wealthier. The wealth effect tends to
make the worker consume more of
most goods – economists call these
normal goods.5 Economists think of
leisure as one such commodity. Thus,
the wealth effect tends to reduce the
quantity of labor supplied in response
to a rise in wages as people wish to
have more leisure. Nonetheless, since
the worker may earn more than before
— if he does not reduce his hours
worked too sharply — both leisure and
consumption will rise.
The rise in wages also implies
a substitution effect: The cost of leisure
has now risen, since each hour of
leisure means an hour less of work,
which is now worth more. As the cost
of leisure rises, demand falls, and this
by itself should increase the worker’s
hours of work. The wealth and substitution effects conflict, and, in general,
there is no way to tell which will
dominate. However, the observation
that average hours worked per person
have not changed in response to a 150
percent rise in earnings has led many
macroeconomists to suggest that at
least for the average or representative
household in the economy, the wealth
and substitution effects have offset
each other. Thus, this offsetting is one
explanation for the observed lack of
trend in hours worked.
WOMEN’S WORK PATTERNS
CHANGED DRASTICALLY
While this net cancellation may summarize the behavior of
the representative household, it does
not accurately reflect the changes in
labor supplied by men and women,
nor the young and old. There have
been large movements in hours worked
across all these groups. However, as we
have already noted, the sum of these
3
www.PhiladelphiaFed.org
Thus, this simple economic model assumes we
do not value jobs directly, but rather indirectly
through the goods available to us as a result of
earning a salary.
4
For example, generic paper towels may not be
a normal good. As your wage rises, you might
switch to a name brand; hence, your expenditure on generic paper towels would fall.
5
Business Review Q4 2004 17
FIGURE 1
Average Weekly Hours Worked per Person
and Real Compensation per Hour Worked*
Index
(1992 = 100)
Hours
120
30
110
Hours per Person
100
20
90
80
70
10
60
50
Compensation per Hour
40
0
1950
1960
1970
1980
1990
2000
* The compensation series is an index of hourly compensation in the business sector, deflated
by the consumer price index for all urban consumers. The index is constructed to equal 100
in 1992.
TABLE 2
The Distribution of Hours Worked by Gender
Year
Average Weekly Hours Worked per Person by Gender
Total Population
Males
Females
1950
22.34
34.18
10.87
1960
21.55
31.93
11.84
1970
21.15
29.72
13.32
1980
22.07
28.70
16.02
1990
23.86
29.11
19.03
2000
23.90
28.34
19.78
6.98
-17.09
81.97
(second panel of Table 4).6 This does
not mean that single women are working less than married women. Rather,
in 1950, the U.S. census shows married
women working far fewer hours than
single women. However, 50 years later,
these differences had largely evaporated as the hours worked by married
women rose to match the initially
longer workweek of single women.
Moreover, most of the change in married women’s hours of market work
happened between 1950 and 1990.
To see this clearly, take, for
example, the weekly hours of married
and single women, between 35 and
44 years of age, in 1950. The census
conducted that year shows that, on
average, married women in this age
group worked about 9.5 hours a week.
Single women in this age group worked
far more: 30.5 hours a week. Now reexamine the weekly hours of women in
the same age group, but 50 years later.
In 2000, married women ages 35 to
44 were working 26 hours, on average. Single women in this age group
worked an average of about 29.5 hours
a week, actually slightly less than their
predecessors 50 years ago.
Across all age groups, the
length of the average workweek of
married women (with spouses present) rose about 200 percent, from
about 7 hours to over 20 hours a week,
between 1950 and 2000 (Figure 2).7
% Change
1950-2000
Generally, the changes in hours worked by
single men and single women, of any age, are
rather similar.
6
Source: Table based on data presented in McGrattan and Rogerson (2004); original source for the
data is the U.S. census.
The census refers to married men (or women)
with spouses present as a separate group from
married men (or women) with spouses absent.
A married person with a spouse present is living in a household with the wife or husband.
Married people with spouses absent include
those with spouses in the military or living in
institutions. The 2000 census indicates there
are 2 million households composed of a married
person with spouse absent. Hereafter, when I
describe a married person whose spouse is present, I will simply describe them as married.
7
movements has had little overall effect
on average hours worked per person.
Separating weekly hours per worker,
we find that hours worked by males
fell 17 percent, while hours worked by
females rose an astounding 82 percent
between 1950 and 2000 (Table 2).
Almost all of the rise in
18 Q4 2004 Business Review
female hours worked is explained by
an increase in the average weekly
hours spent in employment by married
women (Table 3). The weekly hours
worked by married women ages 25 to
54 rose, on average, more than 200
percent! The corresponding figure for
single women is actually -1.3 percent
www.PhiladelphiaFed.org
TABLE 3
Changes in Hours Worked by Married People
Weekly Hours Worked Per Person by Age (in Years)
Status
Spouse
Present*
Total
Gender
Year
15-24
25-34
35-44
45-54
55-64
65-74
75-99
Males
1950
39.68
42.20
43.46
42.06
37.62
23.39
9.74
1960
39.50
42.33
42.77
41.21
35.77
14.74
6.23
1970
37.14
41.86
42.60
40.90
34.73
11.59
4.31
1980
37.80
40.99
42.01
39.79
30.53
8.15
3.08
1990
38.75
42.52
42.88
40.88
28.68
7.69
2.56
2000
37.17
40.99
41.88
40.39
29.17
8.15
3.06
% Change
1950-2000
-6.33
-2.87
-3.64
-3.97
-22.46
-65.16
-68.58
1950
9.03
7.93
9.43
8.43
4.42
1.67
0.52
1960
10.00
9.10
12.35
13.56
8.66
2.27
0.98
1970
14.67
12.23
15.04
16.26
11.77
2.53
1.22
1980
18.95
19.25
20.13
18.61
12.18
2.44
0.79
1990
21.75
24.36
25.95
24.51
14.00
2.84
0.70
2000
20.49
24.49
26.03
27.27
16.99
3.38
1.02
126.91
208.83
176.03
223.49
284.39
102.40
96.15
Females
% Change
1950-2000
Source: Table based on data presented in McGrattan and Rogerson (2004); original source for the data is the U.S. census.
*A married person with a spouse present is living in a household with the wife or husband. Married people with spouses absent include those with
spouses in the military or living in institutions.
Finally, it is important to emphasize
that these figures are hours per person,
not hours per worker. To a significant
extent, the increase in hours worked by
married women is due to their greater
participation in the labor force. In
sharp contrast to the behavior of married women, the average hours worked
per single woman remained relatively
unchanged, rising 11 percent (Figure 3). Over the same period, hours
worked by married men with spouses
www.PhiladelphiaFed.org
present fell eight hours, or 20 percent.
Finally, hours worked by single men fell
7 percent.
In their 2003 paper, Larry
Jones, Rodolfo Manuelli, and Ellen
McGrattan adopted an interesting perspective on these changes in the labor
force. They noted that in 1950, a married couple’s total hours worked in the
market were much fewer than those we
would obtain by summing the hours
worked by the average single man and
woman. Their census data indicate
that this artificial couple (formed by
combining a single man and a single
woman) would have worked, on average, 60.5 hours a week in 1950; their
hours worked would have changed
little over the next 40 years, falling
slightly to a little over 59 hours by
1990. As I mentioned, the total hours
worked in the market by the married
couple together was initially far less,
49.5, in 1950. However, by 1990, differBusiness Review Q4 2004 19
TABLE 4
Changes in Hours Worked by Single People
Weekly Hours Worked Per Person by Age (in Years)
Status
Gender
Year
15-24
25-34
35-44
45-54
55-64
65-74
75-99
Single
Males
1950
20.02
32.81
34.14
32.07
27.19
15.44
6.00
1960
14.92
31.99
30.78
29.19
24.35
9.74
5.07
1970
13.20
30.88
30.30
28.14
22.31
8.76
4.27
1980
16.58
31.80
30.25
27.21
19.90
6.11
2.30
1990
16.75
32.86
30.88
27.25
18.19
5.99
2.21
2000
16.25
33.29
30.68
27.81
19.12
6.42
3.63
% Change
1950-2000
-18.83
1.46
-10.13
-13.28
-29.68
-58.42
-39.50
1950
14.25
30.64
30.53
28.61
22.72
10.31
3.14
1960
10.76
29.46
29.49
29.07
24.40
10.62
3.40
1970
10.44
28.72
27.70
27.69
24.38
8.34
3.08
1980
13.53
30.28
28.89
26.73
20.60
5.20
1.36
1990
14.17
30.75
30.99
28.30
19.11
5.23
1.17
2000
13.86
30.52
29.56
28.53
19.67
5.43
1.51
% Change
1950-2000
-2.74
-0.39
-3.18
-0.28
-13.42
-47.33
-51.91
Females
Source: Table based on data presented in McGrattan and Rogerson (2004); original source for the data is the U.S. Census.
ences in hours worked between these
two pairs — the married couple and
the artificial couple — had largely disappeared. The average married couple
was working 61 hours by then. Thus,
we see that in terms of total hours
spent in the market, married couples
are now behaving much more like
single people. Why has the behavior of
married households changed?
THE HOUSEHOLD
EMPLOYMENT DECISION
To answer the question
about changes in hours worked, we
20 Q4 2004 Business Review
must consider the determinants of
hours worked. Many economic factors
determine an individual’s employment
decisions. The number of dependents
and earning ability are just two characteristics that come to mind. In turn,
these characteristics are themselves
affected by an individual’s decisions
about the number of children to have
and the years of schooling to invest in.
I won’t attempt to discuss the general
economic theory of labor supply, a rich
theory that has been developed over
several decades by many economists.
Instead, I will focus on more recent
macroeconomic models developed
to understand why married women’s
hours of work in the market have risen
so sharply.8
Three basic changes in the
economy have likely contributed to a
rise in hours worked by married women: (1) technological progress that has
made durable consumer goods, such
Given this focus, the models I will discuss will
abstract from many issues that affect individuals’ employment choices but do so more or less
uniformly across individuals of different gender
and marital status.
8
www.PhiladelphiaFed.org
FIGURE 2
Average Weekly Hours Worked per Person
Married, Spouse Present*
45
40
Males
35
30
25
20
Females
15
10
5
0
1950
1960
1970
1980
1990
2000
* A married person with a spouse present is living in a household with the wife or hus-
band. Married people with spouses absent include those with spouses in the military
or living in institutions.
FIGURE 3
Average Weekly Hours Worked per Person
Not Married (Single, Widowed, Divorced,
Married-Spouse Absent*)
25
Males
20
Females
15
10
5
0
1950
1960
1970
1980
1990
2000
Year
* A married person with a spouse present is living in a household with the wife or hus-
band. Married people with spouses absent include those with spouses in the military
or living in institutions.
www.PhiladelphiaFed.org
as home appliances, more productive,
(2) a reduction in the gender wage gap
that yields lower pay for women than
for men for the same work, and (3) a
change in social norms.
Producing at Home or
Working in the Market. The data
do not imply that married women are
working more hours but that they are
working more hours in formal employment, that is, in the market. This has
come at the expense of fewer hours
worked at home. A necessary starting
point for understanding changes in
married women’s market hours is a discussion of how a household allocates
the time of its adult members between
the home and the market. This involves understanding the economic
model of home production.
All households, whether
composed of a married couple or a
single adult, value goods bought in
the market – for example, restaurant
meals, wine, and the inevitable dose of
aspirin – and goods produced at home
– for example, breakfast.
To purchase market goods,
the typical household must work in
the market. The earnings from this
employment allow the household to
consume market goods. Other market
goods a household buys are not directly
used but are themselves inputs in the
production of a different set of goods
produced at home. These inputs are
durable consumer goods, such as
refrigerators and washing machines,
that are used at home. Finally, some of
the household’s earnings may be saved
toward future consumption.
Economists find it useful to
think of three broad uses of time. Of
the hours in a day not spent sleeping, the time may be spent engaged in
home production, market work, or at
leisure. An example of time spent in
home production — labor that does
not earn a wage or salary but contributes to the production of goods at
home — is time spent washing one’s
Business Review Q4 2004 21
automobile or cooking a meal. A clean
car and a cooked meal are both goods
most of us enjoy. An example of time
spent enjoying leisure may include
driving that car across the countryside
or eating that meal. Alas, there are
only so many hours in the day, and
some of those hours must be spent
sleeping or engaged in basic activities
such as bathing. It then follows that,
given an individual’s consumption of
leisure, time spent in home production
tends to reduce the amount of time
an individual can work in the market.
A household must then decide how
much time its adult members should
devote to market work and how much
to home work.
Members of married households have historically specialized
between work in the market and work
done at home. As such, they provided
an illustration of Adam Smith’s famous
theory of the division of labor. Once
married, the majority of women spent
all or most of their time in home
production. Men specialized in market
work. According to Adam Smith’s
theory, two individuals, each specialized in one task, are likely to be far
more productive as a team than when
each individual spends part of his or
her time on each task. This division
of labor is a large part of the reason
that most people do not sew their own
clothes and build their own houses.
This specialization may well hold true
for time a household divides between
market and home production.
Technology may have had
more subtle effects in leading households to specialize between market
and home work. Because technological
progress has enabled the production
of more and more commodities, goods
that once required substantial levels of
home production are now widely and
cheaply available in the marketplace.
A prominent example is prepared food.
Restaurant meals are far more com-
22 Q4 2004 Business Review
mon now than they were 50 or 100
years ago. Home delivery of meals was
almost unknown until the postwar
period. If, some time in the past, many
of the goods consumed by a typical
household involved a substantial level
of home production, the household
would have had to allocate far more
Members of
married households
have historically
specialized between
work in the market
and work done at
home.
time to home production. As more
substitutes prepared in the market became available, the household was able
to devote less time to home production and increase the time it devoted
to market work. If the male adult was
already working in the market, the
female adult could then increase her
hours of market work.
MACROECONOMIC THEORIES
OF THE RISE IN MARRIED
WOMEN’S MARKET HOURS
Several macroeconomic theories seek to explain the pronounced
increase in the hours worked in the
market by married women. Each of
these theories involves a completely
specified model of the economy in
which households’ and firms’ decisions
interact to determine total quantities,
such as total hours worked, and prices,
such as real wages. These macroeconomic models make predictions about
how hours worked by women will
change in response to a change in
some outside factor, such as a change
in preferences or technology.9
The Gender Wage Gap. In
the first macroeconomic theory we
examine, the outside factor is a discrimination tax that leads to a gender
wage gap. Researchers Ellen McGrattan, Rodolfo Manuelli, and Larry Jones
studied the extent to which changes
in married women’s market hours may
be explained by the gender wage gap.
Their analysis does not attempt to
explain the gap between men’s and
women’s wages. Rather, taking this gap
as given, they wanted to evaluate how
much of the difference in hours worked
in the market may be attributed to
the gap. While the ratio of female to
male wages has been less than one
over the entire postwar period, the
gap has been narrowing. In her paper,
economist Francine Blau reported that
in 1969, women who worked full time
earned about 56 percent of what men
earned. By 1994, they earned 72 percent of what men earned. Part of this
difference in pay may be attributable
to differences in occupation and skills.
The remainder is effectively a tax on
women’s hours of work, which we call a
discrimination tax.
Think of the problem faced
by a household that must allocate the
time of its two working-age adults to
home production, market work, and
leisure. If the woman’s market wages
are lower than the man’s, the woman
will spend more time at home. She
may spend some time in market work,
but it must be less than the time the
man spends if his wage is higher.
My survey of the macroeconomic literature
does not include a large body of microeconomic
empirical research into women’s labor supply
decisions. Such research has offered important
insight into why married women’s hours of work
may have changed that is complementary to
the macroeconomic theories we discuss here.
For example, Lawrence Katz and Claudia
Goldin discuss the role of birth control; Mark
Rosenzweig and Paul Schulz study the effects
of changes in fertility; and Lawrence Katz and
Kevin Murphy examine the changes in the difference between male and female wages.
9
www.PhiladelphiaFed.org
Lower market wages for women, at
least those with the same skills as men,
amount to a tax on women’s wages
relative to those of men. By devoting
less of the woman’s time to market
work, the household avoids the tax.
As the gap between the man’s
and the woman’s market wages narrows, the time spent by each in home
and market work should become more
nearly equal. This is exactly what we
have seen. Indeed, as the wage gap
associated with women’s market work
falls, the total amount of household
members’ time spent in the market
may rise.
An interesting aspect of the
gender wage gap theory is that it does
not require a very large gap to explain
the observed disparity between married men’s and married women’s hours.
This is because even a small difference in the wages paid to women leads
women to acquire less human capital,
that is, invest less in education.
In such economic models,
people invest in education because
firms pay higher wages for more educated workers. Since the gender wage
gap reduces women’s wages, women
find it less worthwhile to invest in
education. A gap in women’s wages
relative to those of men reduces the
returns to schooling for women relative to returns earned by men. Lower
investment in education by women
further reduces their potential market
earnings, above and beyond that implied by the gender wage gap. Women’s
lower investment in human capital
reinforces the extent of their specialization in the home.
The theory implies that as
the gender wage gap narrows, women’s
investment in human capital should
rise relative to that of men. There
is evidence for this prediction. For
example, in 1960, women, when
compared with men, were 60 percent
as likely to be college graduates. This
www.PhiladelphiaFed.org
fraction rose steadily until, in 2002,
women were 88 percent as likely to be
college graduates.10
Another interesting feature
of the gender wage gap argument is
that it can explain the large change in
hours worked by married women without incorrectly predicting — contrary
Home production, just like
production in the market, requires
capital. We all understand that firms
combine workers with capital, in the
form of both equipment and structures,
along with materials and energy in
order to produce output. Similarly, in
home production, consumer durable
Another interesting feature of the gender wage
gap argument is that it can explain the large
change in hours worked by married women
without incorrectly predicting concomitant
large movements in hours worked by single
women.
to the evidence — concomitant large
movements in hours worked by single
women. Single women, not having a
partner, cannot specialize in home production. As single women value both
home and market goods, and the latter
cannot be bought without earning
market wages, they will specialize far
less than married women, despite the
wage gap. The theory leaves unexplained the origin of the gender wage
gap and how it might affect incentives
to marry.
Technological Progress at
Home. Economists Jeremy Greenwood, Ananth Seshadri, and Mehmet
Yorukoglu (2003) argue that technological progress is largely responsible
for the rise in married women’s market
employment. They suggest that improvements in labor-saving equipment
used in the home has freed up women’s
time for market work. Thus, like the
gender wage gap theory discussed
above, the technological progress explanation also uses as its basic framework a model of home production.
The data are taken from Table 228 of the
Statistical Abstract of the United States: 2003.
10
goods serve as capital in the production of goods and services made at
home. Stoves, dishwashers, washing
machines, and refrigerators are all
examples of capital used in home
production.
Over the past 100 years, as
electricity has reached more and more
households in the United States, the
technology of home production has
undergone a dramatic change. Households have begun to invest in capital
goods — consumer durables — that
have increased labor productivity in
the home. Investment in household
appliances, as a percentage of gross
domestic product, has nearly doubled
over the past 100 years. As a series of
new household appliances has increased productivity in the home, the
amount of time a worker must devote
to home production, to produce any
desired level of goods and services, has
fallen sharply.
A refrigerator is one example
of a labor-saving consumer durable
that has become common in households. The availability of refrigeration
allows meals to be prepared far in
advance of when they are consumed.
Households with refrigerators are able
Business Review Q4 2004 23
to prepare several meals at one time,
thus reducing the labor required to
cook meals at home. In the 1920s,
almost no households in the United
States had refrigerators. Twenty years
later, about half of all households had
such equipment. By 1960, almost all
households had refrigeration. Richer
families bought refrigerators first; as
the real cost of the technology fell, additional families adopted it.
Electric washing machines
and irons have also sharply reduced
the amount of time people must work
in the home. Consider the following example, taken from Greenwood,
Seshadri, and Yorokoglu. In 1900, 98
percent of households used a scrub
board to wash their clothes. The
process of washing clothes required
water to be transported to the stove,
then heated by burning wood or coal,
which itself had to be brought into the
house. Next, the clothes were cleaned
and rinsed. Afterward, the water used
had to be removed and the clothes
had to be hung on a clothesline to
dry. Finally, the clean clothes had to
be ironed, using flatirons heated on
the stovetop. A study by the Rural
Electrification Authority in 1945 found
that washing and ironing 35 pounds of
clothes required 8.5 hours when done
by hand but only two hours and 16
minutes when done using a washing
machine and an electric iron.
Such examples confirm that
the introduction of household appliances, combined with electricity, central heating, and indoor plumbing, has
dramatically increased the productivity
of home work. Thus, Greenwood and
his co-authors argued that these capital-specific productivity improvements
have allowed the substitution of capital
for labor at home. As a result, the
amount of time required in the home
has fallen, and this has freed up time
for market work, especially women’s
time. (This argument assumes that,
24 Q4 2004 Business Review
historically, the majority of home work
was done by women.) Greenwood and
co-authors found that more productive
capital in the home leads to reduced
time spent working in the home. This
is in contrast to standard macroeconomic models in which more capital
would tend to increase employment
because capital accumulation raises
the value of labor.
Changing Social Norms. A
final explanation for the rise in hours
worked by married women shifts the
focus away from home production and
toward changing social norms. In their
paper, Raquel Fernandez, Alessandra
suggest that the change in married
men’s preferences has come about as
a result of being raised in households
in which their own mothers worked.
They showed that the sons of mothers
who are skilled and who work are more
likely to marry wives who are also
skilled and who work. They concluded
that a few mothers set an example that
led their sons to become more accepting of women working. As these sons
themselves married, their wives found
it easier to work in the market. Thus,
women, who could now work without
hurting their marriage possibilities,
undertook more education in order to
The introduction of household appliances,
combined with electricity, central heating, and
indoor plumbing, has dramatically increased
the productivity of home work.
Fogli, and Claudia Olivetti suggest that
changes in men’s attitudes about their
wives’ working have been important in
bringing about the rise in the fraction
of married women who work.
They discussed evidence
indicating that in the early part of the
century, men strongly disapproved of
married women working, a disapproval
that has lessened over time. In 1938, a
Gallup poll asked, “Do you approve or
disapprove of a married woman earning money in business or industry if she
has a husband capable of supporting
her?” Of the men surveyed, 81 percent
did not approve. However, by 1972, the
percentage of negative responses had
fallen to 38; 10 years later this percentage had decreased further to 25. By
1998, only 17 percent of men surveyed
gave negative responses. Clearly, married men at the end of the century
were more willing to have their wives
work than were men of 60 years before.
Fernandez and her co-authors
earn higher wages. Each generation of
households increased the acceptance
of married women working, and over
time, the fraction of working wives
increased.
The authors provide evidence
of the growing acceptability of marrying educated women, women who are
far more likely to work in the market.
Between 1890 and 1950, the likelihood
that a college-educated woman would
eventually marry, originally much
lower than that for men, had risen to
about the same as that for men. When
studying women born in 1890, Fernandez and co-authors found that 31
percent of those with a college degree
did not marry, while only 7.8 percent
of those without college educations
did not marry. In contrast, during this
same period, there was no comparable
difference in the marriage rate for
men; they had a 10 percent chance of
not marrying, whether or not they had
a college degree.
www.PhiladelphiaFed.org
If we look at women born 60
years later in 1950, who were attending college in 1970, the percentage of
those with college degrees who did not
marry fell to 7.9 percent. For comparison, 5.5 percent of those without
college degrees did not marry. The
probabilities for men were similar.
Fernandez, Fogli, and Olivetti
developed a model in which the sons
of educated mothers, who are more
likely to work in the market, are less
unhappy about marrying educated
women. In other words, there is a
direct transmission of preferences from
mother to son, and sons of educated
mothers find educated women less
unsuitable as partners.
These co-authors’ model predicts that, over time, more and more
women will choose to obtain an education, marry, and work in the market.
Their model matches several facts.
First, both women’s market work and
educational level have risen. Second,
men’s attitudes toward working women
have improved over time as more and
more of them are themselves the sons
of working mothers. Finally, the marriage rate of working women has risen
relative to that of women who do not
work in the market.
CONCLUSION
Over the postwar period
there has been a large change in the
composition of the labor force that is
hidden when one examines the overall
change in total hours worked per
person. The labor force participation
of married women has risen sharply,
while the hours worked by others has
fallen a little. Economists are using
the home production model to better
understand the determinants of these
changes. Other explanations center on
changing social norms.
The working behavior of the
old and the very young has undergone
significant changes over this period.
Economic theory now must integrate
the changes in working behavior across
all ages and marital status. BR
REFERENCES
Blau, Francine D. “Trends in the Well-Being of American Women, 1970-85,” Journal
of Economic Literature, 36, March 1998, pp.
112-65.
Fernandez, Raquel, Alessandra Fogli, and
Claudia Olivetti. “Marrying Your Mom:
Preference Transmission and Women’s
Labor and Education Choices,” NBER
Working Paper 9234 (September 2002).
Greenwood, Jeremy, Ananth Seshadri and
Mehmet Yorukoglu. “Engines of Liberation,” unpublished manuscript, 2002.
Jones, Larry E., Rodolfo E. Manuelli,
and Ellen R. McGrattan. “Why Are
Married Women Working So Much?”
Federal Reserve Bank of Minneapolis
Staff Report 317 (2003) (available at:
http://research.mpls.frb.fed.us/research/sr/
sr317.html).
www.PhiladelphiaFed.org
Katz , Lawrence F., and Claudia Goldin.
“The Power of the Pill: Oral Contraceptives and Women’s Career and Marriage
Decisions,” Journal of Political Economy,
100, 2002, pp. 730–70.
Katz, Lawrence F., and Kevin M. Murphy.
“Changes in Relative Wages, 1963-1987:
Supply and Demand Factors,” Quarterly
Journal of Economics, 57, 1992, pp. 35-78.
McGrattan, Ellen R., and Richard
Rogerson. “Changes in Hours
Worked Since 1950,” Federal Reserve
Bank of Minneapolis Quarterly Review
Review,
22, 1, Winter 1998, pp. 2-19 (available at:
http://research.mpls.frb.fed.us/research/qr/
qr2211.html).
McGrattan, Ellen R., and Richard
Rogerson. “Changes in Hours Worked,
1950-2000,” Federal Reserve Bank of Minneapolis Quarterly Review
Review, 28,1 July 2004
(available at http://minneapolisfed.org/
research/qr/qr2812.html).
Rosenzweig, Mark, and Paul Schulz.
“The Demand for and Supply of Births,”
American Economic Review
Review, 75, 1985, pp.
992-1015.
U.S. Census Bureau, Statistical Abstract
of the United States: 2003 (123rd Edition)
Washington, DC, 2003.
Business Review Q4 2004 25