Phil Davies - Staff Writer
Published December 1, 2003 | December 2003 issue
Harriet Nelson of the TV show "The Adventures of Ozzie and Harriet" epitomized what it meant to be a married woman in America in the early 1950s. Harriet was a typical homemaker (OK, her house was a lot neater than your average suburban manse) who took care of sons David and Ricky while Ozzie was at work. By 1980 Harriet had morphed into Doralee Rhodes, the feisty office worker played by Dolly Parton in the movie "Nine to Five." The Harriet-to-Doralee transition represented a sea change in the American workplace, the entry of millions of married women into factories, retail stores and construction zones alongside single women. Today, married women work in every conceivable occupation from actuary to zoologist. It's their lower-paid, less-educated husbands who often elect to stay at home with the kids, or work part time.
This explosion in the participation of married women in the U.S. labor force over the past half century is well documented in U.S. Census and Department of Labor statistics: Between 1950 and 1990, hours worked by married women in this country almost tripled, from about eight hours per week to nearly 23 hours, while hours worked by single women, single men and married men have stayed relatively constant. The average single woman in 1990 worked about 29 hours—the same number her mother or grandmother worked during the Truman administration. Similarly, males have just slightly changed their workloads, dropping from 41 to 38 hours for married men, and from 32 to 30 hours for single men (see table).
Over the same period, the gender wage gap narrowed considerably. Francine Blau, director of the Institute for Labor Market Policies at Cornell University, has documented that between 1969 and 1994, the ratio of the hourly wage earned by fully employed women to that of men rose from 56 cents on the dollar to 72 cents.
But economists have been at a loss to establish a causal relationship between the surge in work by married women and the diminished gender wage gap, to explain why wives migrated en masse to the workplace while their single sisters marched in place. Social historians and economists have run a number of theories up the flagpole over the past 40 years, none of them entirely satisfactory in light of actual, observed behavior by men and women in the labor marketplace.
Lately, though, economists have gained fresh insights into why Harriet stayed at home, while Doralee struck out into the workforce. In "Why Are Married Women Working So Much?" Fed adviser Larry Jones of the University of Minnesota, Rodolfo Manuelli of the University of Wisconsin and Minneapolis Fed senior economist Ellen McGrattan find that a modest narrowing of the gender wage gap can fully account for the tripling of married women's hours as well as the historic pattern of hours worked by men and single women. A key ingredient in their hypothesis is increased investment by women in education and work experience as the gender wage gap shrinks.
It's a cogent explanation reached by examining the issue in the context of labor decisions made by each partner in a married couple. By taking a kitchen-table view of the merits of wives joining the workforce at different points in history, the economists' model disentangles the truth from a thicket of interrelated and sometimes countervailing influences on labor supply.
In the early 1960s pioneering labor economist Jacob Mincer identified rising overall wages as the main factor behind the surge of women into the labor force in the 20th century. But neither he nor later researchers using more sophisticated statistical analysis to test wage-increase theories could fully account for the quantum jump in hours worked by married women.
In her 1990 book Understanding the Gender Gap, economist Claudia Goldin of Harvard University describes widespread discriminatory practices that discouraged women from working full time. In the years before and immediately after World War II women were often relegated to low-paid, dead-end clerical and retail jobs, and passed over for promotion in favor of men. Wives encountered additional discrimination in the form of "marriage bars"—policies against hiring married women or retaining them after they married. Goldin theorized that the relaxation of marriage bars in the 1950s, when employers became desperate for labor in a booming postwar economy, led married women to increase their participation in the labor force at a faster rate than single women.
Other studies, most of them based on painstaking parsing of labor statistics, attribute the dramatic increase in married women's market hours to later childbearing, the rise of the service sector, changes in societal attitudes or higher female enrollment in vocational schools and universities. But these theories fail to come to grips with the root economic causes of increased workforce participation by married women.
Over the last five or six years economists with a more theoretical bent have taken a crack at the working-wife question, leveraging the power of the microprocessor to assemble and test dynamic equilibrium models—complex mathematical representations—of the economic forces at work in labor decisions. Computer models allow economists to tease out the subtly intertwined drivers behind complex issues that can remain hidden to the unaided eye sifting through raw data. Of course, any computer model is only as good as the assumptions built into it; different assumptions lead to different conclusions, making economic consensus elusive.
In a 2001 study, Claudia Olivetti of Boston University concludes that rising women's wages alone can't explain all that extra time spent working outside the home. Returns to experience—the acquisition and honing of skills on the job—plays a major role, setting up a positive feedback loop in which enhanced job skills result in higher wages, which encourages married women to work even more hours.
No, forget about work experience; married women have Thomas Edison to thank for their careers, argue Jeremy Greenwood, Ananth Seshadri and Mehmet Yorukoglu in a 2002 study, published by the University of Rochester, on the impact of technological progress in the household. The electrification of homes in the 1920s and '30s led to the introduction of a host of labor-saving appliances—electric irons, refrigerators, washing machines, vacuum cleaners—that mitigated the drudgery of housework. In their model, progressively more efficient and less expensive "home durables" freed up time that married women chose to spend in gainful employment.
Both of these explanations seem plausible, but there are flaws in each—predictions at odds with the way the real macroeconomy works. For example, Olivetti's equations significantly boost the working hours of single as well as married women, when in reality, single women haven't increased their working hours. And in Greenwood et al.'s technology, model households spend less money over time on refrigerators, lawn mowers, kitchen remodeling and other home investment goods—a finding at odds with government statistics and Home Depot's business model.
Troubled by such inconsistencies, economists strive to draw closer to the truth by building a more comprehensive model. The crucial distinction in the Jones-Manuelli-McGrattan model—the feature that allows it to accurately mirror how single as well as married women actually responded to greater market opportunity after 1950—is its treatment of a married woman as a partner in marriage, making rational choices about how much time to spend in the market and how much to spend at home, raising children and carrying out household tasks. Thus, their model allows a husband and wife to specialize—a luxury not enjoyed by single people, who must work to support themselves and keep the apartment clean.
In creating this microcosm of a working household, Jones, Manuelli and McGrattan adapted earlier models they had used to test theories about savings behavior and the impact of taxation on labor markets. The model has two key features. First, like McGrattan's previous analyses of the relationships between employment and work in the home, the model levies a tax on market work, in this case not just a government tax but also a wage penalty paid only by women—the gender wage gap. And, building upon the work of Jones and Manuelli, Olivetti and others, their simulation also includes investment in human capital—education and skills learned on the job—as a catalyst for dramatic increases in hours worked by married women.
"We don't want to deviate far from the standard theory that does well in certain predictions about business cycles and savings decisions," McGrattan said in an interview. "We want to stay as close to that as we can and deviate only so as to address the specific question we were after, which is why married women and single women look so different in their behavior."
Model in hand, the researchers set out to test the gender wage-gap theory and two other popular explanations for the post-World War II surge in distaff employment: technological advances in the household, and "inferiority" of home-produced goods relative to those produced in the market.
The notion that married women owe their emancipation to smarter home technology seems entirely reasonable when you consider how far domestic amenities have come in 50 years. Harriet's automatic washing machine was vastly more efficient than the wringer tub her mother used, and Doralee had at her disposal labor-saving devices unheard of in the 1950s—microwaves, dishwashers, freezers for ready-to-eat pancakes and pizza. "The impact of the household revolution was no less than the industrial revolution ... technological progress in the household sector played a major role in liberating women from the home," write Greenwood, Seshadri and Yorukoglu in their 2002 paper "Engines of Liberation."
New and improved home goods, specifically household appliances, allowed wives (who do all the housework in their simulation) to cook, clean and care for children more efficiently, freeing up hours for market work. Noting a 77 percent decline in the inflation-adjusted price of common household appliances between 1950 and 1990, the "Engines" researchers argue that widespread adoption of labor-saving devices helped to increase labor-force participation by women, especially married women. In this model, the growth curve in female labor participation—from about 22 percent in 1950 to 55 percent in 1990—roughly follows the trend seen in labor statistics.
In their analysis, Jones, Manuelli and McGrattan conclude that the "household revolution" may have made quotidian tasks easier, but it wasn't the primary or even a major driver in the mass movement of married women into the workforce after 1950. Even assuming that real prices of cars, houses, furniture, clothing and other household goods have dropped at the same rate as appliances (they haven't), their model fails to generate significant increases in married women's hours from technological advances in the home. "What we find in the model is that improved home technology increases leisure, decreases time in the home, increases time in the market, but only by a very small amount," Jones said.
To reach that conclusion, Jones, Manuelli and McGrattan employ a classic distinction in economics: Are products made at home and those bought in the market substitutes or complements? To economists, all goods—and by extension, the time and labor needed to produce them—fall somewhere on a continuum between the two. Substitutes (like Pepsi and Coke) each provide the same basic satisfaction of wants and needs, so one can readily replace the other in the marketplace. If Pepsi becomes cheaper, demand for Coke drops while Pepsi sales surge. In contrast, complements (like hamburgers and french fries) naturally go together in the market; lowering the price of burgers increases demand for burgers and fries.
The researchers tested the "Engines of Liberation" hypothesis both ways. When housework and paid labor were designated substitutes in their model, smarter home technology actually reduced the hours worked by married women between 1950 and 1990. That makes sense; if working at home satisfies the same family needs as working in an office, and better technology allows a wife to perform those home duties more efficiently, she'll put in more hours at home. But in actuality the opposite occurred: Wives put in fewer hours at home. So if home and market labor are substitutes, as economists generally believe, the "Engines" theory fails miserably.
To give this argument a better chance of success, the model was reconfigured to treat housework and market work as complements: Lowering the price of one would increase demand for both. In this version, a technology-induced boost in home productivity ("See, dear—now I can wash three loads of laundry in less time than it took to do one!") should result in part of the time saved being reallocated to market work.
This version of the "Engines" hypothesis seems to fare better in that it reproduces the pattern of working hours seen in the data in the 1990s (see second row of table). But to do so, the economists had to simulate in the model a leap in home productivity five times greater than the rise that actually occurred in the workplace between 1950 and 1990. And even that artificial level of hyperproductivity didn't narrow the gender wage gap nearly enough. Moreover, even though this model ably generates 1990s work patterns, it does a poor job of generating 1950s patterns.
In debunking the household revolution as an impetus for increased working hours by married women, Jones, Manuelli and McGrattan raise an obvious question: Where did all the time saved by dishwashers, microwaves, floor polishers and the like go? Their model suggests that some of the extra hours were reallocated to working outside the home, but married women also used them to achieve higher standards of cleanliness and style in the spirit of Martha Stewart. That interpretation is shared by economic historians such as Joel Mokyr of Northwestern University, who infers from time-use data studies in the early 1900s (when advances in home technology were most dramatic) that married women employed labor-saving devices to do more housework in roughly the same amount of time—washing clothes more often, or waxing as well as vacuuming the floor.
Another oft-proposed explanation for the increase in work time by married women is that as households grow richer they tend to spend a smaller fraction of their rising income on home activities such as preparing meals, washing clothes and personally caring for children. In economic terms, those activities are "inferior" goods-less desirable as incomes rise. Instead, people eat out, pay a laundry service or engage a nanny. That frees up time that married women can then spend on the job, earning money to pay for those market goods and services. (Homemade macaroni and cheese is a good example of an inferior good. As incomes rise, most people spend a smaller percentage of their money on it; instead meals at restaurants—perhaps penne con parmigiano—take a bigger piece of the budget pie.)
The "Engines" theory assumes that home activities are indeed inferior; in that model married women invariably spend their extra, technology-given hours on either leisure or market work. Jones, Manuelli and McGrattan put this notion to the test, programming into their model an escalating degree of inferiority to gauge the impact on married women's market hours and other market variables.
The results showed that the "No, not meatloaf again!" theory doesn't fly, at least not as the explanation for an upswing in market hours by married women over the past 40 years (see third row of table). It could only reproduce the pattern of hours worked by single women as well as married women if people became fed up with housework, child care and other home tasks very quickly. Such a strong bias against work in the home probably doesn't reflect reality. "We consider that hypothesis a failure because we had to go to such an extreme to get the story to work at all," Jones said.
This scenario also implies that the proportion of household income spent on home durables in the United States falls over time. In fact, spending on home investment goods has remained fairly steady since the 1930s, at about 12 percent of GDP.
The "Engines of Liberation" and inferior goods theories thus fail to account satisfactorily for the storming of the workplace by married women. Jones, Manuelli and McGrattan next turn to the narrowing gender wage gap. Can it explain the increase in labor force participation by married women?
The economists allow that they don't know exactly why the gender wage gap shrank. It has been argued that ebbing discrimination against women has closed the gap over the past 50 years. Changing mores, passage of equal-pay legislation such as Title VII of the Civil Rights Act of 1964 and the feminist movement of the 1970s greatly diminished marriage bars, career tracking, glass ceilings and other employment practices that depressed women's earnings.
But the narrowing of the gender gap can also be attributed to increased productivity of women in a technologically advanced economy less dependent on male brawn. Rosie the Riveter, the icon of women's labor during World War II, flexes an impressive bicep in the well-known poster; she had to be strong to work in the munitions plants and shipyards of the 1940s. But by the 1970s computer-controlled machines had rendered muscle power on the factory floor largely superfluous. And in an increasingly service-oriented economy, most people worked in offices or retail establishments, hefting nothing weightier than a telephone.
Whatever the cause, there's no question that women's wages relative to men's progressively rose, especially after the mid-1970s. So the researchers' task is to determine whether this diminished gap can explain the tripling in married women's working hours, without dramatically affecting the number of hours supplied by other workers.
Based on previous work by McGrattan on the role of home production in business cycles, the Jones-Manuelli-McGrattan model assumes that working in the home and outside it are highly substitutable, or elastic, endeavors. That is, individuals move easily from hearth to workplace and back again, given the right market incentives. Their model also assumes that married couples strike the best economic bargain in deciding whether the wife takes a full-time job, works part time or stays at home to care for young children.
The benchmark, or base model, mimics a shrinking gender wage gap by lowering the employment discrimination "tax" borne by women over time. Diminishing the discrimination tax elevates women's wages, increasing the income married women can contribute to the household and reducing their incentive to avoid the tax by staying at home, or working only part time. Harriet, contemplating earning just over half what Ozzie made doing whatever he did (in the TV show, he just hung around the house), remained content with home life. Forty years later, Doralee, given the opportunity to earn a higher relative wage, chose to work—even if it meant putting up with an obnoxious boss. In contrast, single women can't dodge the discrimination tax; like single men, they must work full time to put bread on the table. So while reducing the gender wage gap raises their standard of living, it doesn't appreciably increase their market hours.
This basic model neatly simulates the 40-year trend of market hours in this country: the big jump by married women, a small drop by married men of about three hours per week, a smaller drop for single men and virtually unchanged hours for single women (see fourth row of table).
But the economists find that these numbers pan out only when investment in human capital is part of the mix. In the model, the narrowing gender gap triggers a 172 percent increase in human-capital investment by married women between 1950 and 1990. (Single women increase their investment by only 36 percent.) Higher relative wages encourage married women to increase their marketability through formal education and on-the-job training, which boosts productivity and raises the price of staying at home, out of the job market.
McGrattan herself is an example of a highly educated wife and mother who can't afford not to work. She holds a Ph.D. in economics, a field that demands continuous employment in order to keep up with new developments in research and teaching. "My skills are not actually very good for home production," she said, laughing. When McGrattan and her husband (an aerospace scientist at the University of Minnesota) had children, they hired a nanny so that they both could keep working full time.
In a version of the model that ignores the role of human capital, a much greater drop in wage discrimination is required to generate the observed pattern of hours worked by the various groups. Thus increased investment in human capital amplifies the effect of small reductions in the wage gap. The more time married women spend in college or on the job, adding to their portfolio of work skills, the more money they earn and the more hours they choose to work. Over 40 years, married women invest more and more time in the accumulation of human capital and market hours.
Where do men, beneficiaries of workplace discrimination against women for decades, fit into the model? Their working hours remain generally static, because a relative increase in women's wages gives them no incentive to work longer hours. For the same reason, they don't invest as much time and money in higher education and other forms of human capital. It's no coincidence that, nationally, 129 women earn college degrees for every 100 men. Instead, men invest in physical capital—cars, real estate, mutual funds—which women tap into in the form of bank and college loans to finance their educations.
Martin Baily, a senior fellow at the Institute for International Economics in Washington, D.C., and a former economic adviser in the Clinton administration, has argued that an influx into the workforce of women with weaker skills than men slowed the pace of productivity growth in the 1970s and early 1980s. The work of Jones, Manuelli and McGrattan bears out that theory; in some versions of their model, overall productivity growth slackens as women increase their participation in the workforce. But, as Jones points out, perhaps bringing down the neighborhood slightly is the price of improving the economic lot of large segments of the population—women, and married men who benefit from additional income brought home by their working wives.
"If the reason we see the productivity slowdown in the data is because of a reduction in discrimination, most people would say, 'maybe that's not such a bad thing,'" Jones said. In any case, the model forecasts a rebound in overall productivity growth as women upgrade their skills through education and returns to experience. That trend was well under way by the early 1990s, according to labor statistics.
The model also predicts that the upward trend of married women's working hours will level off as the gender wage gap continues to close. That's already happened, based on preliminary data from the 2000 census showing that American married women aged 35 to 44 worked slightly fewer hours per week than they did in 1990. This implies that the gender wage gap continues to shrink and may have disappeared completely (although there are no recent data to support that conclusion). If so, women play on a more or less level field in the workplace for the first time in U.S. history. They can work outside the home as much or as little as they want and receive pay commensurate with their labor—a goal of the Civil Rights Act that has taken almost four decades to achieve.
This explanation of why Harriet stayed at home and Doralee commuted to the office isn't perfect; some predictions of the Jones-Manuelli-McGrattan model don't reflect reality. For instance, the model generates a larger drop in single men's hours than is seen in the data from the 1950s through the 1990s.
To better simulate the behavior of single households, the researchers are developing a more complex model that addresses the impact of marital uncertainty on the process of accumulating human capital. The current model assumes that marital status remains unchanged throughout life, ignoring the possibility that a single woman could marry, or a wife divorce (much more likely today than in 1950). "Every time we present this paper people say, 'the threat of divorce is a huge thing ... that's got to be quantitatively important,'" McGrattan said. "The question is, is it?"
In the new model, uncertainty about the future (will I marry? will I stay married?) affects how much women invest in education and work experience. That decision influences their productivity and, in turn, their working hours. If this random, or stochastic, model generates a smaller investment in human capital by women than the current model, single women will end up working fewer hours. And single men may continue to hold their own in a labor market that no longer, as in Harriet Nelson's day, automatically favors men. In less than two generations, the world of work has undergone a seismic shift, and the earth is still moving.