Illustration: Sébastien Thibault

As economics professor Claudia Olivetti wrote in her Ph.D. thesis at the University of Pennsylvania in 2001, “The pattern of married women’s work hours over the life cycle has changed substantially.” (She was drawing comparisons by gender using figures from the 1970s and 1990s.) Since then, her research has focused repeatedly on women and work: how U.S. improvements in maternal health between 1930 and 1960 contributed to women’s expanding participation in labor markets (Journal of Political Economy, 2016); how the growth in service industries was also key (Annual Review of Economics, 2016); and how a range of behaviors—from changing jobs to getting married—help to explain the increasing differences between women’s and men’s earnings over the life course (National Bureau of Economic Research Working Papers, NBER, 2017).

In 1980, according to the U.S. Department of Labor, women earned, on average, 64 percent of what men did. The ratio has since improved for women—who realized almost 80 percent of men’s wages in 2016. A recent study by Olivetti and colleagues at Harvard University, Wellesley College, and the University of Oslo sought explanations for the shortfall that survives.

Working with information from the 2000 U.S. Census long form (distributed to one in six households) and from the Census Bureau’s Longitudinal Employer–Household Dynamics program (LEHD), they analyzed data that profiled some 1.18 million individuals in large metropolitan areas of 23 states, from 1995 to 2008. Combining the two data sets—the LEHD provided details submitted by private employers—allowed the team, for the first time, to match individuals with their places of employment and to study the earnings differentials among them by gender, age, education, and other factors within shared work environments, rather than in an agglomerated labor force. This enabled the researchers to ask more granular questions. For instance, they could see that men switched to higher-paying employers (and to higher-paying jobs with the same employer) more often than women did. To what extent, if any, was the gender-wage gap tied to that fact? And what might the explanation be?

They reported their results through the NBER website and in an article titled “The Expanding Gender Earnings Gap,” published in the American Economic Review in May 2017. Among 25-year-old college graduates, they found, women lagged men in earnings by only 10 percent. But by the time they turned 39, the gap had widened, to 43 percent. A “considerable part” of the widening happened during long-term job stays, and the authors posit “non-linear” (i.e., unequal) payment practices as a probable cause. But nearly 30 percent of the gap owed to the greater frequency with which men changed jobs for better work and pay. And most of the gap—about 80 percent of it—developed between ages 26 and 32, “the time when people start planning for or are having children,” says Olivetti.

Did marital status contribute to compensation inequality? The researchers found the “earnings path” for married and nonmarried men to be “very similar.” And, although they reported an earnings gap of 15.6 percent between single women and single men in their early thirties, the gap did not grow much beyond that. “The widening of the wage gap,” they write, is “almost entirely . . . driven” by a lower rate of growth “among married women.” What’s more, this widening is “particularly strong” when both members of a couple are college graduates and there are children in the home.

The researchers studied non-college graduates, as well, and report some differences. A smaller gap developed between female and male high school graduates during the same years (from about 15 percent to 27.2 percent). Among women and men lacking a high school degree, the gap grew least (4.3 percentage points), but the authors note that for this group the gap at the outset was greater (23.4 percent).

Olivetti, who is also a research associate at the NBER, says that her next step will be to construct a “large and extraordinary” data set with funding from the National Science Foundation. It will “follow couples across time,” she says, tracking their wages “through their marriage’s trajectory, as they have children, say, or encounter a health shock such as an elderly parent who needs care. We want to continue exploring the interaction between what it takes to have a career and what it takes to have a family.”

Erick Trickey is a Boston-based writer.