Picture two women who both want to lose weight and both ask their doctors about the new generation of weight-loss drugs. One gets a prescription and starts taking it. The other, blocked by cost or insurance, stays on the waiting list. A year and a half later, what looks different about their lives, beyond the number on the scale?
That question sits at the center of a working paper by Rebecca Diamond, an economist at Harvard University. She uses the rapid spread of GLP-1 medications, the class of drugs that includes semaglutide and tirzepatide, to ask where exactly body weight matters in a woman’s social and economic life. Her analysis offers evidence that large weight loss changes outcomes mainly at the moments when someone new forms a first impression: a prospective romantic partner, or an employer weighing an application.
A long-standing puzzle about weight and opportunity
Economists have documented for decades that heavier women, on average, earn less, work less, and are less likely to marry or live with a partner, and that these gaps fall more heavily on women than on men. The hard part has always been interpretation. Body weight travels alongside many other things, including health, family background, education, and local job opportunities. So a simple comparison of heavier and lighter women cannot tell you whether weight itself is driving the gaps, or whether weight is just a marker for something else.
Diamond points to two broad explanations that researchers have struggled to separate. Weight might affect a person’s own health, energy, and productivity. Or it might change how other people treat her. Field experiments have shown that otherwise identical job applications get fewer callbacks when the applicant is presented as obese, which makes discrimination at first contact plausible. The open question, as Diamond frames it, is not only whether weight matters but where it matters: inside jobs and relationships people already have, or at the point where new ones form.
GLP-1 drugs create a setting to look at this. Clinical trials show sustained use produces weight loss of roughly 12 to 18 percent, and the drugs reached a broad, non-surgical population quickly. In the data Diamond uses, by the end of 2024, ten percent of adults had started a GLP-1, and six percent of women had started one specifically to lose weight.
How the study was built
Diamond draws on the Understanding America Study, a nationally representative internet panel of about 15,000 U.S. adults run by the University of Southern California. A special module fielded in late 2024 and early 2025 asked respondents about their GLP-1 use, including when they started and why. The panel also tracks employment, work hours, household income, and marital or cohabitation status every quarter, along with periodic questions about health, mood, and loneliness.
Her approach compares women who started a GLP-1 to lose weight with women who said they would like to start one but had not yet done so. Restricting the comparison to women who have already expressed demand for medical weight loss is meant to hold fixed many of the reasons some people seek out treatment. Within that pool, she matched women on pre-treatment body mass, health, income, employment, partnership, and well-being. She also set aside people who started a GLP-1 to manage diabetes, since those starts tend to be triggered by illness rather than a desire to lose weight. She focused on women because far more women than men in the data took the drugs for weight loss. The main analysis rests on 242 treated women matched to 850 comparison women.
To estimate the effects, Diamond used a method called a stacked matched difference-in-differences, which compares how outcomes change over time for the women who started the drug against the matched women who did not. She also reweighted the comparison group so its measured characteristics lined up tightly with the treated group. Because this is not a randomized experiment, the results are best read as associations that follow GLP-1 initiation, though Diamond runs several checks to probe how much hidden differences in trajectory could explain the patterns.
What changed, and what did not
Weight fell, as expected, by between roughly 1.4 and 4.2 BMI points depending on the group and time horizon. The more revealing patterns showed up in the social and labor-market outcomes.
Among women who were single when they started, the probability of being married or living with a partner rose 18 percentage points overall and 29 percentage points after six or more quarters. The gain built up gradually as weight came off. Diamond reads that slow build as a sign that the effect is unlikely to be reverse causation, since a woman who started a drug because she had just begun a relationship would show a jump near the start, not a steady climb over three years.
Women who were not employed at the start also moved into work. Their employment rate rose 13 percentage points overall and 27 percentage points after six or more quarters, with weekly hours climbing by nearly ten hours at the longer horizon. Much of that gain came from women leaving unemployment.
What stayed flat is part of the story too. Among women already partnered when they started, relationships did not dissolve at higher rates. Among women already employed, there was no climb in hours, income per household member, or moves to new jobs. If anything, they changed employers slightly less often, a pattern concentrated among women whose insurance covered the drug, which Diamond suggests may reflect a desire to keep coverage that pays for the medication.
Subjective well-being did not improve in step with the weight loss. Self-rated health did not get better, and measures of life satisfaction, loneliness, and depression were largely flat or moved in the wrong direction. Diamond argues this pattern is hard to square with a general mood-and-motivation story, in which feeling better explains the new relationships and jobs. Reported work-limiting health problems did fall for previously non-employed women at longer horizons, which she says may account for part of their move into work.
The author’s interpretation
Diamond reads the overall pattern as evidence that an important part of the female obesity penalty operates as what she calls a “first-impression discount.” The outcomes that responded are the ones where another person forms a fresh assessment of a woman’s body. The arrangements that did not respond are the ones already in place, where weight is one piece of a much richer set of information accumulated over time.
A few caveats sit alongside the findings. The men in the sample were too few for firm conclusions, but the limited estimates pointed in a different direction: partnered men appeared more likely to leave their partners after starting the drug. Early adopters in the data were disproportionately well off, with higher incomes, and 40 percent paid the full cost out of pocket at about $299 a month. Diamond notes that if medical weight loss changes how women are treated, then unequal access may shape who can sidestep the penalties attached to body size. She also calls it “unsettling to think about using medical treatment as a possible remedy for discrimination.”
The window is short, and earnings among already-employed women did not rise over the period observed. Whether these effects last, and whether earnings gains eventually appear, remains to be seen.
