One of the most common explanations for the gender-wage gap is that women simply don’t ask for higher salaries, while men do. The idea that one should “always negotiate your salary” is standard college-counselor fare. In a blog post about the gender wage gap, the U.S. Department of Labor suggests women “aim higher and negotiate better” as one of the possible remedies.

And at first, it seems innocuous, right? Negotiating is one of the few ways workers in today’s economy can secure better salaries and benefits. Women, the thinking goes, should do it or risk being ground up by the merciless gears of capitalism. Or, you know, make at most 93 cents to the dollar.

Women do tend to negotiate less than men do, and some researchers suggest that’s because they justifiably fear they’ll violate societal norms of demure, communal female behavior—and be punished for it. (As one study depressingly found, “Perceptions of niceness and demandingness explained resistance to female negotiators.”) So how should women proceed? In a new paper in NBER, three economics and management researchers find that advising women to “always negotiate” might not be in their best interest—because, it seems, women seem to already know when negotiations won’t work out in their favor.

For the study, the researchers matched people in pairs, with one representing the employer and one the worker. At the start, the workers contributed either $20, $15, or $10 to a pot of money (the amount was determined by their performance on a math problem), and the firms contributed either $25 or $20. The researchers randomly assigned a “default wage” for the worker—either $4 less than they contributed, $2 less, exactly what they contributed, or $2 more.

The worker-employer teams were entered into one of two types of negotiations, all performed over instant message without revealing age, name, or gender. In the first condition, the worker could choose whether to negotiate their salary, and in the second, they had to negotiate, even if they didn’t particularly want to. Either way, they knew both parties would be penalized $5 if they failed to reach an agreement.

Both genders were more likely to choose to negotiate when their “suggested wage” was $4 less than their contribution—in other words, when they stood the most to lose from not negotiating. When they were forced to negotiate, men and women didn’t differ in their success, with both male and female workers netting a gain of about $0.66.

When given the choice, though, the female participants didn’t opt to negotiate as often as the men did. That might at first seem to support the idea that “women don’t ask” for more money or better benefits, and that’s why they don’t receive them.

But the data from the “choice” condition revealed what was really going on: Women were avoiding the negotiations they knew would not end well for them. The likelihood of women losing money tripled if they were forced to negotiate, rather than given the option. “By opting out of negotiations, women are avoiding substantial financial losses,” the study authors write. “That is, women know when to ask.”

Unlike women, men were not particularly likely to opt out of negotiations that they would probably lose. Thus, being forced to negotiate was neither bad nor good for the men, but it was bad for women, says Christine Exley, an assistant professor at Harvard Business School who co-authored the study along with Muriel Niederle of Stanford University and Lise Vesterlund at the University of Pittsburgh.

We don’t know from the study why this happened. Did the men like negotiating a little too much? Were the women more psychologically attuned to the potential costs and benefits? Since the “firm” participants couldn’t discern the gender of the negotiator, the typical explanations for gender disparities, like the backlash effect or the double bind, are ruled out here. Catherine Eckel, a behavioral economist at Texas A&M University who was not involved in the study, said the study suggests one reason for women’s success in negotiating in the real world could be that the ones who negotiate are the ones who are already good at it. Those who aren’t, don’t, and if they were forced to, they wouldn’t necessarily nab the same career gains.

Still, women who practically never negotiate should probably not take this study as evidence they shouldn’t bother trying. In this study, the overwhelming majority of women negotiated when they were at risk of losing money. What’s more, the negotiation experiment was more transparent than a real-life salary discussion. As the study authors explain, they made clear what the worker risked by not negotiating, but “such a benchmark may not be available in many negotiation settings; women may shy away from financially beneficial negotiations if they undervalue what they bring to the table."

Instead, Exley said the results could be a word of caution about the universal call for women to always be more aggressive in their negotiations. That school of thought suggests “all of you women who aren’t negotiating, you should negotiate, too, because we are all doing so well [in negotiations],” she says. But “the women who are choosing not to negotiate would have lost had they entered the negotiation.”

Linda Babcock, author of the popular book Women Don’t Ask, which details the “high cost of avoiding negotiation,” praised Exley’s findings as “solid.” She said the study points to the importance of negotiation training. “Negotiation skill is not something you’re born with, but can be taught” she said.

Eckel, at Texas A&M, meanwhile, said it’s also worth examining other reasons for the gender wage gap, rather than blaming women for not driving a hard enough bargain. “If there are gender differences, why does that mean we have to fix the women?” she said. “Maybe we can change the situation, the rules of the game?”