SIX

What Do Women Want?

The great question that has never been answered, and which I have not yet been able to answer, despite my thirty years of research into the feminine soul, is “What does a woman want?”

—Sigmund Freud, letter to Marie Bonaparte, 1926

Different Preferences

IF WOMEN ARE NO LESS TALENTED than men, why don't they receive similar rewards? Here are three possible explanations. The first appeals to the idea that women might simply have, on average, different preferences from men: even if they could do any job just as well as men, they might not necessarily want to, and the jobs they want to do might be less well rewarded whoever did them. Within any occupation they might also avoid some of the positions that happen to be most highly rewarded (for instance, they might be more averse to risk or to aggressive competition). There is some evidence in support of this view, though it's far from conclusive.¹ The other two explanations both appeal to the idea that women get a smaller share than men of the value their work creates. It might be that, for a variety of reasons, women negotiate less aggressively with employers and so end up with a smaller share than men of the benefits that their work creates for their employers. There is evidence in favor of this possibility as well, though it is sketchy and its interpretation is controversial. Alternatively, employers might negotiate more aggressively with women, either from explicit prejudice or because of norms and customs, explicit or implicit, that are a hangover from earlier times. There's a large amount of anecdotal evidence in favor of this third possibility, but again, it's not easy to judge how scientifically well-founded it is or how much of the discrepancy in economic rewards it can explain. This third mechanism may interact with the first: women may have somewhat different preferences from men, but aggressive negotiation by employers may lead to women's paying an unreasonably high price for those different preferences.²

The idea that, on average, men and women might have different preferences that influence their professional choices should not really be considered controversial, but it is. Consider, for example, some of the professions where women are systematically overrepresented, such as psychotherapy. It is hard to believe that hidden barriers keep men out of careers in psychotherapy: indeed, there are frequent reports of the vain efforts that are made by psychotherapy educators to attract more male candidates.³ It doesn't follow that whatever preferences make men less likely to be attracted by psychotherapy are innate (though they could be); they could have been formed by socialization in childhood or may simply reflect perceived preferences of others. But whatever their origins, different preferences of adult women and men, on average, for the professional activities of psychotherapists remains much the most likely explanation for the discrepancy in their representation. If that's true for some professions where women are overrepresented, it may also be true for some in which they are underrepresented: airline pilots may be one example, jazz drummers another. If any reader can point me to a scientific study that finds as many girls as boys dreaming in kindergarten of becoming airline pilots when they grow up, I shall gladly recant.

Still, it would be good to have harder evidence for the existence of such differences in preference. The scientific evidence that we do have comes largely from laboratory experiments. It's controversial how fully behavior in the laboratory can explain the way people behave in life outside the laboratory, but nobody seriously thinks that's because subjects in laboratories behave differently from the way they act in normal life. In one sense, of course they behave differently: that's the point of the laboratory setting. In a laboratory, the experimenter aims to screen out other confounding influences so as to focus on the dimension of interest—in this case, some possible preference difference. By making two choices identical in all respects, except that one is riskier than another and has a higher expected payoff, it's possible to see whether the subjects find the trade-off acceptable—and whether women, for instance, are less likely than men to do so.

There's a lot to criticize in laboratory studies—for example, the common tendency to use university students in prosperous countries and generalize from them to the whole population. And the very structured setting may induce kinds of behavior that you wouldn't find elsewhere—though experimental laboratories are not intrinsically more artificial than other kinds of setting people routinely think of as part of the “real” world, such as stock exchanges and company boardrooms. So although we should keep an open mind and never think of laboratory behavior as an infallible mirror of behavior in other settings, when we find systematic differences between how men and women behave in the laboratory, we should be prepared to look hard for similar differences in their behavior outside the lab.

Two kinds of laboratory experiment have been particularly influential in shedding light on women's preferences. The first consists of experiments showing that women tend to be more risk-averse than men.⁴ The fact that these findings have been replicated many times suggests they are capturing something systematic about human behavior. (There is also a related but distinct set of findings that men tend to be more overconfident than women about their ability and likely future performance, which leads to what looks very like risk-seeking behavior.)⁵ There are good reasons to expect females of any species to be more averse to risk than males. The difference in reproductive success between the most successful and the least successful females is much smaller than the difference between the most and the least successful males; consequently, natural selection has favored more risk-taking strategies on the part of males.⁶ And studies of risk taking in many other animals confirm this conjecture.⁷

It seems plausible that if men are willing on average to take more risk, those for whom the risk pays off would end up systematically better rewarded than women of equal ability. They might also prosper because of promotion procedures within firms that are unable to distinguish between talent and luck.⁸ If success is partly due to talent and partly due to good luck, individuals who take bigger risks could, when those risks pay off, appear to have greater talent. (Similarly, when the risks fail to pay off, they will appear to have less talent, which might explain some of the overrepresentation of men at the bottom end of the distribution of society's rewards.) Thus fallible promotion systems may erroneously identify men as having greater talent when in fact they have just had better luck, amplified by their greater willingness to trust to luck.⁹ In fact, the possibility that certain reward systems may encourage excessive risk taking (a possibility that the recent financial crisis has given new plausibility) may mean that in such situations, men may receive higher economic rewards on average when the value they create for others is lower than that created by women. This may happen in a range of ways and not just because of a difference in attitudes toward risk per se: for instance, there is evidence that, depending on the context, women may be more inclined to altruistic behavior than men, and therefore more likely to take into account potentially harmful side effects of their actions.¹⁰ There is also some suggestive cross-country evidence that female government officials are somewhat less prone to corruption than males in the same position, a form of socially valuable behavior for which they remain sadly underrewarded.¹¹

The second kind of experimental finding consists of differences between men and women in preferences for competition. One much-cited study consisted of a test of skill in which subjects could choose whether to be rewarded according to their performance relative to other players (a tournament) or according to their absolute performance (a piece-rate scheme). Nearly three-quarters of men but only a third of women chose the tournament scheme, although there was no difference in performance between men and women under either scheme.¹² Some other studies have shown a tendency for women to perform less well under competitive reward schemes than under piece-rate schemes, while men's performance tends to improve under competition. If rewards in the economy at large favor those who are attracted by competitive contexts and whose performance flourishes in such contexts, this might account for the kinds of difference in economic performance we have seen.

Still, there are two main reasons to be wary of such a conclusion. First, as I emphasize in chapter 4, it is simply not true that modern economic life is all about competing. It's as much about managing cooperation as about competition, and no one has yet come up with even a theory, let alone with evidence, as to which kinds of preference are likely to be most valuable to participants in the modern economy. It's a reasonable guess that a preference for competitive environments would be an advantage to people hoping to make it to board level in a Fortune 500 company. But it remains a guess, and we really don't know to what extent a preference for less competitive environments could be considered a handicap. After all, if you were thinking of employing someone with whom you would have to work closely afterward, would you necessarily want to pick the most competitive person you could find? Other experiments suggest that women have a positive preference for teamwork, and there are surely many contexts in a modern economy where this ought to give them a significant advantage in recruitment.¹³

The second reason for caution is that the experimental findings about gender differences in preferences are quite sensitive to context. For instance, women's performance tends to weaken when they compete against men, not when they compete against other women.¹⁴ Women's preference for teamwork seems to depend in part on their greater optimism than men about the quality of their fellow team members, an optimism that is likely to fluctuate according to what else they may know about them.¹⁵ This sensitivity to context is a general feature of experimental studies of both risk aversion and attitudes to competition. In some laboratory settings the differences in risk aversion are large, while in others they are negligible; this variation makes it difficult to know just how important such differences may be in life outside the laboratory. In some experiments it makes an important difference to women's performance whether they are competing against men, while in others it does not.¹⁶ Similarly, results for adults are often not replicated when the experimental subjects are children, contrary to what one might expect if they were the result of stable underlying preferences.¹⁷ Results found in one country are often not found in another.¹⁸ There's also some evidence that differences in preferences are more likely to be found when the tasks on which subjects compete are considered to be typically male tasks (this is an instance of stereotype threat).¹⁹

Nevertheless, it would be a mistake to dismiss the findings as due solely to context. There is strong evidence, for instance, that genetic or hormonal factors underlie at least some such gender differences, though others have not been convincingly associated with such factors.²⁰ Overall, it seems safe to say that these differences are intriguing but that their importance in accounting for gender differences in economic rewards remains a matter of speculation. Rigorous studies of this question are still in their infancy. So far, the measured impact of such preferences on labor-market outcomes, like the measured impact of personality differences more generally, has been small.²¹

Women Don't Ask?

A second possible explanation for differences in women's economic rewards is that, even if women create as much economic value for their colleagues and employers as men do, women may negotiate less aggressively for their share of that economic value and so may end up with a smaller reward. Women Don't Ask is the title of a deservedly famous book by Linda Babcock and Sara Laschever that presents large numbers of case studies suggesting that an unwillingness to negotiate may cost women a lot: “Negotiating your starting salary for your first job can produce a gain of more than a halfmillion dollars by the end of your career,” they write.²² In later studies with other colleagues, Babcock has shown that women's attitude toward negotiation is itself very dependent on context: for instance, women negotiate more aggressively when doing so on behalf of others, when the negotiation setting provides cues as to what is a “reasonable” agreement, and when it is framed as an opportunity for nonconfrontational “asking” rather than negotiation.²³ She has also shown that women's reluctance to negotiate aggressively may be based not just on inhibition but also on a shrewd assessment of how their behavior will be viewed by others. In an experiment where individuals were asked to evaluate others as potential colleagues, men gave significantly lower evaluations to women who negotiated rather than simply accepting offers, while female evaluators showed no such tendency.²⁴

Like the findings of different preferences for risk and competitive environments, these findings are both plausible and intriguing. But before we accept them as an important factor in explaining gender differences in economic rewards, we need to consider a puzzle. The more convincing the “women don't ask” explanation is in accounting for the lower salaries of women, the less convincing it is as an explanation for the lower representation of women in certain occupations. Someone who's willing to accept a smaller share of the cake as the price of her participation is obviously a more appealing colleague than someone who negotiates hard for a larger share. If it's true that women create as much value for their employers as men do but cost the employers less, the employers would be insane not to prefer employing women.²⁵ To put it another way, “women don't ask” might be a complete and convincing explanation for low female salaries coupled with female overrepresentation in those jobs that pay them low salaries. But it won't suffice as an explanation of low female salaries coupled with female underrepresentation. There must be another reason why employers are not rushing to take advantage of the profit opportunities that these reluctant negotiators are pushing their way.

A Masculine Bubble?

What other explanations are open to us, in the light of what we've seen so far? If women are not, on average, contributing less than men to the environments they work in, either through having less talent than men or through a relative distaste for the environments in which large contributions are to be made, and if the main reason for the discrepancy does not lie in their unwillingness to negotiate as aggressively as men for their share of the gains, only one possibility remains. It is that men, individually or collectively, are negotiating harder with women than they are with other men, whether through prejudice, habit, or conscious coordination.

We saw in chapter 4 that women might be disadvantaged in the distribution of economic rewards because men monopolize a scarce resource, even if women are at least as talented as men in the areas in which their talents can be compared. Is it imaginable that something similar may be happening in a modern economy in which hunting and gathering are no more, in which nothing prevents women from specializing in the production of a scarce resource? In a modern economy, unlike (perhaps) a small hunter-gatherer community, men don't negotiate as a bloc with women as a bloc. Instead, individual employers negotiate with workers, or perhaps with trade unions. There may be pockets of trade-union solidarity or employer coordination, but there exist no mechanisms whereby men as a whole can dictate terms to women as a whole. More men than women represent employers, certainly, but most men are employees, on the other side of the salary bargain. The gender gap in salaries shows up at every level, between male and female employees as well as between female employees and male employers. It's hard to see how men could be colluding to negotiate more aggressively with women, because it's not clear what could be the mechanism that coordinates this collusion.

And without a coordinating mechanism, it's hard in turn to see how the result of millions of individual bilateral negotiations could result in a systematic discrimination against women, or indeed against any other disfavored group. Let's return to the impact of height on earnings and ask whether that could be the result of an irrational prejudice. It seems unlikely, someone might argue, that there could be systematic discrimination in employment and earnings against short people. Whatever Marxists might once have argued, employers are not a single-minded class who collaborate to enforce their will on a reluctant proletariat. There are millions of individual employers with their own idiosyncratic preferences, and if any employers choose to indulge their prejudices by irrationally refusing to employ talented short people except at lower wages than tall people, other employers will soon discover that they can make more money by doing the opposite. It's conceivable that all employers might become infected by the same irrational prejudice, but it would need to be explained how that was possible. A much more likely story is that if short people are less favored by employers, this is because they happen to have, on average, fewer of the various talents that employers need. There won't necessarily be any one single type of talent, and employers won't all be looking for the same ones. But it won't do any employer any good to base hiring decisions on height if height bears no relation to the talents employers really need. So, unless some good reasons can be found for thinking that economic outcomes are systematically biased against short people in a way that would harm the self-interest of the employers themselves, the fact that tests show that tall people do better would be a sign of the reliability of the tests. Or so the argument would go.

This line of argument now looks less convincing than it might once have, particularly in the aftermath of a financial crisis in which it has become clear that irrational prejudices (about the tendency of real-estate prices to go on rising for ever in real terms, for instance) can come to be very widely entrenched even though it would be possible to prosper by repudiating them. Still, the prejudices have to be strongly salient, the kind that get reinforced as people talk about them, and of which there are enough cases in everyone's experience for the prejudice and the anecdote to override the evidence and the science. Many beliefs about health are like this: belief in the efficacy of homeopathic medicine, for instance, which flourishes in the absence of even a shred of rigorous evidence, reinforced by the strong power of placebo effects (for which we still have very little scientific explanation).²⁶ Maybe a belief in the economic efficacy of height could be just such a prejudice, reinforced in turn by placebo effects (for example, if confidence is good for economic performance, and tall people are more confident because they have come to believe that height is right, they will perform better).

As we saw above, the link between height and economic rewards does seem to be based partly, though only partly, on a genuine connection between height and talent. But we've found no evidence at all for a link between gender and talent, certainly not for the multiple and flexible talents that a modern economy demands. Might prejudices about the existence of such a link nevertheless persist in the labor market, like the ineradicable faith in homeopathy? Could it be a kind of masculine bubble that inflates the earnings of men relative to those of women, even though nobody has any evidence that it makes economic sense?

The evidence from detailed studies of gender and earnings gives us a few tantalizing clues. First of all, given women's lower earnings, you might expect to find that the returns on higher education are lower for women than for men. After all, a college degree costs as much for women as for men, and if they earn less afterward, it sounds like a less profitable investment. In fact, most studies find a higher return on schooling for women, which is puzzling. The most likely explanation is that more highly educated women do face some earnings disadvantage relative to equally talented men, but this disadvantage is much less than that faced by less highly educated women.²⁷ So women who invest in education not only improve their skills in a way that is valuable to employers; they also equip themselves with what they need to get a larger share of the value they create for those employers (including, Babcock and Laschever might say, a greater willingness to ask). This makes sense if there's an element of shared prejudice in the disadvantage; the more educated the women against whom the prejudice is exercised, the harder it is for the prejudice to survive.

A second intriguing clue comes from a detailed study of the earnings of MBAs from a top business school, carried out by Marianne Bertrand, Claudia Goldin, and Lawrence Katz.²⁸ They show that although the earnings of men and women after they graduate are fairly similar, they soon diverge, with women earning progressively less than men over time. For instance, men earn, on average, $130,000 at graduation compared to $115,000 for women: while nine years later they are earning $400,000 compared to $250,000 for women. Part of the discrepancy appears to be associated with the fact that the men in the study sample had more training than the women before receiving the MBA. But a large part of the story is that women have more career interruptions and work fewer hours, on average, than the men: for instance, after nine years, 30 percent of women have had at least one spell of not working, compared to less than 10 percent of men. These factors are associated with motherhood: women without children do not have more career interruptions than men, do not work fewer hours, and do not suffer a decline in salary relative to men. The authors are careful to check whether there might be any difference in talent between the women who have children and the women who don't, but they find that women who have children in fact tend to have had slightly higher earnings beforehand than women who don't.²⁹ The outlines of the story are unsurprising: what's interesting about the detail is the very high price that the women pay for these career interruptions. Or rather, it's the price everyone pays for career interruptions: those few men who have career interruptions suffer as great a salary penalty from them as the women do.

The business world analyzed by Bertrand, Goldin, and Katz is not a microcosm of the economy at large. It is a world of the socially and educationally privileged, in which both men and women are trained to negotiate hard for what they can get—and they get a lot. So we must be careful before drawing wider conclusions from the study. Nevertheless, it represents a slice of the US labor market for which concerns about discrimination have often been voiced (the fact that only 2.4 percent of CEOs of Fortune 500 companies are women being a particularly striking example). For this microcosm, then, the picture seems reasonably clear. The different salary dynamics do not appear to be a result of discrimination against women as such: men and women with identical qualifications who make identical decisions about career interruptions do equally well. As the authors put it: “The data do not indicate that MBA women lose more than MBA men for taking time out. It appears that everyone is penalized heavily for deviating from the norm.”³⁰ Instead, the story is one in which the rules for getting ahead emphasize long hours of work, single-minded devotion to the job, and a refusal ever to take a break. MBAs (both men and women) work, on average, sixty hours per week after graduation, with investment bankers averaging a startling seventy-four hours per week. The fact that fewer women want to play by these rules puts women at a disadvantage. In particular, taking career breaks is costly not just at the time but apparently for years and even decades afterward. It's a matter of different preferences, certainly, but also a matter of the high price those preferences entail. And this in turn raises the question of whether these rules make sense in the modern world.

Are the extreme demands made on the participants in the modern workplace really appropriate to the kinds of tasks the twenty-first-century economy requires? The estimates of Bertrand, Goldin, and Katz suggest that people who work fifty-eight hours a week on average are rewarded by a slightly more than proportionate increase in earnings compared to those who work fifty-two hours, even without considering any subsequent differences in prospects for advancement. Are the last hours of their working week really as productive for the employer as that? And if such career breaks underlie the later discrepancies in representation at the top, are the best CEOs really the people who have proved their worth to the company by never doing anything else in their lives but work? Can it really make sense that career breaks taken in someone's twenties and thirties should constrain their opportunities when they are in their fifties and sixties? It's possible that something about modern business organizations really does require such focus and extreme drive. But it's surprising, if so, that the diffusion of information and communication technologies that has been such an important influence of both social and working life, and has driven such a spectacular increase in multitasking in our personal lives, does not appear to be making more of a difference to the necessary single-mindedness of this professional commitment.

We return to these questions in chapter 9, but at this point I want merely to put to rest an argument that is often made in support of such working practices: namely that if they were not efficient, competition would make them disappear. There's a big flaw in this argument, as we saw in chapter 2. As the peacock's tail reminds us, an extraordinary number of highly inefficient social practices survive and spread, both in nature and in human society, because they signal something about those who adopt them. Given the way the signals are interpreted, everyone may have an individual interest in settling on the same equilibrium; but they might all be better off if less wasteful ways to signal their qualities were available. The social codes in which commitment to the goals of the organization can be signaled only by a willingness to work long hours and never take a career break may have outlived their usefulness, but that doesn't mean that individuals can choose to ignore them without penalty.

The Signaling Trap

The modern office is an environment in which mutual signaling and maneuvering are as essential to survival as they were on the African woodland savanna where we first evolved. Busy people go to meetings where they waste their own and everyone else's time purely in order to show how busy they are. Important people spend time pressing the flesh just to remind others of their importance, while unimportant people offer their flesh for pressing in the eternal hope of transcending their unimportance. Senior managers arrive early at work and leave late to signal their diligence to others; junior managers arrive early and leave late in the hope of becoming senior managers who will do exactly the same. Their departure is further delayed by socializing with those whose company weighs on them already, because the prospect is less unattractive than that of imagining their colleagues socializing without them. People who spent their college years broadcasting to others about how much they would love to take time off to work among other people's children in the third world do a sharp about-turn once they become corporate employees: they broadcast no less strenuously their utter antipathy to the idea of taking any time off at all, and certainly not for the benefit of their own children.

Successful business organizations prosper by finding ingenious ways to allow their members to escape mutual signaling and maneuvering just long enough to do the productive work that enables the organization to survive and employ them in the first place. If this smacks of undue cynicism, ask yourself how much of your time yesterday was spent doing things that didn't contribute directly to the value of your work but was directed instead at signaling that value to other people. It's not surprising that decisions about how much time to devote to raising children are riven with anxiety about how these decisions will be decoded by others: some people may pay a large penalty for their decisions because, well, it's always been that way. Changes that might benefit everyone may not be within anyone's individual reach.

It seems likely, too, that the kinds of signaling women feel the need to do in the workplace may be more complex than those required of men and therefore more difficult for their colleagues to interpret. In chapter 2 I mention the idea that many minority groups feel under pressure to engage in “covering”—not exactly hiding their identity, but making it inconspicuous so that it no longer looms large on the radar of colleagues and friends who don't share that particular identity. Kenji Yoshino cites the case of Margaret Thatcher, who trained with a voice coach to lower the timbre of her voice.³¹ Any woman who works in a largely male environment recognizes the challenge of finding a way to project herself as competent and professional while not treating her femininity as a shameful secret. This isn't to say that these pressures are unique to women and ethnic minorities: those middle-aged white males in identical dark suits are toning down something too. Everyone has a hinterland of which their professional colleagues rarely glimpse more than a shadow. But worry about looking masculine doesn't oblige men to undersell many of their best talents in the same way that worry about looking feminine may do for their female colleagues.

If women are stuck in a signaling trap, this may explain not only why they pay such a high price for their career breaks but also why a failure to negotiate may be so costly. I suggested above that someone who's willing to accept a smaller share of the cake as the price of her participation is “obviously” a more appealing colleague than someone who negotiates hard for a larger share—but suppose it isn't obvious after all? Suppose that others routinely assume that if she doesn't negotiate hard, it's because she's not worth so much? Then we might understand why lower salaries for women do not lead to overrepresentation in the jobs that are paying them less. So it becomes urgent to see whether such a signaling trap exists, and if so, how it manages to persist when it leads to the persistent undervaluation of women's talents.

Signaling traps may matter outside the workplace too. If women still want to take career breaks to look after children more often than men do, that may not just be because social codes don't yet put enough value on paternity breaks for men. More subtly, women may be more aware than men of the need to signal conscientiousness through the way they take care of children. Signaling maternal qualities might have been particularly important in hunter-gatherer communities where a woman might need to find other sources of support if the father of her children died, and the need to do so may have become psychologically rooted to a degree that no longer reflects the much safer environments in which children grow up. If so, it's an unfortunate historical development that the family and the workplace have now become so separated. Women caring for children signal a quality— conscientiousness—that employers really value, as we saw in chapter 6. But employers are not present to observe them with their children, and women continue to pay a high price for their absence from the workplace during those years.

Social codes survive because they are enforced in networks of people who live by them. Someone may wonder why, if talented women are underrewarded by modern labor markets, intelligent entrepreneurs (male or female) do not spot a profit opportunity and turn those talents to creative use. The best answer may be that it is one thing for such talented women to exist and another thing for intelligent entrepreneurs to identify and deploy them. The reason this may be difficult is that everyone, male or female, lives and works in networks that shape how we present ourselves to the wider world, and talented but underrewarded women may have less visibility in the networks than their better-rewarded male colleagues enjoy. The importance of networks for our social and economic life is central to our heritage as social primates, and it is the subject of chapter 7.