The 59¢ Fallacy

The earnings gap between men and women can't be pinned on discrimination.


This being an election year, you have probably been seeing people wearing every sort of campaign pin imaginable—the standards like Reagan and Bush, Mondale, and Hart, not to mention all of the more-localized varieties. But you may also have seen a pin that says simply "59¢."

There is nobody with that surname running for office. It is a pin distributed by the National Organization for Women (NOW), and according to a recent fundraising letter signed by NOW president Judy Goldsmith, it symbolizes "the plain, frightening fact…that most women are paid just over half as much as men for the very same work—to be exact, 59¢ for every $1 earned by a man." She continues, "Today, nothing better illustrates the economic plight of American women than NOW's 59¢ campaign button."

What Goldsmith is keying into is the much-discussed "earnings gap." It is an article of orthodox feminist faith that this earnings gap requires aggressive intervention by the government. The argument runs something like this: Earnings differences between men and women are evidence of discrimination in the free market. The earnings gap widened between 1955, when women's earnings averaged 64 percent of men's, and 1977, when women's earning level dropped to 59 percent. Therefore, it is concluded, discrimination against women has increased. Since this discrimination takes place in the free market, the government needs to intervene to protect women.

However, there is a large gap in this earnings-gap argument, as becomes clear if several key questions are considered. Is the earnings gap really due to discrimination? What other factors might account for the earnings gap? And, most important, is it even possible to distinguish discrimination from some of these other factors?

Surely the most striking change concerning women in the labor market since 1955 is that there are so many more of us now than there were then. The labor force participation rate of women the percentage of women who are working or seeking work jumped from 31 percent to 52 percent from 1952 to 1982, which translates into nearly three times as many working women. This huge increase in the number of women working outside the home is important because increases in the supply of something are usually associated with decreases in its price. In this case, the supply of women workers increased in comparison with men, so it should not be surprising to find the wages of women falling in comparison to those of men between 1955 and 1977.

The Supply Side

Suppose, though, that the labor market doesn't really work like the market for apples or houses, and therefore a supply increase could not cause that large a wage change. It's a fair enough question. But then consider a different example that can't possibly be explained by discrimination. The baby boom created a huge increase in the supply of young workers in comparison with the number of older workers in the late 1960s. Did the earnings of the baby-boom new workers fall in comparison with prime or middle-aged workers? The answer is a resounding yes. The earnings of young workers fell from 63 percent of the earnings of middle-aged workers in 1968 to 54 percent in 1974.

This supply increase actually caused a larger fall in earnings than the one that women have experienced. And the baby boomers' financial bust happened over a shorter period of time. In fact, some economists have been surprised that the earnings gap between men and women didn't widen even more than it did! Probably the reason is that not all the women entering the labor force went into "female" jobs. Many entered jobs and professions formerly closed to women, in which they compete primarily with men and not with other women. Because they were not increasing the supply of female-job workers, the women who pioneered in the fields of banking, medicine, construction, and law did not lower the average wages of women.

Even so, a disproportionate number of women did enter traditionally female jobs such as clerical and service work. Clerical jobs actually account for a larger fraction of jobs held by women now than in 1960. The wages in these occupations have not kept pace with wages in other sectors of the economy, at least in part because of the flood of new women workers entering these fields for the first time. Since these are the jobs in which women are most heavily concentrated, it is not surprising that women's average wages have fallen in comparison with men's.

Another characteristic of the newly working woman is that she is likely to be an older person, either returning to work after child rearing or entering the labor market for the first time. This means that she very likely has limited experience in the job market. In fact, women have 10 to 15 years less labor-force experience on average than do men of the same age. Naturally, the wages that inexperienced women can command will usually be lower than those of men or of other women who have worked longer or more continuously. So the average earnings of women have fallen in part because so many women are relatively inexperienced.

This is not to conclude that there is no discrimination against women. That would be silly. But it is equally silly to conclude that discrimination has increased since 1955 because the difference in earnings between men and women has increased over that time. There are many plausible explanations for that fact.

The Choice Factor

Anyone who has had his or her eyes open at all has noticed that women have opportunities today that were unheard of in our mothers' time. Women who would have been afraid to risk the wrath of society and family are venturing into the workplace for the first time. Others of us are breaking into male-dominated fields and finding a degree of success and acceptance that astounds our mothers. (The percentage of lawyers and judges who are female jumped from 4 percent in 1972 to 15 percent in 1982. The percentage of physicians who are women increased from 10 to 15 percent in the same period. And the percentage of women on college and university faculties has risen from 22 to 37 percent just since 1967.) Still others of us are taking even greater risks and starting businesses of our own. We still have some distance to travel in the job market, but it is difficult to take seriously the idea that women face more discrimination today than they did in the 1950s.

Of course, feminists usually don't come right out and claim that women are worse off economically today than they were 30 years ago. That would be too obviously absurd. So NOW president Judy Goldsmith, for example, talks in a fundraising letter of "the economic plight of American women" and urges the 59¢ button on women as "a symbol that vividly demonstrates the intolerable economic discrimination against the status of women in our society."

The insidious thing is that an argument is implied here rather than stated directly. If Goldsmith were to come right out and say, "The earnings gap is totally caused by discrimination," she would have to draw the absurd conclusion that women face more discrimination today than they did in the 1950s. More to the point, her readers might draw that conclusion themselves and then question her initial premise.

On the other hand, it would be an important concession if Goldsmith were to acknowledge that the earnings gap might be caused by something other than male or market malevolence. NOW's symbol, the 59¢ button, would lose some of its point. More importantly, though, the plain fact of the matter is that it is extraordinarily difficult to establish a linkage between discrimination and a difference in wages. The 59¢ logic sidesteps that problem by placing the wage differential on the defensive. It presumes that an earnings difference is prima facie evidence of discrimination.

Actually, many of the factors that contribute to the earnings gap are the result of personal choices made by women themselves, not decisions thrust on them by bosses. The most important example is marriage.

Married women often are unable to relocate to further their careers as much as they would like. Many married women drop out of the labor force to raise children. As a result of both of these factors, married women tend to choose occupations with easily transferable skills that do not deteriorate when unused. This is why many women traditionally have chosen elementary-school teaching more often than college teaching, nursing more often than doctoring, and humanities more often than technical subjects.

These differences between married women and single women (and between married women and men, for that matter) contribute dramatically to reducing the earnings of married women. Thus we find, in a comparison of the earnings of never-married women and those of never-married men, that the women's earnings in 1980 were 89 percent of men's. This figure has been essentially unchanged since the 1960 census. So if one is looking for a "culprit" for the earnings gap, it is far more plausible to pin the blame on marital status than on gender.

Some people might argue that these figures simply reflect the oppressiveness of marriage and the need for radical changes in this institution. There is some merit in this argument. Nevertheless, it is undeniable that many women do choose to get married. Perhaps a large number believe that the pleasures of intimate companionship and raising children are worth some financial sacrifice. We might wish that marriage did not require that women make this sacrifice, and we might work very hard to promote better options. But it is entirely unfair to blame employer discrimination for earnings differences that are essentially the result of choices made by individual women.

Moreover, while marriage may often mean a disadvantage in the job market, that has been changing for the better. Women are now less likely to drop out of the labor force to raise their children. More women are training themselves in technical fields. And in many households, the wife's career needs are determining the family's next move.

So why don't married women have higher relative earnings now? This is certainly a valid question. But these different lifestyles have become widespread only over the last 10 to 15 years. The earnings gap is based on aggregate data that include women ranging in age from 25 to 64 years, most of whom have not really been affected by these lifestyle changes. Among full-time workers 25 to 34 years old, women's earnings were 70 percent of men's in 1980.

So there are a number of differences between men and women in the job market that may account for their earnings differences. Taken altogether, these very reasonable and understandable factors cannot, it is true, account for the entire earnings gap. But when the gap is corrected for these factors, it is not 59 percent but more like 66–87 percent, depending on the study.

It is often claimed that the difference that still remains after all the economic factors have been accounted for must be due to discrimination. That is, discrimination is measured as the residual, or leftover, difference, after all other factors have been taken into account. But this attempt to gauge discrimination is dubious.

The Information Gap

The residual actually measures our ignorance. It includes everything that has not been directly measured but that influences a person's wage. The residual includes things as diverse as good luck and personality as well as discrimination. Common sense tells us that personality makes a huge difference to a person's career success. Ambition, aggressiveness, willingness to take risks, ability to get along with and motivate others, commitment to the job, willingness to assume responsibility—all of these factors contribute to higher wages. In fact, many career magazines for women advise their readers to develop exactly these traits. But none of these factors can be measured, and the residual earnings difference could just as easily be due to differences in these factors as well as to discrimination. The point is that we cannot distinguish discrimination from these other, unmeasurable factors.

Do we really know that women are on average so much less aggressive and less ambitious that their earnings would be 87 percent of men's? The answer is that we don't know and we can't know. We cannot rule it out as a possibility, though, because our ignorance of what does generate a person's earnings is so great.

Clearly, neither feminist fund raisers nor the average well-informed citizen knows this stunning fact: only 40 percent of the earnings of white men can be accounted for by measurable factors. That is, if we look at a population of white men, a full 60 percent of the differences in earnings among them cannot be explained by anything we can measure. Conventional discrimination cannot possibly be an issue in this particular population. Yet the unexplained residual earnings difference within this group swamps the largest difference in male-female earnings that could possibly be due to discrimination.

This is why we cannot rule out the possibility that the entire earnings gap between men and women is due to real personal productivity differences that cannot be measured. The upshot is that the presence of discrimination can be neither proven nor disproven with statistical tests.

Despite these technical problems, which are well known among economists, statistical tests are often introduced as evidence in discrimination lawsuits. An unexplained earnings gap is usually accepted as evidence of discrimination. And if it is accepted as evidence, the plaintiff will almost always win. In effect, this means that there is a presumption of guilt rather than innocence on the part of the employer accused of sex bias. It is true that the earnings difference in a specific employer's work force could be due to the employer's discrimination. But it could instead be the result of something equally unmeasurable. Mere statistics are not enough to tell.

A few years ago, an economist named George Borjas wrote an article in which he examined the salaries at the Department of Health, Education and Welfare (HEW), now Health and Human Services. He subjected HEW's salaries to the same statistical procedure that HEW itself used to demonstrate that universities were discriminating in their employment practices. Lo and behold, he found that HEW discriminated at least as much and sometimes more than the institutions they were charged with regulating. Many people interpreted this study as an attack on HEW salary and promotions policies. Actually, the point was to attack the methodology that HEW used in its attempts to ferret out discrimination in academia. Unfortunately, these procedures are still widely used in court cases.

So the 59¢ button is not really a very good symbol of women's economic plight. It is much more a symbol of a flawed method of correcting some very real problems that exist for some women in the workplace. It is a symbol of the misuse of statistics, both to make a dramatic point in the arena of public opinion and to win in court. It is a symbol of the faith that much of the women's movement places in government intervention as a solution to women's problems.

Unfortunately for that faith, many women are liberating themselves without the help of the law. They are finding their own path through the maze of the world of work and devising their own way to balance all of their financial, personal, and emotional needs. And the movement's loss is the individual's gain; for many of the problems that some women face today will best be solved by the individuals themselves and not by government action.

Jennifer Roback is a professor of economics at Yale University. She has contributed to various publications and is a radio commentator in the Reason Foundation's Perspective on the Economy series.