The small representation of women and minorities among students of economics has been noted for some time. While the proportion of B.A.'s earned by women in psychology rose from 36.7 percent in 1949-1950 to 70.8 percent in 1988-1989, in sociology from 50.6 percent to 68.8 percent, and even in mathematics from 22.6 percent to 46.0 percent, in economics it has increased from only 7.6 percent to 32.5 percent. The share of Ph.D.'s earned by women in 1988-1989 was 56.2 percent in psychology, 50.9 percent in sociology, 26.6 percent in business, 19.4 percent in mathematics, and 19.0 percent in economics. Hence, general sexism in the classroom1 does not appear to be the main culprit, nor do the explanations that mathematics requirements inhibit women's entry into economics or that women are uninterested in business-related fields seem convincing. Instead, one must look to factors specific to economics. Evidence that women students do not perform as well as men in introductory economics courses (John J. Siegfried, 1979; Gordon Anderson, et al., 1994), although they have higher grades overall, further adds to this conclusion. For these reasons there has been considerable interest among feminist economists in the chilly classroom climate, for women and minority students in economics courses. In this paper, the focus is on the small representation of women among economics faculties, the biased subject matter, and the narrow approach of traditional economics. The number of women faculty can only be increased gradually as their representation among graduate students and new faculty hires increases. However, the subject matter can be changed more rapidly, as the consciousness of instructors and authors of textbooks is raised, and the challenge to the traditional economic approach appears to be making more progress than most of us dared to hope only a few short years ago. Thus, in spite of the remaining problems, there is reason to believe that in economics, as in most other disciplines, women's progress will eventually accelerate.

M UCH research has been done in recent years examining the differentials in earnings of men and women in particular occupations.' While these studies have contributed considerably to our understanding of this earnings gap, they tend to accept the occupations as given and do not examine the extent to which men and women have a different occupational distribution or question the reasons for it. In some types of jobs this question does not arise at the time the person enters the labor market. For example, when highly specialized skills are required, as is the case for most professions and crafts, only those trained in relevant skills are recruited. But there are many other positions at the entry level for which only general requirements, such as intelligence, literacy, motivation, or, in some cases, physical strength are needed. Theie are also many higher level positions that are filled through upgrading and promotion from the lower levels. In such labor markets employers, and perhaps employees, are likely to have considerable discretion with respect to placement of particular individuals into either a ladder-type or a dead-end job. If gender plays a significant role in this decision, any measure of achievement and rewards that is restricted to initial placement differences within job categories may seriously underestimate the effect of sex-discrimination on subsequent promotions and earnings. A number of investigations have found segmented markets for labor, each marked by differences in wages and career progression.' One study of male workers specifically noted that onthe-job training and type of starting position are crucial because they are the beginning of a dynamic process which continually affects the employee's earnings. The same study showed that starting in the high and middle skill sectors is consistently associated with higher earnings later on.3 In this paper we shall focus on the extent to which men and women are found in different entry jobs and how this affects their later earnings.4 Finding women in entry jobs which do not lead to high earnings in the long run need not be to their disadvantage if their labor force participation tends to be intermittent. It has been argued (e.g., Johnson and Stafford, 1974; Polachek, 1979) that women take jobs that pay relatively well initially but do not provide valuable training leading to great increases in earnings because they expect to drop out of the labor market in any case. This hypothesis would be supported by findings that women are paid more than men with comparable qualifications initially, and hence earn more for the limited period they expect to remain in the labor market, even though other jobs pay better later on. If, on the other hand, men earn as much or more to begin with, we cannot ascribe their higher earnings in later years to their greater willingness to invest in their own training on the job.

Numerous studies have established that part of the very substantial male-female earnings gap is explained by differences in the amount of human capital workers have accumulated. (See, for example, Jacob Mincer and Haim Ofek, 1983.) Institutional factors have also been found to play a role in determining wages (David Gordon et al., 1982). Occupation further helped to explain the remaining gap, but several researchers have shown that introducing dimensions of work authority by taking into account the individual's position in the work hierarchy explains more of the variation in earnings than does occupation (Martha Hill, 1980). Last, two recent studies (Ferber and Spaeth, 1984; Spaeth, 1985) also included control over monetary resources. This variable added substantially to the explanatory power of earnings regressions, even after human capital variables, institutional factors, and several other measures of work authority had been entered. Like the other studies, Ferber and Spaeth also found that reward structures for men and women are quite different, suggesting the possible existence of discrimination. The question whether women may also be at a disadvantage in achieving control over monetary resources was not investigated. When Hill examined the process of achievement of work authority, she found substantial differences between male and female workers. In this paper we examine whether the same is true for attaining financial control. I. Data and Analysis

The study of discrimination received a major impetus in the 1960's when increasing social attention focused upon race and gender differentials in market outcomes. Gary Becker's The Economics of Discrimination (1957) strongly influenced empirical research by providing a definition of wage discrimination and suggesting a specific way in which it might operate. During the following years new theories were developed and refined in an attempt to explain why there appears to be continued discrimination in spite of market forces presumably operating against it. Similarly, a large amount of empirical work has been done to determine whether and how much discrimination actually exists, and to a lesser extent to test the implications of the various theories. Even so, the hope expressed by Becker in the preface of the second edition (1971) that our understanding of discrimination would increase so rapidly that the materials in his book would become obsolete before another decade began has clearly not been fulfilled. Here, we review what has been learned in the intervening years and suggest some fruitful directions for future research.1 The focus of this paper, like that of most of the empirical research in this area, is on determining the extent of discrimination rather than on testing alternative models of discrimination.