Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Wage Differentials and Institutional Features of Labor Markets What features of the wage-setting process create the differences in wages observed between men and women and between minorities and nonminonties? What explains the extent of job segregation observed? Why do these differences persist? In this chapter we take an institutional view of labor markets in examining these questions, and we argue that it is likely that some portion of the wage differentials observed can be regarded as the result of discrimination, both intentional and uninten- tional. In reviewing~a variety of explanations for the existence of dis- crimination offered by economists, we note that none of them satisfac- torily explains the causes of discrimination, but each suggests that it is plausible that discrimination exists. Finally, we briefly consider the im- plications of the complexity of labor markets for implementing policies to reduce discrimination and its effects. LABOR MARKETS Of today's employed civilian labor force of approximately 100 million, about 90 percent work for wages or salaries (the rest are self-employed). For employees, access to jobs, wage levels, conditions of work, and other aspects of employment are determined by the operation of labor markets. In the conventional mode] of perfectly competitive labor mar- kets' demanders and suppliers of labor possess complete information and total mobility; the bargaining of individual employees and employers and the unfettered adjustment of supply and demand determine the 44
Institutional Features of Labor Markets 45 wage of each worker; consequently, the wage of each worker exactly equals the value of his or her economic contribution (i.e., marginal revenue product). But workers rarely participate in the labor market with full information or mobility' are often not aware of all opportun- ities, and are not likely to have access to all of them (Rees and Schultz, 1970~. Similarly, employers rarely have access to all possible employees and are often constrained by custom, agreements, and other factors. Major institutional constraints include internal labor markets (arrange- ments in which most positions are filled by promotion from within a firm); union agreements that determine hiring rules and pay rates; and the segmentation of labor markets into noncompeting groups, largely on the basis of the sex, race, and ethnic~ty of workers. While institutional economists acknowledge that wage rates observed in the market reflect the forces of supply and demand, they point out that supply and demand are themselves strongly affected by institutional factors. In this view, ng~dities and barriers to mobility characterize labor markets, and ine- qualities in wages between workers with similar qualifications doing similar work are endemic. Institutional analysis differs from the more conventional neoclassical analysis of the operation of labor markets in its emphasis on the im- portance of institutional features and their relative inflexibility in de- termining wages and other conditions of employment. In the judgment of the committee, the institutional view offers a more fruity perspective from which to understand the existence and the persistence of wage differentials between men and women, especially since, as we note in Chapter 2. attempts to explain earnings differences by productivity dif- ferences hare not been very successful. Our discussion? which focuses on institutional features, should not, however, be interpreted as a com- prehensive review of all approaches to understanding the operation of labor markets. The institutional view is only one view of labor markets, and is the subject of much controversy currently in economics and so- c~ology`(Doeringer and Piore, 1971; Wachter, 1975; Gin, 1976; Beck et al., 1~978; Piore, 1979; Hauser, 1980; England, 1982~. COMPARABLE WORTH AND INTERNAL LABOR MARKETS Issues invoking the comparable worth of jobs emerge most clearly and acutely in situations in which a single firm employs a large work force allocated among many different types of jobs. Such situations are furry common, since at least half of the U.S. work force is employed by large-scale employers: 20 percent of all wage workers are employed
~6 WOMEN. WORK, AND WAGES by federal, state, and local governments, and 40 percent (nearly half of those in the private sector) work in private establishments that employ 100 or more workers (U.S. Bureau of the Census, 1979:xxiii). The 500 largest industrial corporations (the Fortune 500) alone employed 16 million workers in 1979, about 45 percent of all workers in manufacturing and mining, averaging about 32,000 workers per company. The largest U.S. industrial employer, General Motors, has more than 800,000 em- ployees, and the largest nonindustnal employer, American Telephone and Telegraph, has about 1 million employees. For most of these workers the conditions of employment are deter- mined largely by administrative rules, promulgated by employers or negotiated by employers and unions (about a quarter of all U.S. non- agncultural workers are members of unions). Once an employee enters a large~scale establishment, he or she becomes part of an "internal" labor market in which job openings are usually filled from within and workers are usually deployed in accordance with established rules and procedures rather than in direct competition with workers in the '`ex- ternal'' labor market. For many workers, movement into (and between) large firms with highly organized internal labor markets is quite limited (Doeringer and Piore, 1971; Edwards, 1979~. Several theories seek to explain why internal labor markets are cre- ated. Doennger and Piore (1971) argue that because modern technol- ogies require specificity of skills and relatively long periods of on-the- job training at the employer's expense, employers attempt to minimize turnover and enhance the stability of the labor force by creating job structures within firms that reward longevity~uch as increments based on seniority and promotions up well-defined job ladders. Economic forces operate on employers, of course, but they operate mainly to encourage employers to m~nun~ze their training costs by minimizing turnover. Thurow (1975) suggests that higher wages are needed to "bnbe'' older workers to teach their skills to younger ones. Gordon (1972) and Stone (1975) suggest that the existence of many different job ladders, each with several rungs, may not be required by differences in skill levels but instead may serve to diode workers (particularly by race, cthnicity, or sex), thus minimizing the collective power of the workers and enhancing that of the employer. Edwards (1979) argues that man- agement creates internal job structures pnmanly D order to provide incentives for workers to perform their jobs, rather than because of skill requirements. Kahn (1976) provides an example of a union taking the Icad role in transforming a capncious, casual labor market in longshonng Into a highly structured one, primarily to reap such benefits as higher wages, sen~onty iDacases, and employment stability. According to all these explanations, job structures that entail many
Institutional Features of Labor Markets 47 rungs on ~ long ladder, each step requiring greater skill or involving more pa, and responsibility' are established. Only for entry-level jobs are wage rates strongly influenced by the competitive forces of supply and demand. The major supply for the jobs higher on the ladder are those workers already in the firm, and the only effective demand for those particular workers is that of their current employer. Some of the jobs have few, if any, analogues in the external labor market and no established market wage rates; rather it is the employer, and possibly workers, who determine appropnate wage rates for the jobs. If internal!, organized jobs were directly open to external market competition, employers would find little use for extensive compensation analysis (except perhaps for area wage surveys to determine the "going waged. The fact that many large firms use such analyses, particularly scat evaluation, to tonic internal fobs to particular "benchmark" jobs jobs that do have external markets and external wage rates (Treiman, 1979 - supports the notion that, within broad limits, such jobs are shielded from immediate external competition. Additional sources of institutional rigidity, arise from the prevalence of custom and tradition In setting wages (see, for example, Phelps-Brown, 1977; Wootton, 1955) and the relative immobility of workers between liens. Since access to higher-paying jobs comes only from within a firm, workers stand to benefit by remaining with that firm, generally do not seek jobs eise- where, and ma, be unaware of other opportunities. . . .. . . . . SEGMENTATION OF THE LABOR MARKET The existence of internal labor markets and other institutional features has led some researchers to the concept of labor market segmentation. The part of the economy characterized by highly articulated internal labor markets is called the primary segment. It consists of industries with advanced technology, large capital investments, unionization, sometimes a degree of insulation from competition In their product markets, or high profit~industnes such as petroleum, chemicals, heavy equipment manufacture, and utilities (Bluestone, 1970; Doennger and Piore, 1971; Edwards, 1979~. Industries that require highly educated workers, such as computer manufacture and service, insurance, and finance as well as most of the public service sector, also tend to have well-developed internal labor markets with established channels for ad- vancement and predictable work rules and hence are also part of the primary segment. The remaining jobs in the economy constitute what has been awed the secondary segment. These jobs have little Pupation from compet- itive market pressures and more closely approximate the textbook mode}
48 - ~ WOMEN. WORK, AND WAGES of a perfect market. In addition, they generally tend to have "low wages and fringe benefits, poor working conditions, high labor turnover, little cnance or advancement, and often arbitrary and capricious supervision" (Doeringer and Piore, 1971:165~. Such jobs tend to be found in highly competitive industnes with low capital investment, little unionization, and low profit~many service-onented industries and such manufac- tur~ng industries as textiles, garment making, and food processing (Blue- stone et al., 1973; Tyler, 1978~. The secondary segment includes many jobs requiring relatively little job-specific skill-attendant, guard, food server, sales clerk, stock clerk, messenger, and cleaning worker; sea- sonal jobs, particularly in agriculture; and increasing numbers of clerical jobs in typing' filing, and keypunching pools. For all these jobs, the most salient distinguishing feature is the relative insecurity or lack of an internal market structure that is, the immediacy with which these jobs are subject to external market forces (Edwards, 1979~. Although primary jobs are thought to prevail in large firms and in- dustries with a historical pattern of structured internal labor markets, and secondary jobs to prevail in enterprises with no formal internal labor markets' both types of jobs may exist side by side. For example, service jobs in hospitals exist alongside medical and nursing jobs; and janitonal and packaging jobs in large firms exist alongside skilled craft jobs. But in neither case do the secondary jobs feed into job ladders leading to the primary jobs. Jobs traditionally held by women teaching, nursing, and secretarial work have some features of organization that are more characteristic of a secondary than a primary pattern. Despite relatively high levels of skill, the wage levels of these jobs are often low. For example, in 1976 in manufacturing industries sampled by the Bureau of Labor Statistics, 12 of 26 clerical occupations were paid less than janitors (Ward, 1980~. Jobs traditionally held by women often have short job ladders. It has been noted that these jobs ohen involve general~zable skills that can be moved from one job to another, the acquisition of which is paid for by women themselves (Oppenheimer, 1970~. One plausible explanation of these attributes is that they have evolved because employers, believing women to have short job tenure and high turnover rates, have been unwilling to bear large training costs or to structure these jobs in such a way as to make their investments in training pay off in greater lon- gevity.~ ' O~cr job stmeturcs for these occupations are possible. Sccrc~ial work, for example, could be organized as a skilled craft with apprenticeships, catry~lc~cl jobs, and career ladders, pardadarly when firm-specific Wills "c important.
Institutional Features of Labor Markets TABLE 12 Distribution of Earnings in Selected Occupations in the Newark Metropolitan Area, January 1980 49 Earnings. Occupation Median Ranger Hourly. Straight Time Electncians. maintenanceS8.63 S6.40 14.00 Mechanics. maintenance (automotive)9.05 5.6~11.80 Tool- and diemal~ers8.78 6.0~11.80 Forklift operators6.69 3.5~9.30 Order fillers4.85 3.30.90 Weekly, Straight Time Stenographers. seniorS250.50 S140.0~340.00 Sccretanes. class A328.00 200.0460.00 Key entry operators class B180.00 120.00 340.00 Draftsmen. class A379.50 260.0600.00 Computer programmers (business) class A352.00 240.~.00 }registered industrial nurses291.50 180.0(~480.00 ~ Exclusive of premium pa, for overtime and for work on weekends, holidays, and late shifts. Lee lowest and highest rates received by the workers surveyed. SOUR - : U.S. Department of Labor, Bureau of Labor Statistics, 1980a. In contrast, it has been claimed that employers respond to the pos- sibility of men leaving their jobs after costly training has been invested in them by structuring the jobs in such a way as to discourage turnover and reward longevity. The consequence of this difference in the way "women's' and "men's' jobs are usually structured is that jobs that traditionally have been held by women often have more exposure to marketplace competition and provide less advancement with seniority and experience than jobs that traditionally have been held by men. Although institutional theorists disagree about some of the causes, a sizable (and growing) empirical literature presents a fairly consistent picture of labor market segmentation.2 Mat jobs are segmente~not only occupations is indicated by the range of wages paid for very sirrular work, a phenomenon that has been well known for some time (Wootton, 1955; Chamberlain and Cullen, 1971~. Table 12 shows considerable var- iation in a sample of occupational wage rates paid by surveyed estab- lishments in one metropolitan area, Newark, New Jersey. For this sam 2 The following summary of the empirical literature on labor market ~cg~DcDtation relics busby on Edwards (1979~.
50 WOMEN. WORK, AND WAGES ple the highest-paid workers generally received more than twice as much as the lowest-paid workers in the same narrowly defined occupations. Other research confirms the existence of high-wage and low-wage firms that tend to pay consistently high or low wages to all employees (Rees and Schultz, 1970; Blau, 1977; Ward, 1980~. This wide variation in wages renects an underlying structure of segmented labor markets. Early research posited the existence of segmented labor markets. Doennger and Piore (1971) found through extensive interviewing that decisions about hiring, pay, and termination are made in very different ways in the primary and secondary segments, the secondary segment being characterized by arbitrariness and the primary segment by sys- tematic procedures. Several historical studies suggest that in job search processes, workers in..the secondary segment, particularly minority workers, have been restricted by lack of information and in many cases by discnmination in access by employers. Minority workers in particular often perceive that only a narrow set of jobs is open to them, and they use very limited community networks to find jobs (Baron and Hymer 1968; Baron. 1971; Glenn, 19801. The appropriate way to define labor market segments quantitatively is a matter of considerable debate. Theory suggests that they should be defined by such characteristics as occupation, industry stability, auton- omy, unionization, and advancement opportunity. In practice, infor- mation on all these relevant characteristics is rarely available in most data sets. and researchers use various, not wholly satisfactory, indicators (such as wage rates and industry or occupation alone). One estimate suggests that roughly a quarter of the U.S. labor force hold jobs in the secondary segment and slightly more than half hold jobs in the primary segment~ivided into an independent primary segment (25 percent) and a subordinate primary segment (30 percent). About 20 percent of the labor force (e.g., self-employed persons, high-level managers) fall outside the schema (Edwards, 1979:166~. The empirical literature further suggests that various labor market processes operate differently in the two segments, although Cain (1976) has questioned the validity of findings of differences between segments when the Endings relate to vanables that are correlated with, or the same as, those used to define the segments.3 Several empirical studies suggest that training, mobility, schooling, and seniority have different payoffs in terms of wages in the different segments (Buchele, 1976; Rumberger and Carnoy, 1980~. Wages in the secondary segment typi 3 1iliS cntiasm is Woo applicable to Buchelc's study (1976), which uses job content vanables as the basis for deeding the segments, then proceeds to mves~gate the wage- sc~g process.
Institutional Features of Labor Markets Sl cally average 70 to 75 percent of those in the primary segment. Job tenure and employment stability are also less certain in secondary jobs. Earnings equations like those based on human capital models, suggest that the returns to years of schooling and experience are Virtually zero in the seconder, segment (Edwards, 1979:16~69~. In the primary seg- ment, by contrast, returns to schooling and experience are generally substantial. at least for men. Osterman found that the earnings of men In the independent primary segment increased 34 percent for each 10 years' additional experience; the earnings of those in the subordinate primary segment increased IS percent; and the earnings of those in the secondary segment increased only 4 percent (Edwards, 1979:175~. Os- tennan (1978) also found that each year of schooling increased a man's earnings by 10 percent in the independent primary segment, 6 percent In the subordinate primary segment, and only 1.5 percent in the sec- ondary segment (Edwards, 1979:175~. Similar differences have also been found by Buchele (1976) and Rllmberger and Carnoy (1980~. Such findings are consistent with the hypothesis that different labor market segments operate in different ways, but they are at odds tenth the general tenor of the neoclassical human capital mode! of the labor market. Neoclassical theorists recognize '`transitional" phenomena that result in flee labor market's being out of equilibrium. For example, computer personnel are "overpaid" until enough workers can train for the position: workers in new oil fields are "overpaid" until more workers amble. Moreover. jobs in rural areas may remain at slightly lower wage kvels than urban jobs because the difference in wage levels is not great enough to overcome the costs of search and relocation. `'Equal~zing"~ or "compensating ' wage differentials are also recognized by neoclassical theorists: risk`. dirty, or unpleasant jobs are thought to earn premiums (relative to others requ~nng similar skills) in order to induce workers to take them. Jobs that require long and costly training also command such premiums (Friedman, 1971; Robert Smith, 1979~. The neoclassical Mew recognizes the existence of noncompeting groups (men and women, for example) in the labor market, but their existence is seen as an anomaly, expected to disappear over time due to the forces of oom- petition. The difference between the neoclassical and institutional Mews thus turns on judgments regarding the importance and systematic nature of market imperfections, not their presence or absence. In the ~nstitu- tional view, such imperfections are seen to be logical outcomes of prod- uct and labor market processes, so large and so pervasive that they dominate the wage-setting process. In the institutional analysis, a worker's marginal productivity or worth to the employer is determined not only by the human capital be or she bungs to the market but also by the way his or her job Is structured.
52 WOMEN, WORK, AND WAGES The institutional approach considers as influences in the determination of wages such factors as the structure of the product market of em- ployers, the arrangement of jobs by emolovers. union ~^i,q1 intensity, and technological factors, In addition to human capital attn- butes. What a job is 'iworth" to an employer depends largely on how the employer chooses to structure it (given the constraints of industry, profit margins, and so forth) as well as on the customs and traditions of the particular workplace. Workers do not operate as individuals in the labor market, but rather as members of groups defined by their relationship to labor market structures, and labor market structures effectively limit the choices open to them: "Over significantly long pe- nods, job structures exist, and workers must live with them as best they can" (Harrison and Sum, 1979:694~. ~- - ~- ~- - - 7 ~ ~4 8 ~_~ V 1 ~ ~ 1 lOB SEGREGATION Job segregation by sex' race, and ethnicity is common in today's labor market. Women and men and minorities and nonminorities often work in different jobs. As Blau (1977) has demonstrated for several narrowly defined clerical occupations, even for occupations that are integrated by sex, men and women work in different liens, with men more likely to be found in high-wage firms and industries. Hence, even when oc- cupations are integrated by sex, the jobs men and women actually hold are segregated by sex.4 Because custom and tradition have in the past assigned subordinate social roles to minorities and women and because labor markets tend to incorporate, mirror, and perpetuate such roles, institutional theorists would expect minorities and women to hold low- paying jobs with limited opportunities. As we concluded in Chapter 2, the evidence on the differences in earnings between men and women suggests both that they cannot be satisfactorily accounted for on the basis of worker characteristics thought to affect productivity and that job segregation contributes to the lower earnings of women. Why women are concentrated in low-paying jobs is a crucial question still left un- resolved. Three different explanations have been offered: women choose, for reasons other than pay, jobs that turn out to pay poorly; women are excluded from high-paying jobs; and the jobs that women hold tend to be underpaid because they are held by women. If the latter two explanations have empirical support, we must conclude that em- ployment discrimination and wage discrimination exist. ' Beck at al. (197S, 1980a, 1980b) have reported empirical findings OR women's work and eam~gs ~ se~ncated labor markets (fee Heuser, 1980, for a cotiguc); for earlier worl: on WOmCD in }o~v-wagc u~dustnes, see Stevemon (1973, 1974, lg7S); for a historical trcatmcat, see HanmanD (1976~.
Institutional Features of Labor Markets CHOICE 53 It is sometimes asserted that women choose to work at certain types of jobs despite the fact that such jobs have relatively low pay rates. A vanety of reasons has been offered. First. women may be socialized to believe that some types of jobs are appropriate and that others are inappropriate for women; socialization may be so effective for some women that it net er even occurs to them to consider other types of jobs. Second, women may have pursued courses of study they thought par- ticularly appropriate to women and in consequence may not have the education or training that would suit them for other available jobs. bird, women may lack information about other available jobs, their pay rates, working conditions. and access to them. Fourth, women may be aware of alternatives. but because of actual or expected family ob- ligations may structure their labor force participation in particular ways. For example' tines ma' be unwilling to invest a great deal of time, effort. Or money in preparing for jobs because they do not expect to remain in the labor force after marriage or after childbeanng. They may be willing to accept lou-pasing jobs. or jobs with limited opportunities for advancement. and hold them until they marry and begin to raise child- ren. Or. in expectation of returning to work after their children are in school or grow n they may choose jobs that are easy to leave and reenter, jobs that do not require the continuous accumulation of skills and con- sequentls do not lead to significant increases in earnings with experience (Mincer and Polachek. 19-4. 1978; Polachek, 1976, 1979~. To accom- modate the dual demands of work and family responsibilities, women may choose jobs with limited demands restricted hours. no overtime work, no travel requirements. etc. Or they may defer to the demands of their husbands career advancement, moving with their husbands from place to place, etc. (Oppenheimer, 1970~. Some of these family- related factors may influence women's willingness to pursue advance- ment in their jobs. Fifth. women may be aware of alternative types of jobs but believe them to be unavailable or unpleasant because of dis- cnm~nation; their labor market preparation and behavior may be af- fected in man, Bass by this perception: the course of study they take; the time, money and effort invested in training; their willingness to accept promotion. etc. It is difficult to assess the relative importance of the choices women make in the labor market and of the factors affecting their choices.5 5 An extensive literature exists on some of the relevant topics, especially socialization. For usefb1 reviews see Maccoby and 3aciclin (1974), Mcdnicic at al. (1975), Scenzoni (1975), Duncan and Duncan (1978~' sod Laws (1979~; for bibliographies of these materials see Asiin ct al. (1971~. Bickner (1974), and Astin et al. (l975~.
54 WOMEN WORK, AND WAGES Many of these "choices" are adaptations to constraints of various kinds. Ethers are the result of complex somal-psycholog~cal processes. Women's choices observed so far may change (and may already be undergoing change) as women's participation in the labor market changes. More and more women are entering the labor market and remaining in it even when their children are very young.6 Their career expectations, their willingness to invest in training. and so on man be aff``rt~1 he tale changes. When efforts have been made to measure these factors, in an attempt to explain the difference in earnings between women and men, the results have been mixed. Using data on women ages 30 44 from the National Longitudinal Study (NI-S), Mincer and Polachek (1974) found that wages varied positively with work experience and negatively with work interruption, and they estimated that half the earnings gap between men and women could be accounted for by their different patterns of labor force participation. They reasoned that work interruptions would be important because women's skills would depreciate while they were out of the labor market. Polachek (1976, 1979) has suggested that women choose jobs the wages of which will be least affected by interruptions (because such jobs have low returns on experienced. Using data for women of all ages from the Pane] Study of Income Dynamics, however, Corcoran (1979) found that continuity of work experience did not seem to have a large effect on earnings differentials between men and women, and, in particular, that labor force withdrawals did not usually lower women's wager their skills did not in general atrophy or depreciate while they were out of the labor market. England (1982), using the NES data for women ages 304, attempted to test the Polachek thesis by estimating the amount of skill depreciation that women with intermittent labor force participation experience in different occupations. She found no corre- lation between rates of skill depreciation and the percent female in occupations, indicating that women probably do not choose their oc- cupations to minimize income loss, as Polachek suggests. a_ ~_~- v,~ Lea: ~ Lb recent years, the largest proportional inacasc m labor force participation has been among manned women: in 1947, married women's partic potion rate was 20 percent; by 1979, Marty 48 percent of all mamcd women were working outside the home (Ralph E. Smith, 1979 4~. Morco~cr, much of this mcrcax Evolves the mercasing participation of women traditionally least lilccly to wori~mamcd women living with their husbands and ninth young children. In 19SO, 11.9 percent of Ted women with husbands and with children under SLY years old were employed; by 19~, about 4S percent were employed ITS. Dcpartmcot of Labor, Bureau of Labor Statistics, 1977:Tables 18 and 22; Bureau of Labor Statistics DEWS "kasc 8~767, Dcc. 9, 1980~.
Institutional Features of Labor Markers EXCLUSI ON' 55 A second explanation of why women are concentrated in low-paying jobs may be that tines are simply excluded from high-paying jobs. While this needs to be established on a case-by-case basis, there are now a large number of documented cases of denial of occupational opportun- ities on the basis of sex (Schlei and Grossman, 1976~. Historically, women were prohibited from certain jobs by laws designed to protect them from exploitation (see U.S. Department of Labor, Women's Bu- reau. 19691. Store recently, in several major cases the courts found that women were excluded from higher-level or higher-paying jobs (EEOC, 1972; Osterman 1978~. Such exclusion may take the forth of denial of employment Or of restriction in opportunities for promotion. While it Is difficult to establish discriminatory intent on the part of employers, it is possible to study the pattern of employment within firms to determine whether men and women are allocated to jobs in a manner consistent with their qualifications. A study by Malkie! and Malkie! (1973) does this. Using data on professional employees of a large cor- poration, they showed that adjusting earnings on the basis of a very good measure of job-related labor market experience plus measures of post-high-school education. rate of absenteeism, marital status, college Seld of study. and personal productivity (measured he number of pub- lications) increased the female/male earnings ratio from 66 percent to beween 75 and 89 percent (depending on the year and type of decom- position used). Adding a 13-categor, index of job level to the regression resulted in an adjusted earnings ratio of almost 1.0 (98 percent), which led the authors to conclude that in this firm there was no discrimination in the form of unequal pay for equal work but that there was discnm- ination in occupational assignment~pecifically, that women were as- signed to tower-lesel jobs than men with the same qualifications. Studies by Talbert and Bose (1977) and Halaby (1979) also conclude that the louder gages of women result from discrimination. Talbert and Bose found that among retail sales clerks, men are more likely to be assigned to "big ticket. departments (furniture, large appliances, etc.) and hence to earn more. Halab>. using 1960 data for managerial per- sonne] of a Verdi large California-based public utility firm,' found that female nnanagers earned about two-thirds of what male managers earned and that adjustments for differences between them with respect to cd- ucation, seniority, previous labor force experience, and number of pre ~ Managenal personnel according to the company definition includes all supervisory personnel. Hence. the sample includes some individuals who would fall into the clerical Ed manual catcgones of the U.S. census classification.
S6 WOMEN. WORK, AND WAGES TABLE 13 Decomposition of Earnings Differentials of Men and Women Among Managerial Personnel in a Large Public Utility Company Component Percentage Dollars Human capital Diffcrcoces In mean levels S.7155 Differences in rate of return 17.1466 Shared - .7-19 TOTAL =.1~2 Position structure Diffcrcnces in job composition 43.81,194 Differences in rank 34.1930 TOTAL . ~.92,124 TOTAL ~ BURN 1~.02,726 SOURCE: Adapted Tom Halaby, l979:Table 9. pious employers accounted for only about 13 percent of the difference in earnings. At any given level of human capital' according to Halaby~s formulation, women tend to be concentrated in lower-level occupations in the firm's hierarchy and at lower ranks within occupations, at sub- stantial detriment to their earnings. Table 13 gives Halaby's decom- position of the total difference in earnings between men and women into portions due to what he considers human capital variables and what he considers job-related factors. More than 20 percent of the difference is due to human capital variables, holding constant jobs and ranks: most of that difference occurs because women get a lower return on these human capital variables than men do; only about 5 percent of the dif- ference occurs because women and men differ in their average levels of human capital. The rest of the earnings gap arises from the fact that women have lower-level positions when human capital is held constant. UNDERPAYMENT OF WOMEN'S WORK The third explanation for the lower pay rates of jobs held mainly by women is that women's work is underpaid because women do That is, that the same work would be paid more if it were done by men. Again, the evidence is sparse, but in several documented cases pay rates were shown to have been influenced by the sex composition of the work force as well as by the content of jobs. The cadence consists of a demonstration that employers Notated their own criteria for determining
Insantnonal Femures of [Labor Markers 57 the re}ati`,e monetary worth of jobs, in order either to implement an explicit decision to pay women or minority workers less than men or whites or to conform to an external standard for establishing pay rates. In both cases the evidence comes from an examination of the job eval- uation procedures used by employers. Job evaluation procedures, which are discussed more fully in Chapter 4 and also in the committee's interim report (Treiman, 1979), involve rating a group of jobs faith respect to a set of "compensable factors," features of jobs tI}at an employer regards as important in setting pay rates. Compensable factors usually include indicators of the skill, effort q and responsibility required by a job and the nature of the working conditions involved. In most firms that use job evaluation procedures the sum of scores on the set of rated factors is used to assign jobs to pay classes. In this context we would regard jobs held mainly or entirely by women as underpaid if their pay rates are lower than those of jobs with the same scores held mainly or entirely by men. Several examples exist of fimns that pay, or have paid, jobs held mainly by women less than jobs with similar job evaluation scores held mainly by men. Some of these pay practices antedate the passage of the 1963 Equal Pas Act and the 1964 Civil Rights Act; they were not illegal then and were in fact widely accepted. Such pay practices would be of historical interest only but for the fact that, in many cases, they continue to govern pay differences between jobs. For example, in 1945 the War Labor Board found. and the companies agreed, that both the Westing- house Electric Corporation and the General Electric Company had re- Juced the pas rates of production jobs held by women to a level below those of jobs held by men Westinghouse by 1~20 percent, General Electric by about 33 percent even though the jobs had similar job evaluation scores (lS'ewman, 1976~. In the late 1930s. General Electric and Westinghouse had established job evaluation systems for the purpose of standardizing wage rates throughout their plants. Most of the production jobs were entirely seg- regated by sex during this period and were known as either female or male jobs. At one Westinghouse plant, female jobs were assigned to five labor grades. 1 through 5; male jobs with the same job evaluation scores as the female jobs were assigned to parallel labor grades, 1 through 5; and five additional labor grades were specified for male jobs with higher scores (see Table 14~. The wage rates for these male and female labor grades were not' however, parallel. In fact, the hourly wage rates were established in such a way as to pay all of the female labor grades less than the lowest male labor grade, despite the fact that the parallel male and female grades represented jobs of comparable
58 WOMEN. WORK, AND WAGES TABLE 14 Wage Rates at a Westinghouse Plant. 1943 Evalua Proba- Qualify- tion tionary ing Standard Point Labor Grade Rate Rate Rate Range Female 1 S0.615 S0.645 S0.675 0 49 2 0.645 0.675 0.705 50 62 3 0.675 0.705 0.735 6~78 4 0.70S 0.735 0.765 7~98 S 0.735 0.765 0.795 9~125 Male Common labor S0.785 ~37 1 (6) SO.?85 S0.785 0.815 38~9 2(7) 0.785 0.815 0.845 5~62 3(8) 0.815 0.845 0.875 6~78 4 (9) 0.845 0.875 0.905 7~98 S (10) 0.875 0.905 0.955 99 123 6 (11) 0.905 O.9SS 1.055 12~154 7(12) O.9SS 1.055 I.OSS 15~199 8 (13) 1.055 1.055 l.lSS 199 239 9 (14) 1.055 1.155 1.25S 24~299 10 (15) I.1S5 1.255 1.405 300 SOURCE: Adapted from Newman. 1976:268. worth according to the company's own criteria. Only the male common labor classification paid less than the highest paid women's jobs. This rank ordering of labor grades and pay differentials remained essentially unchanged until the separate series for male and female jobs were abol- ished in 1965. When these separate series were merged into a single series in 1965, however, the male grades 1 through 10 were simply relabeled 6 through 15, so that the sex differential in pay was preserved. The labor grades were not combined in such :' I: ac In ·-fl^~t t1 original evaluation of these jobs. A similar situation existed at several General Electric Company plants. As a result of several lawsuits brought by the International Union of Electrical, Radio and Machine Workers (lUE), the pay rates of some jobs held mainly by women have been raised. A selection of the job titles compared in one case is shown in Table IS; similar jobs were matched and the issue in contention was their relative worth. As the table shows, the jobs were almost entirely segregated by sex, and the highest paid women's job was paid less than the lowest paid men's job. As a result of the settlement, these women Is jobs were generally raised several grade levels, although only four of them were judged to be of _ _ ~ ~ ~ V · -~ A
Insatutional Features of Labor Markets 59 fully equal worth to similar men's jobs. After the settlement, the union negotiated wage increases for many more jobs held by women, which were not as readily comparable to those held by men but were com- parable to some of the moments jobs that had been raised.8 A different kind of evidence comes from a situation in which job evaluation procedures were used explicitly for the purpose of assessing the existence and extent of pas discrimination based on sex. In 1974 a study was conducted of state government jobs in the State of Washington (Willis, 1974, 1976~. A claim had been made that the existing practice of pegging the pay rates of state employees to prevailing rates in the pirate sector was discriminatory because the private sector traditionally underpaid jobs held mainly by women. To assess this claim, the study compared the pa, rates of jobs held mainly by women with the pay rates of jobs of "comparable worth" held mainly by men. A total of 121 positions, in which at least 70 percent of incumbents were of the same sex, were chosen for evaluation. The job evaluation plan used, which was developed by Norman Willis and Associates of Seattle, was sub- stantialIy similar to the Hay Associates plan (see Treiman, 1979:21-23~. Of the 121 positions. 59 were filled at least 70 percent by men and 62 were filled at least 70 percent by women. Figure 2 shows that in the State of Washington. jobs held mainly by women were paid substantially less than jobs of comparable value, as defined by the job evaluation fo~.ula, held mainly by men: the pay rates for the jobs held mainly by women averaged about 80 percent of the pay rates for jobs with the same number of job evaluation points held mainly by men.9 'Newman (1976:271) ~rites: Me psychology here is interesting. .6Jane, ' a grade 11, did not compare her job with ..John s ' grade 14 job; instead she compared herself with "Sandra," a grade 8 who vitas a plaintiff in the lawsuit. as a result of which "Sandra" also became a grade 11. ..Jane" was sure she was worth three grades more than "Sandra"~but did not know she was worth as much as "John," who bad been performing a "man's" job. 9 For this study it is possible to apply Bimbaum's test (scc Chapter 2, note 17) to confirm that the underpayment relative to their worth, of jobs held by women is not a statistical artifact. For the 121 occupations included in figuec 2, let X equal job worth points, Y equal the average monthly salary, and F equal 1 for the 62 jobs held mainly by women and O for the 59 jobs held maim, by men (computed from Willis, 1974~. For these occupations pYF X iS - . - and IF y is .832. Mat is, it is tic that (1) ~1 - g - extant job worth. jobs held maim, by women are paid less than jobs held maidy by men (the estimated diffcrcace~he metric coeff~cicat of Finis Sl75 per month, a bit more than 20 percent of the average salary for all jobs) and that (2) holding CoDs~t Francs, jobs held mainly by women score higher in terms of job worth points man jobs held mainly by men. Vising Birnbaum's criterion, discrimination is clearly present.
60 ;^ - - ~: - - A, r - D .o es ~ oo _ O O' Ed Os I: - 6 5 MU 3 E D _ ~ .~ ~ ~_ _ L - - - - - .~ o it 3 A, Z ~ - - o .- eii O I ' _ 8 6E; 9, ~ ~ ~ ~ ~ ~ r -~01~8 ·8 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ en ~ Hi ~0 e to ~ to ~ oo ~ e "E _ _ ______ ~ . ~.~.~ ~ 3 e m.> oo ^.0 j 6 -6 ~ ~ ~ ~ t _ _ | ~ [ ! I I e _ . - ~m =~e - ~~~
Insatunonal Features of Labor Markets O For jobs held mainly by non i ~ 1.87{X} + 473 · For jobs held mainly by women i" 1.40IXI+393 1,600 1,400 t,000 800 _ t' ~04 600 400 0 /< a/ o 0 0 c)~ _~ ''c ~ S. O Loo ~° Ado ~- ~; I n I I I ~I 200 300 400 500 sm X=JobWorth Points FIGURE 2 Scattcrplot of monthly salaries by job worth points, for S9 jobs held maims by men and 62 jobs held By by women in the Washington State public service. SOURCE: Computed from Willis, 1974. 61 One of the features of wage differentials between jobs held by men and by women is that they become customary, accepted as the standard rates for the jobs. Evolving wage structures~in a community come to reflect these differentials, which are passed between internal and ex- tcrnal labor markets. Area wage surveys reproduce these differentials. If institutional factors such as those reviewed here do operate in the labor market, then relying solely on area wage rates to establish pay rates for a particuisr find will incorporate those differentials into that firm's wage structure. Those differentials will be based on all factors, including productivity and discrimination, that create differentials in earnings between women and men in the wider market. By use of the "going wage'' as a standard to set pay rates the wages of a (nondiscri- moating firm will be biased by the discrimination of other firms In the market. In the State of Washington case renewed above, the cause of
62 WOMEN. WORK, AND WAGES the "underpayment" of jobs held mainly by women was not a result of an overtly discriminatory act on the part of the employer but simple conformity to the prevailing pay rate of the private sector. DISCRIMINATION IN LABOR MARKETS Although the institutional approach to understanding labor markets is not itself a theory of discnmination, it does view labor markets as inherently rigid and balkanized. The approach thus provides a basis for viewing discnmination as an integral part of both labor market processes and their outcomes. The labor market, incorporating a range of political, social, and economic forces, is seen to create an institutional context in which groups with different interests attempt to stabilize or enhance their positions. In this context discnmination can be understood to be one of the important mechanisms that have contributed historically to the creation of segmented labor markets. In addition, the ways work is organized and the associated patterns of productivity and wages in both the secondary and the primary segment~incorporate and repro- duce patterns of unequal access and disparate rewards for different groups of workers. Rigidities such as custom, tradition, barriers to mo- bility, and administrative rules tend to prevent change and to reinforce established patterns. Thus, hierarchy and inequality, including discrim- ination, are seen to be part of labor markets. In the institutional view of labor markets, discrimination would be expected to take the fonn of segregation into different jobs rather than of lower pay for identical jobs (BIau and Jusenius, 1976; Blau. 1977; Stevenson, 1978~. This is so because internal labor markets in large firms typically function with the aid of work rules that are bureaucratically enforced and uniformly applied and administered; because these rules specify that everyone in the same job category should be treated simi- larly; and because sex, race, and ethnic differences have recently come to be perceived as unfair (and indeed illegal, with the passage of the Equal Pay Act and the Civil Rights Act). r D We must stress, however, that in an institutional framework neither the fact nor the form of discrimination is inevitabl~in fact, important changes have occurred within the past two decades. Minorities and women in earlier times were subjected to discrimination of venous forms: lower pay for doing work identical to that performed by non- m~nonty men; de lure and customary exclusion from some jobs; inad- cquate cUucation because of segregated (and unequal) schooling; and ~ facto exclusion from some jobs because of residential segregation. While none of these forms of discrimination Is absent from contemporary
Insatutional Features of Labor Markets 63 American society, their importance has been reduced by legal, social, and political intervention. The institutional view of labor markets helps to explain how discnm- mation can occur and persist through job segregation that confines women and minorities to low-wage jobs. There are, of course, other News of how discrimination occurs within or through labor markets, although no economic theory attempts to explain the origins of discum- u~ation. Some theorists rely on the concept of "tastes," people's pref- crence for working with or hiring their own kind (Becker, 1971~; others suggest economic advantage is the motive, either of white male workers (Bergmann, 1974; Madden, 1975) or of employers (Reich, 1978~; still others suggest that discrimination occurs because it minimizes the cost to employers of doing business (Phelps, 1972~. Most agree on the pn- macy of job segregation as the mechanism of discrimination. Moreover, since the persistence of discrimination over time cannot be explained by conventional models of the labor market (which posit perfect mo- bility, information, and access), most theories of discrimination incor- porate departures from these assumptions in an attempt to better rep- resent the actual behavior of labor markets. In one of the earliest models of discrimination, Gary Becker (1957, 1971), incorporating a central element of neoclassical theory, relied on "tastes" of employ ers, employees, or customers as the motivating force behind discriminatory behavior. He suggests that people's tastes for discrimination can be satisfied or bought off at certain paces. For ex- ample, if whites do not want to work with blacks, they can be encouraged to do so only by being offered wages that are high enough to offset their distaste for working with blacks. If employers do not want to hire blacks, however, the employers can be encouraged to do so only by hiring blacks at wage rates low enough to compensate for their distaste. If consumers do not want to buy products made or served by blacks, they mI] buy them only if the prices are low enough to offset their distaste. Becker's hypothesis is that in such cases resources are being allocated inefficiently, and competitive market forces would be expected to eliminate the wage and price differentials over time. In another influential mode' of discrimination, Barbara Bergmann (1971, 1974) applied the concept of overcrowding to race and sex dis- crimination. Her mode} is based on the assumption that the distaste for hiring minorities and women is so strong that employers exclude them from many jobs. This has the effect of increasing the supply of labor available to fill the jobs that are open to women and m~nonties, which In turn Knees the market wage for these jobs down relative to what it would be if there were no restrictions on occupational opportunities.
64 WOMEN. WORK, AND WAGES Similarly, the wages of favored groups are higher than they would be otherwise because of their relative undersupply. Again, in this model resources are being inefficiently allocated; the free movement of workers would result in more total product. In postulating a bifurcated labor market Bergmann departs somewhat from the neoclassical concepts of perfect competition, mobility, and information' in that in her mode] men and women and blacks and whites are not able to compete with each other for jobs. Yet, just as in Becker's model, discnmination is expected to disappear over time because any employer who does not exclude minorities or women would have a competitive advantage over those who incur higher production costs by excluding them. Phelps (1972) and others have developed the concept of statistical discrimination. This model, like Thurow's queuing model (1969, 1975) and early models of the dual labor market (Doeringer and Piore, 1971), postulates that to minimize training and turnover costs employers at- tempt to find the most productive workers for jobs requiring stability and long or costly training. If employers believe that women and mi- norities are less productive or have higher turnover rates, they will not hire them. They do not necessarily have tastes for discrimination; they are simply minimizing the costs of screening prospective employees by summarily eliminating those they think are likely to be unproductive or costly workers generalizing about individuals on the basis of their per- ception of group characteristics. Whether or not these workers are less productive, they are barred from the opportunity to take `'good" jobs. The mode! recognizes that employers do not have perfect information and that obtaining information incurs costs. Lee workers thus excluded take other jobs, for which turnover costs are perceived by employers to be less important. Or, if employers hire such workers for good jobs, it is at wage rates low enough to compensate employers for their ex- pected higher costs. Of course, when employers' beliefs are so erroneous that the costs of screening would be more than compensated by the quality of the workers erroneously excluded, discrimination would be expected to abate or disappear over time, because employers who do not hold such erroneous beliefs would have a competitive advantage. As Cain (1976) has noted, the human capital approach, the approach most commonly used in developing models to measure the extent of discrimination In the labor market, does not provide information about the possible mechanisms of discrimination. Such models attempt only to measure the results. As we note in Chapter 2, these studies indicate that women earn much lower returns OR their human capital than do men, and that the earnings differential] cannot be explained by sex dif- ferences In characteristics of workers thought to affect their productivity.
Insanmonal Features of Labor Markets 65 Despite the inability to measure precisely the venous sources of wage differentials and the plethora of postulated mechanisms of discumina- tion, researchers agree that job segregation is an important source of the difference in wages between men and women. In our view, although the concentration of women in lower-pay~ng jobs exists at least in part because of women's choices, it also results from the exclusionary prac- tices of employers and from the systematic underpayment of jobs held mainly by women. Wages are higher in some jobs and lower in others than they would be in the absence of job segregation. In particular, the wage rates of jobs traditionally held by women are depressed relative to what they would be if women had equal opportunity in the labor market. CONCLUSI ON The main conclusion of our analysis of labor markets is that observed market wages incorporate the effects of many institutional factors, in- cluding discrimination. This conclusion has three corolIanes. First, mar- ket wages cannot be used as the sole standard for judging the relative worth of jobs. Second, in order to end discrimination, policy interven- tions to alter market outcomes may be required. Third, because of the complexity of market processes, actions intended to have one result may well turn out to have other, even perverse, consequences. The first corollary requires no additional comment. With regard to the second, we simply note that over the years a variety of strategies designed to alter market outcomes have been proposed and imple- mented. These strategies include programs to encourage additional schooling and job training; programs to encourage the employment of the disadvantaged; improvements in fob information and job search techniques; equal opportunity legislation designed to eliminate the dis- cominatory practices of employers, labor unions, and employment agen- aes; and programs to encourage women and minorities to train for and enter untraditional jobs. In addition, the coverage of the federally man- dated minimum wage has been expanded over the years to protect workers in many of the lowest-paid jobs, such as those In domestic service, agriculture, laundries, restaurants and hotels, and fast food chain~jobs in which women and minorities are overreprcsented. His- toncally, organization by both employees and employers and regulation by government agencies have been the methods most often used to attempt to alter market outcomes. With respect to the third corollary, any policy of intenren~don requires that full consideration be given to the complexity of labor markets and
66 WOMEN, WORK, AND WAGES to all the forces that influence market outcomes. Because so many factors are influential, policies to alter outcomes that focus only on selected factors may not have the intended effects. Equal employment opportunity programs, for example, focus on the demand side of the market by attempting to influence employer hiring, promotion, and compensation policies. But if only some employers are affected by the regulations, those with government contracts, for example, they may hire minorities or women away from employers without government contracts, and the net effect may be that the employment of minorities or women has not increased as much as was expected (for an example in construction see Flanagan, 1976~. Similarly, if government efforts succeed either in raising the wages paid to minorities and women above the levels employers think are warranted or in placing minorities and women in jobs that employers think are ~nappropnate for them, em- ployers may substitute other factors of production. Other changes in the economic environment faced by the employer may also occur simulta- neously with affirmative action efforts and may have countervailing effects. For example, one study claims that since the well-public~zed consent decree involving American Telephone and Telegraph, the rate at which technological change has displaced women workers appears to have increased more rapidly than previously predicted by the company (Hacker, 1978~. Strategies that focus on the supply side of the market are similarly limited if they are not implemented in connection with complementary strategies. Women and minority workers can be trained in new fields, for example, but if employers refuse to hire them, market outcomes will not be altered. Strategies that focus on improving the operation of labor market~by increasing the information available about jobs, for ex- ample assume that both demand and supply are adequate so that once workers and employers find out about each other they can come to mutually beneficial teens. It must also be recognized that in our econ- omy not everyone can have a "goon,' high-paying job. Our economy generates low-wage jobs as well as high-wage jobs; attempts to prevent their being filled in this country may simply result in the exportation of low-wage jobs. The complexity of the labor market does not mean, of course, that market outcomes cannot be altered. It does mean that no single type of policy is likely to be effective by itself and that any strategy to alter outcomes in one part of the labor market must take into account the likely consequences in other parts as well as the structure of the economy itself. The com~uttec is convinced by the evidence, taken together, that women arc systematically underpaid. Policies designed to promote equal
Institutional Features of Labor Markets 67 access to all employment opportunities will affect the underpayment of women workers only slowly. Equal access to employment opportunities may be expected to be more effective for new entrants than for estab- Ushed workers and more effective for those who have invested less In skills than for those who have invested more. Since many women cur- rently in the labor force have invested years of training time ~ their particular skills (e.g., nursing, teaching, librananship, and secretarial work), access to other jobs (e.g., physicianship, plumbing, engineering, or sales) may not be preferred. For these reasons the committee believes that the strategy of "comparable worth," that is, equal pay for jobs of equal worth, meets consideration as an alternative policy of intervention in the pay-setting process wherever women are systematically underpaid. The viability of a strategy of paying jobs in accordance ninth their '~worth" requires, first, that an appropriate mechanism, other than cur- rent market wage rates, can be found to measure the relative worth of jobs to an employer and. second, that wages commensurate with worth can be set and paid by the employer. In cases In which wage costs, productivity, profitability, conditions in product markets, or economic growth permits the new costs to be absorbed, this strategy provides a direct method of redress for wage discrimination due to occupational segregation. Since much of the wage differential arises because women work in low-~'age firms and men work in high-wage firms, however, even a comparable worth approach, applied to single firms, would not entirely eliminate the differential. For this reason and because, given the complexity of labor markets' a comparable worth strategy may have unanticipated and unintended effects, it cannot be viewed as a panacea. Raising the wages of jobs held by women through a comparable worth strategy might have various effects. On one hand, employers might alter the nature of the jobs women hold in order to encourage longer job tenures and lower turnover rates, thereby reducing costs and making their investments in women pay off. Higher wages might also encourage employers to offer women more on-the-job training and skill enhance- ment programs. On the other hand, a comparable worth strategy might reduce employment either because employers shift to alternative, less labor-intensive methods of production or (if the new labor costs were paid and passed on) because consumers might switch to other, less expensive goods or services.'0 We want to point out, however, that the to Gregory and Duncan (1981) investigated the relevance of labor market segmentation theory to Australia s recent efforts to increase the wages of occupations filled mainly by women. They suggest that the wage increases did not negatively affect the numbc' of women employed. in part because many employers of women were sufficicDdy insulated Tom competitive market forces to absorb the higher costs.
68 WOMEN. WORK, AND WAGES strategy of comparable worth is conceptually similar to earlier policies of directly altering wages by raising them above previous market rates, such as the minimum wage and overtime premium provisions of the Fair Labor Standards Act and the Equal Pay Act of 1963. Employers were certainly able to hire the workers they needed at rates lower than those prescribed by the new provisions; they were, however, required by law to pay more. Economists are still debating the merit of these provisions and attempting to identify and measure their effects on the amount and terms of employment (Lev~tan and Belous, 1979~. While further study of the possible effects of a comparable worth strategy is certainly required, the committee believes that policies de- ~sed to alter pay structures so that jobs are rewarded in a nondiscri- minatory manner that is, commensurate with their demands and re- quirements rather than in ways based on the sex, race, or ethnicity of those who hold the~are clearly desirable. It is not, however, an easy task to ascertain for any particular job whether its pay rate includes discriminatory elements. Chapter 4 discusses the potential usefulness of venous proposed procedures for identifying and eliminating discrimi- natory elements of pay differences among jobs within an individual firm.
