A question that emerges after completing any empirical analysis is the degree to which the immediate findings are representative of larger social regularities. Every dataset from the narrowest case study to the largest set of cross-national statistics is going to have some quirks and idiosyncrasies that will introduce a degree of bias. Therefore it is important for any sociologist to step back from her material at the end of her investigation and ask what can be inferred about her dependent variable in general. This particular analysis has been a case study that has attempted to draw some larger theoretical conclusions about the determinants of women’s economic opportunities. Most of the material has come from two relatively unusual firms, the Great Western Railway and the British General Post Office, two Victorian British clerical employers who were extremely large, monopolistic, and unionized as well. Other material came from a larger consideration of the history of office work. These analyses are free from the biases associated with the GWR and GPO, but are still colored both by the limited time frame of the analysis, in this case the years 1870 to 1940, and by the peculiar problems of clerical work, which are different from those of other occupations. Studying a sample of occupations from 1970 to 1980 would hardly have eradicated these problems, since no sample of occupations can ever be complete, and the modern era with its legal changes and its tumultuous transformations in gender roles is certainly no less idiosyncratic than any other. Nevertheless, some explicit attempt should be made to place this study’s findings in a larger context and to speculate on how they relate to an economic structure that is filled with work situations that are totally dissimilar from early twentieth-century offices.
In the present analysis, a fairly large number of arguments have been made on a variety of different topics. It is important to summarize these coherently and identify the major themes that bind together the argument before integrating them into a larger theory of occupational sex-typing. Then these principles will be applied to two larger questions: first, why did clerical work feminize? and, second, how does white-collar work differ from blue-collar work and why are women disproportionately concentrated in the former? The study of the GWR and the GPO suggests answers to both these questions and these hypotheses parallel each other to a significant degree. Third, the conclusion explicitly considers what kind of variables might be significant to a theory of sex-typing that might be obscured by a study of Victorian offices. In particular, the roles of law and of organizational vacancies are explored, although the analysis remains strictly at the level of speculation. Lastly, the question of women’s concentration in low-status jobs is addressed. While this study provided some answers to this question, the theoretical structures that have been suggested cannot account for the universality of female subordination. An alternative hypothesis is put forward that is not immediately and obviously testable but is consistent with the domination of men over women in settings where the synthetic turnover hypothesis would not apply.
The Main Findings of the Analysis
Buffering from Labor Costs
In Chapter 2, it was shown that an important determinant of the percentage of women in a firm is the capacity of an employer to tolerate discrimination. This capacity to tolerate discrimination is affected by many different factors. Discrimination is less likely to occur in firms which are labor-intensive and, all other things being equal, is less likely to be tolerated among occupations that make up the bulk of a firm’s labor force. Thus, firms and offices that are clerical-labor-intensive are more likely to use women clerks, while firms and offices that are non-clerical-labor-intensive are more likely to use male clerks. Organizations that are under pressure to maximize profits are less likely to tolerate discrimination than are organizations whose success is determined by other considerations. In the Post Office, offices whose political duties freed them from the necessity of cost-cutting were more likely to hire male clerks than were offices that were administered strictly by criteria of economic efficiency.
Women were generally confined to low-status positions within each firm. Nevertheless, in these cases, the overall feminization of clerical work cannot be attributed to an increase in the percentage of low-status positions. Real clerical salaries did not decline in this period, either in Britain as a whole or in these particular firms. Furthermore, neither firm experienced a decrease in the ratio of promotions to entry-level jobs. Although there were dramatic changes in the technology of clerical work, individual innovations varied in their implications for the substantive complexity of office work. Thus, in the aggregate, technological change had little net impact on the overall distribution of skills, since the various reforms tended to cancel each other out.
The exclusion of women from high-status positions in these firms cannot be explained using human capital theory in either its pure or its modified forms. Human capital theory argues that women’s anticipated careers are too short to allow them to be profitably trained for skilled or supervisory positions. In these firms, women worked for an average of twelve years, a period sufficiently long to justify training them for nearly any position. Furthermore, management seems to have been less concerned with conserving scarce human capital than with artificially increasing turnover. In firms where there is a preexisting supply of skilled labor, where employees do not significantly increase their productivity over time as they gain experience, and where these employees are paid on tenure-based salary scales, there will be a strong incentive for employers to prefer newer workers to older workers, since newer workers will be considerably less expensive. Under such conditions, employers will attempt to maximize their quit rates by hiring employees whom they believe are unstable or, if this fails, by imposing tenure bars. Both the Great Western Railway and the General Post Office used marriage bars. These policies were explicitly defended by management as ways of keeping wages low and promotion chains fluid. Similar policies were instituted by other white-collar employers, as was shown by surveys of firms in the United States, Britain, and other developed nations. White-collar employers were particularly prone to using marriage bars because they paid with tenure-based salary scales.
Women are often excluded from particular jobs on the basis of organization-specific policies whose primary functions seem to be patriarchical in nature. Although the initial motivation behind the implementation of these policies may have been sexist, their persistence depends on the degree to which they can be made compatible with the requirements of efficient organizational and economic functioning. Two specific examples on the GWR and GPO were night work bans and sex segregation. Both policies were initially implemented out of a managerial concern with traditional sex roles. The ban on night work survived because it forced a significant proportion of men to serve late shifts. This facilitated the functioning of internal labor markets in each firm by intensifying the competition over day jobs. Sex segregation did not survive in most firms, since it seriously impeded communication between employees working on closely related jobs. It survived in the Post Office only due to a rare set of conditions that allowed the GPO to make extraordinary adjustments to the demands of such a regime. Furthermore, once the policy had been in place for a long time, eliminating sex segregation might have allowed experienced women to be promoted over younger high-level men.
The patterns of feminization that occurred in the Great Western Railway and General Post Office cannot be explained in terms of employers’ being prevented from hiring women due to pressure from sexist male employees. Most unions were not sufficiently strong to be able to force management to make concessions on sex issues. This condition almost certainly applied to the white-collar unions on the GWR and the GPO. Nearly every explicit demand concerning women’s employment that was made by these unions was disregarded by management; this applies both to those demands for the exclusion of women that were made by the sexist unions and the demands for the preservation of female employment made by the feminist unions. Informal employee discrimination does not appear to have affected hiring either. Strong workers should have been more capable of informally excluding women than should weak workers. Nevertheless, there was no correlation between the percentage of female clerks and the bargaining strength of male workers.
Women as Labor Control
The patterns of hiring in these two firms can not be explained in terms of the use of women to control male white-collar militancy, since there seems to have been no tendency for women to be less strike-prone than men. The telegraph service of the Post Office attempted such a strategy in 1870 by filling strategic offices with women clerks. However, the female strike rate was essentially the same as the male strike rate, and the Post Office abandoned feminization as a strategy of labor control. Women were no more prone to docility on the Great Western Railway; after controlling for occupational composition, the differences between men and women in their participation in the General Strike of 1926 were trivial.
Alternative Sources of Secondary Labor
Both the Great Western Railway and the General Post Office increased their use of female clerks because of difficulties associated with the use of other sources of inexpensive labor. In the nineteenth century, duties that are now given to female clerks were given to temporary adult male clerks and boy apprentices. Male wage clerks fell into disfavor during the twentieth century because they lacked the human capital required for office work. They were doomed by the rationalization of office management that emerged during the Taylorite era. High human capital requirements also reduced the demand for boy clerks. Firms were only willing to use those boys with relatively high levels of intelligence and education. Victorian firms attracted such individuals by offering them full adult positions after they reached the age of eighteen. However, there were demographic limitations on the percentage of positions that could be given to boys, because of this obligation to absorb male graduates. Rapidly growing firms had no absorption problems, but as growth rates decreased, firms would be obligated to turn elsewhere for their cheap labor. This problem was exacerbated by a growing scarcity of qualified boys as higher education and introductory lower managerial positions drew off boys who previously would have taken clerical positions. Feminization in both the GWR and the GPO was accompanied by some sort of limit on the availability of acceptable juvenile job candidates.
The preceding analyses have important implications for two related larger questions in the study of occupational sex-typing. The first question is why clerical work changed its sex-type from male to female. The traditional explanation is that clerical feminization occurred as a response to several different social and economic changes. Among the most important of these were the increase in the supply of educated women seeking employment, the de-skilling of office work, and the reduction in both managerial and employee resistance to female employment that came from a change in societal norms (Davies 1975). While the increased supply of educated women undoubtedly facilitated feminization, it is unlikely that a scarcity of female job candidates ever inhibited the use of female clerks. The Post Office in 1870 had no problem obtaining a nearly unlimited supply of women telegraph recruits. The only supply difficulties they reported were with finding men. When the Great Western Railway chose not to hire women in 1876, the availability of female candidates did not even emerge as an issue in the debate. The continued use of men both by the GWR and by other private employers in the 1870’s and 1880’s would seem to have been due primarily to demand-side considerations.
De-skilling does not seem to have been a factor either. Many of the technological innovations that were associated with feminization, such as typing, actually represented upgradings rather than downgradings of clerical skill. Women entered clerical work not because there was an increase in the proportion of unskilled jobs, but because managers changed their strategies for filling these unskilled jobs.
There can be little question that male resistance to the use of women clerks represented a very significant obstacle to the progress of clerical feminization. Virtually in every instance that women clerks were introduced to a new setting, there was intensive protest by both managers and workers. The managerial objections were far more important than those of rank-and-file workers on the office floor. Although male clerks protested vigorously, their attempts to restrict or eliminate female employment came to nothing. Since the firms in these studies had unions that were unusually strong and successful, it is not likely that other male clerks were more effective at excluding female workers. In a sense, the feminization of clerical labor can be attributed in part to the overall weakness of clerical workers that prevented them from being able to control entry into their occupation.
Managerial sexism, on the other hand, remained the most potent barrier to clerical feminization. The degree to which this sexism was tolerated depended to a large extent on the degree to which offices had to economize on clerical labor costs. Over the course of the nineteenth and twentieth centuries, firms became increasingly clerical-labor-intense, which forced them to pay greater attention to the expenses of office management. In the era of the Victorian counting house, when the administrative staff of many concerns consisted of fewer than a dozen clerks, it was not terribly economically significant whether these positions were filled by men or women. A firm’s fortunes were generally determined by its performance in other areas. The growing clerical labor intensity of industry came from two sources. On one hand, the growing sophistication of accounting and financial analysis in business necessitated the collection and processing of large amounts of data to facilitate more rational decision-making. This dramatically increased the percentage of staff in manufacturing concerns engaged in administrative rather than substantive duties. At the same time, the development of the service sector created a new set of industries that were almost entirely composed of clerical labor. Banking, insurance, and public administration each employed workforces that were nearly entirely white collar. The growing clerical labor intensity of industry promoted a global concern with efficient office administration. In the twentieth century, managers eagerly adopted many of the reforms of the office Taylorites and reworked their procedures for correspondence, calculating, and data retrieval. At the same time, they began to staff their offices economically and hired women clerks instead of men.1
Synthetic turnover also contributed to the feminization of clerical work. Most clerical employers would have suffered from the joint problem of tenure-based salary scales and relatively flat productivity curves over time. They would not have had the extreme problems of overannuation of the Great Western Railway and General Post Office, since those firms paid above prevailing market rates, but many of them would have wanted to increase their rates of turnover, and, as was seen in Chapter 4, it was quite common for clerical employers to use marriage bars. The use of gender as the criteria for differentiating between short-term and long-term employees virtually guaranteed that a significant share of low-status office jobs would henceforth be reserved for women. It is hard to imagine any other demographic classification besides gender that could have been used for a tenure bar that would have been at all practical in an office setting. Hiring juveniles and then dismissing them upon reaching adulthood provides a career of only two or three years in length. Dismissing people at age sixty-five is a common practice but hardly solves the particular problems of clerical work. Hiring sixteen-year-old women and forcing them to retire at marriage ensures careers anywhere from six to ten years long. This is long enough to ensure the conservation of any firm-specific skills, but not so long as to create a severe crisis of productivity. Thus, ironically, even as tenure bars diminished the opportunities for older women, they gave younger women a near monopoly on clerical positions, since their tenures could be more easily manipulated to fit the needs of office employers. Synthetic turnover thus helped ensure that clerical work would become a female sex-typed job.
Finally, clerical work feminized because of the growing problems inherent in using other sources of cheap labor. There was no increase in the problems associated with using casual males. Undereducated clerks had been a problem in Victorian Britain, and concerns with clerical quality control date back at least to the 1870’s. However, since temporary men were a suboptimal solution to the problem of staffing routine clerical jobs, one would expect that their use would diminish over time, as managers became more sophisticated in the business of office management. Each wave of office rationalization should have made further inroads into the numbers of temporary male clerks until they finally disappeared altogether. For boys, the problem would have been the growing scarcity of qualified candidates as more and more boys turned instead to higher education and starting jobs with explicitly managerial job titles. Increasingly, women represented one of the few sources of highly educated skilled inexpensive labor that would have been available to employers.
A second closely related question is why women are predominantly concentrated in white-collar rather than manual positions. To some extent, this question asks why jobs similar to clerical work became feminized as well, and the answer closely parallels the above discussion. White-collar occupations tend to be labor-intensive rather than capital-intensive. Factories tend to expend a substantial proportion of their budgets on raw materials and technology. Offices devote their budgets entirely to labor. White-collar employers are less capable of tolerating discrimination against women and are more likely to make use of their relative cheapness. White-collar occupations are also relatively poorly unionized. As such, male workers in these positions are less capable of acting in their own behalf to restrict the supply of labor and forbid employers from using secondary sources of labor. White-collar jobs tend to pay on tenure-based salary scales, thus making them more apt than blue-collar jobs to require some form of synthetic turnover. Blue-collar occupations are paid on piece rates or with job-specific wage rates that do not vary by the demographic attributes of the jobholder. Since, in these situations, older workers are not necessarily more expensive than younger workers, employers have no incentive to increase turnover, and there is no motivation to create an underclass of female employees. Lastly, most white-collar jobs require a non-trivial amount of formal education. These jobs are likely to require the verbal and intellectual skills that are taught in schools, unlike blue-collar occupations, where the training often occurs in extracurricular settings, or occurs on the job itself. Because of the high value that is placed on formal education, employers reserve white-collar jobs for workers with a college degree or better. In general, minority members, juveniles, and casual males are less likely to have the educational qualifications that are available in the adult white female labor force. Thus, women should receive a disproportionate share of white-collar positions, while workers in other categories of secondary labor should find more employment in manual occupations, where education is a less important barrier to entry.
The Operation of Occupational Sex-Typing at the National Level
Some of the arguments that were elaborated in the main analysis are of direct relevance to the quantitative literature on occupational sex-typing that has examined national occupational structures. A full-scale empirical test of these theories using national data has not been attempted. Nevertheless, the existing studies do provide an opportunity to examine individual propositions to see how they fare.
The quantitative literature provides further evidence of the relationship between buffering from labor costs and the percentage of females in an occupation. Labor intensity has generally been a significant predictor of occupational sex type. Michael Wallace and Arne Kalleberg, as part of a larger investigation of the structure of dual labor markets, estimated the determinants of the percentage of females among 68 industries. They found that capital intensity was negatively correlated with the percentage of females in an industry. Furthermore, capital intensity was the only variable in their model that was statistically significant at the .05 level (Wallace and Kalleberg 1981).2 Bridges obtained similar findings in his analyses of 80 primary and secondary industries and 320 “industry strata."3 In his earlier research, the only analytic variable that consistently predicted the percentage of women was capital intensity. Fixed assets per worker had the strongest zero-order relations with the use of women and was the only variable besides industry dummies that survived the inclusion of other variables. In his later work, Bridges used a variable called “classic capitalist organization,” which was a combination of firm size and fixed assets per worker. Out of eleven independent variables, classic capitalist organization was the third best predictor of the percentage of females. The best predictor was white collar-blue collar, which, as was argued above, can also be viewed in part as a proxy for labor intensity. The raw correlation between fixed assets per worker and the use of women was much weaker for tertiary industries, with the relationship being insignificant in multivariate analyses (Bridges 1980,1982). This may have been due to the presence of financial and insurance firms in this sector. Such firms tend to own a lot of assets but have operating budgets that are entirely labor intensive. Except for those tests involving the tertiary sector, the performance of measures of buffering from labor costs generally has been very impressive.4
The percentage of females in an industry is also affected by the demographic composition of the rest of the workforce. In some cases, this can be explained in terms of a trade-off between primary and secondary labor. If the use of women was positively correlated with the use of other disadvantaged workers, this would suggest the presence of some factor that was benefiting both groups at the expense of white adult men. If the use of women was negatively correlated with the use of these other groups, it would suggest a trade-off between alternative sources of secondary labor such as was discussed in Chapter 8. In general, one would expect that if one did not control for attractiveness of employment, there would be a positive correlation between female and minority hiring, since the zero-order relationship would primarily capture the overall advantages white males enjoy in the labor market. As one controls for attractiveness of employment and other variables affecting white male employment, the inverse relationship between female and minority employment should become more apparent. In Bridges’ models, both types of relationships make an appearance. In his early work, Bridges found a negative relation between female employment and that of blacks. The zero order relationship was relatively weak, while as one would expect, the relationship after secondary sector was controlled was much stronger. In Bridges’ later work, the second most important predictor of the percentage of females was the percent of males who are between twenty-five and fifty-four. Women are less likely to work in industry sectors whose men are predominantly in their prime working years. The direction of the relationship is embarrassing at first glance, since in a model that included numerous controls for attractiveness of employment, there was a positive rather than a negative relationship with the percentage of juveniles. Nevertheless, the strength and significance of the relationship suggests that the dynamics of trade-offs between these groups needs further study and that there appears to be a very definite relationship between female employment and the dynamics of the hiring of other types of labor.
Bridges’ work also illustrates the relative unimportance of employee exclusion in determining occupational sex type. After controlling for other variables, the path between unionization and the percentage of females was virtually zero. Furthermore, unionization significantly decreased the amount of sex segregation within industries. If unions were restricting female employment, one would have expected the opposite—that is, that in unionized industries, women would be more likely to be confined to a small subset of occupations.
These findings are consistent with the historical evidence on male exclusionary campaigns presented by Alice Kessler-Harris. Male unionists have consistently been opposed to female employment. However, there have been relatively few cases of an employer’s attempting to introduce women but being forced to abandon such plans in the face of organized resistance from labor. Unions have frequently barred women from union membership, only to find that employers have ignored this lack of certification and hired the women anyway. In the nineteenth century, the National Typographical Union banned women from membership and steered work away from female printers. Women printers set up their own union and seem to have prospered throughout the remainder of the century.5 Cigar-makers originally excluded women from their union, but were forced to reverse their policy in the face of an overwhelming feminization by an unsympathetic management. When Detroit streetcar conductors refused to give union cards to female conductors, the women were able to overturn this policy by appealing to the National War Labor Board (Kessler-Harris 1982). It is true that many employers claim that they are unable to hire women due to the policies of their unions. However, many of these statements can be easily interpreted as rationalizations by managers whose own interest in feminization is lukewarm. Thus Bridges’ findings of no significant union effects are likely to be relatively robust.
Environmental Constraints on the Rates of Change of Occupational Sex-Type
The analysis thus far has implicitly assumed that occupational sex-type is determined largely by the desires of management. Most of the arguments that have been made to explain the persistence of relatively expensive male clerks involve either countervailing advantages (synthetic turnover), buffering from the requirement to maximize profits (clerical labor intensity), or the availability of sources of relatively inexpensive male labor (boy clerkships). The only environmental constraint that was discussed was resistance from organized labor, and the position taken was generally negative. Nevertheless, there are some important environmental constraints on the ability of a firm to set its sex-type at will that did not play a role in this study. One of these is law and government policy. The other is the supply of vacant positions.
Postal hiring was subject to a wide variety of political constraints. Nevertheless, most of these tended to be idiosyncratic and were not typical of the problems felt by the average firm in the private sector. The most important legal constraints that have affected the hiring of women have been protectionist legislation and affirmative action. Neither of these was terribly important to the feminization of clerical labor, but each have had obvious consequences in the blue-collar and professional sectors. Any quantitative study of contemporary American sex-typing would have to take such considerations into account. Net of whatever variables are suggested by this analysis, the design should include explicit measures of the timing and enforcement of affirmative action policy. Firms in industries or regions where enforcement activity has been especially rigorous should be more likely to hire women, even if the firms themselves have not received any direct pressure.
The role of vacancies in determining sex-type is less obvious. The discussion has implicitly assumed that an employer can immediately adjust his labor force to reflect current economic and social conditions. In fact, he can only introduce new workers into positions for which there are vacancies. If the number of openings is relatively small, his current sex-type may reflect past rather than present economic conditions. This was graphically illustrated in the Post Office by the length of time it took to get the political appointees out of office. The Orders in Council banning patronage were passed in 1870. The last political hires finally left the Post Office in 1911, fully forty years after the original decision to dispense with patronage was made. Most organizations are somewhat more adaptable than the Post Office. Nevertheless, the effect of this inertia must be factored in when considering any cross-sectional employment data.
The speed at which vacancies arise is determined by three factors: the secular rate of organizational growth; the effect of cyclical factors on organizational growth; and the rate of voluntary quitting by employees. If a firm is growing steadily, there will be an ample supply of new positions that can be filled with members of a new sociodemographic category. If a firm is stagnating, the employer will only be able to put new workers in those positions that arise from attrition. If the firm is shrinking, there will be no new vacancies at all. Furthermore, employers generally have less autonomy over who they fire than they do over who they hire. The employer may not be able to lay off individuals in such a manner as to control the demographic composition of the resulting workforce.
Cyclical firms should be particularly responsive to current needs. Firms that fluctuate in size go through phases of releasing older workers and then obtaining new ones. Even if some institutional arrangement exists allowing former employees to get their jobs back, there are usually enough workers who do not return to create a supply of vacancies that management can fill with a free hand. In the case of extremely unstable employment, such as seasonal labor or casual work, the labor force itself is sufficiently transitory to allow for nearly complete restaffing with every wave. There should be no lagged adjustment at all in these situations.
The role of voluntary turnover in creating vacancies is obvious and need not be belabored. This implies that life-tenure bureaucracies, relatively high-paying firms, and firms that refrain from hiring juveniles will all suffer from some form of inertia in their hiring, since all three types of firm are particularly likely to have low turnover. Internal labor markets thus have the effect of reinforcing a pre-existing in-group’s statistical domination of employment by reducing the number of vacancies for which an outsider can be hired.
An illustration of the importance of these principles comes from Snyder, Hayward, and Hudis (1978), who found that, in the secondary sector, occupations were more likely both to increase and to decrease their percentage of women workers. Secondary sector industries are more likely to be highly cyclical, are more likely to be low-paying, causing workers to leave, are more likely to hire juveniles who tend to quit for personal considerations, and are less likely to have well-developed internal labor markets. Furthermore, to measure secondary status Snyder et al. used the percentage of workers that were part-year and the percentage of workers that were part-time. Both of these measures are particularly sensitive measures of the overall supply of vacancies.
The effect of legal changes is likely to be relatively permanent. Even if future governments were to stop enforcing affirmative action, the bureaucratic procedures that were created for initial compliance are likely to persist, albeit with reduced efficiency. However, the effects of vacancy shortages are relatively transitory. Over the middle term, most employers should be able to make some adjustment in the composition of their labor force, even if the change is less than they might desire. Lag models can make a contribution to the study of occupational sex-typing, but only within the context of a larger theory that specifies the changes that employers are attempting to achieve.
Women and Low-Status Jobs
One of the fundamental questions in the study of women’s work is why women are concentrated in low-status jobs. A major goal of this study was to develop an alternative to the natural turnover hypothesis. This second explanation, the synthetic turnover model, was extremely useful in explaining women’s subordination in the Great Western Railway and the General Post Office. It is also quite helpful in explaining women’s subordination in the white-collar sector. However, the model only works under a fairly narrow range of economic conditions. If the employer does not pay with tenure-based salary scales, or if labor is scarce, or if employees significantly improve their productivity with increased experience over the long term, or if there are efficient sex-neutral mechanisms for eliminating workers whose pay increases have exceeded their productivity, then there will be either no advantage to increasing turnover or no need to use gender to differentiate the labor force and the synthetic turnover model will not hold. The general predictive power of the synthetic turnover model is based on the fact that the complex conjuncture of events that sets the model into action occurs relatively frequently in practice. Many women’s jobs are characterized by tenure-based salaries, a labor surplus, minimal on-the-job training, and an absence of practical sex-neutral tenure bars. However, another explanation must be sought for women’s economic subordination in the extremely large number of empirical settings where the terms of the synthetic turnover model do not apply.
The most compelling explanation of women’s concentration in low-status jobs involves employee discrimination. Most forms of employee discrimination do not directly affect occupational sex-type, since most employees do not control entry into their own occupations. However, there is a significant exception to this principle. A particularly important set of workers who do control entry into their occupation is managers. Managers fundamentally control the setting of occupational sex-type by determining who gets hired for virtually every kind of job. This study has followed the traditional practice of viewing discrimination by managers as being a form of employer discrimination, which can be explained in terms of pure sexism and patriarchical ideology. However, discrimination by managers can also be viewed as employee discrimination.
The exclusion of women from high-status positions benefits the managerial class by restricting the supply of replacement labor for these positions and increasing the economic returns to managers as a whole. Employee discrimination is very easy to ground in economic rationality. Effective discrimination restricts the supply of candidates for any given occupation. This provides job security for pre-existing workers by making it less likely that they will be replaced by an alternative job candidate. This difficulty of replacement in turn gives them bargaining power that they can use to increase the wages for their occupation. Managers are employees just like operatives or laborers, with a rational concern for protecting their careers and economic prospects. Executives defend the overall salary levels of their occupation by restricting the supply of entrants to the entry-level positions in administrative job chains. By hiring only men for junior-level jobs, employers make qualified managerial trainees artificially scarce. This raises the salaries of entry-level positions through simple market competition. Upper-level executives benefit from this inflation because their salaries are generally set to be higher than those given to entering trainees. Thus, if discrimination raises the salaries of entry-level managers, it will put upward pressure on the entire executive salary structure.
A somewhat related argument can be made for professionals. Professionals also control entry into their own occupations. They control entry into whatever academic institutions provide the basic training in the field; they control the accreditation process by which students are allowed to practice the profession; and most importantly, they control the hiring of professionals in the firms in which they themselves are employed. Most professional firms are administrated in a collegial fashion, with all senior partners having a significant vote in determining both hiring and advancement within the organization. Even when professionals work within a corporate or bureaucratic setting, it is unusual for anyone to be hired who does not meet the approval of the pre-existing professional staff.
The motivation for professional discrimination is somewhat different from that of managers. Professional incomes are largely determined by the volume of fees that has been earned by the firm. Although most professionals do receive a nominal salary, a significant proportion of their remuneration comes in the form of a bonus, which is entirely dependent on the prosperity of the firm. Thus professionals do not benefit in the same way managers do from internal labor markets that ensure a positive differential between the salaries of entering and advanced workers in an occupation. Instead, professionals benefit from discrimination because it increases the price they are able to charge for their services. If successful discrimination drives women or minority practitioners from the labor market, there will be a reduced number of firms capable of providing professional services. This reduces competition among the firms already in the labor market and maintains an inflated schedule of fees.
Most managers and professionals do not have such overtly economic motives explicitly in mind when they institutionalize policies of discrimination. In general, the non-hiring of women would be motivated by sincerely held beliefs in patriarchical ideology and a genuine concern with the compatibility of gender and professional roles. However, the fact that discrimination is likely to raise managerial and professional incomes helps to eliminate a potential check on the operation of sexism in hiring. When managers think about the economics of hiring non-managerial workers, their patriarchical concerns are tempered by a consideration of larger economic considerations. Because managers are interested in obtaining the lowest rather than the highest possible salaries for these workers, they have to consider the relative cheapness of women in their calculations. This provides a continual pressure to feminize and create labor economies. When managers or professionals hire themselves, they become concerned not with lowering but with raising the standard of living for their occupation. Any policy that will be consistent with maintaining levels of remuneration will tend to be accepted rather uncritically by employers. Sexism survives because it serves a latent economic function for this class of individuals.
These tendencies are exacerbated because the hiring for both types of positions is relatively buffered from a concern with using sex to minimize labor costs. Managers and executives make up a very small percentage of the labor force of most firms. If levels of executive remuneration were to become extremely high, this would have almost no bearing either on the labor costs or on the overall profitability of the firm. As a result, most corporations tend to be relatively generous in their executive benefits. Six- and seven-figure salaries, stock options, and expensive perks such as private planes are all affordable because the absolute expense they represent is trivial compared to the total corporate budget. Given that these rewards are approved by trustees and directors who would like to see comparable remuneration given to themselves, it is not hard to see why such generosity continues to persist.
A related argument can be made for professionals. One cannot claim that the profits of professional firms are independent of labor costs. Professional fees make up the lion’s share of the budget in most of these firms. However, doctors and lawyers cannot save money by hiring female partners rather than male partners. The professions are one of the few occupations where the fifty-nine cent to the dollar rule does not apply. Female professionals still make less money than male professionals because they are barred from practicing in the more lucrative specialties. Nevertheless, once a women professional is hired by a firm, she is paid the same rate as a man. It is more difficult for a professional firm to deny a woman comparable worth than it is for a corporate employer. Corporations are large authoritarian organizations that can mask pay inequities in a welter of different job titles. The presence of bureaucratic administration provides companies with the ability to segregate women workers by arbitrary job title providing different rates of return to similar work. Professional partnerships are less capable of practicing this kind of discrimination. Because they are small, they do not have the same semantic differentiation of their staff. This tends to provide comparable treatment for all workers within broad categories, such as “summer intern” or “senior partner.” Furthermore, most professional partnerships are run under norms of collegiality. This reduces the viability of having separate regimes for ostensibly equal workers. These factors keep professional firms from paying women less than the male rate for doing equal work. However, they also destroy whatever economic incentive there may be for hiring women in the first place. Since women work at the same rate as men, there is no pressure for professionals to seek economies through feminizing. As a result, they can indulge their sexist fantasies, or they can act to restrict the supply of women into the profession. Both of these will tend to limit female employment.
Thus, although the synthetic turnover model cannot serve as a general explanation of why women are concentrated in low-status jobs, the various arguments of the book provide tools to help fill in this gap. Employee discrimination can only work when workers have the power to control entry into their occupations. High-status workers are the only cases of such exclusion’s being effective. Women are generally excluded from occupations when such exclusion benefits their employers. Thus, women are barred from the occupations of the employers themselves so as to preserve upper-level salaries. Women are excluded from settings that are buffered from the need to use cheap female labor to reduce labor costs. Both professional and managerial occupations enjoy the benefits of such buffering. Thus, the propositions that were derived from a study of variance in female employment can be used to explain the relatively universal phenomenon of women’s economic subordination.
Further Research in Women’s Work
A final lesson of this study is methodological. Those who have taken a radical or critical perspective on gender have in general been too reluctant to test their theories rigorously with either historical evidence or statistical methods. There is generally a despair on the part of progressive writers on women’s status that their problems are particularly intractable to the use of quantitative social science analysis. Sometimes this suspicion comes from the perception that large sample analyses tend to use individuals as the unit of analysis and thus become biased towards micro-level rather than macro-level strategies of explanation. In other cases, the suspicion is based on a despair with the quality of historical data. Archives are often seen as only providing scraps of anecdotal evidence that relate to particular and unique social occurrences. Some fear that the analytical clustering of incidents will do violence to the individuality of each event. Others fear that the data is so incomplete as to make any program of systematic measurement or sampling practically impossible.
These reservations are real, but can be readily overcome by the analyst who is willing to be creative with what the data have provided. The pitfall of perpetually sliding into an individual-level analysis can be avoided by relentlessly seeking larger units of analysis. Instead of comparing particular workers, one can compare firms, industries, or any other aggregation that makes employers rather than workers the key decision-making actors. The analyst who is concerned about riding roughshod over the individual characteristics of his cases can devote careful scholarly attention to the historical context surrounding each. Such systematic attention to the historical record can suggest unanticipated sociological regularities that in fact were operating unobserved in the other cases. The plea of historical individualism is often a disguised argument for the necessity of rich multivariate analyses. Finally, the data that is available in public archives are far richer than most social scientists have realized. Most of the major corporate archives in the United States are unidentified, let alone unexplored. Furthermore, the most banal, useless-looking document can be found on second analysis to contain a wealth of analytical possibilities. The turnover data in this study came from actuarial analyses in pension files. The dependent variable for the Post Office came from personnel lists, which often contained no more information than the name and job title of an office’s clerks. The only data a name will provide is whether that clerk is male or female. A lot of archival data is not quantitative in and of itself. However, there are a lot more numbers out there than most people realize. The decision makers and bureaucrats of earlier eras kept accounts, took censuses, kept membership lists, took inventories, counted errors, and estimated costs in a fashion that allowed them to quantify the problems they faced in their daily lives. Sometimes these accounts were kept voluntarily by the organization; other times they were kept as a requirement of the local or national government, and a copy can be found in the public archives. These figures combined with the qualitative leavings of letters, reports, and public announcements can provide historical reconstructions that are remarkably detailed. An archive will not always generate information on precisely what an investigator is looking for. However, it often contains a wealth of information on important topics that one might have never even considered examining. By being willing to consider what an archive has to offer, and by making special efforts to consider what can be quantified from this material, one can uncover some fascinating and bizarre historical processes that have direct relevance for the intellectual questions of today.
1. An interesting empirical illustration of the importance of clerical labor intensity in promoting clerical feminization is the persistence of the male shipping clerk. Shipping clerk is perhaps the only clerical occupation that has remained predominantly male (Garfinkle 1975). The work of filling out bills of lading and receipts is similar to work done by female clerks in offices; furthermore, the physical demands of the job are not great, since, in most firms, loading and unloading freight is handled by packers or warehouse staff. What differentiates a shipping clerk from an office clerk is that the shipping clerk is usually the only clerical worker in a staff that is primarily blue collar. Although the employer of the shipping clerk, namely a warehouse supervisor or foreman, may claim that women cannot handle the physical demands of such a position, this is really no different from the claims that were made by the Postal Engineers or the Great Western Railway management. What all of these managers have in common is a clerical force that is sufficiently small relative to their total staff that the luxury of male clerks can be afforded.
2. This relationship was not robust to controlling for occupational composition. However, that finding comes from an equation that used fully fifteen variables to predict variance in only sixty-eight cases. The uncontrolled equation used only six independent variables. The small N available in most analyses of industrial attributes makes data attrition a real methodological concern.
3. In Bridges’ terminology, an industrial stratum is either the white-collar or the blue-collar staff of an industry considered separately.
4. A labor cost buffering theory would also imply that secondary sector firms would be more likely to hire women than would primary firms because of their generous supply of organizational slack. This proposition has been disconfirmed in several preliminary analyses. There are several methodological reasons for questioning these findings, and the most that can be said is that the relation between patterns of gender hiring and dual labor markets is empirically unknown. For a discussion of the complex measurement issues involved, see Appendix G.
5. By the twentieth century, printing had become a male-dominated occupation. A closer examination of the decline of the female printer would be an excellent test of union exclusion theories as well as a fascinating case study in its own right.