De-Skilling and Technological Change
Many discussions of clerical feminization have put a great deal of emphasis on the ways in which office technology and strategies for organizing the labor process produced changes in the skill composition of the clerical force. Most discussions of office work claim that clerical jobs de-skilled during the nineteenth and twentieth centuries (Klingender 1935; Mills 1951; Lockwood 1958; Braverman 1974; Davies 1975, 1982; Glenn and Feldberg 1977). The argument is that clerks evolved from a set of management trainees in small personalized counting houses to proletarianized employees in large impersonal bureaucracies. The increase in the size of clerical staffs forced management to reconsider the administration of office work, leading to the development of various strategies for rationalizing and reorganizing the labor process. This resulted in the development of office machinery, the creation of detailed divisions of labor, and the introduction of Taylorite strategies for monitoring output, all of which served to lower the autonomy and promotion prospects of office workers. The end result was the creation of the semi-skilled clerical operative, who makes up the majority of the workforce of modern offices.
This process has been linked, although with some reservations, to the explanation of the origins of clerical feminization. Historically, women have been confined to low-status occupations characterized by low levels of responsibility, inferior rates of pay, and limited prospects for promotion (National Manpower Council 1957; Oppenheimer 1970). If clerical work had systematically been de-skilled, the increasing undesirability of office jobs would have facilitated the use of women for these positions (Davies 1975, 1982; Glenn and Feldberg 1977; Rotella, 1981). However, those who have linked feminization to the degradation of office work have been very cautious in advancing their claims. One reason for caution is that de-skilling cannot provide a complete explanation of sex-type change. Davies, for example, argues that clerical feminization was a complex phenomenon that was caused by a number of different factors, such as changes in the female labor supply, the demand of employers for cheap labor, and changes in national gender ideologies. These would have contributed independently to the assignment of women to office tasks, weakening the overall importance of skill dynamics. This new caution can be seen in the work of Davies and of Evelyn Glenn and Roslyn Feldberg in which unfavorable technological change is treated as a burden affecting female occupations as well as a factor creating female occupations.
However, there is a more fundamental problem with such an analysis. It is not at all clear that clerical work has been de-skilled to any significant extent. There are both theoretical and empirical reasons for suggesting that office skill levels may have been more stable than is customarily acknowledged. The theoretical reasons involve the relationship between capitalist rationality and skill levels. Since Harry Braverman published Labor and Monopoly Capital, it has been customary to argue that managers seek to reduce overall skill levels in pursuit of the lowest possible labor costs (Braverman 1974). However, there is no necessary logical relationship between the goal of economizing on personnel and the use of an unskilled labor force. This point has been made quite cogently by Michael Piore and Peter Doeringer in Internal Labor Markets and Manpower Analysis. Piore and Doeringer argue that most managers attempt to economize on labor by substituting capital for labor to eliminate jobs entirely. Interviews with engineers showed that the criterion used by systems designers to evaluate the impact of technological change on profitability is not the extent to which an innovation will use a cheaper form of labor but the extent to which it will lower the overall size of the labor force. Managers were so unconcerned with the impact of technology on skill levels that they not only neglected to make projections of the type of labor that new innovations would require but used only one base salary for all workers regardless of skill in their calculation of overall labor costs (Piore and Doeringer 1971).
Furthermore, the use of technology to destroy jobs has important implications for skill distributions. An innovation that eliminates positions in the lower end of the status hierarchy will by default raise the average skill level of those workers who remain in the establishment. An innovation that eliminates jobs at the upper end will lower the aggregate skill level. Job elimination thus can have varying effects on the resulting ratio of skilled to unskilled labor. To some extent, menial jobs are easier to eliminate by mechanization, since the work is simple, repetitive, and involves routine physical manipulations. Thus, there should be a slight bias for low-level jobs to disappear first, producing a net trend of upgrading. Since different technologies will have different effects, however, the overall impact of technical change should be somewhat random, producing no consistent pattern of downgrading or upgrading over time.
The empirical basis for the clerical de-skilling hypothesis has come from the historical work of sociologists studying white-collar labor such as C. Wright Mills, Francis Klingender, and David Lockwood. These authors argued that white-collar work had suffered a catastrophic loss of status. However, these conclusions are based on an overromanticized conception of the quality of clerical life in the nineteenth century. According to these accounts, nearly every Victorian clerk was a management trainee who in return for personal loyalty to an entrepreneurial patron could expect job security, future advancement, and job assignments with challenging substantive responsibilities. Gregory Anderson’s work on nineteenth-century offices provides a serious challenge to these accounts by documenting the existence of a large clerical secondary sector. Much of the work in early offices was done not by clerks but by copyists. Copyists were temporary workers, comparable to modern Kelly Girls, who were paid by the piece to do tasks involving writing. The work consisted of simple duties such as making copies of correspondence, addressing receipts, or making entries in ledgers. The job security, pay, and prospects of these positions was marginal. The skills required were negligible, since transcription requires few skills other than literacy and penmanship. The status of permanently established clerks was often not much better. Nineteenth-century commerce was extremely cyclical, and many firms were forced by business conditions into bankruptcy. Small firms of the type Lockwood discusses would have been especially vulnerable to failure during the great financial panics. Anderson documents extremely high levels of immiseration among Victorian British clerks. Unemployment was a persistent problem. Many clerks only avoided unemployment by taking marginal jobs and living in poverty (G. Anderson 1976). When one compares the modern typist with the Victorian clerk, the modern typist compares quite favorably. The typist enjoys far greater job security, and although her duties may not be as complex as those of the semi-managerial trainees, they are certainly more complex than the duties of the copyists.
Just as the traditional account tends to overstate the status of the nineteenth-century office worker, it understates the status of the modern office worker. This is because most comparisons are made between those workers in both periods who are nominally called “clerks.” In the Victorian era, the word “clerk” referred to anyone who worked in an office. The GPO Establishment Books use “clerk” for everyone lower than a department director but higher than a copyist. Many of the superior clerical positions would nowadays be referred to as “administrators,” “accountants,” or “middle-level managers.” The claim that de-skilling has occurred because clerks no longer do the jobs involving supervision and responsibility that they once were assigned is based largely on a semantic difference rather than a substantive change. The real question consists of whether there has been a loss of skill or status among the entire population of office workers, whether the ratio of true managers to true clerks has declined over time. Measuring changes in the status of office workers as a whole is not easily accomplished with national census data. The population of office workers contains multiple occupational titles, whose meanings change in subtle ways over time. For example, the population of individuals called “managers” includes many proprietors of small firms, such as stores or family businesses, that do not use clerks in a formal sense. Such petty bourgeousie would have to be excluded from the analysis. Given the complexities of doing a methodologically rigorous study of national changes in office status, the white-collar historians can scarcely be criticized for confining their attention to the population of nominal clerks. Nevertheless, consideration of the whole population of office workers is critical to the assessment of whether office work de-skilled.
This chapter is an attempt to estimate the extent to which office work de-skilled between 1870 and World War II. If a significant amount of de-skilling occurred, it would not be unreasonable to suggest that this would have facilitated the feminization of clerical work. If the skill required for clerical work increased or remained constant, this would suggest the need to explore alternative explanations for the introduction of women to the office. The data that would allow for a completely rigorous assessment of the national evolution of office skill structures are unavailable. The fragmentary evidence that can be used consists of those materials previously examined in the literature on changes in clerical status in England and America as a whole and materials from the archives of the Great Western Railway and the General Post Office that allow for a discussion of status changes in two particularly important firms.
There are two types of data that have been particularly important to the discussions of national status trends. First, there are data on clerical wages. Large sample surveys of clerical incomes for different dates have been used to identify trends in remuneration. The postal and railway archives provide somewhat similar data. Both organizations left extensive documentation on salary levels. The firm censuses allow the reconstruction of detailed pyramids of pay levels within each organization, along with the number of individuals who were found at each level. These pyramids can be studied over time to determine trends in levels of remuneration. The data also allow the identification of entry-level and non-entry-level positions. Such data can be used to evaluate the prospects of upward mobility within these firms.
The second source of data is the writings of Taylorites and other advocates of scientific management. The usual way these materials are used is as chronicles of the attempts of the scientific managers themselves to impose rigorous regimes of work discipline on the labor force (see Davies 1982). Such a reading can be extremely informative. However, these works can also be read in their own right as descriptions of the skills required to do clerical work. In the process of attempting to reform the office, the Taylorites left a large number of technical discussions of the right way and the wrong way to do a wide variety of clerical tasks. One can use these descriptions of job contents to make an informed assessment of the actual difficulty of performing various office duties. By comparing earlier and later technologies for doing comparable clerical tasks, one may do a subjective assessment of whether such work was upgraded or de-skilled. As in the case of the salary data, there are parallel bodies of material for clerical work at the national level and clerical work within the firm. The latter materials are provided by scientific managers’ in-house analyses of clerical work on the Great Western Railway.
National Trends in Salaries
The most systematic data available on long-term trends in clerical skill level comes from the traditional measure of status levels: salaries. Salaries are a useful index of changes in skill; however, these data must be treated with caution due to the presence of many potential sources of measurement error. Pay levels can increase because of general labor scarcity or because of a peak in labor demand within a particular industry or occupation. Institutional factors such as an increase in union activity will produce non-skill-related raises in wage rates. Salaries have a component that is skill-related. Artisanal workers tend to be highly productive, which raises the ceiling on the range of pay rates that would be economically justifiable. Furthermore, since skilled workers tend to be relatively scarce, they can command premium salaries through market competition for their services. Salaries are thus predictable from skill levels and a number of exogenous variables. One can use salaries to measure skill provided one pays explicit attention to the likely impact of these contaminating factors.
The data consist of a 1909 survey of British clerical salaries conducted by Cannon and Bowley and a 1929-30 survey of clerical salaries in London included in the New London Survey. These were the materials from which Klingender (1935) and Lockwood (1958) drew most of their conclusions for this period.1 These two sets of data are not fully comparable. They refer to different geographical areas and used inconsistent sampling strategies and definitions of their industrial classifications. Furthermore, the periodization is not ideal. There is no information on any changes between 1929 and 1945, a period that included the clerical wage cuts of the Great Depression. Even worse, there is no data for any period before 1909. Not being able to measure differences in status between 1870 and 1909 truncates a very important period in the evolution of office work.
Despite these severe limitations, it is possible to draw some useful conclusions from the data. The findings suggest that the period from 1909 to 1929 saw very little deterioration in clerical standards. Since feminization occurred primarily between 1870 and 1945, with 1909 to 1929 being a critical period in this transformation, this suggests that the level of clerical de-skilling may not have been pronounced.
Table 3.1 is a recalculation of material originally presented in Klingender (1935). The figures presented here are methodologically superior to those in the original; however, a comparable analysis could be made of the published figures. The estimates of average industry-specific levels of salaries in shillings per week come directly from Klingender. Real wages were calculated using the 1909 Sauerbeck-Statist price index.2 The figures generally show clerical salaries remaining constant or improving during the period. Men maintained their economic position, while the women clerks made substantial percentage gains.3
There is one condition under which these data would be consistent with clerical de-skilling. If some exogenous force had independently raised office salary levels, it is conceivable that de-skilling could have dampened the effect of this externality, producing a modest increase where one would have expected an increase that was more dramatic. Office work expanded dramatically during this period; the explosion of new jobs could reasonably be expected to have increased labor demand and thus contributed to an overall inflation of salaries. The effect of such market forces can be exaggerated, however. The expansion of clerical occupations produced a simultaneous increase in female labor force participation. The increasing supply of women would have eased the competition among employers for clerical workers and thus would have dampened the effect of occupational growth on rates of pay. Conceivably market forces could have produced increases in clerical salaries that were anywhere from massive to negligible. It is thus very difficult to examine a specific empirical case of an increase in the pay of office workers and judge whether such an increase is likely to have been caused by expansion, de-skilling, or some combination of the two.
Average Weekly Salaries of British Clerks in Selected Industries, 1909–10 and 1929–30
*Sex Ratios for banking and insurance merged in the original.
†No females hired in this period.
Note: All figures are shillings per week. Real wages adjusted to 1909 shillings using the 1909 Sauerbeck-Statist price index.
Source: Adapted from Klingender (1935). Industry-specific clerical sex ratios from Klingender (1935) and the British Census. These sex ratios are used to create weighted averages of male and female salaries.
In the light of this ambiguity, the following conservative conclusion seems warranted. The data do not show any patterns that positively support the contention that clerical work de-skilled. Had real clerical salaries declined during this period, this would have been a very strong confirmation of the de-skilling hypothesis. Since the natural factors affecting pay rates had an upward bias, the observed declines could only be attributed to the negative effects of skill. This data is also inconsistent with any claims of clerical immiseration. The standards of living of both male and female clerks were generally improving, due to a favorable mixture of market mechanisms and skill dynamics. These findings cannot disprove the hypothesis that a strong expansion in the demand for clerical workers was capable of compensating for a weaker tendency for office jobs to de-skill. Nevertheless, these data provide little support for the bleak picture that is usually painted of changes in the prospects of office workers.4
Lockwood (1958) does provide data that suggest that some deskilling may have occurred after World War II. This finding is of definite interest, but is less important than the data he provides from earlier periods. The usual arguments about clerical de-skilling maintain that the downgrading occurred during the transformation of the offices from small personalized enterprises into large, bureaucratically administered, corporate firms. Furthermore, in these arguments, Taylorism is identified as being central to the process by which deskilling occurred. Both arguments strongly suggest that the lowering of clerical standards should have occurred in the early twentieth century; this is the period in which Taylorism was at its height and the personalized offices of the nineteenth century became transformed into large centralized establishments. By 1945, the fundamental bureaucratization and redesign of clerical work had already occurred. If salaries were robust throughout the early twentieth century and only declined after World War II, this disconfirms most of the traditional accounts. Furthermore, clerical feminization occurred in the late nineteenth and early twentieth centuries. By World War II, the sex-typing of office jobs had been long established. The timing of the decline of clerical salaries is inconsistent with the proposition that the feminization of clerical work was caused by de-skilling.
Salary Trends within the Great Western Railway and the General Post Office
The discussion of national salary trends concentrates on the period 1909-29. While 1909-29 is important in its own right, the absence of the earlier material diminishes the utility of the national data for the discussion of long-term trends. Fortunately, both the Great Western Railway and the General Post Office have excellent data on salary levels that span the period from 1870 to World War II. With these data, one can replicate the analysis and the findings of the previous section. In both cases, clerical salaries rose, although there were exogenous factors that would have predisposed salaries to increase. Evidence exists for some departments that suggests de-skilling, but outside of this subset, clerical skill levels probably did not decline.
Tables 3.2 and 3.3 show the evolution of clerical wages in the Great Western Railway as a whole and in selected departments of the General Post Office. The rail data come from the two staff censuses of 1870 and 1933. The postal data comes from the Establishment Books. The sample used in Tables 3.2–3.5 is slightly different from that used in the rest of the book. The usual analyses exclude managers and professionals to allow for the study of the determinants of clerical sex-typing. In these analyses, however, the ratio of managers to clerks is what is at issue, so all white-collar workers are included.
Table 3.2 shows that real clerical wages more than doubled between 1870 and 1933. The trends in real wages were actually more favorable than the trends in money wages since prices declined during the period. Table 3.3 shows that real wages also improved in the Post Office. However, there was considerable diversity among the offices in the rate of improvement. The most dramatic wage increases came in the telegraph offices. Between 1876 and 1931, real telegraph wages more than tripled. The situation in sorting offices is almost as impressive: sorting salaries more than doubled. There was a set of intermediate offices such as the control offices, the Registry, and the Savings Bank. Lastly, there were offices whose salaries showed very modest improvement, such as RAGO, (the Receiver and Accountant General’s Office), the Money Order Department, and the Returned Letter Office.
Trends in Mean Salary on the Great Western Railway, 1870 and 1933
Note: All salaries are given in pounds sterling per annum. All real wages are calculated in 1870 pounds.
Source: Staff Statistics (1870); GWR Staff Census (1932).
The figures in the middle columns of Table 3.3 actually underestimate the extent of salary improvement. That is because most of these offices feminized between 1871 and 1931. Because of the overcrowding hypothesis, women would have worked for less than men, even if they were doing precisely the same tasks. Correcting for such a bias produces even greater increases in salaries over time. Appendix C describes the procedure by which the sex-adjusted pay rates were calculated. The figures in the right columns of table 3.3 represent an estimate of what wages would have been in the absence of a change of sex in the Post Office. The estimating procedure uses the properties of a small set of occupations that are extrapolated to the dataset as a whole. Therefore the reader is advised to view the sex-adjusted figures as very rough approximations of the true rates. Nevertheless, when sex discrimination is taken into account, most offices can be shown to have doubled their pay rates. The superior gains of the telegraph offices remain unchanged in these calculations. These offices had already feminized by 1876, so the impact of further sex change would have been negligible. However, the dramatic improvements of the telegraph department now appear in most of the other offices as well. Out of all the departments in the Post Office, only RAGO and the Money Order Department show modest rates of salary increase, and even here real wages went up 30 percent.
Trends in Mean Salary in Selected Departments of the General Post Office, Selected Dates
*Receiver and Accountant General’s Office.
Note: All salaries are given in pounds sterling per annum. All real wages are calculated in 1871 pounds in the first half of the table and in 1876 pounds in the second half of the table.
Source: GPO Establishment Books (1871, 1876, 1931).
There were both upward and downward exogenous pressures on salaries in these firms. The Great Western Railway and the postal sorters and telegraphists were highly unionized. Although this unionism was of mixed effectiveness, there was undoubtedly some impact on pay rates, leading to the relatively favorable increases for all three groups of workers. Likewise from 1870 to 1930, both the firms and the British economy as a whole expanded, creating some inflationary pressure through increased labor demand. On the down side, the Post Office eliminated patronage in many of its operations. During the period the civil service moved from a regime of premium wages to one based on market rates. The main device for lowering pay rates was to transfer work being done by higher job classifications to workers with pay scales more in line with private rates. For example, in telegraph work, it was fairly common to transfer work customarily done by supervisors to rank-and-file signallers (Hobhouse 1906). A related development was the consistent introduction of new job titles lower than those that had existed previously with a concomitant knocking down of work assignments one level. Financial offices saw the introduction of male assistant clerks and female writing assistants, who did work previously done by male and female second-class clerks (Evans 1934). These semantic changes would have produced a lowering of pay that was not due to any real reduction in the true complexity of office tasks.
All things considered, the exogenous factors affecting salaries would have had a mixed effect, although those that raised wages probably predominated. Nevertheless, every office in the sample shows a very strong increase in real levels of remuneration. Thus, the conclusions for the GPO and the GWR parallel those made for the clerical force of Britain as a whole. The data consistent with deskilling—an absolute decline in salaries or a modest wage increase coupled with overwhelming external wage pressure—do not occur. Although there can be no certitude, the best estimate would appear to be that skill levels in these firms remained constant. If there was any noticeable de-skilling, it is most likely to have occurred in the GPO’s Receiver and Accountant General’s Office or the Money Order Department, where the pay raises were markedly inferior to those found elsewhere in the sample.
Non-salary Measures of Status in the Great Western Railway and the General Post Office
Salaries provide a useful measure of clerical status. However, they are subject to factors external to the dynamics of the firm, such as changes in the labor market or in price rates. An alternative approach is to study the distribution of relatively good and bad jobs within the firm, as measured by hierarchical position. The skill distribution plays a key role in determining the overall ratio of superior to inferior positions. Entry-level work is considered to be relatively simple compared to non-entry-level work. Presumably, an individual equipped merely with the skills available to the population as a whole can do an entry-level job, while jobs reachable only by promotion require some level of firm-specific knowledge. A highly skilled firm may have a rectangular status pyramid—that is, one in which there are roughly equal numbers of individuals at every level of the hierarchy. A highly unskilled organization will have a small number of supervisors and a large number of subordinates. This will produce a triangular status pyramid, with most of the workers located at the bottom.
The advantage of such a measure is that it actually reflects the ratio of managers to subordinate workers, a critical concept in a claim that there is a relative proliferation of inferior jobs in clerical work. Furthermore, the rate of promotion is largely determined by the shape of the status pyramid, with promotions being relatively more frequent in rectangular firms. Lastly, the measure is independent of global exogenous factors that complicated the analysis of the salary data. The interpretation of these figures is much less ambiguous, which permits the generation of simpler and more straightforward conclusions.
The shape of the status pyramid will be measured by the percentage of workers who are non-entry-level. Where multiple ports of entry exist in a given job ladder, the highest port is considered to be entry-level even if it can be reached by promotion. The percentage of non-entry-level jobs does not capture all of the inter-office variance in skill levels. Differences in skill that come from moving to advanced entry-level tasks are not reflected in the measure, nor does the measure make any distinctions among the superior levels of the hierarchy. There are semantic problems as well. Skill is a continuum in which many conceivable cutting points are possible. Firms with identical human capital requirements could have different percentages of non-entry workers if one firm had a slightly lower cutting point in defining the first promotion.
Furthermore, status and skill are not identical phenomena. Sometimes a high-level worker can be less skilled than the individual he supervises. A hotel manager may be less skilled than the French chefs in his kitchen. A production manager may know less about his machines than the repair personnel. Even when these types of errors are taken into account, however, the percentage of jobs that are non-entry-level still captures a basic split between those workers at the bottom and those at the top of the status hierarchy. Since on-the-job learning and skill represent a significant proportion of the difference between the superior and inferior jobs, the crude perentage of non-entry-level jobs is still useful as a measure of the overall distribution of job complexity in the organization.
The measurement of the percentage of non-entry-level jobs on the Great Western Railway poses some minor methodological problems due to the differences in job classification schemes used in 1870 and 1933. In 1870, sixteen-year-old boys were hired as junior clerks. In two years’ time, they could take an examination to become senior clerks. Thereafter, jobs were formally unclassified, although managers had explicit ideas about the maximum pay and natural ladders of progression associated with different positions. In 1933, the same junior/senior clerk distinction still existed. However, the senior clerks were divided into six grades—special and 1–5—with 5 being the lowest senior position. There were also three grades of adult women clerks, grade 2 being entry-level and grade 1 and special being advanced. There was a grade called “junior female clerk,” but this was not a common port of entry as was the case with junior male clerk.
There are two ways to conceptualize the definition of entry-level positions in the GWR. In one sense, the only entry-level positions were the junior clerkships of both sexes and the women’s grade 2, since the grade 5 male positions were attained by promotion from junior clerk. This definition of non-entry-level is called Assumption A; it explicitly treats the male grade 5 as a superior ranking. However, such a procedure creates an inequity between men and women. A male could become a grade 5 virtually automatically by waiting two years and then passing a test. The overwhelming majority of men passed their senior exam on the first sitting, while many of the remainder passed on the second or third examination. Leaving grade 5 was substantially more difficult. Over half of the adult male clerical force were employed as grade 5’s, and many workers kept this status throughout their careers. Grade 2 was the equivalent job title for women. The overwhelming majority of women worked in grade 2 clerkships with only a small elite being promoted to grade 1. Thus, there is a sense in which grade 5 represents an automatic promotion, with grade 4 being the first discretionary screen. Assumption B treats grade 5 men as entry-level, an interpretation that captures the trivial nature of the senior promotion. There is no literal equivalent of grade 5 clerk in the 1870 data, so a false grade 5 was constructed using salary as a proxy for job grade. (See Appendix D for details of the calculation of the grade 5 equivalent.)
Table 3.4 shows the trend in the percentage of non-entry over time in the GWR. Assumption A and Assumption B both produce very similar findings: clerical work upgraded between 1870 and 1932. Using Assumption A, jobs on the Great Western Railway went from 71 percent high-status to 83 percent high-status. Under Assumption B, the improvement is especially striking. The number of superior jobs grew from 4 percent to 40 percent, a dectupling. The estimate for any given year of the absolute number of high-status jobs changes depending on the methodological procedure used. The number of grade 5 men was so enormous that shifting the definition of grade 5 has an overwhelming effect. Nevertheless, there clearly was a trend for the relative share of inferior jobs to decrease and the share of superior jobs to increase.
Trends in the Percentage of All Clerical Jobs in the Great Western Railway That Were Non-Entry-Level, 1870 and 1933
Note: Assumption A is that grade 5 men and their 1870 salary equivalents are non-entry-level. Assumption B is that grade 5 men and their 1870 salary equivalents are entry-level.
Source: Staff Statistics (1870); GWR Staff Census (1933).
Estimating the percentage of non-entry-level jobs for the Post Office poses few methodological problems since the job titles were essentially comparable throughout the period. However, the empirical findings are much more complex.
Table 3.5 shows an enormous diversity of patterns, with some offices de-skilling, some offices maintaining their status, and some offices showing dramatic improvement. Among the offices whose relative share of non-entry-level jobs rose were the Returned Letter Office, the Savings Bank, East Central Sorting, and the telegraph signalling offices. The West Signalling Office showed the most dramatic improvement, with a near quadrupling of superior positions. Among the offices that lost status were RAGO, the Money Order Department, the Registry, and all control offices. RAGO and Money Order were offices that also received comparatively minor increases in salary. This further supports the claim that de-skilling may have occurred in these particular sections. The registry lost status despite its relatively robust pay levels. The status data probably provide a more accurate picture of this office. The prime function of the registry was the maintaining of the postal archives, a job that required an extensive amount of filing. As will be argued below, filing and records management was one of those aspects of clerical work where Taylorism did produce significant de-skilling. The control office cases are more difficult to interpret. The Establishment Books show that, over time, there was an increasing proportion of clerks to managers in these offices. However, it is unclear whether this represents a de-skilling or a permanent assignment of staff to the control offices to do work formerly done by the general force of the supervised offices. Given the dramatic declines in the percentage of non-entry-level jobs, it is quite likely that both processes were occurring.
Trends in the Percentage of All Critical Jobs in Selected Departments of the General Post Office That Were Non-Entry-Level, Selected Dates
Source: GPO Establishment Books (1871, 1876, 1931).
The data on the percentage of non-entry-level jobs support the findings of the pay discussion. The railway showed a dramatic improvement in clerical pay and status. While exogenous factors could account for a secular increase in pay, a change in the status pyramid from triangular to rectangular is harder to account for by such factors as unionization or a tightness in labor markets. Both of these factors would be expected to raise wage rates globally, rather than altering the distribution of pay within the firm.
The experience in the Post Office was more diverse. Some offices did show strongly suggestive signs of de-skilling, in particular RAGO, Money Order, and the control offices. However, others showed levels of non-entry-level jobs that are constant or rising. As in the case of pay, the opportunities in sorting offices were good and the opportunities in telegraph offices were excellent. Note that “truly clerical” offices did not show a general tendency to de-skill. On both pay and non-entry-level status measures, the Savings Bank and the Returned Letter Office showed signs of significant real improvement. The picture of the Post Office is complex and undoubtedly some de-skilling did go on, but the most reasonable interpretation of these data is that the skill level of clerical work remained constant, with gains in some departments being offset by losses in others.
Mechanization, Scientific Management, and Rationalization
These findings have important implications for the reader with a general interest in skill and technological change. At the beginning of the chapter, two conceptualizations of the evolution of skill were identified, that of Braverman and that of Piore and Doeringer. In the Braverman model, managers use technology to reduce labor costs by subdividing jobs into their simplest components, allowing management to hire the cheapest form of labor. In the Piore and Doeringer model, managers use technology to eliminate jobs. Because job elimination can occur among jobs with a wide range of skills, technological change should have a neutral impact on skill distributions. Some Bravermanian transformations can occur, but they are balanced by upgradings caused by the automation of unskilled positions.
This finding can be supported another way. Within the study of skill, there are two methodological traditions. One approach is to measure skill quantitatively, using measures of job complexity derived from the Dictionary of Occupational Titles (Horowitz and Herrnstadt 1966; Dubnoff 1978; Wright and Singelmann 1982). The second approach is to use case study material to examine the impact of particular technological changes. Some of the most impressive support for the Braverman hypothesis comes from the collection of essays put together by Andrew Zimbalist in which experts in a large number of industries provided an overwhelming array of examples of how capitalist technological change had undermined skill levels (Zimbalist 1979). Likewise, Piore and Doeringer’s own insights grew out of participant observation of a Route 128 firm (Piore and Doeringer 1971). In the case of clerical work, it is extremely informative to turn to qualitative materials and consider the actual impact of the specific technological transformations that affected office work. A consideration of the content of clerical work will show that the effect of technological change on skill was extremely complex and that the present findings are no statistical artifact.
This section attempts to review as comphrehensively as possible the major technical changes in office procedure that were introduced in the late nineteenth and early twentieth centuries. The primary source for most of this analysis is a set of book-length Taylorite works on office management.5 The scientific managers wrote extensively on office technology. Their descriptions of job processes are explicit, detailed, and based on years of acute observation. Given these descriptions, one can often reasonably estimate the skills it would have required to master these various procedures. Another source of information are the documents in the GWR and GPO archives. In the course of doing the research, it was necessary to examine original documents and ledgers that were written in different periods between 1840 and World War II. The records of the mid-nineteenth century are completely different from those found a hundred years later. Victorian records are hand-written in ink; they are kept in ledgers or piles of loosely related, pinned papers; statistical analyses are rare, based on small populations, and not highly disaggregated. Records from the 1920’s are either printed, typed, or mechanically duplicated; the use of ledgers has declined, while the file classifications have become much more refined and detailed; statistical analyses are now common, cover the organization as a whole, and are available at very fine levels of disaggregation. It is not hard to examine a document, determine how it was made, and estimate what skills would have been required in its production. Thus the Taylorite descriptions of the cognitive demands of differing office technologies have been supplemented by my own assessments, based on a subjective interpretation of the physical properties of the papers containing the records of this study.
An examination of such materials can really change one’s perception of scientific management. Ever since Braverman (1974), it has been customary to associate Taylorism with time-motion study and the speed-up. Braverman makes explicit reference in his discussion of clerical work to automated typewriters in which work is passed from clerk to clerk in assembly-line fashion. However, the office management literature paints a picture of Taylorism in the office that is both more mundane and more profound. Time-motion study and the development of micro-controls over the physical motions of workers played a trivial role in the scientific reform of clerical work. The contribution of the Taylorites was the invention of virtually every piece of office equipment we are familiar with today. Their contribution to the typewriter and the adding machine is well known. But few people stop to consider that file cabinets, rolodex cards, hanging manila envelopes with alphabetical tabs, postage meters, loose-leaf notebooks, swivel chairs, staplers, paper clips, carbon paper, and duplicating machines are all inventions of the Taylorite era. They were all absent from the Victorian office; they were all developed to deal with inefficiencies of early offices; and they all are basic to clerical functioning today. Some of these innovations represented de-skilling. Others, however, were labor-saving devices designed to keep clerks from wasting their time on tedious, menial, and time-consuming tasks. The postage meter, for example, eliminated the licking of stamps. Licking stamps cannot be said to be one of the rare and valued skills of a traditional clerical artisan.
One of the more important technological changes in the office was the simplification of arithmetic calculation; the Taylorites were fairly successful in reducing the necessity of doing arithmetic operations by hand. Adding and calculating machines reduced the need for bookkeepers to do complicated manipulations such as adding long, repetitive columns of numbers or doing long division with unwieldy units of analysis such as tons, hundredweights, pounds and ounces, or pounds, shillings, and pence. Basic arithmetic skills could be found in the general population and did not require special training. Being able to do hand arithmetic for hours on end without making a mistake, however, requires a level of patience and special concentration. The loss of this would have represented a modest de-skilling. There existed other alternatives to hand calculation that would have had different impacts on skill levels. The use of published tables such as interest tables would have been an absolute de-skilling since the need for calculation would have been eliminated entirely. On the other hand, the Taylorites often recommended the use of slide rules. Learning to operate a slide rule would have been a wholly new skill that represented a significant upgrading.
There were at least five major innovations in the processing of business correspondence. These were the development of shorthand, dictating machines, the typewriter, mechanical duplication, and the addressograph. Each of these had very distinctive effects on the skill required to produce, record, and mail a business letter.
Shorthand was one of the major artisanal accomplishments of the turn-of-the-century clerk. Although shorthand was not extensively used in the early nineteenth century, over time employers realized the advantages it offered so that, by the late nineteenth century, fluency in shorthand was expected of any professional stenographer (Pitman 1891). The GWR, for example, required all male clerks to learn shorthand within the first few years of the senior clerkship (Salaried Recruitment 1912). Shorthand represented a very significant skill acquisition, since not only did the clerk have to learn the various symbols, but he had to develop speed and accuracy in taking dictation while using the symbols. The competent practitioner benefited from the fact that neither the general public nor the average employer could read a communication written in shorthand. As such, it could be used as a secret language to generate a certain feeling of exclusiveness. The British Museum has an extensive collection of shorthand journals dating from the turn of the century. These are magazines written entirely in different shorthand systems catering to the clerical community. The circulation and importance of these journals is unknown; their readership may have been confined to a small number of hobbyists, or they may have been distributed throughout offices nationally. However, the presence of such journals demonstrates that shorthand represented a body of knowledge that was sufficiently complex and involving that some people took an interest in the system for its own sake. Thus the diffusion of shorthand literacy that took place in the mid-nineteenth century would have represented a major upgrading of the clerical population.
The dictating machine was designed to undercut the artisanal base of shorthand writers by allowing the direct transcription of letters from voice. Had dictating machines been universally accepted, which in general they were not, the dependence of employers on shorthand clerks for dictation would have been completely broken. Before World War II, there were relatively few Ediphones in active operation. Taylorites remarked on the difficulty of working the machines and the preference of many employers for interaction with a live stenographer. Only in the 1970’s—with the development of convenient, high-quality cassette recorders—was shorthand effectively removed from the working skills of the professional clerk.
Typing represented a significant upgrading of clerical work. Most discussions of office work treat typing as a menial occupation because it compares unfavorably with managerial positions. Such a position fails to recognize that the introduction of the typewriter represented a significant upgrading over what had gone before, which had been hand-copying. The only skill required in hand-copying is basic penmanship. Typing represented a quantum shift in the training requirements of copyists by demanding the acquisition of a whole new set of motor skills.6 Not all of this new human capital was provided by the firm; much of the introductory typing training was done by professional business schools and by the typewriter companies themselves (Davies 1982). However, the Taylorites argued quite forcefully that typing schools only provided a rudimentary familarity with the keyboard. Many typists did not know the touch system. Those who did were often extremely slow and inaccurate, requiring long hours of on-the-job practice to develop reasonable standards of speed. The Taylorites devoted a great deal of attention to increasing the rates of production for typists. Time-motion studies were done to discover the fastest way to feed paper into the machine, the best posture for rapid typing, and the most efficient positioning of the material to be copied. Competitive typewriting contests were held every year, and the techniques of the winners were studied carefully. The time-motion studies were attempts to identify, codify, and publicize the procedures used by the most artisanal typists, and they represented a conscious attempt by office engineers to de-skill by shifting the locus of expertise from workers to management. Nevertheless, what is important is that typing was poorly understood by both Taylorites and typists alike. The long training periods combined with a stock of complex techniques for improving performance are what one would expect from a demanding, difficult job. Typing made such intense demands on clerical workers that Taylorites were forced to turn to time-motion study to find compensatory simplifications.
Duplicating machines and addressographs are examples of upgradings that were produced by labor-saving machinery’s being introduced to comparatively unskilled positions. Carbon paper, mimeographs, and hectographs all replaced hand copyists with other operatives who were equally unskilled. However, since a mimeograph operator can turn out multiple copies in a small fraction of the time it takes to produce individual transcriptions, the demand for copyists became severely reduced. Addressograph machines eliminated the task of repetitively hand-addressing envelopes. Once again, a small number of clerical proletarians replaced what had once been a significantly larger force.
Reforms of business correspondence tended to produce upgradings of the clerical force. However, many Taylorite innovations significantly undercut the skill requirements of office workers. One of the most important and least discussed attacks on the stock of clerical knowledge came from the rationalization of filing. Most of the modern equipment used in filing, such as file cabinets, loose-leaf binders, rolodexes, and card indices, were developed by the Taylorites in the early twentieth century. Before loose-leaf filing, business transactions were recorded in bound ledgers, while loose documents were tied together in large bundles and stored separately. Bound ledgers were incredibly inefficient. When a new account was opened, a given amount of space was allocated in a book. If the account outgrew its space, or if part of the account required different treatment, a new section of a book would have to be allocated. The result was that related materials became strung out in a series of isolated, unrelated volumes. Filing was also impeded by poorly designed classification systems for loose documents. It was not uncommon for Victorian firms to file their business correspondence by date of arrival, rather than by client or subject matter. The result was that lost documents were common. Long searches for missing material and random research expeditions looking for information were a standard component of a clerk’s working day. The only convenient way information could be found was if someone had a deep personal acquaintance with the files. An experienced clerk who could remember transactions that had occurred years before was an invaluable asset in producing current business analyses..
The Taylorites developed alternatives to the tradition of bound volumes and loose piles of pinned papers. The introduction of looseleaf records that could be recombined represented a breakthrough in providing file categories that made substantive sense. Files could be adjusted to reflect current business conditions, and mutually relevant materials could be kept together. With the new filing systems, data collection was transformed into a routine activity. This resulted in a significant de-skilling of file clerks, since it eliminated the need to rely on clerks who could remember where everything was recorded. Offices now could be run with inexperienced workers, and it was less important that clerks learn to remember key transactions.
A more well-known innovation that caused de-skilling was the creation of an elaborate division of labor within the office. One of the most common Taylorite recommendations was that managers develop "routines” for most complex tasks; such routines would involve sending work to particular clerks for the execution of single steps within the process. These divisions of labor probably preceded scientific management. On the Great Western Railway, formal descriptions of job duties can be found in the late nineteenth century. At any rate, the fragmentation of the clerical labor process would have reduced a worker’s exposure to a large number of clerical tasks and would have contributed to reducing the clerk’s intellectual command of office procedures.
Another source of de-skilling was the development of standardized pre-printed forms and letters. Printed forms removed the responsibility of the clerk for designing the layout of a given account and ensured that all the necessary information would be included on the document. The development of a rational system of forms occupied a high percentage of an office manager’s attention. The Great Western Railway had an in-house committee of Taylorites who were active from World War I to the 1930’s. Their official assignment was to consider all aspects of clerical functioning, making concrete recommendations of sources of economies. It is highly significant that in the first fourteen years of this committee’s existence it did not execute a single time-motion study. One such study was attempted in 1930 and abandoned because it did not produce any significant savings (Labour Report 1931). The Great Western Clerical Work Committee spent the bulk of its time on the review of printed forms. The overwhelming majority of the minutes in its records refer to the elimination of forms or the alteration of the format of various GWR documents (Clerical Work Committee 1916-28).
It is not hard to justify such a single-minded focus. A printed form was implicitly a rule book that laid out in elaborate detail how any given clerk was to spend his or her time. The printed form spelled out precisely what information was to be collected, what office was responsible for collecting it, and what other information must be obtained at the same time. The development of such forms did not start with the Taylorites, since even the earliest offices have elaborate specifications for the formats of returns and accounts. However, many unspecified data collections can occur in the form of letters and memos, where the writer can report on phenomena in any manner he or she chooses. The proliferation of forms would have reduced the right of workers to compose their own memos and would have greatly reduced clerical skill and autonomy.
Many of the items that have been discussed in the previous pages have been sources of significant downgrading that were produced by clerical mechanization and rationalization. There is an important countervailing factor that would have upgraded the labor force: the increasing importance of repairs and mechanical maintenance. The early Victorian office was unmechanized. Because of this there was very little that could break down or go wrong. Ink wells had to be filled, and pencils had to be sharpened, but there was very little that could not be fixed instantly or, at worst, replaced. The maintenance problem became far more serious when office functioning became dependent on a large number of complex, expensive machines with multiple moving parts. There was a larger number of units that could go down unexpectedly, each one putting a clerk or a roomful of clerks out of work. These problems could be handled by outside repairmen. If the pace of work was to be maintained, however, it was important for workers to be able to do minor repairs themselves. This eliminated the downtime that is always inevitable when workers must wait for an outside technician. The ability to fix equipment on the spot required mechanical skills that had never been part of office work. Most secretaries can fix their own typewriters if they have to. Many can also improvise successfully with a mimeograph or Xerox machine. To some extent, this can be considered a loss of status, since in this respect the new white-collar jobs had blue-collar manual duties. Nevertheless, repair work added a new human capital dimension to office work that made demands on the skills and creativity of the whole labor force. Since this required mechanical aptitude that had no precedent in the Victorian office, this surely must have been an upgrading.7
Conclusion
The above analysis shows that office rationalization affected skill levels in a wide variety of different ways. Some innovations produced significant de-skilling, such as the rationalization of filing and the use of adding machines. Others produced an upgrading, such as the introduction of typing and the mechanization of duplicating. From the qualitative analysis, it is impossible to say whether upgrading or downgrading forces predominated in the office. It is not inconceivable that these forces canceled out, producing an overall effect of little net change. Such an interpretation is consistent with the statistical analysis, which showed different outcomes for different departments that averaged out to an average finding of little net change. To the extent that there is a directionality to the movement, the salary and pyramid data suggest that clerical work upgraded rather than deteriorated.
The analysis in this chapter suggests a reformulation of the relation between sex and job status. If women were generally concentrated in low-status jobs, then feminization could be explained in two ways. The first is that the absolute number of low-status jobs increased, allowing for an increase in the number of jobs appropriate to women. The second is that some other transformation occurred that provided women with a greater share of those low-status jobs that already existed. This chapter suggests that the first process did not occur; what happened was that women were taking menial positions that had in an earlier era been given to men. However, this still does not explain why women are concentrated in low-status jobs in the first place. This question cannot be answered with a study of the evolution of skill levels, but must be addressed by explicitly examining the process by which men and women get sorted into particular jobs. The next section relates the subordination of women in the workplace to internal labor markets and the problems of turnover and legitimation that can occur in such settings.
Notes
1. For the period 1946-56, Lockwood also used reports from the Office Management Association.
2. The figures on total combined salaries were calculated by weighting the male and female salaries by the percentage of male or female clerks within each industry.
3. It is also significant to note that the introduction of women to clerical work did not significantly undercut total wage levels. Women worked for significantly less than men, their salaries generally being only 50–80 percent of those of their male counterparts. However, the secular increase in both male and female wages counterbalanced the gender differential. Feminization helped to contain spiraling clerical costs but did not allow management to lower overall absolute levels of pay.
4. Both Klingender and Lockwood acknowledged the robust character of clerical salaries between 1909 and 1929. Both authors attempted to bolster their conclusions by arguing that, although white-collar salaries had risen absolutely, they had fallen relative to blue-collar wages. The blue-collar argument is irrelevant to the issue of skill. The relative decline argument was based on the assumption that clerks experience deprivation by losing their customary pay differential over manual workers. As Bain points out, this assumes that blue-collar workers are a meaningful reference group for clerks. Manual workers are probably significantly less important in this regard than the clerk’s family, his former schoolmates, and other members of his occupation (Bain 1970). Furthermore, the psychological well-being of white-collar workers is not an index of the objective conditions of white-collar work. What is at issue is whether the human capital requirements of clerical work declined and whether this would have affected the economics of hiring women. The perceptions and world views of clerical workers have little to do with corporate hiring policies.
5. Since the contents of these works are extremely similar, I have dispensed with citations of individual authors in the following section. Appendix E contains a list of the more useful and general references. The reader seeking to validate the argument of the next section can find detailed descriptions of the skills required to operate various kinds of office equipment in any of the books in this bibliography.
6. To give the Victorian clerk his due, it should be noted that many clerks had excellent handwriting. Many of the account books and letters of the GWR in the nineteenth century are filled with striking examples of professional calligraphy. Such handwriting is generally absent from postal documentation and disappears from the GWR records in approximately the 1890’s. Even in the prime of the Victorian age, some of the GWR handwriting was quite mundane. Nevertheless, there do seem to have been talented individuals in the clerical force whose penmanship rose far above the requirements of mere legibility.
7. A non-skill-related aspect of Taylorism, which has been neglected, is their advocacy of clean, warm, well-lighted workplaces. Many early offices were quite substandard, being dark, underheated, underventilated, and provided with dubious sanitary facilities. Cleanliness was also an issue. Even in the 1920’s, the Railway Clerks Association had a benefit fund for phthsis victims; “white lung” was contracted by the inhalation of excessive amounts of loose dust. One can imagine the appearance of an office whose records were capable of giving clerks asthma. The Taylorites scientifically demonstrated that the increased productivity of clerks in decent facilities more than offset the extra cost of housekeeping and fuel. The Taylorite research on the relation between comfort and productivity was an important contribution to the improvement of the physical quality of life in both factories and offices.