This article examines some of the more pertinent details of the feminization of clerical work in the context of early twentieth century Canada and the impact that this had upon gender pay inequality. More generally, we address the question of the conditions under which labor market segmentation, such as the feminization of clerical work, can be expected to adversely affect the relative pay of women. To this end new labor market and related estimates for Canada are developed. The Canadian economy experienced significant economic change during the first three decades of the twentieth century. Output and population grew at unprecedented rates while agriculture became less important, Canada's urban population surpassed its rural, and, as well, new forms of business organization were implemented in the private and public sectors alike (Altman 2001a). It was during these times of dramatic economic and social change that the structure of Canadian women's market employment was transformed in fundamentally important ways. During the 1900-1930 period clerical work became the occupation of choice for a growing percentage of female labor force participants and, in turn, clerical work became increasingly feminized.
A critical finding of this article is that the movement of women into clerical work in the early part of the twentieth century, and the subsequent "feminization" of the occupation, did not result in increasing the gender pay gap in Canada. Instead, it reduced this gap as women's pay in clerical work increased sharply relative to men's from 1900 to 1930. Since clerical work was consistently the highest-paying occupation for women apart from occupations employing a minuscule percentage of the female labor force, by moving into clerical work and thereby feminizing it, women contributed toward reducing the gender pay gap, even as the feminization of clerical work contributed toward increasing labor market segmentation or the differential employment of men and women across occupations. In other words, contrary to what leading scholars have argued, labor market segmentation need not contribute to a deterioration of the gender pay ratio against women. The example of the feminization of clerical work in Canada is a case in point. More and more women were simply finding employment in a well-paid occupation, albeit women were paid less than men here as they were in all major occupations.
In the traditional labor market segmentation models women are forced into low-wage occupations-women have no choice in the matter-thereby producing persistent gender pay gaps between women and men. (1) But in economies where women have acquired the legal and social capabilities to seek and choose employment opportunities, as was true of both Canada and the United States and most of Western Europe by the early twentieth century, we argue that women can be expected to seek employment where wages and other pecuniary and non-pecuniary benefits are greatest. Whether or not this results in more or less labor market segmentation critically depends on the distribution of relative benefits between women and men across sectors. Moreover, we argue that the existence or absence of labor market segmentation does not necessarily speak to the existence or absence of labor market discrimination. Discrimination, where it exists, often lurks in the shadows of any particular manifestation of labor market segmentation.
Since women were paid less than men, a literal reading of our estimates might suggest that by employing more and more women, employers were simply hiring the relatively low-priced labor. In the process, therefore, clerical work was transformed from its relatively high-wage standing, for men and women combined in 1900, into an average-wage occupation by 1930, all the while remaining a high-wage occupation for women. The picture we seek to paint, however, is far more complex. The very nature of clerical work was transformed over the course of the 1900-1930 period. It changed from being a relatively insignificant occupation flush with relatively highly skilled men in 1900 to one of increasing importance to employment in Canada by 1930. By then clerical work consisted of a whole gamut of new job types characterized by different skill levels and was no longer heavily weighted by elite high-paying jobs. Both men and women were being employed in the growing number of clerical positions that, as should be expected, were not at the top of the skill and pay hierarchy. And since most new jobs were not at the top of the pay scale, the relative weight of high-paying jobs fell, which meant that average pay across the many jobs types that comprised clerical work was bound to fall. Moreover, it is not at all clear from the evidence that women represented a supply of cheap labor. What is unequivocally clear, however, is that the jobs a large number of women were hired into served to increase their pay relative to what was earned by male clerical workers, on average. In addition, what women earned as clerical workers lifted them above the average pay rates of a third of all male employees in the Canadian economy.
These findings run contrary to the dominant historiographical school in Canadian women's history which is dominated by the view that labor market segmentation was largely responsible for women's relatively low rates of pay at the turn of the twentieth century and that the feminization of clerical work contributed to the inferior position of women in the labor market. (2) Nevertheless, our results are consistent with the most rigorous empirical work for early twentieth century United States (Goldin 1987) and with comparable results for advanced market economies where countries with the most labor market segmentation, such as Sweden and Norway, are characterized by the least gender pay inequality (Rosenfeld and Kalleberg 1990, 88-89). Thus, evidence of labor market segmentation or increases thereof is no proof, in itself, of the inferior economic status of women. In addition, the most recent research on the subject shows that the female-male gender pay ratio improved in Canada in the first three decades of the twentieth century, increasing from 0.46 percent to 0.53 percent, in spite of the fact that during this period there was no reduction in the significant amount of labor market segmentation prevalent at the time (table 1).
The Duncan index of dissimilarity, a basic measure of differences in the distribution of labor by group, suggests that by 1930, with a Duncan index of 55, 55 percent of women would have had to change occupations for there to have been no labor market segmentation. (3) There was no change in this important measure of labor market segmentation over the course of the 1900-1930 period. In addition, we find that the feminization of clerical work contributed to both reducing pay inequality and increasing labor market segmentation. Had clerical work not been feminized, the labor market would have been less segmented (the Duncan index would have been lower), but the gender pay gap would not have been reduced to the extent that it was. Over the 1900-1930 period, occupational change played a large role in reducing the pay gap without affecting the extent of labor market segmentation as women moved into relatively high wage occupations, such as clerical work, be they female or male dominated. (4) Moveover, Meltz constructed estimates for the Canadian labor force (1965) which reveal that the movement of women into clerical work continued from 1930 to 1960 such that clerical work's share of the female labor force rose from 18 percent to 29 percent and women's contribution to this occupation increased from 45 percent to 62 percent. Over this period, women's rate of pay compared to men's also remained the highest of all occupations that were of importance to women. In addition, the rate of pay for female clerical workers remained far above the average female rate of pay. Most recent research on the Canadian labor market also finds that labor market segmentation has little effect on gender pay inequality. Rather, at the close of the twentieth century, Canada's gender pay gap is largely explained by within-occupation pay inequality (Baker and Fortin 1999; Fortin and Huberman 2002).
Clerical Work in Context
An appreciation of the state of women in the American labor market at the turn of the century helps place the Canadian example in context. In the United States, women's share of clerical work increased dramatically from 1880 to 1890 from 4.4 percent to 19 percent and again from 1890 to 1900 to 30 percent. By 1920, women comprised 49 percent of clerical workers and by 1930, 53 percent. Over this entire period clerical work increased its share of the non-agricultural labor force from 1.7 percent to 9.7 percent. It is important to note that even as clerical work became ever more important to female employment, women's share of employment changed only by a few percentage points from 1900 to 1930, increasing from 18 percent to 22 percent. In 1910 and 1920, their share of employment stood at 21 percent and 20 percent respectively. Related to this is the percentage of women in the labor force, which increased somewhat from 19 percent to 22 percent from 1900 to 1930, while the percent of men in the labor force actually decreased from 80 percent to 76 percent over this period (Carter et al. 1997, series 11-25). Therefore, the growing importance of clerical work in the American economy and especially to female workers took place in the context of a relatively stable gender split in the labor force and represented a restructuring of female employment in favor of office work.
With regard to the pay of American clerical workers, Claudia Goldin also found that the female-male pay ratio in the clerical sector increased from 0.49 to 0.71 from 1890 to 1930, contributing to the overall increase in the American gender pay ratio in this period from...