GENDERED DISCOURSES OF WORK

A virtually omnipresent fact of industrial life is the categorization of jobs as “men's work” and “women's work.” Men and women seldom perform the same jobs on the shop floor, and one cannot help but be struck by this profound gender segregation upon entering a factory. A constant theme in the literature on women factory workers is that employers associate a number of positive features with female labor that are not reducible to lower wages. I use the concept of gendered discourses of work to capture these ideas that employers hold about men and women workers.

Although scholars studying the massive flow of women into factory work in developing countries address the attributes that I refer to as gendered discourses of work, they integrate them differently into their analysis.³

The scholarship that gives gendered discourses of work a causal role in shaping shop-floor divisions of labor usually focuses on a small number of factories and rarely offers theoretical explanations for broader gendered trends in employment. Scholars also seldom trace change over time in gender divisions of labor, even when they pay careful attention to other dynamics on the shop floor. The varying impact of gendered discourses of work on actual divisions of labor is therefore usually overlooked. Those offering theories of feminization, in contrast, rarely incorporate examinations of shop-floor practices of job allocation into their analyses. Although they acknowledge the existence of managerial beliefs about women workers (e.g., docility, dexterity, tolerance for monotony), they place the main causal weight on low wages. Since the causal force of gendered discourses of work are seen as pushing in the same direction as wages, gendered discourses of work are usually subsumed under the wage argument, and the independent causal impact of gendered discourses is lost.

Authors also vacillate between treating gendered discourses of work as managerial subterfuge or as traits that women actually possess (Elson and Pearson 1981; Fernandez-Kelly 1983). For example, at one point Diane Elson and Ruth Pearson argue: “It might seem to follow that the labor force of world market factories is predominantly female because the jobs to be done are regarded as ‘women's work.’ But to note that jobs are sex-stereotyped is not to explain why this is so. After all, capitalist firms are compelled by competitive forces to select their labor force and constitute their division of labor on the basis of profitability, not ideology” (1981, 92). On the following page, however, the authors note that women's “nimble fingers” are the result of training, and they specifically mention how industrial sewing is similar to sewing in the home on domestic sewing machines (p. 93). Elson and Pearson shift from calling these traits “ideological” to agreeing that women have them and to showing how women obtain them.

I argue that whether women possess these traits is irrelevant; the crucial point is that employers believe that they do. As elegantly stated by Leslie Salzinger, these images are important not because they reflect reality but because they produce it (2003, 9). In other contexts, feminist labor historians and sociologists have shown persuasively that gender shapes the way that employers put their economic interests into practice (Downs 1995; Humphrey 1987; Milkman 1987; Rose 1992). These scholars demonstrate that rational economic practice is partially constituted by gender and that employers view productivity and labor control through a gendered lens. Gendered discourses of work are therefore integral to an understanding of the gendered dynamics of hiring practices.

I adopt a synthetic conceptualization of gendered discourses of work, combining a Foucauldian (1990) notion of discourse, R. W. Connell's (1987) praxis-oriented perspective, and poststructuralist feminist approaches (Riley 1988; Scott 1988). I take from Michel Foucault the notion that discourse produces subjects. In this case, the gendered worker on the shop floor is created in part through the discourses produced by management about gendered workers. From a Foucauldian perspective, the issue is not whether women are patient, disciplined, and diligent, but that the subject of the woman worker as a patient, disciplined, and diligent worker is produced through discourse. Poststructuralist feminists have shown the utility of paying attention to how gendered subjects are constructed relationally through discourse, often in the form of binary oppositions. Male and female workers are produced relationally through a series of binary oppositions—for example, careless/careful, lazy/diligent, undisciplined/disciplined, strong/weak, heavy/light. Connell's praxis-oriented approach calls attention to how these discourses, once produced, become embedded in institutions and gain materiality, becoming part of the structure of everyday life and shaping relations within it. The importance of binary oppositions and their embeddedness in everyday practice is most evident in the separation of men and women into different job categories in the factory. When these varied approaches to discourse are combined, the contingency of these discourses is acknowledged (they do not represent “truth”) and their operation in everyday practice in the factory is integrated into the analysis. Discourse both structures how, and forms a lens through which, subjects in the factory view the labor process; discourse is both created and recreated—and possibly changed—by practice on the shop floor.

Although the factory is a site of the construction of gendered meanings, factories are not free-floating entities. Women are already gendered when they enter the factory, through law, culture, and politics. Gendered discourses in the wider society provide a foundation, or point of linkage, to the gendered discourses of work created in the factory. Given the vast ethnic and cultural diversity in Indonesia, there are a host of societal discourses and practices about gender.⁴

With the formation of a national polity and a national state that molds gender relations through law, public policy, and public pronouncements, however, an Indonesian or national discourse on gender has emerged (Blackwood 1995; Suryakusuma 1987, 6). The outlines of a hegemonic national discourse congealed during the New Order period.⁵

The New Order regime that came to power in the late 1960s adopted a discourse of separate but equal gender roles and emphasized the nurturing and selfless qualities of motherhood over other aspects of women's identities (Djajadiningrat-Nieuwenhuis 1987; Suryakusuma 1987; Tiwon 1996). As Saraswati Sunindyo (1998) points out, although women were also defined as mothers or housewives before the New Order, the regime put the force of the state behind this particular definition. The state expounded on women's “special” role in development as mothers and wives (Caraway 1998; Rahayu 1996; Sen 1998), and state-linked organizations, such as Pembinaan Kesejahteraan Keluarga (PKK) and Dharma Wanita, assured that New Order ideology about women was spread through the programs that they oversaw for the state (Sullivan 1994; Suryakusuma 1987; Wieringa 1993).⁶

A core element of national discourse on women is the term kodrat wanita, which is translated in a variety of ways: biological or female destiny (Sunindyo 1998); the intrinsic nature, true essence, or essential nature of woman (Tiwon, 1996); and women's nature (Wieringa 1993). Kodrat is often associated with the idea that women are soft and weak (lemah lembut) and comes up repeatedly in discussions about the proper roles of women. Kodrat is what makes men and women different in spite of all of their similarities as human beings. Essential difference is embedded in the concept, but the implication of this essential difference is contested and has changed over time. Its existence, however, is rarely challenged. Kodrat forms a bedrock dichotomy between male and female, which gets translated into—and is used to justify—gendered practices in society, including in the factory. The subterranean foundation of gendered discourses of work in Indonesia is kodrat.

During my fieldwork in 1998–99, management generally presented the distinctions between men and women in dichotomous terms and often explicitly justified gender divisions of labor by referring to kodrat. Managers considered women to be more careful, diligent, disciplined, and patient, easier to manage or control, and better suited for light and monotonous work.⁷

Men, in contrast, worked quickly but often carelessly, did not follow orders as well as women, were naughty,⁸ and complained more, but they were better than women at heavy work. I often received tautological answers when asking why women or men did a particular job: “It's women's work,” or “It's work that's compatible (cocok) with women,” and so on. For certain jobs, some managers were neutral, observing that for that job, either men or women would do, but supervisors seldom mixed men and women in the same jobs because they believed that if men were put in women's jobs (or vice versa), productivity would fall.

Consequently, men and women rarely did the same work, even when they worked in the same area of production; some sections of production were the exclusive preserve of one gender. Since employers viewed jobs in gendered terms, hiring practices were also gendered. Hiring a man for a woman's job would be disruptive, not only because supervisors considered the worker inappropriate for the job but also because most men resented being stuck in a task performed predominantly by women. Management might try out a man in a women's section, either out of curiosity or because of short-term labor shortages, but according to managers, unless the man was “effeminate” he would usually complain or quit. Managers speculated that the men who quit in these situations were embarrassed to do women's work. Other managers observed that men think that they are above doing women's work and would refuse to do it. Still others noted that men doing “women's work” risked being ridiculed and teased by other men. Masculinization of particular types of work usually proceeds in big leaps because enough men need to be placed in a section to give male workers a basis of solidarity.

In most factories, supervisors initiated hiring requests, a process that involved specifying the qualifications of the desired worker. Along with age and education, gender was a key “qualification.” The supervisor determined whether the jobs that needed to be filled required male or female labor, and requested the appropriate worker. In other factories, explicit or implicit company policy dictated particular gendered hiring practices. In these cases, supervisors did not need to specify gender on a hiring request, as the personnel department already knew, on the basis of which jobs were open, the number of men or women that had to be recruited. In addition to taking into account education and age, marital status was also an important consideration when hiring women. Employers overwhelmingly preferred to hire single women since they could extract a few years of work out of them before they married and started to have children. However, married women comprised a large proportion of the workforces in many of the factories in my sample because many of them married within a few years of being hired and continued to work after marriage.

Gendered discourses of work not only include factors that help managers decide on the best gendered worker for a job but also encompass broader features of interest to employers. Some managers regarded women as being less likely to go on strike, but most focused less on the propensity to strike and stressed instead that women were easier to control. They emphasized that women followed instructions without delay and without talking back, were flexible about doing work outside of their job classifications, stayed at their work stations, and were more punctual than men. The number of strikes rose precipitously in Indonesia in the late 1980s and early 1990s, and many of these strikes occurred in sectors that employed primarily women (Kammen 1997), and so it is unsurprising that many managers considered women to be as prone to strike as men. Managers' perceptions about rates of absenteeism are also important.⁹

For example, if management thinks too many women take maternity leave, then over time they might reduce their reliance on female labor by redefining relatively androgynous job classifications held by women as “men's work.”

Why do gendered discourses of work matter for explaining the feminization of manufacturing work? Feminization involves either a gendered redefinition of work—jobs that men previously claimed are redefined as women's work—or the assignment of new jobs to women rather than to men. But what leads employers to change or to establish a particular gender division of labor? I argue that feminization is one method that employers deploy to enhance productivity and labor control. Since gender is a key organizing principle in the factory, managers see productivity and labor control through a gendered lens. For example, if a predominantly male section is producing a lot of poor-quality work, managers may conclude that the work requires more care and patience than men possess, and they solve the problem by replacing men with women. If male workers give supervisors trouble, managers might attribute this to their gender rather than the supervisor's management style or to the stress or monotony of work in that particular section.¹⁰

Feminization is one method of raising productivity and enhancing labor control.

The structural imperative to remain competitive, however, does not result automatically in specific gendered changes in the labor process. A process of translation must occur whereby managers assess, given the resources at their disposal, how best to deal with the competitive situation that they face. Export-oriented industries, especially in labor-intensive sectors, are more likely to make adjustments to the labor process because they feel the forces of competition most acutely and therefore constantly seek out ways to squeeze more productivity out of their workers. Consequently, rates of feminization are higher in these industries than in more capital-intensive and less competitive industries. Wage pressures are important, but they are mediated through gendered discourses of work.

The cheap labor argument is inadequate for two reasons. First, the wage advantages that employers can secure by hiring women instead of men vary from country to country and in some cases are small or nonexistent.¹¹

In Indonesia, while women's average wages are lower than men's average wages, this figure does not tell us what workers are actually paid in factories. In the early 1990s, the Indonesian government systematized a minimum wage policy, and minimum wages set by sector and by region became the standard for pay in many industrial sectors (Harrison and Scorse 2004; Manning 1998).¹² Minimum wages have reduced male–female wage differences within sectors, most notably in labor-intensive sectors. In my factory survey, the ratio of male to female wages was close to one, and in some factories women's average wages exceeded men's, since employers often offered incentive pay to women in exchange for working during their two days of monthly menstruation leave, and since by law women received more generous overtime bonuses for night shifts than men. Hiring women instead of men also incurred other costs. Indonesian law mandates three months of maternity leave with full wages paid by the employer rather than by the state. Many employers evaded payment of maternity leave benefits, of course, but many paid them as well. Employers that paid maternity leave diminished its potential costs by devising elaborate tactics for reducing the number of women who claimed it. They hired young unmarried women, provided family-planning services at factory clinics, and encouraged rapid turnover through a regime of forced overtime, which caused many young women to resign after marriage. All of these strategies carry costs that are only partially reflected in the wage bill, but in spite of these costs, employer interest in hiring women remained keen.

Second, the cheap labor argument cannot explain the variability and the selectivity of the feminization that occurs. Why do some labor-intensive export sectors feminize more than others? And why do some factories in the same industry feminize while others do not? To answer this question, we must examine factory-level processes of job allocation and assess how gendered discourses of work shape the cost–benefit analysis that employers engage in when they make decisions about hiring and the allocation of work. Before turning to an analysis of these processes, however, it is first necessary to elaborate in greater depth the broad contours of feminization in Indonesia.

FEMINIZATION IN INDONESIA

A commitment to economic development was one of the central tenets of the New Order regime that came to power in the late 1960s. After eliminating the Communist Party and other potential opposition, President Suharto restructured Indonesia's economy to make it more appealing to investors (Winters 1996). He christened himself Bapak Pembangunan—the Father of Development—and set out to distinguish himself from the previous regime, which had failed to foster economic stability, much less growth. In the late 1960s, Indonesia was one of the least industrialized countries in the world for its size (Hill 1990a), and manufacturing employment in medium and large firms totaled only about 840,000 workers (Biro Pusat Statistik 1972). The Suharto regime sought to quicken the pace of economic growth, which led the regime to pursue a variety of new economic policies. Industrialization in Indonesia during the New Order can be divided into three major stages: import-substitution industrialization (ISI) (1968–78), the transition period (1979–86), and EOI (1987–97).¹³

During the first stage, ISI, the regime promoted manufacturing industries that produced primarily for the local market and relied mainly on oil revenue for foreign exchange. Capital-intensive sectors expanded most rapidly, and so employment generation was unimpressive, with employment in medium- and large-scale manufacturing increasing by only 130,000 workers during the 1970s (Biro Pusat Statistik 1972, 1981). In the transition phase, the regime reacted to a foreign exchange crisis by devaluing the rupiah and encouraging local investors in a number of labor-intensive sectors, such as garments, textiles, furniture, and plywood, to expand into the export market (Barr 1998, 9; Hill 1991). Domestic investors took the bait, and employment in these sectors of manufacturing began to grow rapidly (Biro Pusat Statistik 1981, 1986).

As oil exports became a less reliable source of foreign exchange, the Indonesian government intensified its efforts to promote export industries in the mid-1980s. The liberalization of imports in May 1986 and a 31% devaluation of the rupiah in September were the first major steps that set Indonesia on the EOI path. More extensive liberalizations of both trade and investment regulations took place over the next two years.¹⁴

These policies made Indonesia increasingly attractive to foreign investors, and led not only to more remarkable export performance by industries that had already begun to prove their exporting mettle, but also to the rise of footwear and electronics as major new exporters. Employment in labor-intensive export sectors boomed, doubling between 1985 and 1990 from just over 700,000 workers to over 1.4 million (Biro Pusat Statistik 1986, 1991). About 700,000 more jobs were added in these sectors between 1990 and 1995 (Biro Pusat Statistik 1991, 1996). Although employment growth in the labor-intensive sectors outpaced that in capital-intensive sectors, the most capital-intensive also posted impressive gains.

Wider developments in society and politics facilitated the flow of women into factory work. Women's educational levels and labor force participation rates were rising (Oey-Gardiner 1993), and fertility rates were falling (Gertler and Molyneaux 1994). Employers reported in interviews that the ban on night work for women began to be enforced less strenuously in the late 1970s.¹⁵

State discourse on women, moreover, increasingly supported women's economic participation. Until the 1980s, the state had stressed women's “special” role in development as mothers and wives, but in the 1980s, new emphasis was placed on women's role as a human resource in the economic sphere (Caraway 1998; Manderson 1980; Rahayu 1996; Sen 1998; Sullivan 1991, 1994; Suryakusuma 1987; Wieringa 1993). This change in government discourse was provoked in part by the increased interest of international organizations in promoting gender equality and women's role in development. By 1988, the national economic policy statement called for an upgrading in women's skills and education to enable them to take advantage of work opportunities (Sen 1998). Islamic religious authorities, moreover, largely stood by as women flowed into the workforce.

Shifts in industrialization and a social and political environment that placed few restrictions on the exploitation of female labor combined to generate an impressive wave of feminization. When Indonesia embarked on its industrialization drive in the early 1970s, men held the majority of jobs, but by the mid-1990s, men and women each composed about half of the production workforce (see Table 1). In 1971, only 3 out of 25 sectors employed more women than men (tobacco, garments, and other). Over the next 25 years, women's share of the production workforce increased in most sectors, and women became the majority in 6 additional sectors (textiles, footwear, other chemicals, plastic, electronics, and professional/scientific equipment). As feminization theorists have argued, EOI was an important factor driving this transformation. While ISI concentrated investment in relatively capital-intensive sectors and generated little employment growth, EOI favored labor-intensive industries and created enormous job growth. The labor-intensive sectors that were the major success stories of EOI, such as footwear, garments, textiles, and wood-based industries, generated much of the new employment for women. As would be expected by feminization theorists, feminization of the greatest magnitude occurred in sectors with high and medium-high levels of labor intensity, particularly in those that were major exporters.

This gendered transformation, however, is both much deeper and more varied than the EOI narrative suggests. First, while many export sectors were female intensive, this does not hold true across the board. For example, wood industries were overwhelmingly export oriented and very labor intensive, yet men remained the majority in these industries. Even more surprising, the most female-intensive industry in Indonesia was an inward-oriented and relatively capital-intensive sector, tobacco. Second, even among labor-intensive export industries, the degree of feminization that took place varied dramatically. Thus, while significant feminization took place in both textiles and wood-processing industries, the change in women's share of employment was greater in wood products (38%) than in textiles (20%). Third, even when controlling for exporting and labor intensity, women's actual share of employment varied widely. For instance, women were a higher proportion of the workforce in footwear than in wood products, even though both industries were extremely labor and export intensive. Fourth, the scope of feminization extended far beyond export sectors. A quick glance at Table 1 demonstrates that feminization affected almost every sector of manufacturing. And finally, as I will show, the extent of feminization within sectors was often uneven, with some firms feminizing and others sticking with men. In sum, even when controlling for labor intensity and exporting, sectors and firms within sectors varied both in terms of women's share of employment and in terms of changes over time in the share of employment, and even sectors that were relatively shielded from international competition feminized. I argue that in order to understand both the broad contours of feminization and the variability of its enactment, gendered discourses of work must be integrated into the analysis, and careful attention must be given to factory-level processes of job allocation.

GENDERED DISCOURSES OF WORK AND FEMINIZATION IN THE TEXTILE INDUSTRY

Gendered discourses of work are fundamental to gendered transformations of work at the plant level. A precondition for feminization is that managers construct or reconstruct gender divisions of labor in a way that channels work to women. But why do employers make particular gendered hiring decisions, and what leads them to change existing gender divisions of labor? Although gendered processes were at work in all four sectors in which field research was conducted, due to space limitations I will analyze two contrasting cases in one industry, weaving.

In 1971, women composed 37% of production workers in the weaving industry (Biro Pusat Statistik 1972), but over the next 25 years, the weaving industry underwent feminization, and by 1996, women held 52% of the production jobs (Biro Pusat Statistik 1997). Although feminization took place in a number of job categories, the feminization of work on the looms accounts for most of it. Existing scholarship on the weaving industry in Indonesia shows that in the 1960s and 1970s, men were commonly found on the looms (Hardjono 1990; Manning 1979; Willner 1961). In my sample of 10 weaving factories in Central and West Java, 6 were established before 1975, and all but 1 employed men on the looms at that time. By 1998, 7 of the 10 factories employed primarily women on the looms, 2 used mostly men, and only 1 relied exclusively on men. In the middle to late 1970s, some employers began to experiment with using women in some sections of production, and more and more concluded that women could do much of the work in weaving factories, most notably jobs on the looms.¹⁶

Although feminization was widespread, factories adopted widely divergent gender divisions of labor.¹⁷

Even feminizing employers were reluctant to place women in certain areas of production, and some firms did not feminize at all. Factories branched onto one of two paths—a feminizing and a nonfeminizing path. Two examples, Femitex, which feminized, and Machotex, which did not feminize, illuminate how gendered discourses of work play a role in the allocation of work at the plant level. Feminization occurred at Femitex because managers identified a gendered solution to productivity and labor-control problems while at Machotex, managers did not feminize because they doubted that changing the gender division of labor would enhance productivity or simplify labor control.

Femitex, an Indonesian-owned weaving factory in Bandung, employed primarily men when it began production in 1968.¹⁸

At Femitex, the opposition between female/male was expressed in a series of binary oppositions—easy to control/rebellious, light/heavy, patient/impatient, careful/careless—and managers placed women and men in jobs that they considered to best match their respective gendered attributes. Managers also thought gender affected how workers responded to different wage systems, with men working faster than women under piece-rate conditions and women putting in a relatively consistent level of effort regardless of the payment system.

As in all weaving factories, the loom operator job absorbed the most labor. Work on the semiautomatic looms required some physical stamina, and since the weaver had to carry the finished product to the warehouse, managers considered the work on the looms to be heavy and thus more appropriate to men. In addition, since pay was based on a piece-rate system, management favored hiring men, as under this method of payment they believed that men would produce more than women. The piece-rate system, however, generated high levels of conflict between workers and management, since wages were cut for product that did not meet quality standards. Management developed extremely negative opinions about male workers. The factory manager noted that “men cut work, are naughty, aggressive, and more likely to strike.”¹⁹

Management was dissatisfied with male workers, but as long as semiautomatic and manual looms were used, with payment based on a piece-rate system, they were reluctant to employ women on the looms.

In 1973, the introduction of new automatic weaving machines opened the door for a gendered solution to management's labor-control problem. The quantity of production on automatic looms depended less on the diligence and stamina of the operator, and so the company abolished the piece-rate system and initiated a daily wage system. Under a daily wage system, managers thought that women would be more consistent workers than men, and since the looms were automatic, less physical strength was required. After weighing their options, management concluded that replacing men with women would probably not negatively affect productivity and that labor control would be easier. As stated by the factory manager, “Men's production is more than women's, but not by much, and the quality is about the same. But men are naughty so we put women on the new machines and replaced the men in stages on the old machines.”²⁰

Management therefore readily attributed problems of labor control to the gender of the workers filling the jobs, even though shifting work from men to women carried some significant short-term costs. Men had more experience than women in operating looms, and hiring women required investing time and resources into training a new workforce. Managers also did not consider the possibility that much of the labor conflict was a product of the piece-rate system of payment, rather than the gender of the workers, and that men might be more docile if their pay was not tied directly to the amount produced.

Even though men were considered to be more troublesome and women easier to control, this belief did not lead to a wholesale replacement of men with women. As stated by the head of the weaving section, “The most important thing is that women be placed in work that is in line with their kodrat.”²¹

Management considered women to be appropriate only for work that was light and that required patience and attention to detail. Feminization was therefore selective and occurred only in areas of production in which management regarded women to be equally or more productive than men. For example, management never feminized the sizing section, and they continued to use men to carry things between sections of production, because they thought this work was too heavy for women to perform as well as men.

In the 1980s, however, management did feminize an additional section of production, warping: “Men's work performance was poor so we replaced them with women, and the results were better.”²²

A supervisor further explained that “warping work requires patience and care, so it is really more appropriate to women.”²³ Whether women are really more productive or obedient than men is beside the point. Rather, the fascinating dynamic that merits attention is that management attributed individual traits—whether positive or negative—to the gender of the worker. When productivity in warping declined, instead of developing general ways to make workers more disciplined and productive in their respective jobs, management replaced men with women. In the early 1970s, women were only 20% of the workers at Femitex, but by the end of the 1980s, women claimed about 75% of the production jobs.

In contrast to Femitex, Machotex enacted few, if any, changes in the gender division of labor in their factory complexes in two suburbs of Jakarta (Tangerang and Bekasi). Machotex is a locally owned and diversified textile concern established in 1974. In late 1998, they employed an average of 36% women in the weaving factories, and over the years this percentage has not changed. Like managers at Femitex, those at Machotex also considered women to be more appropriate for light work that required patience and attention to detail, and men to be better suited for heavy work. They were less insistent than managers at Femitex about the rebelliousness of male workers, and most managers simply highlighted that women tended to follow orders more quickly than men and were more flexible about doing work outside of their job description. The main difference between Femitex and Machotex was thus not the gendered discourses themselves, but the linking process between these discourses and the jobs in the factory.

At Machotex, managers classified machines as either heavy or light, and then within the jobs categorized as being light, they distinguished between those requiring greater attention to detail and those that demanded less intensity of concentration. They then assigned the appropriate gendered worker to the job. This linking process led to a dramatically different gender division of labor at Machotex than at Femitex. Unlike the managers at Femitex, those at Machotex considered the work on the looms to be heavy, even though the looms were fully automatic, since the operator had to stand for long periods of time and supervise a number of machines that ran continuously. They therefore preferred to hire men as loom operators. Women were overwhelmingly concentrated in the inspecting, warping, and reaching jobs, all of which were considered to be light and to require great attention to detail. When I mentioned to one manager that many factories used women on the looms, he nodded but replied that he would only take women if “forced” to do so by upper management.²⁴

Managers at Machotex were also reluctant to hire more women because they doubted that women were cheaper to employ than men. Menstruation leave and higher premiums on the night shift made women more costly to employ, and since pregnant women were not allowed to work at night, pregnancies caused scheduling problems in the rotating shift system.²⁵

However, management was more than willing to put up with these drawbacks in sections where they deemed that women's special talents resulted in significant productivity advantages in comparison to men: “Women do have more costs associated with employing them, but we weigh that against their strengths, such as patience and attention to detail.”²⁶ Given that management viewed women to be more expensive to employ, it is unsurprising that on occasion, they experimented with using men in women's jobs, but all of these experiments were deemed to be failures. In the inspecting section, one manager recalled that “we tried out men in inspecting but a lot of flawed material got through, so we decided to stick with women, who are more patient and better at catching defective fabric.”²⁷ The failure of these particular men to perform well in these jobs was immediately attributed to the gender of the worker, rather than to the individual worker's shortcomings.

The gender division of labor thus remained ossified for more than 20 years at Machotex. A necessary precondition for feminization, then, is that managers identify hiring women for particular job categories as improving productivity and/or labor control, and this decision is profoundly shaped by gendered discourses of work. Although intense competition, especially in labor-intensive sectors, increases the probability that feminization will occur, employers that face cutthroat competition do not necessarily conclude that hiring women will give them a competitive edge. Feminization is only one conceivable solution to the problems of productivity and labor control that employers face. Indeed, these two firms confound the conventional wisdom about feminization. Femitex exported little of its production, yet it feminized, whereas Machotex exported nearly all of it but adopted no changes in the gender division of labor and continued to employ primarily men. These differences can only be understood by viewing gendered discourses of work as an integral part of the process through which employers make decisions about hiring.

By integrating gendered discourses of work into the analysis, otherwise puzzling practices become intelligible. Why did Femitex not feminize its entire workforce, given that men were considered to be more troublesome than women? Femitex limited the scope of feminization to jobs that it deemed that women performed as well as or better than men; even though men were harder to control, this disadvantage had to be tolerated in jobs in which men's productivity was thought to be much higher than women's. Why did Machotex, in spite of its exposure to global competition, eschew hiring more women? Because they thought that women were more expensive to hire than men, that women were not significantly easier to control than men, and because much of the work was deemed too heavy for them to perform as well as men. Why did Machotex hire some women, even though they considered them to be more costly than men? They hired women in the job categories in which women were thought to be more productive than men. As the two case studies show, the gendered cost–benefit analysis that managers engaged in is far more complex than conveyed by current theorizing of feminization. Wages are only one consideration that managers take into account, and even when they do, women may not be framed as “cheap labor” but rather as “protected labor” that incurs costs that employers are reluctant to bear.

Let us now turn to an analysis of broader gendered dynamics in the Indonesian political economy that contributed to feminization.

GENDERED DYNAMICS: FOREIGN INVESTMENT, STICKINESS, SPILLOVER, AND SNOWBALLING

Factory-level processes of feminization are crucial to macrolevel outcomes of feminization, but there were other gendered forces at work that contributed to the depth, breadth, and variability of the feminization process in Indonesia. In this section, I highlight some broader dynamics that contributed to feminization there. These factors include the inflow of significant foreign investment from East Asia and three interrelated dynamics: stickiness, spillover, and snowballing. Gendered discourses of work are important in all of them.

In the 1980s and early 1990s, rising production costs in the East Asian newly industrializing countries (NICs) led investors from these countries, as well as from Japan, to shift many of their labor-intensive operations to Southeast Asia. Investment from NICs surged in Indonesia and by 1990, they accounted for 30% of approved investments in terms of value and 58% of approved projects (Wells and Warren 1979; Wie 1991). This investment flowed overwhelmingly into labor-intensive and export-oriented manufacturing (Wie 1991). Foreign investors from Asia accounted for a large share of total employment in footwear, garments, and electronic components (see Table 2), and in these sectors, women on average comprised a larger proportion of the manufacturing workforce in foreign-owned than in domestically owned enterprises.²⁸

Koreans dominated in the footwear and garment industries, and Taiwanese investors also had a notable presence in garments. In electronic components, Singapore was the largest foreign investor, followed by Taiwan and Japan. Since there were few local restrictions on the use of female labor, foreign investors were free to deploy their preferred gendered hiring practices in Indonesia. Although Indonesian employers in these sectors also hired primarily women, this dynamic is worth noting for comparative purposes. Foreign direct investors can inject a new gendered dynamic into economies when they adopt different gender divisions of labor than do local producers. The degree of feminization would have been lower in Indonesia without the influx of foreign investors who came equipped with their own gendered visions about the best workers for jobs in these industries.

In addition to the impact of foreign investment, other gendered dynamics emerged as feminization unfolded in Indonesia—stickiness, spillover, and snowballing. As noted in the weaving case study, some factories bucked the feminization trend. This resistance to feminization, which I call stickiness, would be surprising to most theorists of feminization, but it is not so shocking if we take into account that feminizing is often a difficult process for established firms. First, the division of labor is rooted in deeply held practices about the relative productivities of men and women in specific jobs. The example of other firms that have successfully feminized is not necessarily enough to convince managers that hiring more women is a sensible move. Gender divisions of labor can therefore be quite persistent. Second, even if management decides to feminize, shifting work from men to women in jobs that involve a large number of workers takes time because employers are usually unwilling to fire the entire male workforce and must therefore wait for men to resign. Managers are often reluctant to place women in men's jobs while men are still doing the work, as men seldom welcome women into their realms of activity on the shop floor. Feminization was therefore easiest in firms that expanded, since employers could transfer men from jobs that they wanted to feminize to positions still deemed to be “men's work” in the new unit of production. New factories, in contrast, could hire women from the beginning. Given that gender divisions of labor are sticky, older firms should have lower shares of women workers than newer firms.

The firm-level data from the Biro Pusat Statistik (BPS) industrial survey support this argument. Factories that began production during the ISI period tend to be more male intensive today than the firms established during EOI, even though in most cases, even the ISI firms employ on average a higher percentage of women now than in the past. Figure 1 represents employment in a number of sectors according to when production began. The chart shows that the footprint of the past is still evident. Although older firms have feminized—that is, women comprised a larger proportion of employees in 1996 than in the 1970s—they still employed a lower proportion of women in their workforces than the factories established during EOI. This pattern even holds for inward-oriented sectors such as food that underwent explosive growth in the 1980s and 1990s. The newer firms were relatively free of the “sticky” legacy of the gender division of labor, and if they desired, could (and many did) start with a majority female workforce. As a result of this stickiness phenomenon, employment growth, through both the expansion of existing firms and the opening of new firms, is a prerequisite for significant feminization.²⁹

While stickiness slows down feminization, spillover and snowballing help drive the feminization process forward. Spillover captures the dynamic through which practices in a firm or in a sector influence those in another. Firms that feminized in the late 1970s and in the early and middle 1980s played an important role in later feminization, since they upset the association of many kinds of work with male labor. The most common reason employers give for hiring women is that “it's women's work”—but the 1971 statistics and the evidence presented in the case study show that in many instances, men used to do much of this work. These new employers could only consider the work to be women's work because of the earlier feminizers. Investors in later years were more likely to use women from the beginning, as the early feminizers had already shown that it was possible and profitable to hire women.

Spillover works in a number of ways. The first is simply by following common practices in the industry. For example, if new local investors enter into a specific industrial terrain, they will observe the practices of other firms in the area to get a sense for how they want to set up their businesses. Moreover, they will hire experienced managers who know about labor practices in the industry. If many existing firms already employ a lot of women, the chances that this practice will be replicated are high. Second, managers copy new practices that are perceived to be successful. A number of managers in textiles confessed to me that they had experimented with using women in additional areas of production on the basis of the knowledge that their competitor had successfully employed women in those jobs. Third, spillover can also work through the transfer of management from one factory to another, both within a larger group and between companies. For example, at one of the plywood factories in the sample, a manager who was transferred from another factory in the group surveyed the gender division of labor soon after his arrival and determined that women could do some of the jobs that men performed. He immediately began to feminize these job categories. Fourth, since factories are concentrated in industrial zones, the vision of thousands of women ambling through the factory gates undoubtedly led managers in other industries to consider the possibility of employing women workers as well. The existence of feminization in some sectors makes employers in other sectors more likely to consider shifting selected jobs to women as well. The data in Figure 1 also support this assertion about spillover effects; even though ISI firms feminized, the EOI factories employ more women. Spillover effects can move in either direction—feminization or masculinization—but in a previously male-dominated industry, the emergence of many successful feminized firms injects a new dynamic into decisions about the gendering of workforces and makes the choice of women more likely than in the past.

Snowballing takes place when employment grows rapidly in firms that already employ large numbers of women in their workforces. Employment growth in the 1990s was most rapid in female-intensive sectors, and although some of this growth came from newly established factories, much of it was also a result of the expansion of existing firms. Garment factories in Indonesia, for example, often started off as relatively small enterprises that had just one or two assembly lines. As they grew, they added additional sewing lines and expanded the cutting and finishing sections. Since they often already employed at least 80% women in their workforces, the growth of these firms had a snowballing effect in terms of women's employment. Employment growth therefore not only provides an opportunity to feminize but also allows for the replication of existing practices on a larger scale.