2.5 A Gender Audit

While changes in industrialization strategy in developing countries have led to increased absorption of women as wage labourers into the modern sector, there are other questions to be asked as part of an assessment of this job creation from a gender perspective. Does the proportion of women employed remain high or fall over time? What happens to gender pay relativities as exports increase and industrialization proceeds? Have women in those developing countries with the greatest experience of participation in world markets in manufactures attained a more secure and beneficial position in the labour market as a result?

Scarcely any research has been done into these questions. Data limitations make it impossible to give proper answers to these fundamental questions for any one country, let alone in general. But some observations are relevant.

First, the evidence as to whether or not the demand for labour in export manufacturing remains female-intensive over time in quantitative terms is mixed. Wood's (1991) study shows that the high female share in the labour force has persisted in the large East Asian economies even after diversification. On the other hand, there are some later, apparent counter-examples which give cause for concern. These relate to changes in the gender composition of employment in certain EPZs.

EPZs are the epicentre of export manufacturing production in most developing countries and a strong concentration of women workers is found in them (see Table 5). With only one exception, they employ much higher proportions of women than manufacturing enterprises outside EPZs (in the exceptional case, Trinidad and Tobago, EPZ is a misnomer: the zone is an import processing zone). This paper does not attempt to sort out employment in export manufacturing as a whole from employment in EPZs and in foreign enterprises. Overlap within the three categories varies from country to country within the developing world. No estimates exist of the share of the manufacturing labour force involved in export production as such (not surprising, since many firms supply domestic as well as foreign markets); the workforce in EPZs is often taken to stand for the export manufacturing sector as a whole, but this is not correct, since exporting firms, especially but not only in Asian countries, are also found outside EPZs in all regions.²⁴

Nevertheless, it is important to understand the reasons for the fall in the share of women in some EPZs from the very high levels existing previously. In factories in the Mexican maquiladoras (the EPZ zone along the border with the United States) the proportion of women employed fell markedly during the 1980s from 77 per cent of the total workforce in 1982 to 60 per cent in 1990 (Baden and Joekes, 1993). There have also been falls in the share of women in EPZs in Singapore (World Bank, 1992) and Mauritius (from approximately 80 per cent to 60 per cent over the mid-1980s) (Frisen and Johansson, 1993; Roberts, 1992.

A cross-country study of EPZs carried out by ILO/UNCTC hypothesizes that the fall in the share of women workers in EPZs is a general phenomenon (see Figure 2 reproduced from this source) (ILO/UNCTC, 1988). Given that clothing and electronics (the main industries) are more female-intensive than other industries, it is proposed that the share of women workers in EPZs is related to the extent of product diversification, explaining both variations across countries in the share of women workers in EPZs and the decline in the share over time in some locations. Figure 3 shows the general relation between female intensity in the workforce and output composition of EPZs.

Both Mexico and Singapore exemplify their middle-income status and evolving factor endowment in having relatively high shares of more capital intensively produced goods in their basket of manufactures exports. In particular, they have both decreased the share of clothing in their exports to extremely low levels (just over 2 per cent of total merchandise exports in each case in 1993 (GATT, 1994: Table III.42) compared to 5-6 per cent in the early 1980s). The falling female share in the industrial workforce in EPZs in the two cases may thus indeed probably be attributed to product diversity.

The main form of diversification in the Mexican export border zone (maquiladoras) has been expansion of automotive assembly for the United States market. A strong cultural legacy of masculine metal-working means this industry employs mainly male workers, except in ancillary and finishing positions. The transportation equipment industry in the maquiladoras employed 33 per cent women workers, compared to around 80 per cent in electronics and 83 per cent in textiles and garments (ILO/UNCTC, 1988:66). But Mexico is unique among developing countries in having a long land border with a major rich country market, which favours bulky exports. Auto assembly may not flourish elsewhere.

Along with the Republic of Korea and Taiwan Province of China, Singapore has pursued its goal of product upgrading and deepening of the employment structure through a concerted policy of investment in technology and skills, through technology inflows (in capital goods imports, high technology foreign investment, and technical co-operation) and expansion of education in science and technology.²⁵ Singapore's remarkably high share of R&D in gross domestic investment (29 per cent, which far outstrips the rate in any other Asian developing country) (Lall, 1994:20) and the extremely high proportion of manufacturing employment provided by TNC affiliates (58 per cent) (United Nations, 1994) are other aspects of this policy.

While the gender gap in education in East Asian countries, including Singapore, is generally not large compared to other regions, gender disparities are evident in an enrolment gap at tertiary level and in the mix of courses and subjects taken by male and female students at this level. There is a strong bias towards science and technology subjects among male students. The swing back from female intensity in Singaporean manufacturing may thus be attributable, as a proximate cause, to the fact that women workers with the requisite technical qualifications are not available for recruitment to new technical and other skilled grades, because relatively few women have had access to the necessary subject education. More information is needed on the gender composition of employment in these grades, which to be equitably staffed we would expect to see employ 40-50 per cent women in keeping with the gender composition of the total labour force. The extreme female-labour intensity of employment in assembly-type operations is, as we have seen, a symptom of women's inferiority in the labour market and a swing towards more balanced representation would not necessarily be a negative development. As will be seen in Section 4.2, there is some evidence that in the East Asian region women do have a strong presence in scientific and technical positions in some countries in the information processing industry and in some branches of services.

The Mauritian case is quite different from Mexico and Singapore. The fall in the share of women cannot be explained by product diversification and changes in the factor intensity of production or skill mix of employment. There has been no output diversification, indeed the dominance of the clothing industry has increased (accounting for 89 per cent of EPZ production in 1990). The fall in the share of female workers has been attributed to a combination of factors: the exhaustion of the local female labour supply, the introduction of shift working to improve plant utilization levels (but without changes in technology or production methods) and the abolition of the male minimum wage (previously set higher than the female wage rate) (Frisen and Johansson, 1993).

Employers' preference for female labour is unproblematic where the unemployment rate is much higher among women than men (as in most Latin American countries, see United Nations, 1995) or where the general labour supply situation is generally tight (as in East Asian countries). In some other cases, employers' exercise of preference for female labour at lower rates of pay has been a cause of severe social tension. When male unemployment rates are very high, pressures arise for men to be given access to such jobs as are available, invoking aspects of gender identity such as the "male breadwinner" ethic. Such pressures were in evidence in the Dominican Republic in the mid-1980s in areas where men were dismissed in large numbers from the failing sugar industry. Employers were at that time resisting this pressure — more exactly, they were prepared to recruit men but only at the prevailing rate of pay, which men refused to accept without an additional "male worker" premium, despite their high rates of unemployment (Joekes, 1987). It is reported that the Mauritian government was persuaded to drop the high legal minimum wage for male workers because of men's hostility to the exclusive access to modern sector jobs given to women in EPZs (Frisen and Johansson 1993; Roberts 1992). It is a positive development here from a gender perspective that men have taken work, apparently without a wage premium.²⁶

How far do falling shares of women workers in EPZs undermine the female-intensive export industrialization thesis? There is no necessary inconsistency in the short term. If the export orientation pioneered in EPZs spreads to the rest of the economy, and newly exporting firms outside EPZs employ female labour while EPZs move away from the extreme female intensity displayed in their early days, the gap between female labour shares in EPZs and non-EPZs illustrated in Table 5 will be narrowed from both sides, with the overall sectoral trend continuing upwards.

But if non-EPZ firms also follow EPZs in due course and move to shift working, change the skill mix of employment (without training existing female employees and moving them up the job grade hierarchy) and so on, then the female share in the sectoral labour force may peak at a certain point. If women's employment proves to be confined to technologically low grade parts of industry, and change in the composition of the labour force in EPZs is a harbinger of gendered labour demand patterns elsewhere, indicating failure on the state's part to provide gender balance in educational provision and on employers' part to train women employees to higher level work, we may indeed be in a historically limited phase of the mobilization of female labour, with no consolidation of women's position in the labour market thereafter.

We move on now from investigating changes over time in the quantity of trade-related employment open to women to consider the issue of the quality of trade-related employment, including pay. The issue can be broken down into two parts. The first concerns cross-sectional evidence about the nature of women's employment in trade-related activities, compared to non-exporting activities; and the second relates to hypotheses about likely changes over time in the relative wages in the export sector compared to other activities and gender differentials in pay within the sector.

At the most general level, it is clear that new employment options in manufacturing represent a significantly better opportunity for women than the alternatives, especially but not only in the least developed countries. Physical working conditions are generally better in the modern industrial sector than in the informal sector, in terms of levels of dust, heat, light, and noise, with the biggest enterprises providing the best conditions (World Bank, 1992; ILO/UNCTC, 1985). Nevertheless, high-level health hazards exist in some activities, e.g. use of carcinogenic substances and risk of eyestrain in electronics and micro-circuitry production, which have not been adequately assessed. The literature also contains many accounts of the physical hardships entailed for factory women working extremely long hours under arduous conditions (Baden and Joekes, 1993), with no discretion over the timing or extent of work.

With respect to earnings, there is little doubt that even the worst paid factory jobs are a better alternative for women to low or unpaid work in the agricultural sector and to the poorly paid, usually non-contractual, work in urban areas to which their income earning opportunities are otherwise limited (ILO/UNCTC, 1985; Joekes and Weston, 1994). Whether women consider that the extra pay is more than sufficient compensation for the demanding conditions of factory work is another question on which the literature is mixed.

But a proper assessment of the effects of trade expansion requires evidence of employment conditions in trade-related compared to other employment within the manufacturing sector which is not available. There is information, as before, on the situation in TNCs and in EPZs compared with other firms, although, as noted, neither of these is necessarily a good proxy for the manufacturing export sector.

Within the manufacturing sector TNCs pay marginally higher wages than local firms, in both developed and developing countries (UNCTAD, 1992) and there is no reason to think that exporting TNCs do not conform. Several reasons are advanced to explain this: the occupational mix often differs between TNCs and local firms (the wage comparison should be adjusted to take account of this, but data do not allow it); TNCs are larger than the average firm; high product quality requirements in export-oriented firms lead the TNC employer to be prepared to pay a wage premium to workers skilled in particular operational methods to retain their loyalty (another application of the efficiency wage argument); and finally, political caution leads TNCs to wish to stand as good employers, above local criticism (ibid.).

If there is any concentration in women's employment opportunities in TNCs, then, and taking employment in TNCs as trade-related in the broadest sense, trade-related employment may give access to relatively good jobs for women. Whether this extends to improvement in pay differentials by gender is not known. On the one hand, for political reasons TNCs might be expected to observe equal wage legislation more punctiliously than local firms, where equal wage legislation exists. On the other, TNCs are merely incrementally "better practice" employers; it would be surprising if they countered labour market norms to the extent of ignoring the cost advantages presented to them by the prevailing gap in wage levels by gender and paid more equal wages to female employees than do local firms, even where equal pay conventions exist, and certainly not where they do not.²⁷

Information on employment practices and relative wages in EPZs has some limited relevance. Much research into employment conditions and pay in EPZs has been done, from widely varying perspectives. The feminist literature tends to concentrate on employment terms and conditions, but usually from an absolutist position, i.e., neglecting any comparison with conditions in non-EPZ firms. Equally unsatisfactorily, research into relative wages between EPZ and non-EPZ firms takes an aggregate view, but mostly ignores the gender dimension.

As regards, first, overall wage relativities in EPZ firms compared to non-EPZ firms in manufacturing, some sources (e.g. World Bank, 1992) assert that wages in EPZs are uniformly higher. But counter-cases certainly exist, e.g. in India and Mauritius, where wages are "definitely lower than in the surrounding national economy" (Starnberg Institute, cited in Baden and Joekes, 1993:15). This conforms to the idea of an EPZ "salary life cycle" as proposed by ILO/UNCTC: "in its early days, an EPZ generally offers wages which are slightly higher than other local wages. As the EPZ develops the wage differential gradually decreases and after 10 or 15 years it is not infrequent to observe that wages in EPZ industries are somewhat lower than other local wages" (ILO/UNCTC, 1988:102).

Such assessments of comparative wages ought to be adjusted for differences in the gender composition of workforce inside and outside EPZs, given prevailing gender gaps in wages in manufacturing (Baden and Joekes, 1993). If, however, the gender composition of the respective workforces remains constant over time²⁸ this suggests a secular decline in earnings for women relative to men. The fall in the female: male pay ratio over the 1980s in Sri Lanka (see Table 2) may reflect this, because in this case EPZs account for a significant proportion of total manufacturing sector employment (27 per cent in 1987) and 88 per cent of EPZ workers were women. If however, the female share of the EPZ workforce rises over time, this would in itself account for some of the relative decline in average wages. The question remains open.

The very high share of women currently in EPZs makes direct investigation of gender pay relativities inside EPZ firms problematic: occupational segregation may be so complete that, for instance, no men may be employed in the line-operative positions occupied by women. The few case studies of EPZ employment that address gender differences in pay nevertheless all report systematic underpayment of women workers compared to men, where they work in similar jobs. In some countries there is statutory provision for all firms, including those in EPZs, to pay women less than men; in other cases, such as Mauritius, Korea, Malaysia, Taiwan Province of China and the Dominican Republic, payment of overtly lower wage rates for women is recorded (Baden and Joekes, 1993). The only indication that EPZ employers might prove themselves better employers than the norm in gender terms, and challenge prevailing gender inequalities in pay, comes from the Dominican Republic — but, as noted, by attempting to level down men's wages rather than levelling up women's.

The last piece of anecdotal evidence relevant to pay relativities in trade-related employment comes from the new, rapidly expanding clothing industry Bangladesh. The wage gap by gender is extremely small in this case, with the female:male ratio in wages almost unity (0.97) (see Section 4). A female labour force has been mobilized with great speed in circumstances where wage employment for women outside agriculture was previously virtually non-existent.²⁹ One interpretation of the situation is that in order to create the kind of (female) workforce that experience in other countries suggests they will find most satisfactory, employers have offered women high start-up wages, on a par with male wages, even in the virtual absence of other opportunities in the modern sector, and with great gender disparities in income elsewhere in the economy. If this "kick-start" hypothesis is correct, employers may begin to exploit women's vulnerable labour market position once a female labour supply is established, and, as in agriculture, pay female wages below the male rate.

Moving now to changes in pay relativities over time, the first point to be made is that there is a clear prediction from international trade theory that over the long run trade will raise aggregate incomes and wages in developing countries. There is no doubt that incomes and wages in the export-oriented East Asian miracle countries have risen spectacularly over a relatively short time, and given their weight as exporters, that a statistical correlation exists between trade and wage increases. The micro-level mechanisms which might bring about this outcome, and more precisely how wages move in exporting firms compared to others, and the gender dimension of these changes, have not been researched.

The export sector may or may not prove to be a leading wage sector and may or may not affect the gender wage differential. If the export sector expands rapidly with high profitability as a price taker in world markets, then there may be scope for workers to share in the profits, and employers may be prepared to pay a wage premium to secure the best workers and retain their loyalty. Such a tendency might cover both increases in the average wage in manufacturing and in the female:male wage ratio, to the extent that women are concentrated in the export sector and there is a spillover effect which raises the level of wages women expect in other sectors (holding female labour supply conditions constant).

On the other hand, if producers' competitiveness is threatened, or market conditions are uncertain for cyclical or other reasons, exporters may look to ways of cutting costs — especially of labour, and especially perhaps of female labour — and thus, other things being equal, increase the gender wage gap.

Another way of addressing the possibility of changes in the wage gap can be derived from a simple model of dual (male and female) labour markets. In a trade-expansionary context the demand for female labour increases faster than for male labour, so that female wages also rise relatively fast and there is a tendency to "convergence" between female and male wages over time. But the world is not simple and many complicating factors could exist. Female labour supply might increase so rapidly that it offsets any tendency for employers to offer higher wages to women; the composition of the female labour force might change over time to include larger numbers of those groups (e.g. older, married women, as discussed above, or female heads of household) prepared to accept low relative wages, for reasons to do with the nature of the conjugal contract. Finally, the boundaries between male and female segments of the labour market are not fixed. It is clear that in other contexts the gender division of labour is revised in response to changing circumstances, though the fact of division persists (e.g. with respect to environmental change, see Leach, Green and Joekes, 1995). In the labour market context, redefinition of the segments certainly does take place, in terms of changing stereotyping of male and female jobs (Barbezat, 1993). The systematic shift away from recruiting women with changes in the skill composition of employment, discussed above, could be interpreted as a means for ensuring that excess labour supply conditions are maintained in the low wage female segment, countering any movement towards convergence of wages by gender.

The empirical testing of these various possibilities remains to be done. The only study carried out so far is by Tzannatos (1995),³⁰ which deals with very general question of changes in women's earnings over time with economic growth. Information is given on changes in relative wages for female and male workers over the 1980s at both national and sectoral levels for six countries (Côte d'Ivoire, Brazil, Colombia, Philippines, Thailand and Indonesia). The results are reproduced in Table 6. Relative wages for women have risen steadily in all cases, with reductions in the gender wage ratio ranging between 4.5 per cent points (Colombia) to 12.8 percentage points (Philippines). Tzannatos interprets his findings very positively, suggesting that women's labour market situation may be improving in developing countries in the contemporary period much faster than it did historically in the developed countries. The data do not allow the role of trade expansion in the process to be examined rigorously. Women in the two countries where export growth has been particularly rapid (Thailand and Indonesia) have shared in this outcome, but their performance is not exceptional.

The sources of this improvement are identified and vary between countries. Indonesia and Thailand are distinctive for the fact that reduction in the wage gap attributable to changes in the sectoral distribution of women's employment in favour of higher paying sectors (i.e., manufacturing) has been the highest. That the largest sectoral shift in female employment towards manufacturing should be found in the two most export-oriented countries in the sample is quite consistent with the argument of this paper. The increase in the share of the female labour force in manufacturing has been associated with only a very small increase in women's relative earnings within the sector in Indonesia, and a somewhat larger increase in Thailand. This suggests that wages may be higher in the export sector within manufacturing, compared to other branches of manufacturing where women are employed, to a small extent in Indonesia, and more significantly in Thailand.

In any event, data on other countries, presented in Tables 3 and 7, though not exactly comparable, do not fully bear out his findings. Table 3 gives data on female:male earnings in manufacturing (the figure of most interest to the argument presented in this paper) for a number of developing countries in different regions which show that there is no trend towards amelioration over time, most especially not in Hong Kong, Singapore or the Republic of Korea. Indeed, at the national level the Republic of Korea has the largest gender gap in wages on record. Table 7 presents national level data, such as Tzannatos uses. Among the 15 developing countries for which time series data are given, 10 show a rise in the ratio but four show a fall, and one shows no change.

Two lessons can be drawn from this. The first is that clearly more research is necessary on this crucial topic. The second is that the crude data, but also Tzannatos' analysis, do not take account of an important parametric change. The relative educational levels of the female population, who represent the pool from which female labour is drawn, has been rising in most countries outside Africa (see the forthcoming UNDP Human Development Report 1995). In particular, the marked rise in female wages in the Philippines reported in Table 6 may be attributable to the very strong increase in women's educational attainments in that country (for instance, women's enrolment rate at secondary level is now higher than men's (World Development Report, 1993). That is to say, there should in any event be greater parity in women's earnings in the economy as a whole and in manufacturing in particular, with increases in educational levels. Slight rises in the female:male wage ratio might have to be discounted for this reason. The proper index, in other words, for which changes over time and the possible effects of trade expansion need to be investigated is the degree of wage discrimination to which women are subject.

24 The literature on gender aspects of EPZ employment is reviewed in Baden and Joekes, 1993. Wood's cross-country analysis of the causes female intensity in manufacturing employment overall, reported in section 2.4, used national level manufacturing sector data from ILO sources in investigating the feminization of the manufacturing labour force, the statistically proper measure.

25 Singapore, Taiwan Province of China and the Republic of Korea have total enrolment rates in science and technology subjects at university/college level of the order of 50 per cent higher than in Latin America and about five times higher than Egypt, which has the highest rates in Africa (United Nations, 1992: Table III-3).

26 Barbezat (1993) discusses the factors determining change-overs in the gender composition of the workforce by occupation, with respect to issues of the relative length of labour queues among each gender, customary gender biases, etc., similarly to those suggested as relevant here. But the theoretical literature, which is in any case inconclusive as to causation, discusses only developed country circumstances and is confined almost entirely to cases of switching from male to female employment, not the reverse.

27 The concluding section of the paper suggests that there may be some scope for advocacy and international action to influence TNCs' employment practices for the better, especially in settings where equal wage legislation exists.

28 Though, as discussed above, this situation may not obtain.

29 Haiti may be a similar case. The ratio of female:male wages was unusually high (0.87; see Table 4), in a situation where almost the only wage employment available to women was in the short-lived export manufacturing sector, largely producing baseballs for the US market.

30 Also by report an unpublished study by Susan Horton of the University of Toronto, Canada, not available to the author at the time of writing.