2.3 Gender Composition of the Industrial Labour Force and Export Orientation

The employment of women workers in industry seems universally to fit a particular pattern. The demand for female labour is lowest in direct production jobs in heavy industry and highest in light, labour intensive manufacturing production. Even in centrally planned economies, where the overall share of women workers has historically been high because female labour was mobilized into industry as well as into other sectors, the same employment profile of men and women occurs.

One symptom of this pattern of labour specialization by gender is that the clothing industry — the classic labour intensive industry — absorbs a disproportionate share of female workers. More than two thirds of the global labour force in the clothing industry is accounted for by women and the industry accounts for almost one fifth of the total world female labour force in manufacturing (UNIDO, 1993a).

The economic characteristics of production, especially the relative shares of the main production factors, capital and labour, vary among branches of manufacturing. This indirectly helps explain the distribution of female employees by branch as well as the competitiveness of branch products in international markets.

While it is now generally known that in developing countries the production of manufactured exports relies on female labour (United Nations, 1995), the extent to which trade expansion has drawn women into the labour force in developing countries and the underlying economic logic of the process is not so well known.

"Women constitute a high proportion of the labour force in some conspicuous parts of developing-country export-oriented manufacturing (clothing and electronic products, and export processing zones). In developed countries, women are over represented in the sectors on which manufactured imports have been concentrated, and under-represented in the manufacturing sectors which export to developing countries" (Wood, 1991:168).⁷

The association of growth of exports with the feminization of the industrial labour force has held in all developing countries, whatever their level of income, or the previous pattern of female employment, or the qualifications and experience of the female labour force, or — perhaps most striking of all — the cultural norms pertaining to women taking paid employment. Experience in the case study countries, discussed below in SECTION 4, illustrates this clearly.

The main — though not the only — factor behind the differential distribution of male and female workers in different branches of industry is the gender gap in wages in manufacturing, based on discrimination in the labour market.

This is a substantial allegation. Table 3 presents the sort of first line of evidence usually raised in support of it. It shows female earnings as percentage of male earnings in manufacturing in a number of developing countries with values mostly ranging between 50 and 80 per cent. (The data are not available for other countries.) But these are crude data, which reflect not only differences in levels of pay (if any) but also differences between men and women in hours worked (men usually work significantly longer hours, except in some East Asian countries) and occupational distribution of male and female workers (men usually have a higher share of higher paid, higher skill level jobs). Adjustment for such factors has to be done, if like is to be compared with like.

But it is not straightforward to carry out this adjustment. Analytically, the method most commonly used to identify the true wage gap by gender takes data on individuals' earnings (instead of the plant level data collated by the ILO which generates the information in Table 3), together with information on individuals' labour characteristics which may be relevant to their level of pay, such as age and/or years of work experience, educational qualifications, and marital status. "Earnings functions" calculated separately for women and men on this basis can then show whether different wage structures exist for men and women, i.e., whether given attributes (such as education and experience) are valued differently for women and men — in which case gender discrimination in the technical sense exists. The existence of such differing pay structures has to be proved to uphold the claim that women are the source of the cheapest labour. The exercise also makes it possible to estimate what proportion of the crude earnings gap (as in Table 3) is accounted for by the different labour characteristics of men and women in the labour force, and how much is accounted for by differences in wage structure, i.e., by gender discrimination.⁸

Calculation of earnings functions and the estimation thereby of wage discrimination is a major econometric exercise which has only been carried out for a relatively small number of country or sector data sets. In almost every case, a true, discriminatory wage gap by gender is identified (Birdsall and Sabot, 1991; Psacharopoulos and Tzannatos, 1992; Terrell, 1992; Barbezat, 1993). Table 4 is a summary of the findings of the main studies. It shows that in ten of the eleven countries studies represented, more than half the wage gap by gender is attributable to different pay structures applying to male and female workers, and less than half is explained by variations in workers' individual characteristics. In other words, a discriminatory wage gap of 10-25 per cent is the norm.

If wage discrimination is as pervasive and substantial as this suggests, why are women not the preferred source of labour in manufacturing across the board? A mixture of economic and socio-political reasons can be invoked in explanation. First and foremost is the economic reason that the cost of labour is more pertinent to employment decisions taken in labour intensive enterprises than in other types of industry. The labour intensive producer operating in a competitive market (local or foreign) is obliged to keep the wage bill, the main cost of the business, as low as possible. Employing women minimizes the wage bill without sacrifice of labour productivity.⁹

By contrast, enterprises using capital intensive techniques of production, especially but not only when they are in a protected, uncompetitive market situation — as usually obtains under an autarkic industrialization strategy — do not have a strong incentive to seek to minimize labour costs within total production costs. The total cost of labour is, by definition, low relative to the cost of capital. Indeed, the employer may pay a premium to labour (a high level "efficiency wage") in order to secure worker loyalty and keep the rate of labour turnover low, important where labour productivity is high because of the capital intensity of the production process and there is an associated high opportunity cost of disruption to production. The level of union organization is generally higher in this sector, even though there is no strong evidence that unions are actually effective in raising wages¹⁰ as opposed to other benefits. In political economic terms, workers in the capital intensive sector are able, through the general strength of their wage bargaining position, to secure a share of the rentier profits actually or potentially available to enterprises behind protective barriers (Balassa, 1988).

As well as these narrowly political economic factors, there are reinforcing social reasons to explain why the "labour aristocracy" that emerged in capital intensive sectors should have been male. Men and women's differing, socially constructed characteristics as workers match employers' differing requirements of their workforce in different sub-sectors, to produce a concentration of male and female workers in heavy and light manufacturing respectively. The main dimensions of differentiation flow from a variety of distinctive, socially constructed constraints on women as part of the gender order. These affect women's "employability" with respect particularly to their working hours, to their perceived suitability to do heavy and dangerous work, and to their perceived lesser labour force commitment and skills.

First, the technical characteristics of operation of capital intensive, heavy industrial processes interact with a set of common restrictions against women working at night or doing heavy or dangerous work. Highly capitalized production activities require to be in continuous utilization to amortize the capital invested as quickly as possible; in some cases, e.g. chemical process industries, there are additional technological imperatives for continuous operation of the plant. In either case, two or three shift working is required of the workforce, normally on a rotating basis. Women are commonly forbidden by statute in developing countries from working varying shifts; where reasons are cited, they concern domestic care obligations along with the impropriety of women travelling at night (Anker and Hein, 1986).

Similarly, cultural norms concerning gender identity (again sometimes expressed in labour legislation) tend to prohibit women from carrying out work where heavy lifting or dangerous tasks are involved, as in metal working industries, for example. These restrictions also have a cultural rather than physiological basis. Women do carry out much other hard physical labour, e.g. in road construction gangs in South Asia: if the "lifting" limitation related purely to actual physiological capability, no additional screening criteria to this effect would need to be applied. The notion of "dangerous" work is similarly closely associated with concepts of "masculinity".

"Protective" labour legislation, prohibiting women from night working and heavy lifting, has the paradoxical result of protecting men's jobs: it limits women's access to an important class of jobs (Anker and Hein, 1986), mostly in heavy industry, rather than protecting women's health and safety in employment, as ostensibly intended.

Secondly, the characterization of women workers as "unstable" workers, "lacking commitment" to the labour force and lacking skills, are interrelated allegations. The main basis for them all lies in women's periodic withdrawal from employment, even if only on a short-term basis, for childbearing, and for the alleged lack of priority women give to job performance subsequently when they thereafter assume more domestic responsibilities (caring for family members). This is supposed to result in higher absentee rates among women, lack of flexibility to perform overtime and so on.

Women's intermittent withdrawal from the labour market has other consequences, notably a lower level of skill among women workers. This is because it reduces women's lifetime earnings from employment (irrespective of the wage rate), which in turn leads to a lower level of investment in the education of girls by their families compared to boys, and to a lower level of investment in training of women workers by employers compared to male workers.

A gender critique of these various claims reveals that many of them are spurious and that none is straightforward and empty of gender bias. They are best interpreted as constituting the structure of underpinnings and rationalizations for discrimination against women in the labour market.

Many employers in fact pre-empt any effect that women's withdrawal from employment might have on production by unilaterally truncating their period of employment. In many instances, employers require women to leave after giving birth, or even on marriage (Barbezat, 1993). Furthermore, some employers clearly find high turnover in the labour force advantageous. In industries, such as semi-conductors, where the demand for labour fluctuates sharply over time with the economic cycle, high turnover reduces costs. Women are employed partly because their alleged high turnover makes for a "flexible" workforce; by extension, women are seen as less likely to protest layoffs. Enquiries into short-term absentee rates often reveal no gender difference, with women's absences for reasons of health or domestic duties matched or outweighed by men's absences due to excessive drinking (Anker and Hein, 1986). And "it is no longer unusual in many countries¹¹ to find that women's educational attainment is, on average, higher than men's" (Barbezat, 1993). In other countries, however, overall levels of female educational attainment are not significant for employers. A female workforce with educational attainment as high or higher than male workers may be available to manufacturing employers, always at lower wage cost, if other employment opportunities for women are sparse.¹²

The pattern of female labour force participation found historically in industrialized countries, whereby women did indeed withdraw altogether from employment when they had children, resuming employment years later when they were grown, was always much less common in developing countries and is becoming ever less so (Barbezat, 1993).

As with the issue of arduous and dangerous "male" jobs, the question of the respective skill levels of men's and women's work is problematic in many ways. The skill grading of jobs — an apparently clear-cut concept — also turns out to be complex and problematic and linked with the gender of the operative as much as with any neutral objective definition. That is to say, jobs are more likely to be defined as unskilled when held by women than by men (Phillips and Taylor, 1980). There is very little information on skill levels and training requirements in jobs of different kinds in industry. For instance, there is little evidence that the training periods for male workers in modern capital intensive industry are inherently longer than for women workers in labour intensive manufacturing. Where training periods for women are short (e.g. in garments) it is sometimes the case that women enter the workforce with some initial competence in the relevant skills, gained in the home environment or in school, which because shared by all women, is not appropriately rewarded by the employer.¹³ The gender bias in the process of skill definition feeds into the grading of jobs. Differential grading of men and women workers is a major contributory factor in the gender wage differential.

Nevertheless, these various beliefs about the character of women's labour force participation persist. They contribute to the exercise of "statistical discrimination" against women in employment based on gender stereotyping rather than fact¹⁴ and to employers' normal unpreparedness to pay women equal wages to men. Employers' negative attitudes towards women workers are particularly important in developing countries where there is often a very large surplus of workers interested in obtaining jobs in the modern sector (Anker and Hein, 1986).

Finally, the phenomenon of differences in the sectoral pattern of employment of female and male workers in itself may help to perpetuate the existence of gender discrimination in a vicious circle. It contributes to segmentation of the labour market along gender lines, and the possibility (actuality) of different wage structures and wage outcomes in the different sectors as the particular market conditions work themselves out. Both occupational segregation and within-occupation wage differentials contribute to the overall difference in wages by gender (Barbezat, 1993).

Labour market segmentation models have been quite influential in attempts to conceptualize the determinants of the gender wage gap.¹⁵ But recent research shows that they have limits as an explanatory factor. Reducing occupational segregation may not reduce the wage gap proportionally, and conversely increased segregation may not be incompatible with reductions in the wage gap over time. Thus, while the degree of occupational segregation by gender has been decreasing in most regions, it has been stable or increasing in some South and East Asian countries and yet the size of the wage gap by gender in the manufacturing sector has been decreasing in all cases (Tzannatos, 1995). In the Asian region, women's employment concentration has increased in better paying occupations within the industrial sector. The rapid recruitment of women into labour intensive exporting sectors and increased segregation that follows does not necessarily of itself imply an increase in the gender wage gap.

In the light of this analysis, it is clear why developing countries' abandonment of fully autarkic industrialization strategies and their entry to international markets in manufactures has been accompanied by an increasingly female-intensive pattern of employment.

The strength of this general preference for women workers in export industry is graphically illustrated in the scatter diagram shown in Figure 1, reproduced from Wood (1991). This shows the relationship between changes in manufactured export performance and changes in the percentage share of women in the total manufacturing workforce between the early 1960s and the mid-1980s. As countries have raised the share of products exported to the North in total manufacturing value added, the female intensity of the industrial labour force has risen. There is clearly a very strong, positive relationship.¹⁶ Among all the countries for which data was available, there is no case where a rise in export share was achieved without a rise in the proportion of women in the industrial workforce.

The feminization of the industrial labour force with trade expansion has clearly relied on the increased availability of female labour, and female labour force participation rates have risen in many parts of the developing world, especially in East Asia and Latin America (Schultz, 1990). But Wood's study shows that there is no underlying tendency for the share of women in the industrial labour force to have risen over time. In other words, the increased absorption of women workers in manufacturing has been driven by changes in the trade performance of developing countries.

The universality of the feminization of the industrial labour force in developing countries' exports is perhaps surprising in one respect: have the large gender gaps in education in developing countries not hampered the process? Wood believes that primary education is an essential attribute in the modern sector labour force (Wood, 1994). Women in most developing countries have been discriminated against in educational provision. But restricted opportunities for women in wage employment, together with a higher propensity for more educated women to seek wage employment, mean that, as suggested above, in many developing countries, even those with minimal total educational provision, a pool of women with some education is available to industrial enterprises.

Nevertheless, educational levels may help explain variations in the composition of manufactures exports among developing countries. The educational level of the workforce is lower in clothing (and other classic, relatively low technology operations) than in electronics. There is evidence to this effect in studies of the Dominican Republic, the Republic of Korea and Mexico (see Baden and Joekes, 1993). Production methods in clothing manufacture have been subject to some technological improvements over time (computer aided design, mechanized cutting, etc.) which have affected certain job grades¹⁷, but in the main production line machinery and methods and the tasks carried out by production line workers have not been radically changed from the early days of mass production. Production of electronics products, by contrast, takes place in high-tech conditions, sometimes under stringent atmospheric controls, with operatives working to graphically complex specifications with precision machinery. That a higher level of education is required in the workforce for production of electronics products than in clothing seems intuitively plausible.

While all developing countries with export capacity in manufactures export clothing, the production of office and telecommunications equipment for export is highly concentrated in those regions of the developing world where the levels of female education are relatively high. The East Asian nations largely totally dominate developing country exports of electronics products. The Asian countries are also notable for their high levels of female education: average years of schooling for women in the population as a whole are, for example, 6.7 in Korea, 5.4 in Hong Kong, and 5.0 in Malaysia¹⁸, compared to 2.7 years among all developing countries (0.9 years in the least developed countries and 9.6 in industrialized countries) (UNDP, 1992). By contrast, in countries where clothing is the proportionately largest product category in manufactured exports, the average level of women's education is lower [e.g. Bangladesh (0.9), Tunisia (1.2), Morocco (1.5), and Indonesia (2.9)] (ibid.).

The argument that education of women increases a countries' export propensity in manufactures has not been proven statistically (Berge and Wood, 1994). Thus it cannot yet be added to the list of the many social benefits of women's education, which has convinced policy makers in a number of developing countries to give female educational provision higher priority (for a succinct recent statement, see World Bank, 1995a). The government in Pakistan, for example, has committed itself to a programme to improve female enrolments in the belief that continuation of the present, extremely poor levels of provision of education for girls will penalize the country's international competitiveness in the future.¹⁹ Electronics exports from developing countries now outstrip exports of clothing. If their faster rate of growth continues, the association between export capacity and the availability of an educated, female workforce may become important in future, and countries which fail to provide education to women will find themselves unable to break into this increasingly important part of the world market.
 

2.4 Trade Regulations and Manufactures Exports

Over the past forty years the increase in value of world trade has exceeded world output growth in every year except one (1982). The growth of trade has undoubtedly been facilitated by the trend towards trade liberalization, which is usually viewed as one of the great economic achievements of the Bretton Woods era. In the late 1940s world tariffs on manufactures averaged 40 per cent. Under the auspices of successive rounds of the GATT, the figure has been brought down to 5 per cent (Stevens, 1994). The implementation of the Uruguay Round, the last round of global trade negotiations under the GATT, will bring down the average tariff level even further.²⁰

In developing countries, the shift in domestic policies in favour of openness to trade and the rapid expansion of export capacity in the industrial sector had the result that (starting of course from a very low base) there was a more rapid rate of increase in manufactures exports from developing countries than from developed market-economy countries from the mid-1970s to 1990. Without the restraints and discriminatory provisions of world trade regulations as they impact on developing countries, the gains would probably have been even greater.

Continued discrimination against developing countries takes several forms.²¹ First, levels of tariff are higher the more processed the product (tariff escalation). For example, the average developed country tariff on fresh fruit is seven per cent compared to 17 per cent for processed fruit. Tariff escalation is a disincentive to developing countries to adding value to their primary materials by moving into agricultural processing and manufacturing. It is not yet clear whether the Uruguay Round will reduce tariff escalation.²² Second, quotas (quantitative restrictions on imports to industrialized countries' markets) are far more common on developing countries' exports than on developed countries' exports. Non-tariff measures cover 28 per cent of all developing country exports to the North overall; more than half of clothing exports are restrained in this way. Finally, strict product standards applied to imported goods (regarding e.g. food ingredients, labelling) can have an arbitrary character and seem intended as much to be obstacles to intending exporters as protection to consumers. In terms of manufactures alone, and disregarding quotas, even after the implementation of the Uruguay Round, developed countries will still be imposing tariffs averaging 4.3 per cent on developing countries' exports, compared to 3.9 per cent on imports from other developed countries.²³

The two main categories of developing country exports to industrialized countries, clothing and electronics products, face special trading rules. As noted, the labour intensity of clothing production makes the clothes produced in low wage locations strongly competitive in rich country markets, and the ease of market entry for new clothing firms makes it difficult for established rich country producers to exercise monopoly powers and keep out market rivals. Hence, the weight of producer influence has gone into the creation of one of the most intricate and restrictive of all sets of trade rules, the MFA (Multifibre Arrangement).

The MFA has constituted the most notorious exception, to date, to the liberalizing tendencies in world trade. At last, the Uruguay Round contains a commitment on the part of industrialized countries to phase out the MFA. But the schedule is slow (a 10-year period of phasing out) and qualified, e.g. it allows for the imposition of "transitional safeguards' on certain product groups. Action is allowed against individual exporting countries if it can be demonstrated by the importing country that overall imports of a product were entering the country in such increased quantities as to cause serious damage — or to threaten it — to the relevant domestic industry and that there was a sharp and substantial increase of imports from the individual country concerned (GATT Secretariat, 1994). This opens the door to protectionist measures again, of the kind invoked frequently under the MFA in the past by industrialized countries against developing countries.

The MFA comprises a series of rolling bilateral arrangements between the main groups of industrialized countries (the United States, the European Community and Japan) and individual developing countries, containing lists of items of clothing subject to quota. Thus year by year a developing country would face a new set of restrictions on, for example, the number of pairs of men's trousers, or women's dresses, that it could legally export to a particular country destination. Specifications were laid down in terms of the permitted fabric composition of garments as well as the items of clothing concerned. Within countries, various methods were employed for allocating the quotas amongst local clothing producers, ranging from export permit auctions (in Hong Kong) to industry association-government negotiations (in the Dominican Republic). In countries which produced close to their quota allocation, a cat and mouse game developed of producers inventing new items/fabric mixes to step outside the itemized restrictions, at least until the list was revised (as in Hong Kong); or scanning other countries' lists and investing in locations where particular quotas were under-filled [one example being Korean investors who set up operations in the Dominican Republic to gain extra access to the North American market (Joekes, 1987)].

Quota-hopping investment of this kind, which has intensified in recent years, had a double-edged effect. While it may have been a significant factor in helping establish the clothing industry in new locations, it penalized countries which were too successful in such ventures by imposing a ceiling on their expansion. While experienced producers such as Hong Kong were adept at finding their way round the MFA restraints, albeit at the considerable cost of retooling production year by year, and some countries (e.g. the Dominican Republic) became skilful at obtaining a favourable deal for their clothing producers in annual quota negotiations with the United States, middle ranking producers had the growth and consolidation of an indigenous clothing sector blocked (especially if their exports were dominated by quota-hopping foreign investors). The clearest gains under the old system were experienced by the newest entrant countries, whose chances of breaking into the industry at an internationally competitive level may have been facilitated by the provisions of the MFA.

GATT economists predict a 60 per cent increase in trade in clothing with the phasing-out of the MFA. Most of that gain will accrue to relative newcomer countries, such as Pakistan, Bangladesh, China and India, which would sooner rather than later have hit restraints under the MFA, where they had not already done so (Atkinson, 1994). Weaker countries which might have been next in line to benefit (e.g. perhaps Viet Nam) may therefore find it more difficult to enter.

Another, little noticed bias has existed in the MFA against the establishment of indigenous production capacity in clothing in developing countries. The quota restraints of the MFA were in some instances overridden by trade provisions (such as "Super 807" of the United States trade laws) which applied trade taxes and quotas in full only on articles wholly produced in the exporting country or in a third country. Articles which included materials or components originating in the importing country itself had preferential access in that duties were levied only on the residual value of the product). Thus, for instance, the much greater expansion of the Tunisian compared to the Moroccan clothing industry, despite the fact that wages are significantly higher in Tunisia, is attributed to its far higher share of sub-contracted production in total output (World Bank, 1993b). This feature of the trading régime has acted against the forces of comparative advantage which would have favoured the expansion of poorest country exporters.

Trade in electronics products has faced less restrictive trade regulations. This is largely because of the greater worldwide prevalence of intra-firm trade in this industry, as discussed above. More exactly, it reflects the greater "globalization" or presence of transnational corporations in this field than in clothing. Production of some electronics components was devolved by TNCs to developing countries from the very early days of production, on labour cost grounds. There was, until the East Asian producers entered the field on their own account, no North-South competition as such. Relocation of components production to lower wage developing countries was in fact an instrument of intra-Northern TNC rivalry. Accordingly, TNCs did not press their governments to impose trade restraints on imports of electronics components or assembled products, since they largely represented intra-firm transactions. Super 807-type provisions were introduced under pressure from TNCs to prevent trade duties being levied on intra-firm trade and on trade between parent and sub-contractor. It may not therefore be entirely due to the rapidity of technological progress that trade in electronics equipment has increased so rapidly, far to supersede clothing exports from developing countries. The greater openness of trade in this area has been a facilitating factor.

Finally, the quirks of trade restraint have shaped the pattern of export production and women's employment opportunities in industry in another way. The attraction of Export Processing Zones (EPZs) — where women's wage labour opportunities in manufacturing have been notably concentrated (see Section 2.5) — in developing countries has rested mainly in the fact that EPZ producers are exempt from paying import duties on raw materials and components. Exporters based in EPZs using imported materials of industrialized country origin were able both to import materials free of duty and, if in a position to take advantage of Super 807-type arrangements, to have the final product imported back into that country free of duty except on the value added share, roughly represented by the labour cost element. In countries where high protective tariffs otherwise applied, this was a considerable saving. Elsewhere, and as tariff levels in developing countries have been reduced, the fiscal aspect weighed less heavily in producers' choice of EPZs as operating sites, as against their administrative convenience, infrastructural services etc.

7 Nevertheless, Wood finds that in developed countries, increased trade with developing countries has, perhaps surprisingly, not worked to diminish the share of women workers in industry. Various possible explanations of this are given, to do with a generalized pressure for substitution of female for male labour throughout industry, change in occupational structures, etc. The issue is not pursued any further in this paper.

8 Studies of the sources of the gender gap in earnings commonly show for example that there is a divergence in wages by gender at the point of marriage. Men are awarded an often explicit "breadwinner" premium on marriage, but women's rate of pay is reduced in the stated expectation that their domestic obligations will diminish their work attendance and work performance (Gannicott, 1986 for Taiwan Province of China; Appleton et al., 1995 for Africa; World Bank, 1995 for developed countries). Another cause of the divergence, reinforcing the female:male earnings difference among older workers, is that women commonly lose pay seniority after any break for maternity (Terrell, 1992).

9 A few studies have been done in the clothing industry comparing male and female workers' individual productivity; they show little difference, even in a few cases that women have the edge (Joekes, 1987). Moroccan clothing producers believe that the quality of women's work is superior to men's, while men's work rate is higher (Joekes, 1982); the same views are also observed among Brazilian employers (Humphrey, 1987).

10 See Turnham, 1995 and the research findings on a number of Latin American countries reported in Psacharopolous and Tzannatos, 1992.

11 This includes Jamaica and several countries in Latin America.

12 Again this is a somewhat speculative statement, based on case study evidence. There is a remarkable lack of systematic information about the educational level of the workforce, in different sectors and overall, in developing countries. Data on educational attainment relate to the population as a whole rather that to those in the labour force or in employment.

13 This is the "nimble fingers" argument about one aspect of gender discrimination, particularly associated with the work of Elson and Pearson (see Elson and Pearson, 1981).

14 The concept was developed by Arrow as one of a number of alternative, non-mutually exclusive causal models of gender discrimination in the labour market (see Barbezat, 1993 and Birdsall and Sabot, 1991, for expositions).

15 They appealed particularly to institutional economists, but by focusing on market processes were not anathema to neo-classical economist.

16 Using Ordinary Least Squares, the regression equation for these data has a statistically significant coefficient (0.31), and an R2 of 0.45. The intercept passes through zero, meaning that any secular tendency for the share of women workers in industry to increase can be discounted.

17 These posts tend to be filled by male workers. This is part of the reason why the labour force in exporting industries is not completely female. The other is that in clothing (but not in electronics) some men continue to be taken on in production line positions (paid a higher wage) to act as agents of social control: they break up female solidarity and are used to set a faster work rate (Joekes, 1982).

18 The figure was 3.1 in Singapore in 1990 but probably much higher among younger women, because the situation is changing so rapidly: female enrolment in secondary education now exceeds the male rate (UNDP, 1992).

19 Of course, commitment to increase the appallingly low level of educational provision to women in Pakistan should not need to rely on instrumental arguments such as these.

20 The Uruguay Round, concluded in December 1993, had as one of its provisions the replacement of the GATT itself with a new organization with greater powers, the World Trade Organisation (WTO). The Round was marked by a much greater participation of developing countries in the negotiations than in previous rounds. It marks the consolidation of the principles of rule observance and non-discrimination in global trade, the commitment to eventual removal of existing restrictions and extension of liberalization to new sectors (e.g. agriculture, services, intellectual property) and the ending (except for the least developed countries) of preferential treatment of developing countries. See Stevens, 1994 for a summary analysis of the impact of the Uruguay Round on developing countries.

21 The illustrative examples in the rest of the paragraph are drawn, unless otherwise stated, from Joekes and Weston, 1994.

22 Jane Kennan, personal communication.

23 The downside of the progressive liberalization of tariffs to developing countries is that it erodes the preferential agreements that many groups of countries had secured for access to developed country markets. Two examples are the Caribbean Basin Initiative, giving lower tariff rates into the USA for products originating in Caribbean countries, and the series of Lome Agreements between the EC and the ACP countries (mostly former British and French colonies in the Caribbean, Africa and the Pacific region).