Abstract
ABSTRACT
Macroeconometric studies generally find fairly robust evidence that gender equality has a positive impact on economic growth, but reverse findings relating to the impact of economic growth on gender equality are far less consistent. The high level of aggregation at which these studies are carried out makes it difficult to ascertain the causal pathways that might explain this asymmetry in impacts. Using a feminist institutional framework, this contribution explores studies carried out at lower levels of analysis for insights into the pathways likely to be driving these two sets of relationships and a possible explanation for their asymmetry.
INTRODUCTION
The long-standing preoccupation with economic growth in international policy circles, combined with the more recent interest in gender equality, has given rise to a number of macroeconometric studies, some seeking to estimate the impact of gender equality on economic growth, others focusing on the impact of growth on gender equality. In a recent paper, Luisa Natali and I synthesized the findings from both sets of studies (Naila Kabeer and Luisa Natali 2013). We found the evidence that gender equality contributed positively to economic growth to be fairly robust, holding across a range of different countries, time periods, and model specifications. The evidence for the reverse relationship was less consistent and generally confined to high-income countries.
One possible reason for this asymmetry in impacts is that the two sets of studies do not always use the same measures of gender equality. Efforts to estimate growth are influenced by a long tradition of growth modelling in neoclassical economics and confine themselves to a few measures of gender equality: education, employment, and sometimes wages. Efforts to model gender equality, on the other hand, are relatively new and draw on a wider range of equality measures, including well-being, rights, and political participation.
A second, more interesting reason may be that the two sets of relationships do not operate through the same causal pathways. However, the highly aggregated cross-country regressions that typically feature in these macro-level studies provide very few insights into what these pathways might be. While they are useful in establishing broad empirical correlations, more in-depth, lower-level studies are needed to illuminate possible pathways.
A review of such lower-level studies was carried out by Esther Duflo (2012). She found the evidence for both sets of relationships to be weak and inconsistent. She concluded that gender equality might be a desirable goal in and of itself, but there was no guarantee that it would contribute to economic growth and, indeed, could be detrimental to it. However, a number of limitations to her study suggest that a fresh look at the evidence is warranted. First, her analysis is firmly rooted in the neoclassical economic view of human behavior as motivated by the desire to maximize individual utility. Her evidence is thus interpreted through a highly individualized, rational-choice lens with little attention to the wider context in which choices are made.
Second, in line with the emerging “new” development economics, of which she is a leading exponent, her evidence is almost entirely confined to micro-level studies using randomized control trials or quasi-experimental approaches: only these are considered to rule out the unobservable biases which invalidate other approaches. Her analysis thus rests on a very thin empirical base.
My objective in this contribution is twofold. I want to draw on feminist institutional economics to provide an alternative account of human behavior. In particular, I want to explore the influence of broader contextual factors on the forms of agency women are able to exercise and how this might offer a very different interpretation of the empirical evidence to that offered by Duflo (2012). I also want to go beyond her narrow focus on experimental evidence. Although, as Florent Bedecarrats, Isaebelle Guérin, and François Roubaud (2015: 16) point out, economists of this school engage in the “virtually ritual” denigration of studies that do not use their methodological approach, effectively wiping clean “the memory banks of past knowledge.” The fact is that the practical difficulties of designing real-life research meeting the laboratory-like requirements of experimental approaches seriously undermines their claims to methodological superiority (Angus Deaton 2010). This contribution will take a more pluralist approach to the question of knowledge generation. Not only is this in line with a feminist economics tradition, but it also provides a better understanding of the complex questions addressed here.
GEOGRAPHIES OF GENDER, VARIETIES OF GROWTH
Let me start out by locating the concepts of gender inequality and economic growth within the relevant feminist literature. Feminists agree with neoclassical economists that individuals make choices and exercise agency within the limits imposed by their personal circumstances, but they also draw attention to the constraints posed by structural distribution of rules, norms, resources, and identities between different groups; along lines of gender, class, race, caste, and so on; and the inequalities of power and privilege these generate (Nancy Folbre 1994). Ann Whitehead's distinction (1979) between “gender-ascribed social relationships” and relationships that are “bearers of gender” provides a useful way of capturing the gendered aspects of these collective structures of constraint and their variations across the world.
Gender-ascribed constraints are rooted in the customary norms, beliefs, and values that characterize the “intrinsically gendered” relationships of family and kinship. They spell out dominant models of masculinity and femininity in different societies, allocating men and women and boys and girls to different roles and responsibilities on the basis of socially-constructed aptitudes and dispositions. Variations in the gendered division of labor between productive and reproductive work observed in different regions of the world partly reflect variations in these gender-ascribed constraints. The general assignment of primary breadwinning responsibilities to men helps to explain higher male labor force participation rates (MLFPRs) across the world. But while most societies ascribe primary responsibility for unpaid reproductive labor to women and girls, there is remarkable variation in socially-ascribed expectations regarding their contribution to production, and hence in female labor force participation rates (FLFPRs).
This has given rise to a distinctive “geography of gender” in patterns of labor force participation. Esther Boserup (1970) provides an early attempt to capture these regional patterns. She suggested that developing regions in the world could be broadly divided into men's and women's farming systems on the basis of the gender intensity of production in agriculture and trade. Female farming systems were largely found in Sub-saharan Africa, Southeast Asia, and parts of Latin America. While women within these systems were responsible, like women elsewhere, for unpaid reproductive labor, they were also expected to share breadwinning responsibilities with men, often managing their own farms and enterprises to do so. As a result, these regions were characterized by higher FLFPRs than average for the developing world.
At the other end of the spectrum were male farming systems found in the Middle East and North Africa (MENA) region and much of South Asia. These had lower FLFPRs than the rest of the world. Not only were women in these contexts expected to confine themselves to unpaid domestic work, but there were strong norms of secluding women, restricting their mobility in the public domain. Cultural expectations of women's dependency and associated son preference mean these regions were characterized by the phenomenon of “missing women.” Deficits of women in the overall population reflecting forms of gender discrimination that lowered women's life expectancy relative to that of men (Amartya Sen 1990).
As Boserup observed, the norms, values, and practices associated with the “private” domain of family and kinship were frequently reinforced by gender-related constraints within the public domains of states and markets. She pointed to government policies that privileged male household heads in the distribution of land rights and agricultural services, as well as employer preferences for male labor as examples of this. Purportedly impersonal “public” institutions thus became “bearers of gender” when they reproduced or exacerbated preconceived notions about masculinity and femininity through their rules, norms, and practices.
The world has changed considerably since Boserup's classic work, changes that encompass aspects of this geography of gender. This contribution will be examining the extent to which economic growth has been a factor driving these changes, but for now, let me turn to the feminist literature on growth.
Feminist scholars have pointed to the fundamental gender asymmetry in mainstream conceptualizations of growth (Marilyn Waring 1989). Growth is defined as the increase in the goods and services produced in a country within a given year, valued at market (or imputed) prices – with one important caveat: it is restricted to those goods and services that are, or could be, bought and sold in the market place. The production of these goods and services is classified as “economic activity” and, as the labor force data tell us, is largely carried out by men. The goods and services produced through unpaid domestic labor for consumption and use by the family, including those which reproduce the “economically active” labor force on a daily and generational basis, are explicitly excluded from definitions of economic activity and growth. As time use data tells us, this work is largely done by women. In other words, the economic growth of a country is defined and measured in a way that arbitrarily excludes the essential, but invisible “economic inactivity” that goes into making it happen.
This has led feminist scholars to argue that a key criterion for assessing the gender implications of growth strategies must be the extent to which they address the consequences of this asymmetry. This means asking questions about the quantity and quality of economic opportunities associated with different growth strategies, about the gender distribution of these opportunities, and about the extent to which state policies serve to reinforce or offset gender inequalities in the ability to benefit from new opportunities (Diane Elson and Nilüfer Çagatay 2000).
Strategies for growth have varied over time. Most developing countries embarked on import-substituting, state-led industrialization strategies after gaining independence in the postwar period, using Keynesian macro economic management to achieve nationalist ends. The oil shocks of the 1970s and the ensuing debt crisis in many of these countries coincided with the rise of neoliberal ideologies within wealthy donor countries and their dissemination across the world through the lending policies of the international financial institutions. The result has been a major shift across the world toward export-oriented growth strategies based on the liberalization of markets, trade, and capital flows at national and international levels.
Drawing on the feminist criteria outlined above, it could be argued that there was a strong male breadwinner bias to the early state-led import-substitution strategies. They promoted large-scale, capital-intensive industries, privileged men's employment, and tied access to social security to full-time, life-long employment (of the kind largely enjoyed by men). At the same time, the state also took responsibility for the social welfare of its citizens, and there was a gradual, though highly uneven, expansion in welfare services in most developing countries that created formal employment opportunities for educated women and supported many more in their reproductive roles.
The shift to neoliberal policies has been associated with a very different set of outcomes. Efforts to create flexible labor markets in order to attract “footloose” capital and to enable countries to compete in international trade have led to the dismantling of state protections previously accorded to labor. Where these cutbacks in state welfare provision have not been accompanied by an equivalent expansion in affordable market provision, it has been the unpaid labor of women, particularly among low-income groups, that has had to make good the shortfall. The male breadwinner bias of earlier policy has been partly attenuated by women's access to jobs in export-oriented sectors, but the reduction of the role of the state has meant the loss of one of the few sources of “good jobs” available to women in developing countries, as well as a decline in state support for their reproductive responsibilities.
To sum up, therefore, the feminist economics literature suggests that the interrelationship between economic growth and gender equality is likely to be mediated by a variety of contextual factors, including the broad policy regimes, within which growth strategies are embedded, as well as the structures of local patriarchy. It thus rules out the possibility of making a priori generalizations about how the interrelationship will play out in different contexts. However, as the analysis in this contribution will show, the resilience of certain aspects of patriarchy means that we continue to find elements of continuity in the midst of much change.
THE IMPACT OF GENDER EQUALITY ON ECONOMIC GROWTH: EXPLORING THE CAUSAL PATHWAYS
Stephan Klasen (1999) has suggested two main causal pathways through which gender equality in access to resources might contribute to growth. The first, a family-mediated pathway is based on the assumption that, given their association with reproductive responsibilities, women are more likely than men to invest the resources under their control in their children's human capital, thereby increasing the productivity of the next generation of workers. The second, the market-mediated pathway, is premised on the assumption that innate abilities are randomly distributed between men and women so that equalizing the gender distribution of resources and opportunities will maximize the productivity of the human resources available to an economy.
There are clearly different time frames involved in these two pathways. While women's access to resources may indeed translate into higher levels of investment in children's human capital, its impact on economic growth will not materialize until these children join the labor force. That it does materialize is suggested by the finding reported by various time series studies, cited by Kabeer and Natali that, along with greater gender equality, increases in the overall level of education and in the education levels of the working age population are also significant determinants of growth. The impact on growth through market-mediated pathways, on the other hand, is likely to materialize within a shorter period of time since increasing gender equality implies that the allocation of resources and opportunities within the economy will reflect the distribution of individual talent rather than the dictation of gender norms and should show up in increased productivity.
Gender equality and human capital investments
A solid body of micro-level studies supports the hypothesis of the family-mediated pathway (see Supplemental Online Appendix 1, Table 1, available on the publisher's website). Studies from a variety of different contexts suggest that women's access to a range of valued resources, including education, employment, land, cash transfers, and credit, is associated with increased investments in family welfare, including children's health and education. As Matthias Doepke and Michéle Tertilt (2011) conclude in their review of this literature: “Even though each individual study has certain shortcomings … the fact that a variety of studies using different data sources and empirical methodologies arrive at essentially the same conclusions strongly suggests that these findings are robust features of the data” (15). Duflo makes the additional point that “the correlation with mothers’ education and earnings is almost always found to be stronger than the corresponding correlation with fathers’ education and earnings” (2012: 1065).
Various explanations of these findings are possible. Neoclassical household bargaining models explain them in terms of gender differences in preferences. Women's greater access to valued resources increases their bargaining power within the household and ensures that household resources are allocated in closer alignment with their preferences. The implicit assumption in these studies is that women's preferences are more altruistic than men's.
Duflo challenges this interpretation. In the absence of controls for biases of various kinds, she argues the positive association observed between women's access to resources and investments in children's human capital may simply reflect their husbands’ “unobserved characteristics”: “if he is progressive enough to allow his wife to seek employment, then this same progressive attitude may make him treat his children better” (2012: 1065).
However, her evidence to support the male altruism hypothesis (or refute the female) is weak. For instance, she cites a study from Indonesia showing that maternal education was more strongly associated than paternal with increased age at marriage and reduced fertility rates, but emphasizes that there was no difference when it came to reduced child mortality. She cites a study from China showing that maternal education had a greater impact than paternal in improving infant health outcomes, but suggests that the close correlation between the two made it difficult to separate their relative contributions.
Her third example, a study of the Old Age Pension scheme in South Africa (Eric V. Edmonds 2006) appears more straightforward. She cites its finding that adolescent children were more likely to be in school when they lived in a household with a male recipient of the pension than a female recipient and observes, “In this case it is when men receive the pension that they make the decision favourable to well-being and development” (1074). In fact, a closer reading of the findings suggests a less straightforward account. Among households with nearly-eligible elderly members, children's education was found to be higher in households with an elderly woman than with an elderly man, with girls’ education level generally higher than that of the boys. The actual receipt of a pension by elderly men brought boys’ education to the levels enjoyed by boys in households with nearly eligible women. Mean while, the actual receipt of a pension by elderly women had no significant impact on education level. At the same time, households with both men and women pension recipients reported an increase in education of both boys and girls.
One problem with both sets of interpretations is that they treat preferences as random and idiosyncratic: evidence suggesting female altruism can consequently be countered by evidence refuting it. An alternative interpretation is that the systematic nature of the association between women's access to resources and children's welfare documented by the literature suggests a structural element to the formation of preferences. It is not “women” per se who drive these associations, but women in specific familial relations, most often mothers and sometimes grandmothers. This finding attests to pervasiveness of “ideologies of maternal altruism,” social constructions of motherhood that assign special responsibility for children to mothers, as one aspect of the gendered structures of constraint noted earlier (Whitehead 1981).
At the same time, and the two interpretations are not mutually exclusive, in contexts where women are cut off to a greater extent that men from access to resources and relationships beyond the household, their long-term interests may be bound up to a greater extent than men with the long-term interests of their families, suggesting a self-interested dimension to their apparently altruistic behaviour (Whitehead 1981). Indeed, one reason to question any essentialist notion of “maternal altruism” as an explanation for mothers’ behavior is the fact that their investment in children is often gender-biased rather than egalitarian. This is most marked in regions characterized by strong son preference, where women's status within the household and their security in old age depend on producing sons, ensuring their survival, and winning their loyalty (Mead Cain, Syeda Rokeya Khanam, and Shamsun Nahar 1979). In such contexts, women's access to education may reduce overall child mortality, but it can raise mortality rates for their daughters (Monica Das Gupta 1987).
Clearly the possibility that men's access to resources also contributes to children's welfare cannot be ruled out, but it may be that their contributions are better captured through overall household income, given their breadwinning roles, or through their contributions to joint forms of household consumption (Anne Whitehead and Naila Kabeer 2001). Equally, we cannot rule out the possibility that women's access to resources can lead to the deterioration in children's welfare. For instance, mothers in low-income households who must take up wage employment in order to earn a living often keep their older daughter back from school to look after younger siblings, take their young children to work with them in the field or roadside, or simply leave them at home unattended. However, these diverse findings cannot be reconciled through debates about whether men are more altruistic than women. They must be grounded in explanations that take account of variations in gender roles and responsibilities in different contexts, the identities and interests these generate, and how these might be influenced by class and other aspects of household position.
Gender equality and market efficiency
The evidence for the market-mediated pathway tells a more complicated story. First of all, women's domestic responsibilities make it more difficult for them to engage in market-oriented activity of the kind that enters into calculations of economic growth. Second, the monetary returns of women's labor when they do engage in such activity are generally lower than those enjoyed by men (Supplemental Online Appendix 1, Table 2). In other words, unless there is greater gender equality in respect to the returns on economic activity, an increase in men's activity rates is likely to lead to higher levels of income at both household and national levels than to an increase in women's rates.
The literature features three sets of explanations for the inequality in returns. The first relates to gender differentials in endowments of various kinds. Studies on gender disparities in agricultural productivity draw attention to differentials in landholdings, quality of land, use of inputs, access to agricultural extension, among others. Studies on gender disparities in wages highlight differentials in education, work experience, job tenure, childcare responsibilities, and occupation. Studies on gender disparities in enterprise profits point to gender differentials in education, size of enterprise, use of inputs, access to credit, age of firm, line of business, and familial responsibilities.
A second set of explanations focus on gender differentials in returns to endowments. Studies on gender differentials in agricultural productivity find that, along with differentials in landholding, location, and use of agricultural inputs, they also reflect differentials in returns these endowments yield. Studies on wage disparities note that returns to education and experience are often higher for men than women for a given level of education and experience. Finally, studies into gender-differentiated returns to enterprise report that part of the disparity reflects differential returns to assets, experience, and training.
Some important insights into gender disparities in returns to endowments are provided by recent experimental studies in Ghana and Sri Lanka, which offered equal amounts of transfers in cash and working capital to men and women in small-scale enterprises. In the Sri Lanka study, entrepreneurs who are men reported an increase in their monthly returns after receipt of transfers; there was no change in returns for women's enterprises (Suresh de Mel, David McKenzie, and Chris Woodruff 2009). The Ghana study allowed for greater variation in the scale of small businesses (Marcel Fafchamps, D. McKenzie, S. Quinn, and Chris Woodruff 2011). Businesses made up of men reported positive returns as a result of the transfers, regardless of the scale of business, but returns to women's businesses varied by the scale of enterprise. Women in smaller-scale businesses reported zero returns, but women at the larger-scale end reported positive returns, larger than the average men's returns.
Efforts to explain these findings highlighted the relevance of structural constraints. The Ghana study found that women who reported positive returns not only had larger-scale enterprises than those who did not, but started out with higher initial investments, had higher levels of education, came from wealthier households, had greater access to formal credit, and were more likely to choose their line of business for its profit potential than ease of entry. In other words, their class advantage partly offset disadvantages associated with gender.
The Sri Lanka study drew attention to the gendered character of the occupational structure. A division of sample enterprises into predominantly male, predominantly female, and mixed activities revealed that the higher the percentage of women associated with an activity, the lower its returns. Meanwhile, the higher the percentage of men associated with an activity, the higher its returns. The small minority of women in “mixed” sectors were more likely to invest their transfers productively and report positive returns than the majority working in women-dominated sectors, while a small minority of men in “mixed” industries were less likely to either invest their transfers productively or report positive returns than those working in male-dominated sectors.
These findings take us to the third set of explanations for gender disparities in returns to labor: the widely observed gender segregation of market opportunities along horizontal and vertical lines. One measure of horizontal segregation is the classification of occupations by their “gender dominance” (as in the Sri Lanka study). This was attempted by Richard Anker, Helinä Melkas, and Ailsa Korten (2003) using 1990 data on non-agricultural occupations from forty-three countries. They found that 60 percent of men were in occupations that were 80 percent male, while 30 percent of women were in occupations that were 80 percent female. Of the 7,300 occupations that made up the study, only 10 percent were 80 percent female, declining to 3 percent in the MENA region.
Vertical segregation has been measured by the percentage of women in high-level, decision-making posts (legislators, senior officials, and managers). Using 2000 data from fifty-one countries to estimate, Anker (2005) found that women comprised an average of just 28 percent of these posts, varying from less than 10 percent in the MENA region, to 15 percent in Asia, and 30 percent in the OECD countries and Latin America. Within Asia, estimates ranged from 23 percent in Southeast Asia to just 10 percent in South Asia.
These findings relate to formal employment, which is largely dominated by men. Other studies suggest similar segregation exists within the informal economy where a disproportionate share of the female workforce is concentrated (International Labour Organization [ILO] 2013). In fact, gender disparities appear to be larger in the informal economy. For instance, using data from five middle- and low-income countries, Martha A. Chen, Joanne Vanek, Francie Lund, James Heintz, Renana Jhabvala, and Chris Bonner (2005) estimated that that hourly earnings in informal waged work, as a percentage of hourly wages in formal private-sector waged work, varied between 77 percent in Egypt, 63 percent in El Salvador, 60 percent in Costa Rica, and just 50 percent in South Africa.
Two types of explanations were put forward by the World Bank (2012) for this pervasive stratification of the opportunity structure: first, gender discrimination in the labor market; and second, the voluntary selection on the part of men and women in response to their different domestic responsibilities. Insights from the feminist literature require us to qualify each of these explanations.
First of all, discrimination goes deeper than generally allowed for in the mainstream literature (Deborah Figart 2005). Economists generally decompose gender gaps in earnings into the “explained” component, which reflects gender differentials in observed endowments and characteristics and the “unexplained component,” which reflects “unobserved variables,” including gender discrimination. In reality, gender differentials in the observed variables are themselves frequently the product of gender discrimination, both within and outside the labor market.
Most women do not choose to educate themselves less than men – this was a choice made by their parents, reflecting either discrimination or a rational response to gender inequalities in returns on educational investment. Nor do women choose to own less or poorer quality land, be visited by fewer extension agents, or exclude themselves from formal credit and training opportunities. Such inequalities may reflect local customs governing the distribution of land. It may reflect discrimination on the part of employers, banks, and government officials who determine access to credit, agricultural extension, and training opportunities. Or, it may reflect statutory provisions which impinge on women's economic agency: for instance, a World Bank/International Finance Corporation (IFC) survey of 141 countries (2011) found widespread evidence of legal discrimination in the distribution of property, along with restrictions on women's ability to travel, start up their own business, or open a bank account without male consent.
Second, it is not clear that women's primary responsibility for unpaid domestic work is always entirely “voluntary.” Some may welcome their socially assigned responsibilities, and others may simply accept them as a “given” feature of their lives, but there is also evidence to suggest active dissent. Jayati Ghosh (2009) reports that successive National Sample Surveys in India record increasing proportions of women who say that they perform unpaid domestic work out of compulsion rather than choice. According to Maitreyi B. Das (2006), around 92 percent of women not currently in the labor force give this response. Qualitative research has documented the protracted negotiations that take place over the division of unpaid household responsibilities when married women seek paid work. It highlights the extent to which many women feel that they have little choice in the matter if they want to keep their marriages intact (Naila Kabeer 2013).
There are no simple explanations for the persistence of gender disparities in market returns on labor; nor are the explanations likely to be the same everywhere. However, their persistence and pervasiveness suggests that the efficiency argument for gender equality will have to be cognizant of the underlying structures of constraint that shore up these disparities. When, for instance, the World Bank (2014a) points to the growth dividends likely to accrue for policies which uphold women's land rights, facilitate their access to markets, promote their transition into high-value cash crops, provide community-based childcare centers and strengthen their ability to hire labor, the policies themselves may make a great deal of sense, but the inequalities they seek to address are not just random phenomenon to be eliminated through sensible policy recommendations. They are the surface manifestations of deeply-entrenched structures of patriarchal power and are likely to encounter considerable resistance from those who benefit from these structures. We see this in the subversion of the goals of gender-equality policies by those charged with implementing them (Bina Fernandez 2012). We also see it in the resistance that women encounter when they attempt to enter the labor market (John Sender, Carlos Oya, and Christopher Cramer 2006), when they gain access to credit (Aminur Rahman 1999), or when they seek to take up political quotas (Ulrike Mueller 2015).
THE IMPACT OF ECONOMIC GROWTH ON GENDER EQUALITY: EXPLORING THE CAUSAL PATHWAYS
As was pointed out earlier, the macroeconometric evidence for the impact of growth on gender equality was found to be weak and inconsistent. Standard economic theories would predict otherwise. For instance, a market-mediated explanation would suggest that the competitive market forces engendered by growth should increase the costs of gender-discrimination for employers and equalize opportunities (Gary Becker 1971). A family-mediated explanation would argue that the benefits of growth should ease scarcity-related constraints which otherwise force households to discriminate against less productive members (Duflo 2012). These pathways cannot, however, be taken for granted since they depend on the extent to which growth can generate competitive markets and benefit poorer households. This suggests we need to pay attention to different patterns of growth.
Despite the hegemonic status occupied by neoliberal ideas in shaping strategies for growth across the world, countries have not adopted these strategies at the same time, at the same pace, or to the same extent. Structural adjustment policies were imposed earlier in Latin America and sub-Saharan Africa (SSA), areas hardest hit by debt crisis in the 1980s. In contrast, oil-producing countries in the MENA region were able to use their oil revenues to bolster their public sectors, while larger economies like those of India and China were better positioned to carve out their own liberalization trajectories compared to smaller, aid-dependent economies. We would expect these variations in patterns of growth and in the role of the state to differentiate the impact of economic growth on gender equality.
The other reason to expect variations in the impact of growth is that the forces that create scarcity and wealth in a society are not necessarily the same forces that create and perpetuate the gendered structures of constraint. For instance, analyzing the relationship between growth and women's activity rates, Issis Gaddis and Stephan Klasen (2011) found that the “historically contingent initial conditions” which gave rise to earlier variations in female labor force participation rates remain far more important as determinants of current variations than their rates of growth. In terms of the geography of gender sketched out earlier, we find that these initial conditions include the extent and rigidity of patriarchal constraints in different contexts.
Growth, gender and the quantity of employment
At the global level, economic growth appears to have led to greater gender equality in labor market opportunities. The shift to neoliberal strategies has been accompanied by the “feminization of labor markets” as women increased their share of the labor force, while cross country analysis has noted the positive association between export-oriented growth and women's employment, particularly in middle- and low-income countries (Steven Kapsos 2005; James Heintz 2006).
However, country-level studies suggest that the female-intensity of employment associated with export-led growth varies considerably by type of export (Stephanie Seguino and Elissa Braunstein 2012). Women have gained from wage employment generated by export-oriented, labor-intensive manufacturing (for example, garments and textiles) and services (for example, tourism) and from large-scale, non traditional agricultural production. Traditional agricultural exports associated with smallholder farming, such as cocoa in Ghana and cashews in Tanzania, generally benefit male cultivators, as do non traditional agricultural exports based on out-grower schemes. Women are generally drawn into these industries as unpaid family labor on farms owned by men. Women are also under-represented in wage employment generated by capital-intensive industries, such as mining and petroleum-related production.
In the rest of this section, this contribution compares pairs of countries to illustrate how the impact of growth on the gender distribution of economic opportunities is mediated by variations in patterns of growth, the role of the state, and local structures of patriarchy. Growth rates and labor force data for these countries are provided in Supplemental Online Appendix 2.
We start out by comparing growth in Egypt and Morocco. Both are from the MENA region, a region which exemplifies Boserup's male farming system and continues to have the lowest FLFPRs in the world. One explanation for the persistence of these low rates is the predominance of capital-intensive, oil-based economies in the region, their reliance on an overwhelmingly male labor force, and the lack of incentives to diversify into labor-intensive export manufacturing and agriculture likely to benefit women (Ragui Assaad 2004). The result has been the reinforcement of the powerful male breadwinner ideologies and restrictions on women's mobility that prevail across the region (Valentine Moghadam 1998, 2001). In contrast, it is argued that oil-poor countries in the region were forced to turn to export-oriented manufacturing of garments and textiles. They stand out in the region for their higher levels of employment of women, demonstrating that patriarchal norms are not impervious to economic imperatives.
However, our comparison of Egypt and Morocco suggests a more complicated story. Egypt is one of the countries whose reliance on oil and oil remittances inhibited diversification into labor-intensive, women-dominated exports, while Morocco represents one of the oil-poor countries that did diversify early on in its development trajectory. Yet Egypt's FLFPRs rose from 25.9 percent in 1970 to 37 percent in 2000 (a growth of 43 percent), while those of Morocco rose from 32.6 percent to 43.6 percent (a growth of 34 percent) (World Bank 2004). The hypothesized association between patterns of growth and women's labor force participation was complicated in these countries by the varying role of the state.
Egypt embarked on structural adjustment programs in the late 1970s, but it continued to expand its public sector until the late 1990s. The public sector currently accounts for 30 percent of employment (World Bank 2013). Described as the country's “only equal opportunity” employer, it offered educated women guaranteed access to employment, together with wages and working conditions considerably better than those in the private sector (Mervat F. Hatem 1994). Consequently it accounted for 37 percent of women's employment, compared to just 27 percent of men's (World Bank 2004). At the same time, there was very little demand for women's labor in the undiversified formal private sector. As a result, downsizing the public sector in the late 1990s led to a decline in formal employment among educated women, further exacerbated by the “de-feminization” of formal private employment. Women's share of overall formal private sector employment declined from 14 percent to 10 percent (Assaad 2004).
Morocco embarked on structural adjustment around the same time as Egypt, but it had begun to downsize its already smaller public sector by the 1980s. The sector currently accounts for just 11 percent of women's and 8 percent of men's employment. Morocco also turned very early on to export-oriented garment manufacturing. Despite lower growth rates than Egypt in the 1990s, women's share of private sector employment rose overall from 18 percent to 24 percent by 1999, increasing for almost every category of employment. Their share of employment in textile and garment manufacturing had increased from 39 percent in 1990/91 to 63 percent by the end of the decade (World Bank 2004; Assaad 2004).
A second pair of country comparisons illustrates the impact on women's employment when neoliberal reforms are mediated by very different economic structures and patriarchal constraints. In terms of Boserup's analysis, Ghana represents women's farming system par excellence with women's family labor making up 36 percent of the agricultural labour force and 80 percent of trade. Bangladesh, then part of Pakistan, was characterized as a male farming system with equivalent estimates of 13 percent and 2 percent, respectively. These contrasting patterns are still discernible.
In Ghana, FLFPRs have risen fairly steadily from already high rates of 57 percent in 1960 to 87 percent in 2006. Men's activity rates remained above 80 percent for most of this period. The public sector had accounted for around 14 percent of employment at the end of the 1980s, but downsizing meant this declined to 9 percent over the following decade (Sally Baden 1994). Women lost out disproportionately; while they accounted for 21 percent of public sector employment in the early 1980s, they made up 35 percent of subsequent public sector retrenchments (Baden 1994). Recent figures suggest the public sector provides employment for just 2 percent of the women's workforce and 7 percent of men's (Chen et al. 2005). Private wage employment, whether formal or informal, accounts for around 6 percent of employment for men and 2 percent for women (Heintz 2005). The bulk of employment opportunities are thus informal.
Men have been the direct beneficiaries of the export-oriented expansion of oil, timber, and cocoa. They dominate wage employment in the oil and timber industries and are the primary cultivators of cocoa, the main agricultural export. But the long-standing tradition of independent women's entrepreneurship in Ghana has allowed large numbers of working women to run their own-account businesses in the off-farm sector. This was not only the most important source of employment for women in 1998 and 2006, but it also had lower risk of poverty than other forms of informal employment available to them (Heintz 2005).
Bangladesh, along with the rest of South Asia, historically reported extremely low FLFPRs, but these have gradually increased from 4 percent in the 1970s to 36 percent in 2010. The persistence of cultural restrictions on women's mobility in the public domain means that it is largely women from poorer households who are found in paid work. Despite a 30 percent reservation of jobs for women in public employment, they made up just 6–7 percent of workers in the mid-1980s (Sally Baden, et al. 2004). Current estimates suggest that formal employment accounts for 15 percent of men's labor force and just 8 percent of women's.
Women have been the direct and highly visible beneficiaries of the shift to export-oriented, labor-intensive manufacturing of garments where they make up approximately more than 80 percent of the labor force. According to a recent estimate, 15 percent of women aged fifteen to thirty (the relevant age group) are in the garment industry (Rachel Heath and A. Mushfiq Mobarak 2014). They have also benefited, but far less visibly, from the expansion of microfinance lending, targeted largely at women. Informal self-employment accounts for much of the increase in the women's labor force; but this is largely in unpaid family labor, which rose from 19 percent of women's employment in 1995 to 56 percent in 2010, so it is now the single most important source of employment available for women (Rushidan I. Rahman and Rizwanul Islam 2013).
Finally, the experiences of India and China allow us to compare how women have fared in the two countries that reported the fastest rates of growth of the world's major economies. As with the rest of the South Asia region, India has historically reported extremely low FLFPRs. In Bangladesh, women from poorer households and marginalized groups report the highest rates of labor force participation, along with a minority of educated women able to access “good” jobs in salaried employment. However, unlike Bangladesh, FLFPRs have been persistently low for several decades and have indeed declined from 32 percent in 1994 to 27 percent in 2010. Men's rates rose from 68 percent to 73 percent during the same period (World Bank 2014b).
Economic liberalization has generated some visible gains for women. Women make up 70–80 percent of employment in the country's few export-processing zones where garment and textile manufacturing predominates (Ghosh 2009). They have also made gains in segments of the new IT-enabled sector: for instance, they make up around 60 percent of workers in call centers. However, the overall story is one of low and declining participation. This partly reflects the bimodal employment pattern with the unionized sections of men's labor force employed in large-scale and highly protected establishments, while women predominate smaller, unprotected units where work is less easy to document (Dipak Majumder and Sandeep Sarkar 2009; Nidhiya Menon and Yana Rodgers 2009). The trend toward subcontracting work to these smaller establishments in order to escape labor legislation began in the 1970s, but it has intensified with economic liberalization (Ghosh 2009). In addition, the decline in FLFPRs is believed to reflect women's status-related withdrawal into unpaid family labor, which also eludes measurement (Indira Hirway 2012).
The Chinese experience differs in many ways. While pre-revolutionary China was characterized by a strong public–private divide of the kind associated with men's farming systems (Tamara Jacka 1992), the gender-egalitarian policies of the socialist state meant that China embarked on its liberalization process with extremely high FLFPRs and smaller gender wage gaps relative to most countries (Joyce P. Jacobsen 1998). The rural population was largely engaged in collectivized agriculture, while most working age men and women in urban areas worked in state-owned enterprises.
The impact of economic reforms has been mixed. Women made up 39 percent of the work force in state-owned enterprises in the 1990s, but they made up 61 percent of retrenched workers. The withdrawal of state support for women's unpaid domestic responsibilities meant longer periods of unemployment among laid-off women with young children (Sarah Cook and Xiao-yuan Dong 2011). They were also more likely than men to end up in lower paid and less skilled categories of work and work in the informal economy (Simon Appleton, John Knight, Lina Song and Qingjie Xia 2002; Yueping Song, and Xiao-yuan Dong 2011).
Elsewhere the story was more positive. While women's land rights were undermined in the transition from collective to family farming, they made gains in the growing rural industrial sector, making up 41 percent of its workforce in 1996 (Gunseli Berik, Xiao-yuan Dong, and Gale Summerfield 2007). Their share of formal employment in urban areas increased from 33 percent in 1978 to 39 percent in 1994 (Xiao-yuan Dong, Jianchun Yang, Fenglian Du, and Sai Ding, 2006), partly because of increased opportunities for younger, educated women in private enterprise and self-employment (Gale Summerfield 1994). Many found employment with the expansion of female labor-intensive commercial crops, such as tea, while others, mainly young rural migrants, were employed in the country's labor-intensive export manufacturing sector. Consequently, FLFPRs in China have remained among the highest in the world in 2006 (Berik et al. 2007).
Economic growth and the quality of employment
Women have made clear gains in terms of the quantity of employment generated by neoliberal patterns of growth, but their gains are far less clear in relation to its quality. There has been some decline in the horizontal segregation of the labor market with rising FLFPRs, but little evidence of a commensurate decline in its vertical segregation (Anker et al. 2003). While at the global level, there are roughly equal percentages of men and women in what the ILO classifies as “vulnerable employment” (own-account work and unpaid family labor), women are far more likely than men to be in vulnerable employment in developing countries, with a higher percentage of them in unpaid family labor where they receive no income of their own (ILO 2010).
Cross-country analysis of gender wage differentials suggests that increases in per capita GDP, trade, and FDI have been associated with a reduction in the gender wage gap, but this effect is confined to high-/higher middle-income countries (Remco Oostendorp 2009). The ILO (2007) has highlighted the difficulties of generalization on this issue. It points out that while there has been a decline in the gender wage gap in manufacturing in some countries (Costa Rica and the UK), there has been an increase in others (Egypt, Sri Lanka, and El Salvador). Moreover, where the gender gap narrowed, this was frequently due to a decline in men's wages rather than an increase in women's.
There are a number of reasons why the quality of women's jobs has not increased in line with their increasing share of the labor market. The deregulation of labor markets as part of the liberalization process has meant that women gained jobs at a time when jobs have generally become more casual and informal (Guy Standing 1999). Furthermore, their main gains in the formal economy have been in the labor-intensive, price-sensitive sectors of the export market where precisely the lower costs associated with women's labor relative to men's has driven these sectors’ profitability and explained the preference for women's labor (Stephanie Seguino 2000). While gender segregation of the labor market, and the crowding of women into a limited number of occupations, undermines their general capacity to bargain for better wages, their bargaining power is further undermined by the high levels of capital mobility that characterizes these sectors. Efforts by women workers to demand better terms and conditions have been frequently met either by the flight of capital to other locations or the subcontracting of the work to second and third tier firms in the global value chain where conditions are more informal and wages are lower (Naila Kabeer 2015).
Economic growth and women's well-being and agency
Although the review of the macroeconometric literature by Kabeer and Natali failed to find evidence that growth had any systematic impact on various measures of gender equality in well-being and agency, one finding that stood out in this literature was the role of women's labor force participation as a driver of positive change on some of these other measures. As Seguino (2006) concluded on the basis of her analysis of Latin American data:
Despite the fact that female share of the labour force includes both employed and unemployed women as well as paid and unpaid work, it is clear from these results that women's economic activity improves their well-being. Whether the bargaining power that this confers on women to negotiate with male members of the family, or because women directly generate income, the effect is positive and significant in most cases here (21).
That it is likely to be an “agency” rather than a purely “income” effect is suggested by additional evidence that women's economic activity is associated with an apparent transformation in values and attitudes in the larger society. Seguino (2007) used successive waves of the World Values Survey to analyze the impact of economic growth and women's economic activity on attitudes relating to gender equality among women and men. She found that, along with economic growth, women's share of employment and of the labor force were both positively associated with gender-egalitarian attitudes, more strongly expressed by women but also evident among men.
In this section, we explore what lower-level studies can tell us about the relationship between economic growth and women's well-being and agency. While we have already noted the positive association between women's access to valued economic resources and children's well-being, this is in conformity with socially expected patterns of behavior on the part of women, a “compliant” form of agency. Here, we are seeking evidence that women's access to economic resources translates into forms of agency that have the potential to challenge the gendered structures of constraint or transformative forms of agency.
The literature on the empowerment implications of women's access to valued resources has highlighted some of its positive impacts on women's sense of self-worth, decision-making power, mobility in the public domain, ownership of savings and assets, respect within the community, and so on (Supplemental Online Appendix 1, Table 3; Lucia Hanmer and Jeni Klugman 2015). But while the macro-level literature highlights the general importance of women's labor force participation as a determinant of their agency, the micro-level literature nuances this finding further. It suggests that it is the terms on which women enter the labor force, particularly the extent to which they are able to control the proceeds of their own labor, rather than participation per se, that determine the transformative potential of their work (Whitehead 1985; Miet Maertens and Johann F.M. Swinnen 2008; Naila Kabeer, Ragui Assaad, Akousa Darkwah, Simeen Mahmud, Hania Sholkamy, Sakiba Tasneem, and Dzodzi Tsikata 2013).
Therefore, one reason that economic growth has not translated more systematically into impacts on women's well-being and agency may relate to the wide variations in the forms of employment it has generated (discussed above [Supplemental Online Appendix 2, Table 3]). In addition, we find that some aspects of gender inequality have either proven resistant to change or else have changed in perverse and unexpected ways, bearing out our earlier point that forces that perpetuate gender inequality are relatively autonomous from those that drive economic growth.
The first of these relates to the gender division of unpaid labor within the home. As we have seen, this is one of the key constraints on women's labor market participation. While economic theory predicts that increasing returns to women's work in the market place should lead to some reallocation in the division of labor in unpaid work within the home, for most women, the evidence suggests that this reallocation either does not take place and they ended up working a “double shift,” or else it is reallocated to another female family member, perhaps the eldest daughter. The same normative structures that associate femininity with caring roles appear to define masculinity in terms of distance from these roles.
The resulting gender inequality in hours of work put in by working men and women holds to varying degrees across the world, but the variation appears to reflect the role of the state and local cultural norms rather than rates of growth. For instance, a report by the United Nations Research Institute for Social Development (UNRISD; 2010) found that, among higher income countries, gender distribution of total working hours was equal only among the Nordic countries, with the largest gaps reported in southern European countries. Estimates provided by WDR 2012 covering thirty-five countries at different levels of development found women devote 50 percent more time than men to housework and childcare in Cambodia and Sweden, but give three times more in Italy and six times more in Iraq.
A second aspect of intra-household inequality that appears resistant to changes in women's economic position is domestic violence. While the prevalence of such violence appears fairly universal, its incidence varies considerably across countries, reflecting both variations in the cultural norms defining masculinity and femininity, as well as variations in public policy to address the problem. It might be expected that improving women's access to material resources would strengthen their bargaining power within the domestic domain, but, in fact, the findings are highly mixed (Supplemental Online Appendix 3). Indeed, it is very often women's desire to take up paid work, particularly paid work outside the home, which triggers domestic violence. As Rachel Jewkes (2002) suggests, a common trigger for intimate partner violence across different contexts is the transgression of gender norms and the failure to fulfill cultural expectations of good womanhood/successful manhood. Norms regarding men's role as family breadwinners are pervasive across cultures, although there is considerable variation regarding women's economic roles. Improvements in women's access to work, credit, or property are most likely to trigger violence in contexts in which men's sense of manhood and self-worth is closely bound up with the idea of women's dependence.
A final area in which intra household gender inequalities have not only proven resistant to growth but been perversely affected by it relates to the “missing women” phenomenon noted earlier. While improvements in women's life expectancy relative to those of men has led to a decline in the incidence of “missing women” in most countries that had reported them previously, this decline is only partly related to rates of growth (Stephan Klasen and Claudia Wink 2003). In countries with some of the fastest rates of growth in recent decades, such as India and China, the phenomenon of “missing women” has given way to that of “missing daughters,” abnormally high ratio of sons to daughters at birth. It appears that the resilience of son preference in these contexts is leading families to reconcile their desire for fewer children with their desire for sons by resorting to female-selective abortion, made possible by the dissemination of new ultrasound technologies. Once again, we are reminded that the impact of economic growth is mediated by local structures of patriarchal constraint in ways that confound easy predictions about its strength and direction.
CONCLUSION
The aim of this contribution was to search out possible causal mechanisms in the empirical literature that could explain the asymmetry in the two-way relationship between economic growth and gender equality reported in the macroeconometric literature. Interestingly, the mechanisms it has identified can be nicely summarized by what Gayle Rubin, in her classic text, described as “the endless variety and monotonous similarity” of patriarchal structures across the world (1975: 160).
The positive impact of gender equality on economic growth can be seen as one manifestation of this “monotonous similarity”: the near-universal assignment of primary responsibility for unpaid care work within the family to women. Across the world, it appears that increases in women's access to valued resources is likely to translate into children's well-being and education, and hence into the productivity of future generations of workers. This positive impact can materialize, regardless of whether the increase in question is absolute or relative to men. However, women's ability to contribute more directly to economic growth through their productive activities depends to a greater extent on improving gender equality in access to the necessary resources and opportunities, including greater equality of opportunity to participate in forms of work that are included in estimates of Gross National Product (GNP), and hence in estimates of growth.
Turning to the reverse relationship, our analysis suggests a number of reasons for the weak and inconsistent impact of economic growth on gender equality documented in the macroeconometric literature. First, it is the pattern, rather than the pace, of growth which determines the gender distribution of economic opportunities, and this varies considerably across the world. Second, the state's role in actively promoting greater gender equity in the gains from growth has varied from country to country. The third reason takes us back to Rubin's observation: the “endless variety” of patriarchal structures that mediate the impact of growth means that women are better positioned to claim their share of the benefits of growth in some contexts more than others.
To conclude we would agree with Duflo's conclusion that the case for gender equality should be made on intrinsic rather than instrumental grounds, but we arrive at our conclusion through a different set of arguments. We do not rule out the instrumental argument on the grounds that it lacks evidence, as Duflo does, but rather because the evidence in support of it relies strongly on women's conformity to socially ascribed maternal roles. This constrains their ability to participate in economic activity on equal terms with men and to participate more generally in the public life of their community.
This still leaves us with an important question: how do we achieve greater equality on the range of different dimensions that are essential to a holistic vision of gender justice? As we pointed out, there is no shortage of policy recommendations which could contribute to this vision. What we lack is the political momentum to translate these recommendations into concrete outcomes. As far as how such momentum could come about, we must once again part company with Duflo. She focuses on historical data from the US, suggesting that economic growth created the conditions for progress on women's rights. The explanation for this which she finds “theoretically convincing” relies, not surprisingly, on individual rational choice, with male agency in the form of paternal altruism once again driving change. The argument here is that, with economic development, men were either more willing to surrender some rights to their wives to ensure their children were better educated or else their interests as husbands (wanting all the rights) began to conflict with their interests as fathers (wanting to protect their daughters against their future husbands).
We are more convinced by contemporary evidence that highlights the importance of women's collective agency. We noted evidence that economic growth appears to promote gender-egalitarian attitudes on the part of both women and, to a lesser extent, men. Changing attitudes clearly create favorable conditions for changing practices, but it is not enough. Instead, what recent cross-country analysis suggests is that collective action by women, both through their autonomous organizations and in alliance with other groups, has proven to be the single most important driving force in promoting public action on a range of women's rights (Mala Htun and Laurel Weldon 2010, 2012). It would thus appear that it is to women's ability to mobilize collectively that we must look if changes in social attitudes are to be translated into concrete progress on gender equality.
Many thanks to Jana Miletzki for assistance in the literature search and to Eva-Maria Egger for assistance in the literature search and construction of tables.
