Policy Research Working Paper 11211 Women in Sciences and Engineering Why Leveling Employment in Industrial Sectors Matters for Economic Diversification in Developing Countries Yselle Flora Malah-Kuete Désiré Avom Luc Désiré Omgba Africa Region Office of the Chief Economist September 2025 Policy Research Working Paper 11211 Abstract While the importance of export diversification is empha- overall, women’s employment in different economic sectors sized in development patterns, the gender issues in this positively affects the level of export diversification in devel- process are rarely covered. This paper empirically assesses oping countries. However, this positive effect is not evident the contribution of women’s sectoral employment to the in the primary sector, although it is significant when women export diversification process, using a panel model of 125 are employed in the industrial and service sectors. Specific developing countries from 1991 to 2018. The results training in sciences, engineering, and technology through obtained after several simulations and robustness tests to education is the underlying mechanism that explains the account for potential sources of endogeneity show that, results. This paper is a product of the Office of the Chief Economist, Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at florayselle27@yahoo.com. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Women in Sciences and Engineering: Why Leveling Employment in Industrial Sectors Matters for Economic Diversification in Developing Countries Yselle Flora Malah-Kuete Faculty of Economics and Managment, University of Yaounde 2, 135 Yaounde, Cameroon, florayselle27@yahoo.com Désiré Avom Faculty of Economics and Managment, University of Yaounde 2, 135 Yaounde, Cameroon, Email: davom99@gmail.com Luc Désiré Omgba, Corresponding author BETA-CNRS, University of Lorraine, 23 rue Baron Louis, 54000 Nancy, France, Email: luc-desire.omgba@univ-lorraine.fr Authorized for distribution by Aparajita Goyal, Lead Economist, Africa Region, World Bank Group Keywords: export diversification, economic transformation, women's employment, STEM education, gender and development, developing countries. Code JEL: J16, J24, F1, O14, O54. Note: This paper was independently developed by the author during her consultant appointment at the World Bank’s Office of the Chief Economist for the Africa Region under the WB-AERC Fellowship. It benefited from constructive feedback received during internal seminars with World Bank staff, including the author’s TTL, Mr. Woubet Kassa. Acknowledgments: We thank the participants and discussants of the World Bank seminar in April 2024 in Washington, D.C., for their insightful comments and suggestions. The opinions expressed herein are those of the authors and do not necessarily reflect the official position of the World Bank or its affiliated institutions. Introduction Economic diversification is a fundamental issue for developing countries to cope with constraints associated with the dependence on commodity exports (Prebisch,1950; Singer, 1950; Imbs and Wacziarg, 2003; Cadot et al., 2010). Theoretically, it has been argued that diversification promotes growth by increasing factor productivity, especially labor productivity (Romer, 1990). Diversification also broadens the possibilities for spreading investment risks (Acemoglu and Zilibotti, 1997). Non-exhaustive empirical studies have produced diverse but complementary results. They show that diversification strengthens the resilience of economies (Karanfil and Omgba, 2023), promotes job creation (Ben Hammouda et al., 2006), reduces output volatility, and stabilizes the macroeconomic framework (Acemoglu and Zilibotti, 1997; Sachs and Warner, 2001; Gylfason and Nganou. 2014; Ross 2018, Lederman et al., 2019). This importance of diversification for development has led to an extensive body of literature to identify its determinants. As a result, several determinants of export diversification are presented in that literature. This includes income level (Imbs and Wacziarg, 2003; Klinger and Lederman, 2006), international trade (Cadot et al., 2010), geographic distance to major markets (Parteka and Tamberi, 2008; Agosin et al., 2012), and institutions (Gylfason, 2006; Gelb, 2010; Omgba, 2014; Boschma and Gianluca, 2015). However, the non-exhaustive and non- consensual nature of these determinants, due in particular to the context of the study, the availability of data, and the methodologies used, opens up new research perspectives. In this regard, the role of women's participation in the workforce is not addressed in the aforementioned literature. Despite the incomplete evaluation, in developing countries, particularly in Africa, women, whether formally educated or not, participate in a broad range of economic activities in all sectors of the economy, both formal and informal. According to data from the International Labour Organization (ILO) (2021), women comprise over 50 percent of the workforce in formal micro, small, and medium-size enterprises in Sub-Saharan Africa. According to the Food and Agriculture Organization of the United Nations, women produce up to 80 percent of the food for household consumption and local sales channels (FAO, 2021). Additionally, many successful business models can be referenced in several countries. This is the case of "Nana Benz," women from Togo who developed and controlled a vast regional and global commercial and distribution network specializing in loincloths. So far, an important strand of literature focuses on the effect of gender inequalities on export diversification, examining how the gaps between men and women in terms of health, education, empowerment, and labor force participation influence this diversification (see Kasente et al. (2002); Kazadjian et al. (2016, 2019); and Kilolo et al. (2023)). While this paper joins the aforementioned studies on the important role of women on economic diversification, this paper empirically assesses the effect of women's participation in the workforce on export diversification in developing countries, a point not addressed in previous studies. In depth, Goldin (1987) and Goldin (1994) underscore the significant impact of women's participation in diverse economic sectors. These works provide a compelling framework for our study, which focuses on the influence of women's employment on export diversification in developing countries. Furthermore, this study aligns with the literature that describes women as a powerful engine for international trade and economic growth. Laperle-Forget and Cuneo (2024) emphasize their participation as entrepreneurs, producers, and workers at different levels 2 of global value chains, significantly contributing to productivity, innovation, and competitiveness in global markets (World Bank and WTO, 2020). While recognizing their relevance, this study does not simply examine how including women on an equal footing with their male counterparts in the economy could boost diversification. It also explores how and by what mechanisms women's employment and active participation in the labor market in different sectors (agriculture, industry, and services) can affect the distribution of exports. Accordingly, we use various econometric tools to estimate a panel model of 125 developing countries over the period 1991-2018. The main result of this study is that women's employment, specifically in the primary sector of agriculture, livestock, and fisheries, is negatively correlated with export diversification. In other words, when women are employed in the primary sector, they are associated with higher export concentration. On the other hand, in the industry and service sectors, they contribute significantly to export diversification. The rest of the paper is organized as follows. Section I presents the findings of the empirical literature and the conceptual framework. Sections II explains the empirical strategy. Section III interprets the results, analyzing their sensitivity and robustness. Section IV tests a transmission channel. I. LITERATURE REVIEW AND CONCEPTUAL FRAMEWORK The literature focusing on the role of women in the export diversification processes in developing countries is rare. Literature has been more focused on the role of women in the process of economic development at a global level. In that area of research, three main axes of literature are highlighted, particularly those associated with economic growth. The first axis assesses the effect of female employment on economic growth. The interest in gender-specific determinants of growth has accompanied the shift from exogenous to endogenous growth models. For example, Galor and Weil (1996) combine a growth model based on the endogenous labor supply of women and men with a household model that outlines a couple's choices between unpaid household activities and paid work. The authors conclude that women's participation in the labor market provides the household with additional income, which makes more significant savings possible. Through a backward-looking effect, the increase in savings increases the capital stock per worker and, consequently, output. Indeed, the more actively women participate in the labor market, the faster the national economy grows. The second axis examines the effects of gender disparities and inequalities on trade and economic development (Kleven and Landais, 2017; Ben Yahmed and Bombarda, 2020). For example, Klasen (2000) finds from a sample of 109 countries that gender disparities harm economic growth. The author concludes that a reduction in these disparities in the labor market not only benefits women but is also significant in financial terms. Similarly, in a sample of Middle Eastern and North African countries, Klasen and Lamanna (2009) show that high levels of gender discrimination in employment artificially limit the "talent pool" of a country's workforce. As a result, it hinders the country's ability to compete internationally. For Duflo (2012), closing gender gaps in education, political participation, and employment opportunities has implications for many other society-wide outcomes. It is a "prerequisite" for achieving the Sustainable Development Goals. Kazandjian et al. (2016) find that gender inequality decreases the variety of goods that countries produce and export, particularly in developing countries. They identify two transmission channels. The first is that gender gaps in opportunities, such as lower school enrollment rates for girls relative to boys, limit the potential pool of human capital 3 available in an economy. The second is that gender gaps in the labor market hinder the development of new ideas by reducing the workforce's efficiency. The third axis analyzes the opposite relationship, i.e., the consequences of economic development on women's participation in the labor market. The theory of the "marginalization of women" put forward by Boserup (1970) describes how development, specifically industrialization, modifies the traditional social and productive roles of women. This theory postulates that in developing countries, women become marginalized with development. Women's participation in the labor market follows a U-shaped curve. Initially, development pushes women out of the labor force, in part because of increased market opportunities for men and social barriers that prevent women from entering the paid labor force. However, as countries develop, women's educational attainment increases, and women re-enter the workforce as wage earners in mostly white-collar jobs. This theory has since been the subject of several research discussions and controversies (Psacharopoulos and Tzannatos, 1989; Goldin, 1995). Regarding the specific link between women and the diversification process, to the best of our knowledge, existing studies have been more interested in the effect of gender parity and inequalities in the labor market or education on export diversification. This is exemplified by Kazadjian et al. (2016), who demonstrate that gender inequality decreases the variety of goods produced and exported by countries, especially in low-income and developing countries. Similarly, Kilolo et al. (2023) find, in a sample of developing countries, that gender parity in secondary school enrollment and the political empowerment of women have a positive effect on export diversification, in contrast to gender wage inequality, which has a negative effect. These authors posit that human capital, particularly gender gaps in opportunity, such as lower educational enrollment rates for girls than boys, is the primary transmission channel. Indeed, a vast body of literature recognizes the role of human capital in diversification. For example, a study by Agosin et al. (2012) argues that human capital is the key determinant of economic diversification and helps drive growth. Regarding the role of women's human capital in diversification, Wood and Berge (1997) support the hypothesis that an educated female labor force is a determinant of the growth of manufacturing exports. For Aghion and Howitt (2006), gender disparities in education and health undermine the labor force's potential for export diversification. The study by Amin et al. (2015) specifies that gender disparities in education hurt economic growth, particularly in poor countries, which hinders the process of economic diversification. For Kazandjian et al. (2019), the gap between educational opportunities for women and men harms the potential pool of human capital, which limits diversification and slows technological absorption and innovation. This study thus extends these research findings. Beyond the proven effects of gender parity or inequality, both in the labor market and in education and politics, we posit that women constitute an available and crucial workforce in various sectors of the economy. Therefore, we focus on the effect of sectoral female employment on export diversification in developing countries. The conceptual framework of our study is outlined in Figure 1, delineating the reasons and mechanisms by which women's employment in various economic sectors can lead to export diversification in developing countries. This framework is based on an analysis of existing literature on the specific contributions of women in business production, market expansion, and commercial networking. Women's employment in different sectors, such as agriculture, 4 industry, and services, plays a crucial role in export diversification, operating through several interdependent mechanisms. Figure 1. Conceptual framework: The role of women's sectoral employment in export diversification Source: Authors, based on conceptual insights from Kabeer (2012), World Bank and WTO (2020), and Walters et al. (2022). Firstly, integrating women into these sectors serves as a role model, especially when they hold positions within a given sector. They become role models for young girls and women, thus encouraging them to pursue education and careers in these fields. Additionally, women's employment can provide additional financial resources to families, enabling children, especially girls, to access better education. Moreover, when educated and trained, women contribute to increased productivity and innovation (Kabeer, 2012; Walters et al., 2022 ). By actively participating in the workforce, women not only directly contribute to production but also enhance management practices and decision-making within enterprises (World Bank and WTO, 2020). Furthermore, increased productivity and innovation can be influenced by women's labor force participation in various sectors. For example, in agriculture, women often bring forth more efficient and sustainable production techniques (FAO, 2011). In industry and services, they develop skills in commerce and marketing that open up new markets for local products (UN Women, 2015). These skills are essential for diversifying economic sectors and increasing exports. Moreover, when they are economically empowered, women are often more inclined to create and participate in trade networks and community financing structures (Anderson and Baland, 2002; Agénor et al., 2021). These networks facilitate access to credit and capital, enabling small enterprises to grow and explore new international markets. Increased productivity, improved commercial skills, and the establishment of solid networks lead to sectoral diversification. Enterprises led by or predominantly employing women tend to diversify their products and services, enabling them to penetrate new markets and reduce their dependence on a few traditional exports (Busse and Nunnenkamp, 2012). 5 Finally, sectoral diversification and the opening of new markets directly contribute to export diversification. Countries that integrate women more into key sectors often see an increase in the variety of exported goods, enhancing their economic resilience to fluctuations in commodity prices and external financial shocks (Hausmann et al., 2010). II. EMPIRICAL STRATEGY 2.1. Empirical model To test the proposition of interest in this study, i.e., the potential relationship between female employment and export diversification in developing countries, we rely on the following econometric model: = + + + (1) Where is the export diversification index in country for year , the measure of female employment and is the parameter of interest to be estimated, is the vector of control variables and the vector of associated parameters, is the country's time-varying error term. 2.2. Econometric technique In the econometric literature, panel data are usually estimated using either the fixed effects or the random effects estimator. Nevertheless, these estimators do not account for certain violations of the linear regression model, in particular, autocorrelation, heteroscedasticity, and cross-sectional dependence in the data (White et al., 1980; Newey and West, 1987), which may arise from aspects such as social norms, proximity to the periphery, or sheep-like behavior as in the context of our research interest. To address these problems, Parks (1967) proposed using generalized least squares (GLS), which allows a more flexible covariance structure for disturbances and random effects. This method allows for the modeling of linear cross-sectional time series models and estimation when there is autocorrelation within each panel, as well as cross-sectional correlation or heteroscedasticity between panels. However, because the Parks method can lead to an underestimation of parameter variability (Beck and Katz, 1995), “feasible” generalized least squares (FGLS) are used in this study because they assume that the error is known rather than estimated. Furthermore, according to Olive (2017), many linear models with serially correlated errors (such as AR (1) errors) and many linear mixed models can be fitted with the FGLS estimator. Equation (1) is therefore estimated under an FGLS setting. However, since the FGLS estimator does not account for the endogeneity of the explanatory variables, various additional tests and econometric techniques will be performed in robustness analysis to address this potential shortcoming. 2.3. Data The analysis covers 125 developing countries (DCs) (low-income, upper-middle-income, and lower-middle-income) observed from 1995-2018. Panel dimensionality is conditional on data availability for both measures of female employment and export diversification. In this study, the role of women in the export diversification process in DCs is assessed via the total and sectoral employment rate of women. According to the ILO definition (ILO, 1990), employment is defined as persons of working age (15 years and over) who were engaged in some form of activity in the production of goods or provision of services for pay or profit, whether they worked during the reference period or did not work due to temporary absence from work or working time arrangements. We consider the three main sectors of the economy. 6 The agriculture sector includes agriculture, hunting, forestry, and fishing activities, according to Division 1 of International Standard Industrial Classification (ISIC 2), Categories A-B (ISIC 3) or Category A (ISIC 4). The industrial sector includes mining and quarrying, manufacturing, construction, and utilities (electricity, gas, and water), according to Divisions 2-5 (ISIC 2), Categories C (ISIC 3) or Categories B-F (ISIC 4). The service sector includes wholesale and retail trade, restaurants and hotels, transportation, warehousing and communications, finance, insurance, real estate and business services, and community, social and personal services, according to ISIC 2 Divisions 6-9, ISIC 3 Categories G-Q or ISIC 4 Categories G-U. Total female employment measures the proportion of the female labor force. These data are drawn from the ILOStat database (2022). Regarding the export diversification index, the empirical literature distinguishes several indices: the Herfindhahl-Hirschman Index (HHI), the Theil Entropy Index, the Gini Index, and the Ogive Index (Cadot et al., 2011, 2013; Djimeu and Omgba, 2019). Unlike concentration indices such as the HHI, diversification indices reflect the number of products in a country's export basket and the distribution of their shares. An example is the Theil index. However, according to Tran et al. (2017), overall, export concentration indices provide relatively similar rankings to those of diversification and thus can be used interchangeably. For example, Agosin et al. (2012) and Avom et al. (2020) alternate these indices as robustness analyses in their studies and arrive at relatively similar results. In this paper, we use the HHI in the main specification. As a robustness analysis, the Theil diversification index is used to test the reliability of the results obtained with the HHI. The HHI is normalized between 0 and 1. A value of 1 indicates a highly concentrated export structure (i.e., less diversification) and vice versa. It is calculated as follows: 2 1 �∑ =1� � −��� = (2) 1−�1⁄ Where is the index of country j, the value of exports of the product , = ∑ and the number of product groups according to the Standard International Trade Classification (SITC) revision 3. Data is from the United Nations Conference on Trade and Development (UNCTAD) United Nations Comtrade Database. As for the control variables, we retain GDP per capita and its quadratic value to test the non- linear relationship between economic development and diversification highlighted in the seminal study by Imbs and Wacziarg (2003). In addition, we include trade population density to account for the effect of country size, FDI (Ben Hammouda, 2006), and terms of trade (Agosin et al., 2012). The complete list of variables used, as well as definitions and data sources, are given in Appendix Table A1. III. ECONOMETRIC RESULTS Firstly, we discuss the results of the estimation of the basic model, followed by alternative specification, and finally, the additional robustness results. 3.1. Preliminary findings Table 1 presents the estimation results of equation 1 using an FGLS estimator. As independent variables of interest, we consider, in turn, total employment (column 1), agricultural employment (column 2), industrial employment (column 3), and employment in services (column 4). 7 Table 1. Female employment and export diversification Dependent variable: HHI (1) (2) (3) (4) Total female employment 0.0007*** (0.0002) Female employment in Agriculture 0.0010*** (0.0002) Female employment in Industry -0.0024*** (0.0003) Female employment in Services -0.0001 (0.0003) GDP per capita (log) -0.0856* -0.0559 0.0021 -0.1281*** (0.0455) (0.0450) (0.0438) (0.0446) GDP per capita^2 (log) 0.0024 0.0015 -0.0031 0.0049* (0.0029) (0.0029) (0.0028) (0.0029) Trade -0.0001** -0.0001** -0.0001 -0.0002** (0.0001) (0.0001) (0.0001) (0.0001) FDI -0.0008 -0.0006 -0.0013** -0.0008 (0.0005) (0.0005) (0.0005) (0.0005) Population density (log) -0.0499*** -0.0498*** -0.0464*** -0.0512*** (0.0018) (0.0017) (0.0017) (0.0018) Terms of trade -2.89e-16*** -2.72e-16*** -2.62e-16*** -2.68e-16*** (7.36e-17) (7.17e-17) (7.64e-17) (7.08e-17) Constant 1.0111*** 0.8377*** 0.7246*** 1.2361*** (0.1775) (0.1779) (0.1661) (0.1692) Observations 2238 2238 2238 2238 Wald chi2(7) 1438.22*** 1414.23*** 1833.65*** 1251.53*** Note: Standard deviations in parentheses. p < 0.10, p < 0.05, p < 0.01. FGLS estimates. Source: Authors’ estimations using data from ILOSTAT, UNCTAD, and World Bank (1991–2018). Column 1 shows a positive relationship between export concentration and female employment in developing countries. When we disentangle this effect between agriculture, industry, and services, the positive effect remains in agriculture. In contrast, female employment in industry reduces export concentration. As for women's employment in services, although the coefficient is negative, suggesting that women's employment in this sector tends to reduce concentration, the effect is not significant. Taken together, these initial results highlight the importance of female employment for export diversification in developing countries. In the low-productivity sectors of developing countries, such as agriculture, the presence of women tends to reinforce the negative effect of this sector on diversification. On the other hand, in sectors of economic transformation, such as industry, women's employment reduces the concentration of exports. In developing countries, limited access to resources and certain cultural norms may explain the positive effect of women's agricultural employment on the export concentration index. More specifically, when women are victims of unequal access to productive resources such as credit, land, or agricultural technologies, this can limit their ability to diversify into new, high-value- added products. As a result, they remain involved in farm products traditionally exported from the country, thus contributing to increased export concentration. Furthermore, in some societies, cultural norms require women to be confined to specific agricultural tasks, which limits them to subsistence farming activities rather than commercial activities for export. 8 As for the negative effect of industrial employment on the concentration of exports, this can be explained by the improvement in quality and added value. Indeed, women are generally recognized for their sense of aesthetics and their attention to detail and precision. In this sense, employing women in the industrial sectors could enhance the quality and design of exported products, making them more attractive in the markets. This would encourage diversification into new, higher-quality products. 3.2. Alternative specification This section analyzes how the above results are sensitive to the level of diversification and heterogeneity of the sample, particularly concerning natural resource dependence. 3.2.1. Does the stage of diversification matter? To verify whether or not the intensity of the effect of women's employment on export diversification in developing countries is sensitive to the diversification stage, we opt for a quantile regression. We postulate that a country's stage of diversification can have an impact on the effect of women's employment on export diversification. For example, in countries that are still in the early stages of their diversification process, women can contribute new skills, expertise, and additional labor that are essential for the development of new sectors. On the other hand, in countries at a relatively advanced stage of diversification, the employment of women can not only help to maintain and strengthen the level of diversification but also boost the competitiveness of exports through innovation and the differentiation of exported products. The quantile regression allows us to assess the effects at different levels of the conditional distribution of the HHI series, not only at the mean of the series, as shown by the results obtained with the FGLS estimator. Figure 2. Distribution of the HHI Note: The index is normalized between 0 and 1, with higher values indicating stronger export concentration. Source: Authors’ calculations using UNCTAD export data. 9 The results obtained are summarized in Table 2. Following the distribution represented in graph 2, we focus on the effect on the median (τ = 0.5), first and last deciles (τ = 0.1 et τ = 0.95), and the first and last quartiles (τ = 0.25 et τ = 0.75). For clarity, we present only the coefficients of the variables of interest. The control variables are introduced into the model but are not reported in the table. Two additional lessons can be drawn from these results. Firstly, the negative relationship between total female employment and the export concentration index is not homogeneous and depends on the quintile. Secondly, when broken down by sector, the negative relationship between employment in industry and concentration is confirmed in all quintiles. In contrast, the positive relationship in agriculture is only confirmed from the median quintile onwards. All this suggests a robust effect of women's employment in industry for diversification, which offsets the positive effect of women's employment in agriculture below the 75th quintile. Thus, regardless of the country's level of industrialization, female employment is positive for export diversification. Table 2. Quantile regression results: Female employment and export concentration Dependent variable: HHI τ =.10) τ =.25 τ =.50 τ =.75 τ =.95 Total female employment -0.0004*** -0.0001 -0.0003*** 0.0008 0.0033*** (0.0001) (0.0002) (0.0001) (0.0006) (0.0007) Controls Yes Yes Yes Yes Yes Constant 0.9021*** 0.6984*** 1.0095*** 2.5496*** -0.2082 (0.1063) (0.1678) (0.0492) (0.5114) (0.9106) Observations 2238 2238 2238 2238 2238 Pseudo R2 0.1565 0.1501 0.1414 0.0844 0.0305 Female employment in Agri. -0.0001 -0.0001 0.0003*** 0.0039*** 0.0061*** (0.0001) (0.0001) (0.0001) (0.0005) (0.0004) Controls Yes Yes Yes Yes Yes Constant 0.6227*** 0.7956*** 0.5966*** 0.3406 -2.6705*** (0.1206) (0.1645) (0.0938) (0.4630) (0.6805) Observations 2238 2238 2238 2238 2238 Pseudo R2 0.1565 0.1500 0.1400 0.1011 0.0696 Female employment in Indu. -0.0005* -0.0008** -0.0013*** -0.0028* -0.0071*** (0.0003) (0.0003) (0.0002) (0.0015) (0.0014) Controls Yes Yes Yes Yes Yes Constant 0.5572*** 0.6779*** 0.6596*** 2.2804*** -0.4227 (0.1215) (0.1543) (0.0838) (0.5460) (0.5406) Observations 2238 2238 2238 2238 2238 Pseudo R2 0.1592 0.1503 0.1458 0.0843 0.0436 Female employment in Serv. 0.0002 0.0002 -0.0001 -0.0031*** -0.0082*** (0.0001) (0.0002) (0.0002) (0.0008) (0.0012) Controls Yes Yes Yes Yes Yes Constant 0.7305*** 0.7957*** 0.8034*** 2.1668*** -2.2125*** (0.1134) (0.1509) (0.1040) (0.4413) (0.3806) Observations 2238 2238 2238 2238 2238 Pseudo R2 0.1566 0.1493 0.1403 0.0884 0.0438 Note: Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01. Quantile regressions estimated at τ = 0.10, 0.25, 0.50, 0.75, and 0.95. All models include control variables. Source: Authors’ estimations using ILOSTAT, UNCTAD, and World Bank data. 3.2.2. Sensitivity of results to panel heterogeneity Next, we test the sensitivity of our results to the presence of possible heterogeneity due to the dependence of certain countries in the sample on natural resources. Indeed, as Ross (2018) has 10 shown, natural resource extractive industries are generally male-intensive sectors. Thus, a country's heavy dependence on natural resources can lead to an enclave effect (Isham et al., 2005), where women are excluded from employment opportunities in these specific sectors. As a result, their participation in export diversification may be hindered due to the concentration of male employment opportunities in these extractive industries. We decompose the sample by considering, on the one hand, countries whose total rent from natural resources is greater than 5 percent of GDP (column 3), and on the other, countries where this rent is less than 5 percent of GDP (column 4) (Avom et al., 2020). The results are shown in table 3. Analysis by sub-sample reveals that in countries with high dependence on natural resources (columns 5-8), total employment of women as well as women’s employment in industry and services are negatively correlated with export concentration. This suggests that women’s employment may play a major role in curbing export concentration in resource countries. However, in the agriculture sector, this leads to increased export concentration. In contrast, in countries with low dependence on natural resources, total female employment, and more specifically in services (columns 1 and 4), favors export concentration, while employment in agriculture and industry reduces it. Table 3. Sensitivity of results to sample heterogeneity Dependent variable: HHI Countries total resource rent ≤ 5 % GDP Countries total resource rent >5 % GDP (1) (2) (3) (4) (5) (6) (7) (8) Total Fem. Emp. 0.0010*** -0.0008*** (0.0001) (0.0002) Female Emp. Agri. -0.0003*** 0.0035*** (0.0001) (0.0003) Female Emp. Indus. -0.0026*** -0.0028*** (0.0003) (0.0006) Female Emp. Serv. 0.0018*** -0.0039*** (0.0002) (0.0006) GDP per capita (log) 0.1065*** -0.0067 0.1201*** -0.0094 -0.1615** 0.1788** 0.0336 0.0820 (0.0384) (0.0327) (0.0382) (0.0368) (0.0821) (0.0786) (0.0829) (0.0784) GDP per capita^2 (log) -0.0103*** -0.0044** -0.0114*** -0.0050** 0.0108** -0.0067 -0.0009 -0.0013 (0.0024) (0.0021) (0.0024) (0.0023) (0.0055) (0.0052) (0.0055) (0.0052) Trade 0.0003*** 0.0001*** 0.0003*** 0.0003*** -0.0004* -0.0005*** -0.0004** -0.0006*** (0.0001) (0.0000) (0.0001) (0.0001) (0.0002) (0.0002) (0.0002) (0.0002) FDI -0.0013** 0.0001 -0.0011* -0.0005 -0.0003 -0.0010 -0.0013 -0.0010 (0.0005) (0.0005) (0.0006) (0.0005) (0.0010) (0.0009) (0.0010) (0.0009) Pop. Density (log) -0.0186*** -0.0275*** -0.0146*** -0.0227*** -0.0490*** -0.0508*** -0.0567*** -0.0529*** (0.0015) (0.0009) (0.0016) (0.0014) (0.0045) (0.0040) (0.0043) (0.0041) Terms of trade -1.57e-16 -2.40e-17 -7.89e-17 -4.99e-17 -1.37e-16** -1.78e-16*** -2.12e-16*** -1.32e-16** (1.27e-16) (1.15e-16) (1.26e-16) (9.95e-17) (5.84e-17) (6.01e-17) (6.97e-17) (5.75e-17) Constant 0.0713 0.6970*** 0.0961 0.6359*** 1.2872*** -0.4471 0.5244* 0.3077 (0.1550) (0.1272) (0.1488) (0.1445) (0.3042) (0.3017) (0.3021) (0.2874) Observations 1341 1341 1341 1341 1004 1004 1004 1004 Wald chi2 (7) 980.70*** 340.87*** 805.82*** 735.70*** 839.95*** 459.40*** 874.38*** 478.41*** Note: Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01. Columns (1) to (4) correspond to countries where total natural resource rents are ≤ 5% of GDP; columns (5) to (8) include countries where rents exceed 5% of GDP. All regressions use FGLS estimates and include the full set of control variables. Source: Authors’ estimations using ILOSTAT, UNCTAD, and World Bank data. 3.3 Robustness analysis As a robustness analysis, we test whether the previous results are robust to the use of another measure of export diversification, to the addition of other explanatory variables in the model and to the resolution of endogeneity issue. 11 3.3.1 Another measure of export diversification To ensure that our results remain robust to the use of other export diversification/concentration proxies, we use the Theil index, which expresses the number of products exported by a country and the intensity of export concentration on products. By construction, a low value of the index reflects a high degree of export diversification and vice versa. Formally, and disregarding individual and temporal indices, the indicator is calculated as follows: 1 N Export _ value  Export _ value  Theil = ∑ .ln   N j Average _ Export _ value  Average _ Export _ value  (3) With j representing the product index and N the total number of products. The results of the estimation of equation (1) using the Theil index as the dependent variable are presented in Table 4. Overall, they corroborate those obtained previously, namely that the total employment of women, and more specifically in agriculture, reduces export diversification. The employment of women in the industry and service sectors, on the other hand, tends to promote diversification. However, contrary to the results obtained with the HHI in Table 1, the effect of women's employment in the service sector is statistically significant with the Theil. Table 4. Female employment and Theil's diversification index Dependent variable: Theil index (1) (2) (3) (4) Total female employment 0.0045*** (0.0008) Female employment in Agriculture 0.0035*** (0.0007) Female employment in Industry -0.0044*** (0.0017) Female employment in Services -0.0067*** (0.0014) GDP per capita (log) -0.9811*** -1.0395*** -1.1136*** -1.1018*** (0.2301) (0.2315) (0.2319) (0.2235) GDP per capita^2 (log) 0.0383** 0.0444*** 0.0443*** 0.0497*** (0.0152) (0.0153) (0.0154) (0.0148) Trade 0.0016*** 0.0015*** 0.0017*** 0.0015*** (0.0003) (0.0004) (0.0004) (0.0003) FDI -0.0070** -0.0087*** -0.0069** -0.0067** (0.0032) (0.0033) (0.0034) (0.0032) Population density (log) -0.2273*** -0.2385*** -0.2559*** -0.2397*** (0.0111) (0.0110) (0.0117) (0.0104) Terms of trade -3.42e-15*** -3.24e-15*** -3.37e-15*** -3.18e-15*** (4.67e-16) (4.53e-16) (4.79e-16) (4.54e-16) Constant 9.3211*** 9.5293*** 10.3315*** 10.1156*** (0.8692) (0.8743) (0.8566) (0.8224) Observations 1819 1819 1819 1819 Wald chi2(7) 1895.78*** 1625.32*** 1653.63*** 1839.71*** Note: Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01. The dependent variable is Theil’s entropy index of export concentration. A higher Theil index indicates lower diversification. Positive coefficients thus reflect increased concentration, while negative coefficients imply enhanced diversification. All models are estimated using FGLS and include standard controls. Source: Authors’ calculations based on ILOSTAT, UNCTAD, and World Bank data. 3.3.2 Additional covariates It could be that the previously established relationship between female employment and export diversification is simply related to the effect of a third factor. In this context, the addition of control variables could limit this bias. 12 The synthesis of the literature on the determinants of export diversification allows us to retain several other control variables in addition to those used so far. For example, Cabral and Veiga (2010) study the determinants of export diversification in Sub-Saharan Africa and find that human capital and the quality of governance, especially corruption, are significant determinants of diversification in the subregion. Fonchamnyo and Akame (2017) also find that agricultural and manufacturing value added are important determinants of diversification in Sub-Saharan Africa. In their study on the impact of oil booms on export diversification, Djimeu and Omgba (2019) control for geographical factors and find that the country's distance from the equator is also a determinant of diversification in oil producing countries. Omgba (2014) in his study on the institutional underpinnings of export diversification patterns in oil producing countries, also controls for legal origin to account for the historical and institutional context of the countries. Avom et al (2020) focus mainly on political factors, particularly monetary, and find that the choice of exchange rate regime has an impact on the level of diversification in African countries. Drawing on this literature, we introduce the elicited variables into the model in turn to test whether the effect of women's employment remains unchanged. The results obtained are summarized in Table 5. They remain robust to the control of corruption, human capital (secondary school enrollment rate), agricultural and manufacturing value added, legal origin, exchange rate regime (fixed versus floating) and the country's distance from the equator. Table 5. Robustness check with extended controls: Female employment and export concentration Dependent variable: HHI (1) (2) (3) (4) Total female employment 0.0018*** (0.0002) Female employment in Agriculture 0.0011*** (0.0002) Female employment in Industry -0.0033*** (0.0004) Female employment in Services -0.0020*** (0.0003) Control of corruption -0.1087*** -0.1117*** -0.1114*** -0.1108*** (0.0061) (0.0062) (0.0062) (0.0062) Agriculture value added -0.0019*** -0.0008 -0.0001 -0.0012** (0.0005) (0.0005) (0.0005) (0.0005) Manufacture value added -0.0359*** -0.0398*** -0.0353*** -0.0405*** (0.0018) (0.0018) (0.0017) (0.0018) Exchange rate regime 0.0239*** 0.0107** 0.0170*** 0.0091* (0.0057) (0.0054) (0.0054) (0.0054) Polity2 -0.0039*** -0.0042*** -0.0052*** -0.0043*** (0.0005) (0.0005) (0.0005) (0.0005) Distance 0.0485*** 0.0443*** 0.0490*** 0.0451*** (0.0035) (0.0035) (0.0034) (0.0035) Legal origin -0.0511*** -0.0339*** -0.0416*** -0.0356*** (0.0047) (0.0044) (0.0044) (0.0044) Other controls Oui Oui Oui Oui Constant 0.7585*** 0.8409*** 0.5443** 1.1409*** (0.2502) (0.2441) (0.2526) (0.2403) Observations 1331 1331 1331 1331 Wald chi2(7) 1552.51*** 1704.03*** 2007.01*** 1642.87*** Note: Standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01. All models include additional controls such as corruption control, sectoral value added (agriculture and manufacturing), exchange rate regime, polity index, geographic distance, and legal origin. Other standard controls are also included. Estimates are based on FGLS. Source: Authors’ computations using ILOSTAT, UNCTAD, and World Bank data. 13 3.3.3 Endogeneity issues It could be that our variable of interest, female employment, is endogenous because it is simultaneously determined by other variables in the model, such as GDP per capita and FDI. Indeed, the increase in GDP per capita that generally reflects economic growth can lead to an increase in employment opportunities for women, as companies tend to hire more labor to meet the increased demand for goods and services. In addition, economic growth can stimulate specific sectors where women are often employed, such as services, retail, and processing industries. Furthermore, FDI can stimulate the creation of new businesses and the expansion of industries in host countries. This can increase women's employment opportunities, particularly in labor-intensive sectors such as manufacturing, services, and retail. To take this into account in our analysis, we use the generalized method of moments (GMM) system estimator developed by Blundell and Bond (1998). This estimator has the advantage of dealing with the endogeneity of all the explanatory variables by using their lagged values (in level and or in first difference) as instrumental variables. This estimator is even more appropriate for this study because it respects the dimensionality conditions that would require the total number of cross-sectional units, i.e., countries (N), to be larger than the study period (T). In this study, N = 125 and T = 23. In addition, this study uses the one-step System GMM estimator rather than the two-step GMM estimator because it is more efficient. The first lag level of all explanatory variables is reported as the "internal" instrument, as they are all likely to be correlated with the error term. The following two tests are performed for instrument validation: the Arellano and Bond (1991) test for first- and second-order serial correlations, and the Hansen (1982) test for overidentifying restrictions. The results are stacked in Table 6. They indicate that even after controlling for endogeneity problems related to the simultaneous error, we continue to find a robust positive effect of female employment in the agriculture sector on the HHI, and a negative effect of female employment in the industrial sector. Regarding the results of the post-estimation tests, the p-values given by the AR (2) test of serial correlations indicate that there is no evidence of a significant second- order serial correlation. Similarly, the null hypothesis of Hansen's (1982) test of overidentifying restrictions is not rejected. Table 6. System-GMM estimates: Dynamic effect of female employment on export concentration Dependent variable: HHI (1) (2) (3) (4) L.HHI 0.0007*** 0.0008*** 0.0005*** 0.0006*** (0.0002) (0.0001) (0.0001) (0.0001) Total female employment 0.0013** (0.0006) Female employment in Agriculture 0.0030*** (0.0005) Female employment in Industry -0.0043*** (0.0008) Female employment in Services -1.18e-8 (0.0007) Controls Yes Yes Yes Yes Constant 0.2080*** -0.2770*** 0.2695*** 0.0736 (0.0331) (0.1017) (0.0194) (0.0917) Observations 2156 2156 2156 2156 Instruments 47 47 47 47 Countries 110 110 110 110 AR1 Probability 0.002 0.004 0.001 0.005 AR2 Probability 0.152 0.138 0.154 0.135 14 Hansen Probability 0.207 0.203 0.306 0.258 Note: Robust standard errors in parentheses. p < 0.10, p < 0.05, p < 0.01. Estimates are based on the two- step System-GMM estimator using lagged levels and differences of the explanatory variables as instruments. All specifications include standard control variables. AR(1) and AR(2) refer to tests for first- and second-order autocorrelation in residuals. The Hansen test assesses the validity of instruments. Source: Authors’ estimations using ILOSTAT, UNCTAD, and World Bank data. IV. TESTING A CHANNEL Previous analyses have shown that women's employment, specifically in the primary agriculture sector, reduces export diversification, while employment in the secondary industrial sector promotes diversification. In this section, we discuss the mechanism by which women's employment does or does not contribute to export diversification in DCs. To determine why female employment in agriculture increases concentration while jobs in industry and services reduce it, we highlight the role of competence and skill as measured by scientific research. There are two reasons for choosing scientific research as a proxy for competence. Firstly, studies such as those of Aghion and Howitt (2006) argue that skills acquired in primary and secondary education are essential for existing technology, while those acquired in tertiary education contribute significantly to technological innovation. Secondly, the percentage of researchers in a specific discipline is a better indicator of skills acquired at the tertiary level than some traditional proxies, such as the tertiary enrollment rate. The latter measures the number of students enrolled in a cycle. At the same time, the percentage of researchers reflects the number of high-level scientists researching, experimenting, and advancing the discipline (UNESCO, 2021). The underlying hypothesis is that the employment rate in an activity sector affects the level of production or scientific research in that industry, which determines the level of export diversification. Indeed, the higher the employment opportunities in a sector of activity, the greater the incentive for individuals to undergo training and acquire experience in that field to be hired in the short to medium term. On the other hand, scientific research is an important factor in export diversification. As highlighted in endogenous growth theory and new trade theories (Krugman, 1995; Romer, 1990; Grossman and Helpman, 1991), human capital accumulation allows countries to change their specialization patterns from commodities to manufactured goods or services with more significant knowledge input. Specialized human capital will enable firms to adapt existing goods and technologies to the national environment, ultimately leading to the competitive production of more goods and more significant varieties (Agosin et al., 2012). To test this mechanism, we measure competence through scientific research approximated by the percentage of women researchers in Agricultural and Veterinary Sciences, Engineering Sciences and Technology, and Social Sciences. Figure 3 shows a positive correlation between women's employment in industry (and services) and scientific research in DCs. The higher the employment rate in these sectors, the more it contributes to increasing the percentage of scientific research by women. On the other hand, in the agriculture sector, the increase in the employment rate of women reduces the level of scientific research of the latter. This is consistent with certain prejudices that would make us believe that it is unnecessary to undergo an extensive formal education to get a job in the agriculture sector. On the other hand, the percentage of women researchers in the three disciplines mentioned above is negatively correlated with the HHI, suggesting that women's scientific research, in particular, contributes to the diversification of exports in developing countries. 15 Figure 3: Correlations between female employment, scientific research, and export concentration Note: The upper panel shows the correlation between female employment by sector and the share of female researchers. The lower panel relates female research participation (total and by field) to export concentration (HHI). Higher HHI values indicate lower export diversification. Scientific research is measured as the share of full-time researchers who are women. Source: Authors’ calculations based on data from ILOSTAT, UNESCO UIS, and UNCTAD. 16 Empirically, we introduce the different proxies of scientific research into the model (Table 7). We find that female researchers in agricultural and veterinary sciences significantly reduce export concentration and cancel out the positive effect of female agricultural employment on concentration (columns 1-2). Table 7. Cross-sectional evidence: The mediating role of scientific research in the link between female employment and export diversification Dependent variable: HHI (1) (2) (3) (4) (5) (6) Female employment in Agriculture 0.0024*** -0.0001 (0.0004) (0.0008) Female employment in Industry -0.0027*** -0.0065*** (0.0010) (0.0013) Female employment in Services -0.0035*** -0.0133*** (0.0010) (0.0017) GDP per capita (log) 0.0870 -0.1642 0.0612 -0.2792 0.1807 0.3814*** (0.0685) (0.2337) (0.1104) (0.2402) (0.1103) (0.1099) GDP per capita*2 (log) -0.0055 0.0080 -0.0065 0.0172 -0.0118 -0.0182** (0.0051) (0.0162) (0.0072) (0.0167) (0.0074) (0.0075) Trade -0.0001 0.0002 -0.0000 0.0006** -0.0002 0.0007*** (0.0001) (0.0002) (0.0003) (0.0003) (0.0002) (0.0003) FDI -0.0004 -0.0088*** 0.0003 -0.0149*** 0.0013 -0.0344*** (0.0026) (0.0028) (0.0033) (0.0035) (0.0024) (0.0046) Population density (log) -0.0448*** -0.0522*** -0.0415*** -0.0325*** -0.0514*** -0.0814*** (0.0048) (0.0101) (0.0062) (0.0076) (0.0058) (0.0090) Terms of trade -5.36e-16*** -5.85e-16*** -6.46e-16*** -7.44e-16*** -5.19e-16*** -3.21e-14*** (7.16e-17) (4.31e-17) (7.65e-17) (4.60e-17) (7.36e-17) (1.65e-15) Researchers in Agricultural and Veterinary Sciences -0.0011** (0.0006) Researchers in Engineering and Technology -0.0012*** (0.0003) Researchers in Social Sciences -0.0014** (0.0007) Observations 110 39 110 38 110 37 R-squared 1853.12*** 74.42*** 123.08*** 11126.72*** 256.91*** 4280.49*** Note: This table presents cross-sectional regressions based on the availability of country-level data on scientific research. Standard errors are reported in parentheses. p < 0.10, p < 0.05, p < 0.01. Each column progressively includes female employment in different sectors and the share of female researchers in corresponding fields. The dependent variable is the export concentration index (HHI). All models are estimated using FGLS. Source: Authors’ calculations using data from ILOSTAT, UNESCO UIS, and UNCTAD. This result can be explained by the fact that when they are formally educated to a high level and have sufficient skills in agricultural sciences, women can bring specific knowledge of traditional and local crops and skills in processing agricultural products. This can lead to more agricultural products being exported, which would reduce dependence on a few specific products, thereby reducing the concentration of exports. On the other hand, the percentage of women researchers in engineering, technology, and social sciences also reduces export concentration. It amplifies the reducing effect of women's employment in industry and services on export concentration (columns 3-4 and 5-6). Conclusion This paper provides contemporary empirical evidence of the role of women's employment in export diversification in developing countries. Starting from disaggregated sectoral data and various indices of diversification of the productive structure, we show that the participation of women in the labor market in terms of employment rate is a significant determinant of the observed level of export diversification. More specifically, women's employment in the agriculture sector further promotes concentration, whereas it tends to reduce it in the industrial 17 and service sectors. The effect on the concentration of exports nevertheless shows variations in amplitude depending on the level of diversification and countries' dependence on natural resources. The paper also discusses transmission mechanisms, focusing on the level of skills measured by the number of female researchers in the agricultural and veterinary sciences, engineering sciences and technology, and social sciences. Women's employment in the agriculture sector is incompatible with the pursuit of studies, particularly in training in agricultural and veterinary sciences, where they are victims of stereotypes according to which they do not need to undertake extensive study periods to work in agricultural and livestock activities. This may explain why women in many developing countries are confined to rudimentary production activities intended to meet primary needs and only marginally to processing. On the other hand, the skills necessary to operate in the industrial and service sectors require and encourage them to take training to increase their chances of being recruited in this highly competitive sector. As demonstrated in the case of men, the training of women in the technical and engineering professions constitutes a factor favoring the process of diversification. These results suggest that developing countries need to increase the share of women in engineering sciences and technology and in recruitment in the industrial and manufacturing sector. By encouraging women in technical and engineering training where they are mostly absent, a favorable targeted policy should give the necessary impetus to a virtuous circle that would allow women to also participate in the transformation of the goods they produce and thus to be actors in the diversification process, while integrating value chains. Such a policy would both increase the pool of qualified women for certain trades in the industrial sector and reduce discrimination against women. References Acemoglu, D., and Zilibotti, F. (1997). Was Prometheus unbound by chance? Risk, diversification, and growth. Journal of political economy, 105(4), 709-751.Agosin, M. R. (2008). Export diversification and growth in emerging economies. Cepal Review. Agénor, P. R., Ozdemir, K. K., & Pinto Moreira, E. (2021). Gender gaps in the labour market and economic growth. Economica, 88(350), 235-270. Aghion, P., and Howitt, P. (2006). Appropriate growth policy: A unifying framework. Journal of the european Economic Association, 4(2-3), 269-314. Agosin, M. R., Alvarez, R., and Bravo‐Ortega, C. (2012). Determinants of export diversification around the world: 1962–2000. The World Economy, 35(3), 295-315. Amin, M., Rezaei, S., and Tavana, F. S. (2015). Gender differences and consumer’s repurchase intention: the impact of trust propensity, usefulness and ease of use for implication of innovative online retail. International Journal of Innovation and Learning, 17(2), 217- 233. Anderson, S., and Baland, J.-M. (2002). The Economics of ROSCAs and Intrahousehold Resource Allocation. The Quarterly Journal of Economics, 117(3), 963-995. Arellano, M., and Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The review of economic studies, 58(2), 277-297. Avom, D., Kuété, F. Y. M., and Owoundi, F. (2020). Le régime de change contribue-t-il à la diversification des exportations en Afrique?. Revue d’economie du developpement, 28(2), 31-67. Beck, N., and Katz, J. N. (1995). What to do (and not to do) with time-series cross-section data. American political science review, 89(3), 634-647. 18 Ben hammouda, H., karingi, S.N., Njuguna, Angelica, E., Sadni-Jallab, M., 2006. Diversification: towards a new paradigm for Africa’s development. ATPC project paper Economic Commission for Africa. Ben Yahmed, S., and Bombarda, P. (2020). Gender, informal employment and trade liberalization in Mexico. The World Bank Economic Review, 34(2), 259-283. Blundell, R., and Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of econometrics, 87(1), 115-143. Boschma, R., and Gianluca, C. (2015). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, vol. 44(10), pages 1902-1914. Boserup, E., (1970). Women’s Role in Economic Development. New York: St. Martin’s. Busse, M., and Nunnenkamp, P. (2012). Gender Disparity in Education and the International Competition for Foreign Direct Investment. Feminist Economics, 18(2), 1-31. Cabral, M., and Veiga, P. (2010). Determinants of export diversification and sophistication in sub-saharan Africa. Universidade Nova de Lisboa, Nova School of Business and Economics. Cadot, O., Carrère, C., and Strauss-Kahn, V. (2011). Export diversification: what’s behind the hump?. Review of Economics and Statistics, 93(2), 590-605. Cadot, O., Carrère, C., and Strauss‐Kahn, V. (2013). Trade diversification, income, and growth: what do we know?. Journal of Economic Surveys, 27(4), 790-812. Djimeu, E. W., and Omgba, L. D. (2019). Oil windfalls and export diversification in oil- producing countries: Evidence from oil booms. Energy Economics, 78, 494-507. Do, Q. T., Levchenko, A. A., and Raddatz, C. (2016). Comparative advantage, international trade, and fertility. Journal of Development Economics, 119, 48-66. Duflo, E. (2012). Women empowerment and economic development. Journal of Economic literature, 50(4), 1051-1079. FAO. (2011). The State of Food and Agriculture 2010-2011: Women in Agriculture - Closing the Gender Gap for Development. Fonchamnyo, D. C., and Akame, A. R. (2017). Determinants of export diversification in Sub- Sahara African region: a fractionalized logit estimation model. Journal of Economics and Finance, 41(2), 330-342. Galor, O., and Weil, D. N. (1999). From Malthusian stagnation to modern growth. American Economic Review, 89(2), 150-154. Gelb, A. (2010). Diversification de l’économie des pays riches en ressources naturelles. Washington DC: Fond Monétaire International. Goldin, C. (1987). Women’s Employment and Technological Change: A Historical Perspective. Computer Chips and Paper Clips: Technology and Women’s Employment, 2(1987), 185-222. Goldin, C. (1994). The U-Shaped Female Labor Force Function in Economic Development and Economic History. NBER Working Paper, (w4707). Goldin, C. (1995). Career and Family: College Women Look to the Past. NBER Working Paper, (w5188). Grossman, G. M., and Helpman, E. (1991). Trade, knowledge spillovers, and growth. European economic review, 35(2-3), 517-526. Gylfason, T. (2006). Natural resources and economic growth: From dependence to diversification. Economic liberalization and integration policy: Options for Eastern Europe and Russia, 201-231. Gylfason, T., and Nganou, J. P. N. (2016). Diversification, dutch disease and economic growth: Options for Uganda. In Economic diversification policies in natural resource rich economies (pp. 118-147). Routledge. 19 Hansen, L. P., and Sargent, T. J. (1982). Instrumental variables procedures for estimating linear rational expectations models. Journal of Monetary Economics, 9(3), 263-296. Hausmann, R., Hwang, J., and Rodrik, D. (2010). What You Export Matters. Journal of Economic Growth, 12(1), 1-25. ILO (1990). The promotion of Self-Employment. International Labour Organization: Geneva. ILO (2020). Report on employment in Africa (Re-Afrique): Meeting the youth employment challenge. ILO Regional Office for Africa. International Labour Organization: Geneva Ilzetzki, E., Reinhart, C. M., and Rogoff, K. S. (2019). Exchange arrangements entering the twenty-first century: Which anchor will hold?. The Quarterly Journal of Economics, 134(2), 599-646. Imbs, J., and Wacziarg, R. (2003). Stages of diversification. American economic review, 93(1), 63-86. Isham, J., Woolcock, M., Pritchett, L., Busby, G. (2005). The Varieties of Resource Experience: Natural Resource Export Structures and the Political Economy of Economic Growth, The World Bank Economic Review, Volume 19, Issue 2, 2005, Pages 141–174. Kabeer, N. (2012). Women’s Economic Empowerment and Inclusive Growth: Labour Markets and Enterprise Development. SIG Working Paper. Karanfil, F., and Omgba, L. D. (2023). The energy transition and export diversification in oil- dependent countries: The role of structural factors. Ecological Economics, 204, 107681. Kasente, D., Lockwood, M., Vivian, J., and Whitehead, A. (2002). Gender and the expansion of non-traditional agricultural exports in Uganda. Shifting burdens, gender and agrarian change under neoliberalism, 35-65. Kazandjian, R., Kolovich, L., Kochhar, K., and Newiak, M. (2019). Gender equality and economic diversification. Social Sciences, 8(4), 118. Kilolo, J. M. M., Bah, M. M., & Mastaki, J. L. N. (2023). Does women empowerment foster export diversification? Evidence from a sample of developing countries. The Journal of International Trade & Economic Development, 1-34. Klasen, S. (2000). Does gender inequality reduce growth and development? Evidence from cross-country regressions. Klasen, S., and Lamanna, F. (2009). The impact of gender inequality in education and employment on economic growth: new evidence for a panel of countries. Feminist economics, 15(3), 91-132. Kleven, H., & Landais, C. (2017). Gender inequality and economic development: fertility, education and norms. Economica, 84(334), 180-209. Klinger, B., and Lederman, D. (2006). Diversification, innovation, and imitation inside the global technological frontier. World Bank policy research working paper, (3872). Krugman, P. (1995). Increasing returns, imperfect competition and the positive theory of international trade. Handbook of international economics, 3, 1243-1277. Lederman, D., Pienknagura, S., and Rojas, D. (2021). Latent trade diversification and its relevance for macroeconomic stability. The World Bank Economic Review, 35(1), 58- 91. Newey, W. K., and West, K. D. (1987). Hypothesis testing with efficient method of moments estimation. International Economic Review, 777-787. Olive, D. J. (2017). WLS and Generalized Least Squares. In Linear Regression (pp. 163-173). Springer, Cham. Omgba, L. D. (2014). Institutional foundations of export diversification patterns in oil- producing countries. Journal of Comparative Economics, 42(4), 1052-1064. Parks, R. W. (1967). Efficient estimation of a system of regression equations when disturbances are both serially and contemporaneously correlated. Journal of the American Statistical Association, 62(318), 500-509. 20 Parteka, A., and Tamberi, M. (2008). Determinants of export diversification: An empirical investigation. Universita Politecnica delle Marche, Dipartimento di Economia Quaderno di Ricerca, (327). Prebisch, R., (1950). The Economic Development of Latin America and its Principal Problems, reprinted in Economic Bulletin for Latin America, Vol. 7, No. 1, 1962, 1- 22. Psacharopoulos, G., and Tzannatos, Z. (1989). Female labor force participation: An international perspective. The World Bank Research Observer, 4(2), 187-201. Romer, P. M. (1990). Capital, labor, and productivity. Brookings papers on economic activity. Microeconomics, 1990, 337-367. Ross, A. (2008). The new geography of work: Power to the precarious?. Theory, culture and society, 25(7-8), 31-49. Ross, M. L. (2018). The politics of the resource curse. The Oxford handbook of the politics of development, 200. Sachs, J. D., and Warner, A. M. (2001). The curse of natural resources. European economic review, 45(4-6), 827-838. Singer, H. W. (1950). The distribution of gains between investing and borrowing countries. In Milestones and Turning Points in Development Thinking (pp. 265-277). London: Palgrave Macmillan UK. Tran, T. A. D., Phi, M. H., and Diaw, D. (2017). Export diversification and real exchange rate in emerging Latin America and Asia: A South–North vs. South-South decomposition. The Journal of International Trade and Economic Development, 26(6), 649-676. UN Women. (2015). Progress of the World's Women 2015-2016: Transforming Economies, Realizing Rights. UNESCO (2021) UNESCO Science Report: the Race Against Time for Smarter Development. S. Schneegans, T. Straza and J. Lewis (eds). UNESCO Publishing: Paris. Walters, C., Mehl, G., Piraino, P., Jansen, J. D., Kriger, S. (2022). The impact of the pandemic- enforced lockdown on the scholarly productivity of women academics in South Africa. Research Policy, Elsevier, vol. 51(1). White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica: journal of the Econometric Society, 817-838. Wood, A., and Berge, K. (1997). Exporting manufactures: human resources, natural resources, and trade policy. The Journal of Development Studies, 34(1), 35-59. World Bank and WTO (World Trade Organization). 2020. Women and Trade: the Role of Trade in Promoting Gender Equality. Washington, DC: World Bank; Geneva: WTO. Appendix TableA1. Descriptive statistics Correlation matrix Variable Obs Mean Std. Dev. HHI 2941 0.378 0.220 1 Empl. 3000 46.466 18.92 0.3099* 1 GDP pp 2847 7.657 0.972 -0.1621* -0.7665* 1 GDP pp2 2847 59.586 14.669 -0.1602* -0.7681* 0.9978* 1 Trade 2673 76.538 37.154 0.0159 -0.2741* 0.2574* 0.2502* 1 FDI 2829 4.045 7.051 0.0198 0.0021 0.0516* 0.0524* 0.2758* 1 Dens. Pop. 2984 3.937 1.279 -0.2489* -0.0367* -0.0152 -0.0168 -0.0696* -0.0396* 1 Legal Origin 2736 1.912 0.695 0.0900* 0.0880* -0.0341 -0.0237 -0.2509* -0.0809* -0.0993* 1 Note: statistics are calculated on a sample of 125 developing countries (upper-middle income, lower-middle income and low income) covering the period 1995-2018. Table A2. Definitions and sources of variables 21 Variables Definition Sources Herfindahl-Hirshman export concentration index, which expresses the HHI sum of the squares of the shares of all commodities in total export UNCTAD earnings. Negative entropy export diversification index, which expresses the Theil number of products exported by a country and the intensity of the IMF concentration of exports on products. Total women Proportion of a country's female labor force in employment. ILOSTAT employment Women employment Employment rate of women in agriculture, hunting, forestry and fishing, ILOSTAT in agriculture measured as a percentage of total female employment. Employment rate of women in mining and quarrying, manufacturing, Women employment construction and utilities (electricity, gas and water), measured as a ILOSTAT in industry percentage of total employment. Employment rate of women in wholesale and retail trade, restaurants and Women employment hotels, transport, storage and communications, finance, insurance, real ILOSTAT in services estate and business services, as well as community, social and personal services, measured as a percentage of total employment. Logarithm of gross domestic product per capita, converted into GDP per capita international dollars using purchasing power parity rates. Data are in WDI constant 2017 international dollars. Sum of exports and imports of goods and services, measured as a Trade WDI percentage of gross domestic product. The sum of equity capital, reinvested earnings, other long-term capital FDI and short-term capital as reported in the balance of payments. Data as a WDI percentage of GDP. Logarithm of mid-year population divided by land area in square Population density WDI kilometers. Dummy variable taking the value of 1 if a country has a legal system La Porta et Legal origin derived from that of France, and 0 if not. al. (1999) Perceptions of the extent to which public power is exercised for private Control of corruption ends, including small and large forms of corruption, as well as the WGI 'capture' of the state by elites and private interests. Net output of the agriculture, forestry and fishing sector corresponds to Agriculture value ISIC divisions 1 to 3, after adding up all output and subtracting WDI added intermediate inputs. Dummy which measures the type of exchange rate regime in force, coded Ilzetzki et Exchange rate regim as 1 for fixed and 0 for floating regimes. al. (2019) Democratization index which captures the spectrum of authority of the Polity2 political regime on a scale of -10 (hereditary monarchy) to 10 CSP (consolidated democracy). Geographical Distance of the capital from the equator, measured as abs (Latitude)/90. Rodrik et al. distance 2004 Researchers in Percentage of total full-time female researchers in agricultural sciences. agricultural and UNESCO veterinary sciences Researchers in Percentage of total full-time female researchers in engineering sciences engineering and and technology. UNESCO technology Researchers in Social Percentage of full-time female researchers in the social sciences. UNESCO Sciences Source: authors. Table A3. List of countries Afghanistan, Albania, Algeria, Angola, Argentina, Armenia, Azerbaijan, Bangladesh, Belarus, Belize, Benin, Bhutan, Bolivia, Bosnia and Herzegovina, Botswana, Brazil, Bulgaria, Burkina Faso, Burundi, Cabo Verde, Cambodia, Cameroon, Central African Republic, Chad, China, Colombia, Comoros, Democratic Republic of Congo, Republic of Congo, Costa Rica, Côte d'Ivoire, Cuba, Djibouti, Dominican Republic, Ecuador, Arab 22 Republic of Egypt, El Salvador, Equatorial Guinea, Eritrea, Eswatini, Ethiopia, Fiji, Gabon, The Gambia, Georgia, Ghana, Guatemala, Guinea, Guinea-Bissau, Guyana, Haiti, Honduras, India, Indonesia, Islamic Republic of Iran, Iraq, Jamaica, Jordan, Kazakhstan, Kenya, Democratic People’s Republic of Korea, Kyrgyz Republic, Lao People's Democratic Republic, Lebanon, Lesotho, Liberia, Libya, Madagascar, Malawi, Malaysia, Maldives, Mali, Mauritania, Mauritius, Mexico, Moldova, Mongolia, Montenegro, Morocco, Mozambique, Myanmar, Namibia, Nepal, Nicaragua, Niger, Nigeria, North Macedonia, Pakistan, Panama, Papua New Guinea, Paraguay, Peru, Philippines, Romania, Russian Federation, Rwanda, Samoa, São Tomé and Príncipe, Senegal, Serbia, Sierra Leone, Solomon Islands, Somalia, South Africa, South Sudan, Sri Lanka, Suriname, Syrian Arab Republic, Tajikistan, Tanzania, Thailand, Timor-Leste, Togo, Tonga, Tunisia, Türkiye, Turkmenistan, Uganda, Ukraine, Uzbekistan, Vanuatu, Viet Nam, Republic of Yemen, Zambia, and Zimbabwe. Source: authors. 23