Policy Research Working Paper 10425 Liberian Women Count Evidence from a Macrosimulation of the Gender Dividend Mahesh Karra Mitja Del Bono Joshua Wilde Wendy Cunningham Sarika Gupta Social Protection and Jobs Global Practice April 2023 Policy Research Working Paper 10425 Abstract Liberian women make significant economic contributions participation, intra-sectoral wages, and sector of employ- yet are constrained from contributing even more due to ment were closed, gross domestic product would be 11.5 their exclusion from productive opportunities. This study percent higher. If further reforms were undertaken to develops a macrosimulation model of the Gender Dividend equalize education and reduce fertility rates to a net-repro- that estimates the economic contributions of women and duction rate, gross domestic product would be 23.7 percent the societal costs incurred by excluding them. Using macro- higher. Finally, if the model also accounts for the value of economic, demographic, and survey data from Liberia, the non-tradable production, gross domestic product would analysis finds that women were responsible for 39 percent of be US$5.89 billion, or 45.3 percent higher than today’s market-based output produced annually in 2020, equal to estimates, with women being responsible for 53 percent US$1.08 billion, and contributed another US$530 million of the labor market output. These estimates reinforce the in non-tradable sources of production, namely, housework need for a unified policy agenda that actively invests in and domestic chores. Using the macrosimulation model, women’s human capital and work-related opportunities the study estimates that if the gender gaps in labor force simultaneously. This paper is a product of the Social Protection and Jobs Global Practice. 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 wcunningham@worldbank.org, sgupta@worldbank.org, and mdelbono@worldbank.org. 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 Liberian Women Count: Evidence from a Macrosimulation of the Gender Dividend Mahesh Karra,1 Mitja Del Bono,2 Joshua Wilde,3 Wendy Cunningham,2 Sarika Gupta2 1: Frederick S. Pardee School of Global Studies, Boston, MA, USA 2: World Bank, Washington, DC, USA 3: Max Planck Institute of Demographic Research, Rostock, Germany Key words: gender, gender dividend, women’s empowerment, inequality, demography, human capital, employment, time allocation, economic growth, macrosimulation, Liberia JEL: J11, J16, J22, J24, E24, O40 1 Introduction Women are the bedrock of the economy and of society in Liberia. As of 2016, women comprise 50 percent of the Liberian labor force (Johanssen da Silva, 2021), supplying 76 percent of labor for cash crop production and 93 percent of the labor for food crop production (GoL 2009). About 80 percent of women own small enterprises, compared to 57 percent of men (Johanssen da Silva, 2021). In addition, women in Liberia are responsible for overseeing household activities, particularly in rural communities where they engage in household food security and are responsible for caregiving of children and other household members. Despite their contributions to economic production, women are systematically excluded from fully contributing since they are clustered in lower productivity and lower earning sectors of the labor market. Approximately 86 percent of women are employed in petty trade or in the agriculture sector, which will remain a low productivity sector until broader market reforms are implemented (Johanssen da Silva, 2021). About 94 percent of women are informally employed compared to 79 percent of men (Liberia Institute for Statistics and Geo-Information Services, 2015). Among the non-farm self-employed, 55 percent of enterprises are owned by women, though their enterprises are smaller and less well established than those of men, and less than 5 percent of women hold formal sector jobs (Johanssen da Silva, 2021). In Liberia, women earn about 14 percent less than male counterparts with similar skills and engaged in similar work. Liberian women’s limited economic contributions are due to a range of disadvantages. Women lag behind men in human capital accumulation, which is marked by significant disparities in educational attainment despite recent progress in increasing girls’ schooling. The median years of schooling is 3.4 years for women compared to 6.5 years for men. The threat of violence likewise impedes the movement of women for economic, educational and civic activities (Gaussman et al, forthcoming). Fertility in Liberia is relatively high at 4.2 births per woman of reproductive age, and 30 percent of Liberian women have their first child before the age of 18 (Cunningham, et al 2022). Women receive less financial credit and own fewer physical assets and land. The challenges that women face to resource ownership, schooling, and better health are compounded by a weak infrastructure, restrictive social norms, and poor institutions. Despite electing the first female Head of State in Africa, women in Liberia remain significantly restricted in terms of their relative capacity for decision-making and sociopolitical autonomy; for example, only 16 percent of women who are married or in a union make their own choices over their contraceptive use and health care (Liberia Institute of Statistics and Geo-Information Services (LISGIS), 2014), and women only hold 12 percent of national parliamentary seats in Liberia (World Bank, 2022). Women’s severe disadvantage in Liberia begs the question as to where Liberia would be in terms of development if women were to have similar political, social, and economic opportunities as men. The realization of economic potential through increased investments in women and girls, also known as the “gender dividend,” emphasizes that societies could be more productive and equitable if gender gaps, particularly in the labor market, were closed. Global estimates of the gender dividend have indicated that hundreds of trillions of dollars have been lost as a result of persisting social, political, and economic gender gaps (Wodon et al., 2020). Indeed, the Government of Liberia, civil society, and Liberian citizens recognize the contributions that women have made, and can make, to society, but the decision-makers are yet to allocate public resources and efforts to substantially address the gender gaps that are holding Liberia back. Its Vision 2030 document identifies “a society of gender equity… and opportunities for all.” The development plan for 2018-2023 also recognized the need to address gender gaps as a means to reduce poverty and create 2 prosperity (Government of the Republic of Liberia 2018), but progress has been slower than hoped. A better appreciation of the losses incurred by inaction may help to spur greater investments. This study intends to quantify the economic costs of excluding Liberian women from productive opportunities by estimating the benefits from closing key gender gaps. We posit that Liberian women’s current contributions are under-valued in the policy-making process, and that social, political, and institutional discrimination against women creates an economic cost not only to women individually, but also to society at large. To quantify the current economic contributions of women in Liberia, we adapt the macrosimulation model developed by Canning, Karra, and Wilde (2017), hereafter referred to as CKW 2017, to construct a model of the gender dividend in which we 1) estimate the economic returns to closing gender gaps, particularly in the labor market; and 2) incorporate the value of women’s unpaid and domestic work into our estimate of production.1 We identify the potential economic contributions that women would make if there were equality of opportunity and capability by gender (i.e., a convergence or closure of gender gaps) across a range of factors that have been identified as key determinants of the gender dividend and inputs to economic growth and productivity. These factors, which include educational attainment and human capital, labor force participation, demographic factors such as fertility, wages and returns to labor, among others, are included in the model as maneuverable policy levers. We find that Liberia’s gender dividend is large: women make significant contributions to Liberia’s economy and could contribute even more if gender gaps were narrowed or closed. Currently, women are responsible for 39 percent of market-based output produced annually in Liberia, equivalent to USD 1.08 billion. If the gender gaps in labor force participation, intra-sectoral wages and sectoral representation were closed, GDP would be 11.5 percent higher. If further reforms were to equalize education and reduce fertility rates to a net-reproduction rate, GDP would be 23.7 percent higher. Finally, if we also account for the value of non-tradable production, GDP would be 5.89 billion or 45.3 percent higher than today’s estimates, with women being responsible for 53 percent of the labor-market output. The rest of the paper is organized as follows. Section 2 briefly presents our model of the gender dividend, and Section 3 describes the key modifications to the CKW 2017 macrosimulation framework, including: 1) the addition of the key factors for closing gender gaps; and 2) the inclusion of two kinds of non-tradable production. Section 4 presents the data sources used to calibrate the model to the Liberian context. Section 5 presents and discusses the key findings from the macrosimulations, and Section 6 concludes. 2 Conceptual Framework and Methodology 2.1 Models of the Gender Gap: A Literature Review Over the last 30 years, advances in endogenous growth theory have increasingly considered “human” dimensions of development, with a number of studies looking beyond the exogenously determined factors of production and growth (Roemer, 1996; Lucas, 1988; Barro & Sala-i-Martin, 1995). This discourse pushed for a deeper exploration of first-order determinants of economic output and their estimation in empirical studies. As part of this process, gender was identified as a fundamental driver of growth (Löfström, 2006; Boserup, Tan, & Toulmin, 2007; Natali & Kabeer, 2013), with a number of studies 1 This differs from Wodon et al (2020) in three principle ways. First, this paper uses a macrosimulation model that explicitly models the interaction of gender inequalities in different domains of girls’ and women’s lives while Wodon et al (2020) use a reduced form model. Second, Wodon et al (2020) estimate global impacts while this paper focuses on Liberian women. Third, this model focuses on the aggregate economic consequences of gender gaps while Wodon et al (2020) consider economic, but also human capital implications, of the gender dividend. 3 highlighting women’s direct and indirect contribution to long-run economic growth (Galor & Weil, 1996; Klasen, 1999). A range of modeling approaches have been used to quantify the economic loss caused by structural and systemic differences between men and women. Following the seminal work of Galor and Weil (1996), an overlapping generations framework by Agénor (2012) estimated the monetary valuation of a decline in the cost of childrearing (particularly sons), a decline in gender biases in the workplace, an increase in women’s bargaining power, and increased mothers’ time allocated to daughters. This model finds that policy regimes that increase public spending may contribute to increase in women’s productivity, women’s time allocation to work, and economic growth, particularly if investments were made to improve infrastructure (such as roads and access to water and sanitation). A simulation approach by Cavalcanti and Tavares (2007) examined differences in output and income by varying the degree of gender-based wage discrimination, taking as endogenous savings, fertility, and labor force participation rates. The authors found that a 50 percent increase in the gender wage gap led to a 35 percent decrease in income per capita, noting that up to 60 percent of the gap in output per capita between countries could be attributed to continued gender inequality. In Niger, a Computable General Equilibrium (CGE) model based on a LINKAGE model2 was applied to estimate the economic opportunity from gender equality and to isolate the monetary impact of gender differences.3 The study concludes that reduced gender gaps have the ability to increase GDP in 2030 by 12.6 percent under scenarios of reduced fertility and close to universal education, and by 8.9 percent if increased labor force participation and productivity are modeled (World Bank, 2019). Pennings (2022) constructs the Employment Gap Index (GEGI) that translates the inequalities in employment between men and women into the potential monetary gains (in GPD-per- capita). On average across the countries studied,4 the GEGIs estimate that long-run GDP per capita would be 19 percent larger if gender gaps could be closed by increasing female employment to that of men. With an application of a Lucas span-of-control model of occupational choice with heterogeneous entrepreneurial endowments, Cuberes and Teignier (2016) examine the impact of inequality between men and women in entrepreneurship and labor force participation on productivity and per-capita income, noting average income losses of 15 percent due to gender gaps. Devadas and Kim (2020) use the World Bank Long-Term Growth Model (LTGM), which is built on a Solow-Swan growth model, to estimate the effect of increasing labor force participation and education for women on GDP growth. For the indirect impact of closing gender inequality, the study uses an extension of the LTGM-TFP model in which the total factor productivity (TFP) growth rate is a function of a TFP determinant index and initial TFP level. The authors find that the GDP-per-capita loss from gender gaps amounts to 7.9 percent in low-income countries, 21.3 percent in lower-middle income countries, 12.3 percent in upper-middle income countries and 6.4 percent in high-income countries. In a similar exercise, Wodon et al. (2020) employ a comprehensive and domain-specific framework to assess the economic opportunity associated with closing the gender gap by estimating the impact of inequalities using regression analyses or statistical differences. The authors highlight the importance of closing these gaps by separately estimating the impacts and economic costs of gender inequality across five domains: 2 See van der Mensbrugghe (2011) for a complete description. 3 In the case of Niger, costs calculated are related to earnings, population growth, under-five mortality and standing, and savings for education. 4 The study has a full variant that differs from the basic variant. Rather than using aggregate employment gaps, it uses a weighted average of employment gaps in “better employment” and “other” types of employment. The study uses 185 countries for the basic GEGI and 159 for the full GEGI. 4 (1) earnings; (2) educational attainment, child marriage and early childbearing; (3) fertility and population growth; (4) health, nutrition, well-being, and violence; and (5) agency, decision-making, and social capital. 2.2 Simulation Modeling: A History Attempts to assess the effect of structural factors of economic development can be divided into three categories: aggregate (macroeconomic) statistical analyses, microeconomic studies, and simulation exercises. In this paper, we favor a simulation approach. We outline the rationale for this decision below; however, given that the literature in each of these areas is vast, our summary is, by necessity, selective. Aggregate macroeconomic regression models have been used to analyze the effect of demographic or social changes on macroeconomic outcomes. Since the early 1990s, many analyses have followed the “growth regression” model popularized by Barro (1991) and Mankiw, Romer, and Weil (1992). Typically, terms representing population growth, labor force growth, or dependency ratios are included as right-hand side variables, while GDP per capita growth rates are the dependent variables of interest (see Kelley and Schmidt, 2005; Bloom and Canning, 2008). The aggregate macro-economic regression models suffer from three major drawbacks. First, they fail to elucidate the complex relationship and heterogeneities behind observed outcomes and social and demographic factors. Second, few studies that take an aggregate approach to estimate the effects of population growth on economic outcomes have adequately dealt with the issue of identification. For example, changes in fertility rates – a channel through which women may affect economic outcomes – are not only directly affected by economic outcomes, but also are affected by other variables that, themselves, may have direct effects on the economy, such as education, health, institutional characteristics, and cultural factors. Given likely problems of omitted variables and reverse causation, the studies are unable to draw inferences from the conditional correlations in the growth regressions.5 Finally, it is difficult in a macro- regression to appropriately account for the time frame in which the changes are occur, due to auto- correlation among both dependent and independent variables. A second approach to examining the relationship between social or demographic factors and economic outcomes has been to turn to a finer level of analysis, through micro-economic studies. The unit of observation is at the household- or individual-level rather than at country or regional levels. The methodology often allows for identification to be achieved in a way in which it cannot using macro data (see Joshi and Schultz, 2007; Schultz, 2009; Miller, 2010; and Rosenzweig and Zhang, 2009). These studies also fall short for three reasons. First, many of the effects run through channels that are external to the household—either via externalities in the classic economic sense (e.g., environmental degradation) or through changes in market prices, such as wages, land rents, and returns to capital (Acemoglu, 2010). Second, even if one ignores external effects, aggregating the channels by which demographic and social changes affect economic outcomes is challenging. Finally, as in the macroeconomic literature, the long- time horizon over which the effects of demographic and social changes are manifested limits the ability of a single study to capture them. A third alternative is a single simulation model. If the researcher knows the structural mechanisms that relate economic and social or demographic variables and has information to correctly and causally parameterize each mechanism, a single simulation model can be constructed that can effectively deal with 5 While a handful of studies have attempted to circumvent the identification problem in the macroeconomic context using instrumental variables (see Acemoglu and Johnson (2007), Li and Zhang (2007), and Bloom et al. (2009) for well-known examples), instruments at the macro level tend to be fall short due to the complex nature of aggregate relationships. 5 the issues of aggregation and general equilibrium.6 However, there are several drawbacks to simulations. While simulations are able to incorporate many theoretical channels into their analysis, adding too many channels could create a situation where the model becomes too complicated to uncover the underlying assumptions (Ahlburg 2002). Similarly, Kelley (1988) cited the difficulty of constructing credible long- range demographic forecasts, potential changes in policy or institutions that may occur over the forecast interval, and the lack of available data to specify and validate such a model as potential limitations. Given the relative opacity of these models, the popularity of simulation approaches waned in academic and applied policy circles after the mid-1980s. However, with renewed interest in human capital and overlapping-generations approaches, and the increased aversion to regression modeling with poor identification strategies, simulation modeling has made a comeback. For example, the national transfer accounts (NTA) model of Lee and Mason (2010), the macroeconomic-health model of Ashraf et al (2009), and, most recently, the CKW model of the demographic dividend (Ashraf et al., 2013; World Bank, 2015; Karra et al., 2017) have reinvigorated the field. Work by macroeconomists interested in long-run growth has extended this approach to create fully “micro-founded” (i.e., micro-foundation-based) computable general equilibrium models to analyze the interaction of social and demographic and economic outcomes (e.g., Doepke, 2004). 2.3 The Gender Dividend Model: Extending CKW 2017 Our model of the gender dividend that attempts to estimate the current and foregone economic contributions of Liberian women is based on the CKW 2017 demographic-economic macrosimulation framework. The CKW model was specifically developed to estimate the size of the “demographic dividend”, or the increase in income per capita as a result of fertility decline, in low - and middle-income countries. Predecessors of this model were used to evaluate the macroeconomic effects of disease (Ashraf et al., 2009). In principle, the CKW modeling framework can be used to evaluate the dynamic macroeconomic effects of any structural economic change which affects a nation’s demographics, productivity, or productive factors such as land, labor, or capital. As such, it is an ideal framework for evaluating the macroeconomic effects of increased gender equity, since these changes would impact many productive channels beyond fertility decline (the primary channel of interest in the CKW 2017 model). As depicted in Figure 1, the CKW model considers a two-sector production schedule that is driven by demographics, human-capital, physical capital, and labor supply inputs. The coefficients that we use to parameterize the model are drawn from the published literature and summarized in Appendix 3. We account for aspects of gender equality that have proven important for economic growth with persuasive evidence. Gender-relevant mechanisms that are not accounted for in the adopted framework – due to modeling complexity and insufficient coefficients to parametrize them therein – but are important precursors to economic output are the systemic access barriers to formal financial services such as saving and credit products (Demirguc-Kunt, Klapper, & Singer, 2013), the role of gender-based violence (Day, McKenna, & Bowlus, 2005) and cultural norms regarding gender roles (Hiller, 2014). Figure 1. Generalized Modeling Framework (CKW 2017) 6 The intellectual ancestor of modern economic–demographic simulation models is the Coale –Hoover model (Coale and Hoover, 1958), which studies the effect of fertility change in India and has been followed by many others since (see Enke, 1971; Simon, 1976; Ahlburg, 2002; Kelley, 1988; Lee and Mason, 2010; Ashraf et al., 2013; and Karra et al., 2017). 6 Fertility Population by Age FLFPR Male Labor Supply + Female Labor Supply Education Height (Health) Labor Supply Savings Effective Labor Physical Capital Land Traditional Sector Production + Modern Sector Production The Non-Tradable Raw Materials Production 2.4 Demographic Module and Effective Labor The CKW 2017 model posits a demographic foundation to the gender dividend model, namely, it begins by projecting the size and gender composition of the labor force, by age- and sex-specific population at each five-year period. The model uses fertility rates by five-year age groups in each five-year period. These projections are further used to calibrate the age-and sex-specific mortality rates and the sex ratio at birth.7 The labor supply8 at each five-year time period is then calculated by assuming that the population enters the labor force at the age of 20 and exits the labor force at the age of 65. 9 The age- and sex- specific workforce is calculated by weighting the age-group population with the corresponding age- and sex- specific labor force participation rates in the respective five-year period at baseline.10 We assume that age- specific male participation rates are fixed at this level over time, but we modify the age-specific female labor force participation rate in each period to reflect the impact of fertility change and women’s substitution between childcare and work on total female labor supply. In this manner, the age-specific female labor force participation rates are endogenized with a factor that reflects the impact of fertility on the labor supply of women, as has been calculated by Ashraf, Weil and Wilde (2013). Fertility and demographic change influence the accumulation of human capital, captured by the educational attainment and health of the age-specific cohort. These are respectively calculated by the 7 A low variant, a medium variant and a high variant are available from the World Population Prospects of the UN (2010). Since the comparison of the projected output is undertaken at the next available period from the baseline, the difference between the variants becomes negligible. Here, the low variant is adopted. 8 The original CKW model was a fully supply-side model, meaning there is an implicit assumption of full employment. In the context of many LMICs, this full employment assumption likely does not hold in the short run. However, there are a number of theoretical and practical reasons to not independently model labor demand. Most importantly among these, modeling labor demand in a traditional manner (by deriving labor demand curves from the production function) is inappropriate, as this also assumes full utilization and costless movement of factors -- the same underlying rationale behind a full employment assumption. As a result, any modeling of labor demand outside this traditional method would be necessarily ad hoc, and likely introduce even more untenable assumptions into the analysis, and add to the complexity of the model without guaranteeing more accurate results. In addition, the model already considers several labor market inefficiencies by introducing employment sectors with different wage regimes (average product wages in the traditional sector versus marginal product wages in the modern sector), further blunting the necessity of independently modeling labor demand. 9 While the lower-bound for the labor force is commonly 16 or 18, we use age 20 to align with the five-year cohorts that we are using for population projections. 10 Labor force participation rates are obtained from the International Labour Office’s ILOSTAT database (International Labour Office (ILO) 2013). 7 weighted sum of the average years of schooling and adult height for each age- and sex-specific cohort and in each period separately. Health is proxied by the height of the adult, which is correlated with the productivity in adulthood (Schultz, 2002). The age and sex-specific estimates of educational attainment and health are then aggregated into a weighted average that reflects the human capital stock accumulated for the entire workforce at each period. In following the CKW methodology, fertility-to-education and fertility-to-health elasticities are then incorporated into the analysis, where human capital accumulation investment, as the number of children decreases, are modeled. Similar to the CKW approach, the evolution of fertility is characterized by a feedback channel with the education of each female-specific cohort, whereby the projected fertility rates decrease as the years of education attained increase. The modeling of the education-to-fertility channel is appropriate in Liberia as fertility decline has shown to be correlated with the educational attainment of women since the 1980s (Parr, 1995). To this end, the education-to-fertility and fertility-to-education channels are critical in the context given the large bandwidth for improvement in female education for Liberia. 2.5 Capital Accumulation and Savings We continue to follow the CKW approach by applying a Solow framework for capital accumulation and assuming that the evolution of capital stock is dependent on the aggregate output weighted by the savings rate and a standard depreciation parameter. The estimation strategy follows the linear specification of Bloom, Canning, Mansfield et al. (2007) by modeling the periodic savings rate as a function of preceding savings, income and the old-age dependency ratio. Accordingly, savings behavior hinges on the demographics as we assume that it is highest for prime-age workers and declines with age. The income component is likewise crucial considering that savings increases as income grows considerably in a developing country setting. The model further assumes that investment is limited to domestic savings, following the original CKW methodology. International capital flows are not accounted for in this model, and we recognize that their inclusion could produce a more accurate valuation and change the size of the demographic and gender dividend. However, from an empirical perspective, the value of adding these flows to the model is unclear. Wilde and Karra (2022) evaluate different capital accumulation scenarios within the CKW model framework – including one which includes international capital flows – and find that the historical co-evolution of demographics, income, investment, savings, and capital accumulation more closely follows a pattern similar to a constant domestic savings rate rather than one that incorporates international capital flows. This finding is consistent with micro evidence put forth by Deaton (1992). In addition, since capital markets are not segregated by gender, gender inequalities in capital accumulation are not incorporated into the model, and it is therefore unlikely that a different modeling of investment dynamics would change the central conclusions of this exercise. Finally, on a more practical level, the unavailability of the sex- and age-specific capital accumulation rates needed for this modeling strategy limits its estimation. 2.6 Aggregate Production The production module of the model follows the Lewis (1954) model of economic development, which is comprised of three key inputs: estimated effective labor, capital stock, and land, which is a proxy for the natural resource stock. The supply of labor is parametrically allocated into sectors of production by gender, as later described. We assume that the economy is comprised of a modern sector, a traditional sector and 8 a raw material sector.11 The traditional sector is comprised of the informal sector including agriculture while the remaining workforce is allocated to the modern sector. The aggregate production in the modern sector is parametrized by a standard Cobb-Douglas production function which uses physical capital, allocated labor (demographics), and human capital in the form of average years of schooling in the workforce (as a proxy for education) and average height of the workforce (as a proxy for health). The aggregate production in the traditional sector is likewise composed by a Cobb-Douglas production function, with agricultural land12 and allocated labor as inputs.13 The assumed irrelevance of capital in the traditional sector production is aligned with evidence on the low marginal capital intensity of the agricultural and informal production processes in Sub-Saharan African countries (Schmidhuber, Bruinsma, & Boedeker, 2009). By the same token, the use of unskilled intensive production and the almost non- existent availability of estimates of its employed capital is also well documented (Pratap & Quintin, 2006). Earning-relevant characteristics of workers are likewise under-rewarded in the traditional sector so human capital is not included, either. Finally, production from natural capital and resources is integrated through a constant additive term, as its production is assumed to be entirely dependent on the natural resource endowments of the country, with no physical or human capital or allocated labor contributing to its production. As in CKW, all other aspects of the production function (productivity, the role of government, elasticities of substitution between factors, etc.) are assumed to be constant over time within each sector. Therefore, any additional effects of gender egalitarianism through these channels are not captured by this model. Another possible critique of the model is the lack of sub-national estimation of the gender dividend. While subnational heterogeneities in the gender dividend are of first-order interest to many policy makers and academics, none of the previous iterations of the CKW model allowed for such estimation due to intractable complexities regarding migration. For example, while it is intuitive to think that demographic dynamics -- such as a lopsided age structure due to regional differences in fertility or mortality -- may lead to a larger or smaller demographic or gender dividend, in the presence of intra-regional migration fertility rates in one region may sever this relationship. This is particularly acute in many LMICs, which have very high rural-urban migration patterns. In addition, the lack of sub-national data on gender wedges in labor markets and on other key economic inputs (e.g., capital stock) also prohibits the estimation of regional heterogeneities. 11 The model of production diverts from that used by the CKW, which considers a modern, capital-intensive sector (i.e., manufacturing) and a traditional, labor-intensive sector (i.e., subsistence agriculture) as the latter is comparable with informal production processes in developing settings. 12 The endowment of natural resources is assumed to be fixed across all time periods. While using arable land as a proxy for natural resources is problematic (such land could be reduced due to depletion, desertification, or other environmental processes, as well as endogenous changes in land use), modeling such processes is complex and not first-order related to the gender dividend. Therefore, we abstract from these changes. 13 Our choice of a Cobb-Douglas production function was chosen primarily for tractability. Such a choice assumes a unit elasticity of substitution between all factors of production within each sectoral production function, which is not trivial given the importance of the relationship between movements of labor and subsequent changes in wages, by gender, in our model. One may be concerned that such a unit elasticity may over-estimate the reduction in wages from changes in labor supply. However, this unit elasticity of labor within sectors masks a large elasticity of substitution between labor and other factors across the aggregate economy, since labor can change sectors if the wage begins falling too quickly. As a result, we are hesitant to use a different within-sector production function (such as CES), which will not only complicate the model further, but also underestimate the relationship between changes in labor and changes in the wage in the aggregate production function. 9 3 Departures from the CKW Model Our model of the gender dividend departs from the established CKW framework in several ways to more effectively capture gender differentials and gender-specific contributions across the production process. The adjustments made to the CKW methodology, including the applied equations and their detailed descriptions, are described in Appendix 1.14 We highlight two specific amendments. 3.1 Labor Allocation by Sector and Gender A key departure from the CKW framework is our characterization of the differential role of gender on human capital attainment, labor supply, and production. In particular, our gender-specific approach requires that modern sector and traditional sector wages endogenously adjust by gender and within their respective sectors, and in turn determine equilibrium male and female labor supply allocations across the two sectors that employ workers. This modification is motivated by the fact that gender disparities within the labor market in developing countries and the relative disadvantage of women, both in terms of the type and the quality of their employment and labor market potential, are widespread (Filipiak, Kolev, & Arbache, 2010). This is the case for Liberian women, as discussed above, where a sizeable share of the disadvantage is often characterized by the fact that: 1) Liberian women are more often employed informally as compared to men; 2) women are significantly underrepresented in more productive sectors of employment; and 3) the female-to-male difference in labor income is substantial. In following CKW 2017, modern sector wages by gender are represented as the marginal product of labor for an additional worker with average levels of human capital. Since we assume that the traditional sector is not capital intensive, we allow the wage by gender to be determined instead by the average product of labor. The condition for which resources are shared equally among members in developing countries setting and especially in informal and more traditional settings is the basis for the sectoral differences that are applied in this model and in other macro-economic applications (Cypher, 2014; Lewis, 1954). The inefficient allocation of labor between sectors and the barriers to entry to the more productive (i.e., the formal) sector is here parametrized by a constant term, which is set to explain any baseline differential and is held constant over time. In our model of the gender dividend, we update this approach to equilibrium wage and labor equalization across sectors by also allowing for initial gender-specific differences and frictions in wages, labor supply allocation, and mobility between sectors. The parametrization occurs in a similar fashion to the sectoral equalization as two gender-specific constant terms calibrated to initial gender-specific differences at baseline are adopted and are held constant over time. The resulting sectoral wage is given by the average of male and female wages in each sector, weighted by the relative proportion of men and women in that sector. 3.2 Non-Tradable Production In our model, we do a monetary accounting of these “non-tradeable” activities and include their value into aggregate production. Conventional aggregate production measures often underestimate the economic contribution of women by neglecting to account for the unpaid provision of services, which theoretically could have been purchased on the market. We include non-tradable activities to better reflecting both the market and the intra-household contribution of a population to its economy. 14 The model by Karra, Canning & Wilde (2017) includes detailed equations for all processes that are adopted, which are not repeated in Appendix 1. The modeling processes are described in the following section, but Appendix 1 includes adjustments made to the original technique. 10 Aggregate non-tradable production is estimated through a minimal opportunity-cost framework in which the annual output of an individual is proxied by the income the worker would have earned if she had been allocated to the traditional sector of the labor market. The cumulative value of the non-tradable production is given by the yearly number of hours devoted to the domestic chores and housework activities by gender. The approach applies the average traditional-sector wage, as not doing so would transfer the gender wage inequities from the market to the valuated output (Chadeau, 1985). The model does not predict or impute the wages respective to earning-related characteristics such as the employment status or age-group of the different shares of the population partaking in non-tradable activities as this is deemed problematic for an aggregate standpoint (Hill, 2009). Instead, the alternative that is adopted ascribes the traditional-sector wage as the single and minimum remuneration with which the non-tradable production is valued by the economy. 4 Parameterization 4.1 Data Sources A wide range of data are used to calibrate the model. In following the CKW 2017 approach, the demographic module is comprised of total fertility rates (TFR) and projected quinquennial population by five-year age group that are gathered from the World Population Prospects of the United Nations (2010). The module of two-sector production is divided between modern sector and the traditional sector. The data on sectoral share of labor is taken by the Liberian Labor Force Survey (LFS) of 2010 and the sectoral share of GDP are taken from the 2007-2013 estimated average of Abid (2016). The amount of arable land and the value rendered by the natural resources are likewise inputted into the production module and taken from the World Development Indicators (WDI) database. The capital stock estimates and the Gross Domestic Product (GDP) statistics are drawn from the Penn World Tables (Feenstra, Inklaar, & Timmer, 2015). Baseline data on labor force participation rates (LFPR) by sex and five-years age groups are obtained from the ILOSTAT repository (ILO, 2020). The earnings by sex which are taken for the parametrization of wages are taken by the Household Income and Expenditure Survey (HIES) 2014 data (Liberia Institute for Statistics and Geo-Information Services, 2015). Baseline data on human capital is composed of the average educational attainment (in years) by sex and five-years age groups, obtained from the Barro-Lee dataset (Barro & Lee, 2013), and the average height (in meters) by sex and five-years age groups, obtained from the Demographic and Health Survey (DHS) by the Liberia Institute of Statistics and Geo-Information Services, The Ministry of Health and Social Welfare/Liberia, National AIDS Control Program/Liberia, and ICF International (2014). Baseline estimates of age-specific savings rates are gathered from Bloom, Canning, Mansfield, & Moore (2007). The non-tradable production module is composed of the average time (in hours) allocated to housework and domestic chores by sex, obtained at baseline from the United Nations Global SDG Database of the United Nations Statistics Division (2020) and UN Women (2019). To these measures, the time allocated to fetching water or firewood are added and obtained at baselined from the HIES 2014 from LISGIS (2015). Appendix 1 describes each source of data. 4.2 Calibration and Convergence We use parameters generated in CKW 2017 and impute additional parameters for our extension to the model. The CKW 2017 model parametrizes the reduction in labor market participation due to an additional child, the endogenous response of fertility to changes in education, the impact of fertility on 11 human capital, the effect of adult height on worker productivity and wages, standard estimated values for production factor shares and the effect of schooling and height on earnings. Appendix 1 describes the parameter values that were used to calibrate the model and the sources from which these values were obtained. Several additional parameter values were used in the expanded version of CKW 2017 model. Firstly, we parametrize the sectoral gender-specific friction and the gender-wage gap. This parametrization is drawn from a secondary-data analysis of the HIES 2014 from LISGIS (2015). The secondary source facilitates the estimation of earnings per unit worked (month, week and hour) for each employed person. We recognize that earnings measurement at the micro-level has a noisy and volatile nature15 but it has the advantage of being detailed to the intended population of interest for the adoption at the macro-level in this study. We find that the sectoral friction is noticeable in the higher wages of the modern sector across gender. The wage of female workers in the modern sector is 72.2 percent higher than the wage of female workers in the traditional sector. The average wage of a male worker in the modern sector is 46.2 percent higher than the average wage of a male worker in the traditional sector. Likewise, we find that women have lower wages than men in both the modern and the traditional sector. The average wage of a male worker in the modern sector is 6.5 percent higher than the average wage of a female worker in the same sector. And the average wage of a male worker in the traditional sector is 25.5 percent higher than the average wage of a female worker in the traditional sector.16 The modern-traditional divergence is substantially accentuated, particularly for women, and the gender-pay gap is observed in both the modern sector and the traditional sector to differing degrees. Secondly, we use the adapted CKW 2017 model to estimate gender-specific contributions to output when we incorporate a series of convergences. This exercise seeks to inform policymakers of the potential monetary gain of the gender dividend. Namely, the modeling of these estimates is intended to demonstrate how interventions to close the gender gap within domains such as family planning, women’s educational attainment, returns to labor and the like can ultimately generate economic returns that would not occur under business-as-usual circumstances. For each selected input to production (e.g., educational attainment), we estimate the change in productive output when the existing gender gap within that factor is mechanically closed. We run various exercises for each selected input to production.17 First, we allow a complete convergence between men and women for each domain and which we refer to as the ideal case. We regulate the fertility to an instant-replacement rate for the ideal case, or a quantity necessary to ensure a net reproduction rate of 1.0 starting in the 2015-2020 period (United Nations, 2010). Second, we allow for partial convergences within selected domains through benchmarking, which allow a comparison to averages found either in the Sub-Saharan Africa region or in the OECD region. Some of the benchmarking with Sub-Saharan Africa and OECD regions are not estimated due to domain-specific constraints related to data availability or limited applicability of the convergence. For the sectoral difference, we further hypothesize a complete convergence with depreciation of modern-sector wages and appreciation of traditional-sector wages driven by the surplus of female workers contributing to modern-sector production and incurred by the triggered convergence. We conclude with a simultaneous convergence by allowing the 15 For instance, during an economic downturn (e.g., Ebola, COVID-19) low-skilled workers are the first to be discharged and suffer from temporary dismissals or breaks in employment. Since more affluent workers are the ones to remain in the workforce, the measurement of wages or earnings at that time may be biased upwards. Other types of biases exists, such as mis-reporting from respondents or seasonal fluctuations. 16 These estimates slightly differ from the wage gap discussed in the introduction since the latter condition on worker characteristics. 17 See Appendix 4 for the list of estimates modeled by category. 12 labor-relevant domains to be completely closed and by accounting also for the remaining domains modeled. We continue by parametrizing the additive sub-module for which convergence is not necessarily relevant or accounted for, namely the non-tradable production. We parametrize the former by drawing statistical measures from the HIES 2014 and the UN Women (2019). According to these sources, we calibrate the time spent on non-tradable production activities to be different between men and women and therefore modeling the higher allocation of female time to unremunerated activities. The female population aged 15+ is estimated to allocate 8.1 percent of their time to non-tradable activities (or 1.95 hours a day) compared to 3.9 percent (or 0.93 hours a day) allocated by men. Note that under a more equitable redistribution between men and women, the total value of non-tradable production would still be higher for women than men since the female population aged 15+ is larger than the male population aged 15+. Our approach to modeling closures in the gender gap for a given factor assumes that convergence occurs instantaneously in the starting period, and the effects of this convergence are therefore immediately reflected in subsequent periods. To some extent, this modeling feature reflects the upper bound of the potential impact of closing gender gaps since the effects of an instantaneous convergence are able to accrue over the longest time horizon. 5 Results We present several of estimates, including macroeconomic measures as well as estimated labor-market and earning-relevant statistics. We begin by depicting the current economic contribution of Liberian women, which we refer to as the “status quo.” This is followed by the estimates of the gender dividend for each sub-module or characteristic studied. For each, we describe the initial condition of the sub-module and provide the macro-economic estimate, depicted as a share of GDP, GDP per capita, GDP per worker and GDP per sector-gender output. In relevant cases, estimated differences in the labor-market composition or wages are given. Finally, we present the estimates given by the simultaneous convergence of all the individually studied inputs to output. Appendix 4 summarizes the convergence experiments for each domain while Appendix 5 presents the results of each convergence experiment. Table 1 summarizes the experiments and estimates presented throughout this section. Table 1: Convergence Estimated by Domain, by the Calibration Exercise Category Estimated increase to GDP, % Calibration 8.4 Closure Education 4.9 SSA 8.65 OECD 3.1 net-reproduction rate Fertility 0.57 SSA 3.7 OECD LFPR 2.3 Closure 4.8 Closure Pay-Gap 2.0 OECD Sectoral 5.3 Closure 13 2.5 Closure+ Non-Tradable 19.3 Additive of NTP 11.9 Closure LFPR, Pay-Gap and Sectoral 9.6 Closure+ 23.6 Closure Complete Convergence 22.1 Closure+ 45.5 Additive of NTP Notes: Closure represents complete convergence of the respective gap. Closure+ represents appreciation/depreciation is applied to wages due to the displacement of workers between sectors. SSA and OECD stands for closure of the gap between Liberia and the averages found in Sub- Saharan Africa and OECD countries, respectively. Additive represents the additional value added by the non-tradable sector. The “additive of NTP” presented in the “Complete Convergence” section is the sum of the closure scenario estimate and the value of NTP. 5.1 The Status Quo: Women’s Economic Contribution in 2020 Liberian women contributed an estimated USD 1.08 billion to the Liberian economy in 2020 (Table 2). The estimate is derived from demographic, educational, and economic parameters. According to medium- variant UN projection, the population of Liberia was 5.06 million individuals. The working age (20-65) population was 45.4 percent of the entire population and had an average educational attainment of 4.6 years. Economically, GDP was USD 4.05 billion, GDP was USD 800, and per-worker GDP was USD 2,125 in 2020. The additive economic contribution of natural resources is estimated at USD 1.27 billion, or 31.4 percent of GDP. Women’s contribution represented 26.8 percent of GDP and 39 percent of labor’s contribution to GDP. These initial estimates are referred to as the “status-quo” that will serve as the benchmark for the study. Table 2. Macro-Economic Statistics for Baseline Scenario (2020) 1. GDP 1.1. GDP (USD) $4,049,744,636 1.2. GDP per capita (USD) $800 1.3. GDP per worker (USD) $2,125 2. Contribution of Women 2.1. in USD $1,084,390,549 2.2. as % of GDP 26.78% 2.3. as % of Labor Output 39.04% 3. Value Added by Sector 3.1. traditional, female (USD) $647,951,018 as % of GDP 16.00% 3.2. modern, female (USD) $436,439,531 as % of GDP 10.78% 3.3 traditional, male (USD) $735,858,976 as % of GDP 18.17% 3.4. modern, male (USD) $957,620,009 as % of GDP 23.65% 14 Source: Authors’ simulation based on expanded version of CKW 2017 model. When considering the status quo by sector, women’s contribution to the modern sector was USD 436.4 million or 10.8 percent of GDP while women represented 32.7 percent of its workforce with an average yearly wage of USD 1,582 (Table 3). In contrast, men’s total contribution to the modern sector was more than double that of the female contribution and amounted to USD 957.6 million or 23.65 percent of GDP. Men represented 67.3 percent of the workforce allocated to the modern sector and had average yearly wage of USD 1,686. Table 3. Labor Market Statistics for Baseline Scenario (2020) 3. Decomposition by Labor Supply 3.1. Total Labor Force 1,905,372 3.2. Female Labor Force 888,824 of which, modern (%) 20.69% of which, traditional (%) 79.31% 3.3. Male LF 1,016,547 of which, modern (%) 37.25% of which, traditional (%) 62.75% 4. Decomposition by Earnings 4.1. Modern-sector wage $1,652 4.2. Modern-sector, female wage $1,582 4.3. Modern-sector, male wage $1,686 4.4. Traditional-sector wage $1,030 4.5. Traditional-sector, female wage $919 4.6. Traditional-sector, male wage $1,153 Source: Authors’ simulation based on expanded version of CKW 2017 model. The contribution of women to the traditional sector was USD 647.9 million or 16 percent of GDP. The traditional sector employed 70.5 percent of the workforce in Liberia.18 With an average annual wage of USD 1,030, the traditional sector generated 34.1 percent of total GDP. Women represented 52.5 percent of the traditional sector workforce and had an average annual wage of USD 919. In contrast, the contribution of men to the traditional sector was USD 957.6 million or 18.7 percent of GDP. Men represented 47.5 percent of the workforce in the traditional sector and had average yearly wage of USD 1,153. Women’s lower pay in both sectors, and their greater employment in the traditional sector underpins their smaller contribution to GDP than men’s. That said, women generated about one-quarter of Liberia’s output. Their contribution would have been even greater if gender gaps to productivity were smaller. 5.2 Women’s Potential Contribution to GDP If Gender Gaps Narrow Closing the education gender gap would increase Liberia’s GDP by 8.4 percent (Figure 1). In Liberia, the female workforce aged 20-65 years old has an estimated educational attainment of 3.22 years on average, 18 A lack of consensus on the definition of the formal and informal sectors leads to significant differences in the point estimate for the modern/traditional shares of the workforce. For Liberia, the share of the labor force in the formal sector ranges from 13 percent (Johanssen da Silva, 2021) to 30 percent (ILO, 2020). 15 in comparison to almost a doubled male attainment of 6.03 years on average.19 Increasing women’s mean education by 2.8 years to eliminate the gender education gap would increase wome n’s output by an estimated USD 338 million, making women’s total contribution to GDP equivalent to USD 1.25 billion or 28.6 percent of GDP (up from 26.8 percent under the status quo). The estimated growth due to the closure of the education gender gap is driven by a variety of changes. Wages increased by 0.3 percent across the gender-disaggregated sub-sectors. However, the variation in the sectoral composition of the economy by gender contributed most to the growth in women’s output. The modern-sector output of women increased by 62.4 percent and reached USD 708.8 million as the entire share of the workforce in the modern sector would have increased from 29.5 percent to 41.7 percent. At the same time, women’s participation in the traditional sector would have declined, with their contribution to traditional output declining by 15.9 percent from the status quo scenario. The dividend would be less, but still considerable, if the gender gap in education were equal to the average found in Sub-Saharan Africa (SSA). Whereas women in SSA have 77 percent of the years of education achieved by men20 (Barro & Lee, 2013), such share stands at 41 percent for Liberia. By reaching the level found in SSA, the Liberian GDP would increase to USD 4.25 billion and the aggregate, per-capita and per-worker output would experience a growth of 5 percent. Even larger gains would be achieved if Liberia’s gender education gap were to mirror that of OECD countries. The gender hierarchy in educational achievement in the OECD has recently reversed, where more (young) women than ever are attaining formal education (OECD, 2015). If the women-to- men divergence found in the OECD would be achieved in Liberia, GDP would be 8.65 percent higher. Figure 1: Selected macro-economic changes due to complete convergences of the education gender gap (in USD, growth in %) $5,000 Millions 4,388 $4,500 85% 62.4% $4,000 65% $3,500 31.8% 45% Growth (in %) $3,000 $2,500 25% 8.4% 1.8% $2,000 5% $1,500 1,254 -15.9% -18.4% 1,262 -15% $1,000 709 545 600 -35% $500 $- -55% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations 19 To allows for the 5-year cohort projections, the Barro & Lee 2013 data were used. The value is slightly lower than the more recently measured (2016 HEIS) estimates presented in the introduction section. 20 The sample adopted is the one for which countries were available in the latest estimates of the dataset and excluded: Comoros, Djibouti, Eritrea, Ethiopia, Madagascar, Mayotte, Réunion, Somalia, Angola, Chad, São Tomé and Príncipe, Western Sahara, Burkina Faso, Cabo Verde, Côte d'Ivoire, Guinea, Nigeria. 16 Closing the gender labor force participation rate would increase GDP by 2.3 percent. Generally, women of working age (20-65) have high activity rates, with an average labor force participation rate of 77.8 percent, as compared to 88.6 percent for men.21 Yet if we close the 10.7 percentage point difference, the workforce would grow from 1.90 million to 2.04 million (Figure 2). Ultimately, GDP would reach USD 4.14 billion and the contribution of women would rise to USD 1.13 billion from USD 1.08 billion. The influx of workers into the labor-market would not be equally absorbed, instead displacing female and male workers between sectors to differing degrees. The share of workers in the traditional sector would increase from 70.5 percent to 73.9 percent and the share of workers in the modern sector would decrease from 29.5 percent to 26 percent. Notably, the share of female in the modern sector would increase from 32.7 to 40 percent while its male share would decrease from 67.3 to 60 percent. Since the gender gap in labor force participation is similar in Liberia to those in Sub-Saharan Africa and the OECD, we do not present the results here.22 Figure 2. Selected macro-economic changes due to complete convergence in the LFPR gender gap (in USD, growth in %) $4,500 20% 4,142 Millions $4,000 15% 15.2% 15.9% $3,500 10% $3,000 10.0% 5% Growth (in %) $2,500 3.5% 0% 2.3% $2,000 -5% $1,500 1,252 -10% $1,000 747 809 809 506 $500 -15% -15.6% $- -20% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations Fertility rates equal to the net reproduction rate of 1.0 would increase GDP by 3.1 percent (Figure 3). Overall, Liberian women have 4.47 children over the course of their reproductive years.23 We initially investigated to what extent a fertility rate that is equal to an instant-replacement level alters the economic 21 These point estimates were drawn from ILO 2010, constructed to allow for 5-year cohort averages. 22 Estimates on these benchmarks are available upon request to the authors. 23 This figure is drawn from the World Population Prospects, which estimates population size, in five- year increments, over the century. These rates are slightly higher than the estimates derived from the recently released DHS 2019/2020, reported earlier in the paper. This number decreased from 6.8 in 1970-1975 and it is projected to reach 3.9 in 2025-2030 (United Nations, 2015). There is significant variation in the average fertility rate across regions, education levels, and wealth quintiles (DHS 2019/2020). 17 contribution of women and the aggregate output in the economy. The variant set the fertility parameter to 2.35, a quantity necessary to ensure a net reproduction rate of 1.0 starting in the 2015-2020 period. The economy of Liberia was estimated to experience a 3.1 percent growth if fertility were equal to its instant- replacement rate (Figure 3). Per worker output was estimated to grow by 1.59 percent. The growth due to lower fertility rates was driven through several channels. The average years of education was predicted to grow from 4.6 to 5.0 while we expected the entire supply of labor to increase by 28,600 workers. The variation in output due to lower fertility rates was further driven by an increased allocation of the workforce to the modern sector and more representation of women within it. The modern sector was estimated to expand its share from 29.5 percent to 32.5 percent which led to a modern sector output growth of 18.3 percent for women and 8.8 percent for men. The dividend would be less, but still economically meaningful, if the fertility rate in Liberia were equal to the regional average (4.1 children per woman) in 2015-2020. A reduction of an average of 0.37 children per women in Liberia would lead to an economic expansion of 0.57 percent in GDP and women’s contribution would increase by 0.28 percent. Alternately, when using OECD fertility rates in the same years,24 GDP would grow an additional 3.71 percent in comparison to our baseline, reaching USD 5 billion. Figure 3. Selected macro-economic changes due reduction of fertility rates to instant- replacement rates (in USD, growth in %) $4,500 23% 4,176 Millions $4,000 18% $3,500 18.3% $3,000 13% Growth (in %) $2,500 8% $2,000 8.8% $1,500 3% 1,150 3.1% 1,042 $1,000 634 712 0.8% 516 -2% $500 -2.1% -3.3% $- -7% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations Closing the gender pay gap within sector would increase GDP by 4.8 percent (Figure 4). Liberia is not an exception to the global disparity that exists between male and female remuneration at work and especially in Sub-Saharan Africa (UN Women, 2016). Whereas some of such disparity can be explained by observable attributes of the female workforce such as attainment in education, the gender pay-gap is also reflective of wage discrimination between men and women. In Liberia, male workers earn 6.55 percent 24 The value adopted is 1.95 and sourced from the UN Population estimates given that the same source is used for Liberia and SSA estimates, yet this differs from recent estimates given by the OECD which put the average in 2019 to 1.6 children per woman (OECD, 2021). 18 more than women in the modern sector and 25.5 percent more than women in the traditional sector. We calibrate the average male and female wages to USD 1,686 in the modern sector and to USD 1,153 in the traditional sector. If women and men earned the same within sector, the value of traditional sector production was predicted to increase by 11.9 percent and modern sector production would increase by 2.1 percent. Women’s total contribution to GDP would rise to USD 1.28 billion, growing from 26.8 percent to 30.1 percent. Gains would be even higher if the gender wage gap in Liberia replicated that of the OECD. While female-to-male earnings were similar between Liberia and Sub-Saharan Africa,25 the pay-gap between men and women was 11.6 percent for employees in OECD countries (OECD, 2022). When we calibrated the average gender pay-gap found in the modern sector in Liberia from 25.5 percent to 11.6 percent, the GDP grew to USD 4.13 billion, or by 1.99 percent. Only the contribution of the female-modern sector was affected by this calibration, growing by 12.5 percent. Figure 4. Selected macro-economic changes due to complete convergence of the gender-pay gap (in USD, growth in %) $4,500 4,244 30% Millions 25.5% $4,000 25% $3,500 $3,000 20% Growth (in %) $2,500 15% $2,000 $1,500 1,278 10% 6.5% 958 $1,000 4.8% 813 736 3.3% 5% 465 $500 0.0% 0.0% $- 0% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations If men and women were equally represented in each sector, GDP would increase by 5.3 percent or to USD 4.26 billion (Figure 5). The underrepresentation of women in the high-productive sector is substantial, as outlined in section 5.1. Equalizing the gender distribution within sectors would decrease the share of women in the traditional sector from 61 percent to 46.7 percent and increase the share of women in the modern sector from 39 percent to 53.3 percent. The number of women contributing to modern- sector production would increase from 184,000 to 331,000 while the number of men would remain unchanged. Women’s total contribution to GDP would be USD 1.29 billion or 30.45 percent of GDP and 43.4 percent of the aggregate production of labor. Allowing for modern and traditional sector wages to shift in response to an inflow of women to the modern sector led to a smaller estimated increase in GDP, of 2.48 percent, as compared to estimates when wage appreciation and depreciation were not accounted for. We parametrized the adjustment with 25 Estimates on this benchmark are available upon request to the authors. 19 the averaged elasticities of factor demand (Behar, 2004).26 In the aggregate, the modern-sector wage was predicted to decrease from USD 1,652 to USD 1,485 and the traditional-sector wage was predicted to increase from USD 1,030 to USD 1,084. The modern sector remained more productive than the traditional sector by about 18 percent. Overall, the aggregate output, per-capita output and per-worker output were projected to increase and reach USD 4.15 billion, USD 820 and USD 2,178, respectively. We do not present results for sectoral distributions that approximate those of OECD or Sub- Saharan Africa. For OECD, we could not find complete and reliable statistics that differentiate the informal sector shares by gender. The Sub-Saharan Africa benchmark was not deemed valuable since the share of women and men in Liberia who are traditional sector workers is similar to that in Sub-Saharan Africa. Calibrating the model with SSA data gave results of a similar magnitude to the Liberia status quo.27 Figure 5. Selected macro-economic changes due to complete convergence of gender-sector gap (in USD, growth in %) $4,500 4,264 80.1% 90% Millions $4,000 70% $3,500 50% $3,000 Growth (in %) $2,500 30% $2,000 5.3% 3.7% 0.0% 0.0% 10% $1,500 1,298 -20.9% -10% 958 $1,000 786 736 513 -30% $500 $- -50% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations If non-tradable production were included in the aggregate production calculations, GDP would be 19.3 percent higher, reaching USD 4.83 billion. The value of non-tradable production is estimated at USD 780 million – equivalent to nearly 16.1 percent of GDP – with women responsible for 68 percent of the total value. In the first parametrization – which can be considered the status quo - the female population aged 15+ is estimated to allocate 8.1 percent of their time to non-tradable activities (or 1.95 hours a day) compared to 3.9 percent (or 0.93 hours a day) allocated by men. The estimated opportunity cost of the non-tradable production is USD 530.2 million for women compared to USD 250.3 million for men. Women’s contribution to GDP would be USD 1.61 billion if their non-tradable production were accounted for, as compared to USD 1.08 million when only measuring market-based output. The full accounting would increase women’s contribution to total output to 33.4 percent to GDP (compared to 26.8 percent without accounting for non-tradable production). 26 With a 1 percent increase/decrease in the share of a given sector, the respective wage decreases/increases by 0.35 percent. 27 Estimates on this benchmark are available upon request to the authors. 20 The total value of non-tradable production would not differ if we equalized the intra-household allocation of time between men and women. Under this variant, the contribution of the female population would be USD 392 million (or 50.2 percent) compared to USD 388 million (or 49.8 percent) of men. Women’s aggregate contribution would continue to exceed men’s as the female population aged 15+ was larger than the male population aged 15+. Alternatively, if we allowed the wage rate to depreciate as the traditional sector absorbed hours that were released from the non-tradable sector, the economic contribution of the non-tradables sector would be lower than the status quo. We translate the number of daily hours devoted to non-tradable production into the number of workers that they represent.28 The additional workforce absorbed by the traditional sector would bring its wage down by 14.4 percent and by using this calibration, the non-tradable sector would be valued at USD 551.5 million instead of USD 780 million. 5.3 Women’s Potential Contribution to GDP with Multiple Convergence GDP would increase by 11.5 percent or reach USD 4.51 billion (Figure 6) if labor-relevant gender gaps were simultaneously closed. The analysis continues by considering the effect of equalizing the gender gaps in labor-force participation, wages and sub-sectors shares. The supply of labor would grow from 1.90 million to 2.04 million workers and the contribution of women would be USD 1.62 billion, or 36 percent of GDP and 50.1 percent of the aggregate production of labor. A 9.2 percent growth and 11.1 percent growth from the initial estimates, respectively. The sectoral and gender decomposition indicated that most of the expansion would be driven by an increased allocation of the workforce to modern sector production which would grow by 86 percent for women alone. The aggregate output would grow by 9.6 percent (instead of 11.5 percent) if we would allow wages to change due to the displacement of the additional jobs created between sectors. Given the tight link between labor market earnings and poverty, the closure of the three labor-relevant gaps would reduce poverty by 8.4 percent and reach a headcount of 40.7 percent.29 Figure 6. Selected macro-economic changes due simultaneous convergence of labor-relevant gender gaps (in USD, growth in %) 28 The number of workers is 543,000 and it is estimated by dividing the total amount of hours devoted by the population to non-tradable production by 8 which is the standard number of hours devoted to a job in a day. 29 Poverty in Liberia is widespread and 44 percent of the population lived under the international poverty line of $1.90 per day in 2016 (World Bank, 2021). 21 $5,000 98% 86.0% Millions 4,514 $4,500 78% $4,000 $3,500 58% $3,000 Growth (in %) $2,500 38% 25.4% $2,000 1,624 11.5% 10.0% 18% $1,500 $1,000 9.2% 813 812 809 809 -2% $500 -15.6% $- -22% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations A complete convergence of gender gaps was estimated to increase GDP by 23.6 percent (Figure 7). Given the endogenous relationships between the sub-modules developed, it was not appropriate to simply sum the effects of the individual convergences. The model calculated that Liberia’s GDP would be USD 5.01 billion if all convergences would occur simultaneously. This specification maximizes the potential of women across the inputs of production in the model as their contribution would be 1.87 billion or 37.4 percent of GDP and 50.2 percent of the aggregate production of labor. A 10.6 percent growth and 11.1 percent growth from the initial estimates, respectively. The modern sector would generate 49.2 percent of the economic wealth and women would contribute 50.1 percent of it. The average wage of women would grow by 7.3 percent in the modern sector and 26.4 percent in the traditional sector. The average wage of men would grow too, by 0.7 percent in both sectors of production. Similar to the above- described relationship between GDP and poverty, the increase in aggregate output led by simultaneously closing the gender gaps modeled would allow poverty to reduce by 17.4 percent and reach a poverty headcount of 36.7 percent from the initial 44.4 percent. Figure 7. Selected macro-economic changes due simultaneous convergence of all gender gaps (in USD, growth in %) 22 $6,000 215% Millions 5,007 $5,000 183.0% 165% $4,000 115% Growth (in %) $3,000 65% 1,874 $2,000 1,228 1,235 37.4% 638 15% $1,000 23.6% 634 28.2% -1.5% -13.8% $- -35% GDP Female Trad. Mod. Trad. Mod. Female Male Source: author’s calculations If we would allow wages to depreciate due to the displacement of workers between sectors in our simultaneous convergence model, the growth would be 22.1 percent instead of 23.6 percent. When adding non-tradable production to the full gender convergence, GDP would equal to USD 5.89 billion. With such addition, women’s potential would be maximized in the Liberian economy as their contribution to GDP would rise to 41.4 percent while the value added to the labor market would be 52.8 percent. 6 Conclusions The role of gender equality is a subset of a broader body of scholarship on the relationship between intergroup inequality and growth (Kalecki, 1954). Perhaps most significant in this body of research is the finding of a two-way causality between gender equality and economic growth outcomes. If gender equality is found to be a stimulus to growth, then policies and interventions that promote greater gender equality can contribute to a virtuous cycle of growth. In this study, we showed that the current contribution of women to the Liberian economy is under- valued, creating an economic cost to not only women individually but also to society at large. We adapted a macrosimulation model to construct a model of the gender dividend, which quantifies the current economic contributions of women in Liberia. We then identified the potential economic contribution that women would be able to make if there were equality of opportunity and capability by gender across education, fertility, and labor market factors that have been identified as determinants of the gender dividend and inputs to economic growth and productivity. We also incorporated into the production estimate the contributions that are made from undertaking housework and domestic chores. We created a series of estimates by varying the parametrization of the model. We showed that the current contribution of women – using data as recently as 2020 – represented 26.8 percent of GDP and 39 percent of the aggregate production of labor. The contribution of women to the modern sector was solely 31.3 percent while the contribution of women to the traditional sector was 46.8 percent. If we account for the value of home-based non-tradable output, GDP would be 19.3 percent higher, reaching USD 4.83 billion and with women contributing 33.4 percent of the output. 23 We showed that the potential of women under equality of opportunities was considerable and that their current contribution is undervalued. The highest return of single-variable convergences to GDP was given by closing the gap in education, which would increase GDP by 8.3 percent. Most of the increase would be due to an increased share of (more educated) women in the higher-paying modern sector and a shift out of the lower-paid traditional sectors. A reduction of fertility rates to the net reproduction rate of 1.0 would increase GDP by 3.1 percent, due to the impacts of lower fertility on higher female educational attainment, increased female labor supply, and an increased allocation of women to the modern sector (though men’s modern sector employment would also grow). While individual convergences in labor are limited, simultaneously closing gender gaps in labor- relevant domains would have led to an increase in output equivalent to 11.5 percent of GDP. Closing the gender labor force participation rate would have increased GDP by 2.3 percent, the modest value is attributable to the small difference in women’s and men’s labor force participation rates in 2020. Closing the gender pay gap within sectors would have increased GDP by 4.8 percent and women’s contribution to GDP would have risen to 30.1 percent. If men and women were equally represented in each sector, GDP would have increased by 2.48 percent once accounting for the change in wages due to a changing sector size. If we considered all convergences simultaneously, GDP would have increased by 23.7 percent, reaching USD 5 billion. Given the endogenous relationships between the sub-modules developed, it was not appropriate to simply sum the effects of the individual convergences. However, the macrosimulation model allowed the convergences to simultaneously work through various channels, which reduced the chance of double- or under-counting. Women would have been responsible for 1.87 billion or 37.4 percent of GDP and 50.2 percent of the aggregate production of labor. When adding non-tradable production to the full gender convergence estimates, GDP would equal USD 5.89 billion. With such addition, women’s potential would be maximized in the Liberian economy as their contribution to GDP would rise to 41.4 percent while the value added to the labor market would be 52.8 percent. The predictions found in this study are comparable to studies that have attempted to estimate the economic losses due to gender inequality on employment and productivity. The 11.5 percent increase in GDP (and GDP-per-capita) that was generated by eradicating labor-relevant inequalities between men and women in our model closely resembles the estimated 8.9 percent increase found in estimates from Niger (World Bank, 2019) and the 19 percent increase in GDP-per-capita found in a multi-country study on employment and productivity (Pennings, 2022). On the other hand, our estimate of an 8.3 percent increase in output per capita from closing the gender gap in education was almost three times larger than the increase found by Devadas & Kim (2020) even though our study estimated similar average losses from the difference in labor force participation rates between men and women; the Devadas & Kim (2020) study estimated a 1.4 percent loss from gender gaps in labor force participation, which is about 0.9 percentage points lower than the 2.3 percent loss that we estimated for Liberia. Our approach is not without its limitations, many of which have been discussed throughout this paper as well as in prior studies that have used macrosimulation models for inference. A key constraint of our model is in its limited ability to infer general equilibrium effects that are not explicitly incorporated through our identified channels of interest.30 While there may be value added to modeling these relationships more 30 An example of this limitation can be observed in our conceptualization of the labor market in which labor supply allocations across sectors by gender may in turn have dynamic impacts on wages by gender and by sector, to the extent that wages in the formal sector may fall over time as more skilled women enter the labor market and are allocated to that sector. 24 realistically, we recognize that the inclusion of more realistic dynamics will increase complexity while simultaneously reducing tractability and transparency of the findings. Our model may therefore be more useful for providing insights of the underlying relationships and interactions between factors than for the predictions themselves. To this end, we recommend that our model predictions not be taken as precise measures of the gender dividend but rather be interpreted as illustrative of the potential relative impacts of these factors on key economic outcomes of interest. Furthermore, we assume that a closure of gender gaps occurs instantaneously in the starting period and the effects of this convergence are immediately reflected in subsequent periods. A more realistic approach would be to model convergence in the factors over multiple periods to reflect the time-delays in implementing exogenous policies and the time needed to realize results. This approach would need to specify the conditions under which convergence occurs over time, which would require either empirical evidence or assumptions about convergence velocity, acceleration, and functional form, among others, which are not available in the literature. Our study findings highlight a range of potential economic and social benefits that can be accrued from closing gender gaps and achieving a gender dividend, but it does not account for the cost of interventions. Interventions to close these gaps would require significant investments and coordinated efforts from a wide range of actors and stakeholders across a wide number of sectors. The inclusion of costs would allow us to generate a benefit-to-cost ratio, thereby giving us a better idea of the returns to investment. Yet formally modeling the costs of these investments is beyond the scope of this study. Taken together, our approach to conceptualizing and estimating the gender dividend provides empirical evidence to reinforce the need for a unified policy agenda that simultaneously actively advances women’s opportunities in many areas of life. While Liberia’s policy documents speak to this goal, they yet remain unrealized. The findings of this paper are intended to provide an order of magnitude that may speak to the decision-makers that set spending priorities. This goal has implications for global development at large but also, and more importantly, serves to promote the rights and well-being of women as equal and productive members of society. 25 7 Bibliography Abid, M. (2016). Size and Implication of Informal Economy in African Countries: Evidence from a Structural Model. International Economic Journal, 30(4), 571-598. Acemoglu, D. (2010). Theory, General Equilibrium, and Political Economy in Development Economics. The Journal of Economic Perspectives, 24(3), 17–32. Acemoglu, D., & Johnson, S. (2007). Disease and Development: The Effect of Life Expectancy on Economic Growth. Journal of Political Economy, 115(6), 925–985. Agénor, P.-R. (2012). A Computable OLG Model for Gender and Growth Policy Analysis. Centre for Growth and Business Cycle Research: Discussion Paper(169). Ahlburg, D. A. (2002). Does Population Matter? A Review Essay*. Population and Development Review, 28(2), 329–360. https://doi.org/10.1111/j.1728-4457.2002.00329.x Ashraf, Q. H., Weil, D. N., & Wilde, J. (2013). The Effect of Fertility Reduction on Economic Growth. Population and Development Review, 39(1), 97–130. https://doi.org/10.1111/j.1728- 4457.2013.00575.x Barro, R. J. (1991). Economic growth in a cross section of countries. Quarterly Journal of Economics, 106(2), 407–443. Barro, R., & Lee, J.-W. (2013). A New Data Set of Educational Attainment in the World, 1950-2010. Journal of Development Economics, 104, 184-198. Barro, R., & Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw-Hill. Baten, J., de Haas, M., Kempter, E., & Meier zu Selhause, F. (2021). Educational Gender Inequality in Sub-Saharan Africa: A Long-Term Perspective. Population and Development Review, 47(3), 813-849. Behar, A. (2004). Estimates of labour demand elasticities and elasticities of substitution using firm-level manufacturing data. SALDRU/CSSR Working Papers. Bloom, D. E., & Canning, D. (2008). Global Demographic Change: Dimensions and Economic Significance. Population and Development Review, 34(Supplement), 17–51. Bloom, D. E., Canning, D., & Fink, G. (2009). Disease and Development Revisited. NBER Working Paper 15137. http://www.nber.org/papers/w15137 Bloom, D. E., Canning, D., Fink, G., & Finlay, J. E. (2009). Fertility, female labor force participation, and the demographic dividend. Journal of Economic Growth, 14(2), 79–101. https://doi.org/10.1007/s10887-009-9039-9 Bloom, D. E., Canning, D., Mansfield, R. K., & Moore, M. (2007, January 1). Demographic Change, Social Security Systems, and Savings. Journal of Monetary Economics, 54(1), 92-113. Boserup, E., Tan, S. F., & Toulmin, C. (2007). Woman's Role in Economic Development. Routledge. Campaña, J. C., Gimenez-Nadal, J. I., & Molina, J. A. (2015). Gender Differences in the Distribution of Total Work-Time of Latin-American Families: The Importance of Social Norms. IZA Discussion Paper Series(8933). Canning, D., Raja, S., & Yazbeck, A. S. (Eds.). (2015). Africa’s Demographic Transition: Dividend or Disaster? The World Bank. Cavalcanti, T., & Tavares, J. (2007). The Output Cost of Gender Discrimination: A Model-Based Macroeconomic Estimate. CEPR Discussion Paper(6477). CEDOVIP. (2013). Economic Costs of Domestic Violence in Uganda. CEDOVIP and UKaid. Coale, A. J., & Hoover, E. M. (1958). Population Growth and Economic Development in Low-Income Countries: A Case Study of India’s Prospects. Oxford University Press. 26 Cuberes, D., & Teignier, M. (2016). Aggregate Effects of Gender Gaps in the Labor Market: A Quantitative Estimate. Journal of Human Capital, 10(1). Cunningham, W., Gupta, S., Johanssen da Silva, S. (2022). An Assessment of Gender Gaps in Liberia through a Women’s Empowerment Lens. Washington: The World Bank. Cypher, J. M. (2014). The Process of Economic Development. New York: Routledge. Day, T., McKenna, K., & Bowlus, A. (2005). The Economic Costs of Violence Against Women: An Evaluation of the Literature. United Nations. Deaton, A. (1989). Saving in Developing Countries: Theory and Review. The World Bank Economic Review, 3(suppl_1), 61–96. https://doi.org/10.1093/wber/3.suppl_1.61 Deaton, A. (1992). Saving and Income Smoothing in Cote d’lvoire. Journal of African Economies, 1(1), 1–24. Demirguc-Kunt, A., Klapper, L., & Singer, D. (2013). Financial Inclusion and Legal Discrimination Against Women: Evidence from Developing Countries. Policy Research Working Paper(6416). Devadas, S., & Kim, Y. E. (2020). Exploring the Potential of Gender Parity to Promote Economic Growth. Research & Policy Briefs From the World Bank Malaysia Hub(39). Doepke, M. (2004). Accounting for fertility decline during the transition to growth. Journal of Economic Growth, 9(3), 347–383. Doepke, M., & Tertilt, M. (2016). Chapter 23—Families in Macroeconomics. In J. B. Taylor & H. Uhlig (Eds.), Handbook of Macroeconomics (Vol. 2, pp. 1789–1891). Elsevier. https://doi.org/10.1016/bs.hesmac.2016.04.006 Enke, S. (1971). Economic Consequences of Rapid Population Growth. The Economic Journal, 81(324), 800– 811. https://doi.org/10.2307/2230318 Feenstra, R. C., Inklaar, R., & Timmer, M. P. (2015). The Next Generation of the Penn World Table. American Economic Review, 105(10), 3150-3182. Ferrant, G., Pesando, L. M., & Nowacka, K. (2014, December). Unpaid Care Work: The missing link in the analysis of gender gaps in labour outcomes. Filipiak, E., Kolev, A., & Arbache, J. S. (2010). Gender Disparities in Africa’s Labor Market. Washington D.C.: Agence Française de Développement and the World Bank. Galor, O., & Weil, D. N. (1996). The gender gap, fertility, and growth. American Economic Review (86), 374- 387. Garenne, M. (2008). Fertility Changes in Sub-Saharan Africa. DHS Comparative Reports No. 18. Calverton, Maryland, USA: Macro International. Government of the Republic of Liberia. Pro-Poor Agenda for Prosperity and Development (PAPD). A Five-Year National Development Plan Towards Accelerated, Inclusive, and Sustainable Development (July 2018 – June 2023) (2018). Monrovia, Liberia. Hill, T. (2009). Time Use, Gender and Disadvantage in Australia: Conventional Income and ‘Full Income’ Approaches to Estimation. The Economic and Labour Relations Review, 20(1), 13–34. Hiller, V. (2014). Gender Inequality, Endogenous Cultural Norms, and Economic Development. The Scandinavian Journal of Economics, 116(2), 455-481. ILO. (2017). SWTS Country Brief: Liberia. Geneva: International Labour Organisation. ILO. (2020). ILOSTAT Database. Geneva, Switzerland: International Labour Office (ILO). J-PAL Policy Bulletin. (2017). Roll Call: Getting Children Into School. Cambridge, MA: Abdul Latif Jameel Poverty Action Lab. Johanssen da Silva, S. M. (2021). Transforming Jobs in Liberia. Washington: The World Bank. Joshi, S., & Schultz, T. P. (2007a). Family planning as an investment in development: Evaluation of a program’s consequences in Matlab, Bangladesh. Economic Growth Center Discussion Paper, 951. 27 Karra, M., Canning, D., & Wilde, J. (2017). The Effect of Fertility Decline on Economic Growth in Africa: A Macrosimulation Model. Population and Development Review, 43(S1), 237–263. https://doi.org/10.1111/padr.12009 Kelley, A. C. (1988). Economic Consequences of Population Change in the Third World. Journal of Economic Literature, 26(4), 1685–1728. Kelley, A. C., & Schmidt, R. M. (2005). Evolution of recent economic-demographic modeling: A synthesis. Journal of Population Economics, 18(2), 275–300. https://doi.org/10.1007/s00148-005-0222-9 Klasen, S. (1999). Does Gender Inequality Reduce Growth and Development? Evidence from Cross- country Regressions. Policy Research Report on Gender and Development. Knowles, J. C., & Behrman, J. R. (2005). The economic returns to investing in youth in developing countries: A review of the literature. HNP Discussion Paper. KPMG. (2019). The workplace impacts of domestic violence and abuse. Kuznets, S. (1967). Population and Economic Growth. Proceedings of the American Philosophical Society, 111(3), 170–193. Lee, R. D., & Mason, A. (2011). Population Aging and the Generational Economy: A Global Perspective. Edward Elgar Publishing. Lewis, W. (1954). Economic development with unlimited supplies of labour. The Manchester School, 22(2), 139-191. Li, H., & Zhang, J. (2007). Do high birth rates hamper economic growth? The Review of Economics and Statistics, 89(1), 110–117. Liberia Institute for Statistics and Geo-Information Services. (2015). Household Income and Expenditure Survey (HIES) 2014. Monrovia, Liberia. Liberia Institute of Statistics and Geo-Information Services - LISGIS, Ministry of Health and Social Welfare/Liberia, National AIDS Control Program/Liberia, and ICF International. (2014). Liberia Demographic and Health Survey 2013. Monrovia, Liberia: LISGIS and ICF International. Liberia Institute of Statistics and Geo-Information Services - LISGIS/Liberia, Ministry of Health and Social Welfare/Liberia, National AIDS Control Program/Liberia, and Macro International. (2007). Liberia Demographic and Health Survey 2007. Monrovia: LISGIS and Macro International. Liberia Institute of Statistics and Geo-Information Services (LISGIS), Ministry of Health, and ICF. (2021). Liberia Demographic and Health Survey 2019-20. Monrovia, Liberia: Liberia Institute of Statistics and Geo-Information Services (LISGIS), Ministry of Health, and ICF. Liberia Institute of Statistics and Geo‐Information Services (LISGIS). (2010). Report on the Liberia Labour Force Survey 2010. Monrovia: ILO & Liberia Institute of Statistics and Geo‐Information Services (LISGIS). Löfström, Å. (2006). Gender equality, economic growth and employment. Stockholm: Swedish Ministry of Integration and Gender Equality. Lucas, R. (1988). On the Mechanics of Development Planning. Journal of Monetary Economics(22), 3-42. Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A contribution to the empirics of economic growth. Quarterly Journal of Economics, 107(2), 407. Miller, G. (2010). Contraception as development? New evidence from family planning in Colombia. The Economic Journal, 120, 709–736. Natali, L., & Kabeer, N. (2013). Gender Equality and Economic Growth: Is there a Win-Win? IDS Working Paper(417). OECD. (2015). Education at a Glance 2015: OECD Indicators. Paris: OECD Publishing. 28 OECD. (2022). Gender wage gap (indicator). Retrieved April 7, 2022, from OECD Data: https://data.oecd.org/earnwage/gender-wage-gap.htm Parr, N. J. (1995). Pre-Marital Fertility in Liberia. Journal of Biosocial Science, 27(1). Peet, E. D., Günther, F., & Fawzi, W. (2015). Returns to education in developing countries: Evidence from the living standards and measurement study surveys. Economics of Education Review(49), 69-90. Pennings, S. M. (2022). A Gender Employment Gap Index (GEGI) : A Simple Measure of the Economic Gains from Closing Gender Employment Gaps, with an Application to the Pacific Islands. Washington, DC: World Bank. Pratap, S., & Quintin, E. (2006). The Informal Sector in Developing Countries: Output, Assets and Employment. WIDER Working Paper Series(RP2006-130). Roemer, P. (1996). Increasing Returns and Long-Run Growth. Journal of Political Economy(94), 1002-1037. Rosenzweig, M. R., & Zhang, J. (2009). Do Population Control Policies Induce More Human Capital Investment? Twins, Birth Weight and China’s “One-Child” Policy. The Review of Economic Studies, 76(3), 1149–1174. Rubiano-Matulevich, E., & Viollaz, M. (2019). Gender Differences in Time Use: Allocating Time between the Market and the Household. Policy Research Working Paper. Schmidhuber, J., Bruinsma, J., & Boedeker, G. (2009). Capital requirements for agriculture in developing countries to 2050. In FAO Expert Meeting on “How to Feed the World in 2050.”. Rome: FAO. Schultz, T. P. (2002). Wage gains associated with height as a form of health human capital. Yale Economic Growth Center Discussion Paper(841). Schultz, T. P. (2009a). How Does Family Planning Promote Development? Evidence from a Social Experiment in Matlab, Bangladesh, 1977–1996. Yale University, Economic Growth Center, New Haven, Conn. Schultz, T. P. (2009b). Population and Health Policies. IZA Discussion Paper No. 4340. Simon, J. L. (1976). Population Growth May Be Good for LDCs in the Long Run: A Richer Simulation Model. Economic Development and Cultural Change, 24(2), 309–337. https://doi.org/10.1086/450871 Tuccio, M., & Paci, P. (2016). Why is growth less poverty reducing in Africa. Jobs and Development Conference. UN Women. (2016). Leave No One Behind: A Call to Action for Gender Equality and Women’s Economic Empowerment. Geneva: UN Women. UN Women. (2019). Women Count: Liberia. Retrieved September 2020, from UN Women Data: https://data.unwomen.org/country/liberia UN Women. (2021). Liberia: Country Gender Equality Profile. Geneva: UN Women. United Nations. (2010). World Population Prospects 2010, Online Edition. Rev. 1. Department of Economic and Social Affairs, Population Division. United Nations. (2015). World Fertility Patterns 2015 – Data Booklet (ST/ESA/ SER.A/370. Geneva: Department of Economic and Social Affairs, Population Division. UNSD. (2020, August 4). Global SDG Indicators Database, 4th August 2020. Retrieved August 11, 2020, from United Nations Statistics Division. WHO. (2012). Combating sexual and gender-based violence in Liberia. Retrieved from World Health Organization. Wilde, J. K., & Karra, M. (2022). Savings and the Demographic Dividend. Wodon, Q. O. (2020). How Large is the Gender Dividend? Measuring Selected Impacts and Costs of Gender Inequality. Washington, DC: World Bank Group. Wodon, Q., Onagoruwa, A., Malé, C., Montenegro, C., Nguyen, H., & de la Brière, B. (2020). How Large Is the Gender Dividend? Measuring Selected Impacts and Costs of Gender Inequality. The Cost of Gender Inequality Notes Series. Washington, DC: World Bank. 29 World Bank. (2019). Economic Impacts of Gender Inequality in Niger. Washington, DC: World Bank. World Bank. (2021). Poverty & Equity Brief: Africa Western & Central: Liberia. Washington, DC: World Bank. World Bank. (2022). World Development Indicators: Proportion of seats held by women in national parliaments (%) - Liberia. 30 8 Appendixes Appendix 1: Data Sources Data Type Source Baseline population by age and sex, 2010 World Population Prospects Baseline age-specific fertility rates, 2010-2100 World Population Prospects Years of education by 5 year age-sex groups, 2010 Barro & Lee, 2013 Adult height by 5 year age-sex groups, Liberia DHS, 2013 Labor force participation by 5 year age-sex groups, 2010 ILO Sector- and gender-disaggregated earnings HIES 2014 Hours devoted to domestic chores and housework UN Women, 2019 Hours devoted to collecting firewood or fetching water HIES 2014 Output, 2005 WDI Output, 2010 WDI Oil Output, 2010 WDI Capital stock, 2010 WDI Agricultural land, 2010 WDI Proportion of GDP between sectors, 2010 Abid, 2016 Proportion of labor between sectors, 2010 ILO Appendix 2: Departures from CKW 2017 Non-Tradable Production It is assumed that there is no overlap between (1) hours devoted to domestic chores and housework and (2) hours devoted to collecting firewood (or charcoal) or fetching water. The total daily female contribution through non-tradable production ( ) is given by: = ( × ℎ1 × (1 + 1ℎ ) × . Where = female population aged 15-100 at time ℎ1 = share involved in fetching water/firewood and housework and domestic chores 1 = hours devoted to fetching water/firewood by 1 × ℎ1 1ℎ = hours devoted to housework and domestic chores by 1 × ℎ1 . = hourly wage at time t in the traditional sector. Labor Allocation across Sectors by Gender Our gender-specific labor model requires that the modern sector and traditional sector wages, which endogenously adjust by gender = {, } and within their respective labor markets, will in turn determine equilibrium male and female labor supply allocations across the two sectors that employ workers. Total male labor supply and total female labor supply are shared across the modern and traditional sectors such that = + (1) = + (2) In addition, we have the following labor market-clearing conditions: 31 = + (3) = + (4) = + (5) The wage per worker in the modern sector at time , , is set to be equal to the marginal product of labor in the modern sector for an additional worker with average levels of education and health, or in log terms (6) = [(1 − ) ] In contrast, we assume that the traditional sector is less developed and is more labor intensive with little to no capital endowment, thereby resulting in the wage per worker in the traditional sector at time , , being determined by the average product, or in log terms: (7) = Since the wage in the traditional sector is determined at the average and not on the margin, in equilibrium there will be too many workers in the traditional sector. In addition, there may be migration costs or other barriers to entry into modern sector jobs, which are parametrized by the term , that will contribute to an inefficient allocation of labor across sectors. In equilibrium, workers will migrate between sectors and wages will adjust such that − = (8) Here, is a constant that is set so as to explain any baseline differential in sector wages and is then held constant over time. Similarly to the equalization of wages across sectors in equilibrium, we also allow for differences in wages within each sector by gender, which effectively captures the sector-specific gender gap in wages. In particular, we specify that wages between men and women in the modern and traditional sectors differ by the terms and , respectively, such that − = (9) − = (10) The terms and are also constants that are calibrated to the initial differences in baseline sectoral wages between men and women and are then held constantly over time. We also specify that wage (and hence labor supply allocation and migration) across each sector for each gender is determined by − = (9) − = (10) where the gender-specific wage frictions between the modern and traditional sectors, for men and for women, are specified and calibrated for the aggregate sectors using baseline data on wage differences. Finally, the sectoral wage is set to be the weighted average of male and female wages in each sector, weighted by the relative fraction of men and women in that sector (11) = ⋅ + ⋅ (12) = ⋅ + ⋅ Equilibrium Solutions We calculate our equilibrium conditions in sequence, first solving for aggregate wages and labor supply by sector, and then solving for gender-specific labor distributions and gender-specific wages within each 32 sector. If we replace modern sector and traditional sector aggregate wages with their respective aggregate wage-output equilibrium conditions and substitute modern sector and traditional sector output with their respective production functions, we obtain: − = ( − )− where (1 − ) ⋅ + = ⋅ 1 1 For = 3 and = 6, we can explicitly solve for as 1 3 6 6 ( √ + 4 − = ) 2 We can verify that 0 ≤ ≤ , and we calibrate the value of so that initial labor stock in the modern sector, , matches the data. We then fix to that value in all subsequent simulations. We note the following identities: = ⋅ , > 1 = ⋅ , > 1 = ⋅ , > 1 = ⋅ , > 1 = ⋅ , > 1 These relationships imply that in terms of = = = = ⋅ In solving for modern sector wages for males, we first find that: − = ⋅ + ⋅ Therefore ( − 1) + = [ ] ⋅ For traditional sector wages for males, we find that: ( − ) ( − − + ) = ⋅ + ⋅ − − ( − ) ( − − + ) = ⋅ + ⋅ − ⋅ − ( − ) ( − − + ) = [ + ] ( − ) ⋅ ( − ) ⋅ ⋅ ( − ) + ( − − + ) = [ ] ⋅ ⋅ ( − ) 33 ( ⋅ ⋅ − ⋅ ⋅ ) + ( − − + ) = [ ] ⋅ ⋅ ( − ) + ( ⋅ − ) − + ( − ⋅ ) = [ ] ⋅ ⋅ ( − ) We know that = , so ( − 1) + =[ ] ⋅ ⋅ + ( ⋅ − ) − + ( − ⋅ ) =[ ] = ⋅ ⋅ ( − ) Canceling out the and terms: ( − 1) + + ( ⋅ − ) − + ( − ⋅ ) = ⋅ ⋅ ( − ) Solving for : 2 ( − ) + ( ⋅ − ) − ( − ) = ( − 1) + + ( ⋅ − − ) We can confirm that 0 < < < and can use the solution for to solve for the other labor supply and wage allocations by sector and gender. 34 Appendix 3: Calibration Rel. Symbol Value Description Source(s) 1 π 0.02 Effect of fertility on female labor supply Ashraf et al. (2013) 2 θ_E 0.2 Effect of fertility on childhood education Joshi & Schultz (2007); Rosenzweig & Wolpin (1980) 3 ψ -0.15 Effect of women’s education on fertility Osili & Long (2008) Giroux (2008); Joshi & Schultz (2013); Kravdal & Kodzi (2011); Stevens et 4 θ_H -0.00067 Effect of fertility on adult height al. (2012); Victora et al. (2008) 5 α 0.33 Capital share of output in modern sector Hall & Jones (1999) 6 β 0.167 Land share of output in traditional sector Kawagoe et al. (1985); Williamson (1998, 2002) Banerjee & Duflo (2005); Oyelere (2010); Psacharopoulos (1994); 7 γ 0.1 Economic returns to schooling Psacharopoulos & Patrinos (2004) 8 λ 0.08 Effect of health on output Schultz (2002, 2005) 9 δ 0.07 Depreciation rate of capital Schmitt-Grohe & Uribe (2006) 10 ϕ_1 0.758 Effect of lagged savings on current savings Bloom et al. (2007) 11 ϕ_2 0.133 Effect of wage rate on savings rate Bloom et al. (2007) 12 ϕ_3 -0.006 Effect of squared wage rate on savings rate Bloom et al. (2007) Effect of ratio of old to working age 13 ϕ_4 -0.209 Bloom et al. (2007) population on savings rate 14 υ +-0.356 Elasticity of factor demand Behar (2004) 15 τ -0.326 GDPpc-to-poverty elasticity Tuccio & Paci (2016) 35 Appendix 4: Estimations by Domain and Sensitivities Table A4.1 Convergence Estimated by Domain Category Model No. Calibration 1.1 Closure Education 1.2 SSA 1.3 OECD 2.1 net-reproduction rate Fertility 2.2 SSA 2.3 OECD LFPR 3.1 Closure 4.1 Closure Pay-Gap 4.2 OECD 5.1 Closure Sectoral 5.2 Closure+ Non-Tradable 6.1 Additive of NTP 7.1 Closure LFPR, Pay-Gap and Sectoral 7.2 Closure+ 7.3 Closure Complete Convergence 7.4 Closure+ 7.5 Additive of NTP Notes: Closure represents complete convergence of the respective gap. Closure+ represents appreciation/depreciation is applied to wages due to the displacement of workers between sectors. SSA and OECD stands for benchmarking with averages found in Sub-Saharan Africa and OECD countries, respectively. Additive represents the additional value added by the non-tradable sector which is summed to the status-quo. Table A4.2 Sensitivity Tests Model of Appl. Description Value 1.1 Economic returns to schooling 4.2% 1.1 Economic returns to schooling 12.5% Effect of fertility on female labor 3.1 supply 0 6.1 Non-tradable wages depreciated Notes: refer to Appendix 4 for the status-quo values of coefficients. 36 Appendix 5: Results Table 5.1. Convergence Estimates for Education and Fertility Education Fertility 1.1 1.2 1.3 2.1 2.2 2.3 Baseline Closure SSA OECD net-repr.rate SSA OECD 1. GDP 1.1. GDP (in USD) $4,049,744,636 $4,388,335,127 $4,253,190,626 $4,400,110,268 $4,175,874,219 $4,072,908,921 $4,199,841,582 1.1.1. Growth (in %) 8.36% 5.02% 8.65% 3.11% 0.57% 3.71% 1.2. GDP per capita (in USD) $800.71 $867.66 $840.94 $869.99 $825.65 $805.29 $830.39 1.2.1. Growth (in %) 8.36% 5.02% 8.65% 3.11% 0.57% 3.71% 1.3. GDP per worker (in USD) $2,125.44 $2,303.14 $2,232.21 $2,309.32 $2,159.22 $2,125.66 $2,169.19 1.3.1. Growth (in %) 8.36% 5.02% 8.65% 1.59% 0.01% 2.06% 2. Women's Contribution 2.1. in USD, bil. $1,084,390,549 $1,253,722,388 $1,184,810,896 $1,259,793,203 $1,150,278,839 $1,102,102,586 $1,159,794,558 2.2. of GDP (in %) 26.78% 28.57% 27.86% 28.63% 27.55% 27.06% 27.62% 2.2.1. Growth (in p.p. %) 1.79% 1.08% 1.85% 0.77% 0.28% 0.84% 2.3. of Labor Output (in %) 39.04% 40.23% 39.74% 40.27% 39.61% 39.35% 39.61% 2.3.1. Growth (in p.p. %) 1.19% 0.70% 1.23% 0.57% 0.31% 0.57% 3. Sectoral Value Added 3.1. traditional, female (in USD) $647,951,018 $544,919,521 $587,085,006 $541,193,305 $634,051,255 $650,632,344 $628,268,957 3.1.1. Growth (in %) -15.90% -9.39% -16.48% -2.15% 0.41% -3.04% 3.2. modern, female (in USD) $436,439,531 $708,802,867 $597,725,890 $718,599,899 $516,227,584 $451,470,242 $531,525,601 3.2.1. Growth (in %) 62.41% 36.96% 64.65% 18.28% 3.44% 21.79% 3.3 traditional, male (in USD) $735,858,976 $600,342,350 $653,652,082 $595,736,673 $711,563,750 $735,955,225 $704,207,490 3.3.1. Growth (in %) -18.42% -11.17% -19.04% -3.30% 0.01% -4.30% 3.4. modern, male (in USD) $957,620,009 $1,262,395,286 $1,142,852,546 $1,272,705,289 $1,042,156,527 $962,976,008 $1,063,964,432 3.4.1. Growth (in %) 31.83% 19.34% 32.90% 8.83% 0.56% 11.11% Source: Authors’ simulation based on expanded version of CKW 2017 model. 37 Table 5.2. Convergence Estimates for LFPR, Pay-Gap, Sectoral and Non-Tradable LFPR Pay-Gap Sectoral 3.1 4.1 4.2 5.1 5.2 Baseline Closure Closure OECD Closure Closure+ 1. GDP 1.1. GDP (in USD) $4,049,744,636 $4,142,044,348 $4,243,588,292 $4,130,475,421 $4,263,837,535 $4,150,064,054 1.1.1. Growth (in %) 2.28% 4.79% 1.99% 5.29% 2.48% 1.2. GDP per capita (in USD) $800.71 $818.96 $839.04 $816.67 $843.04 $820.55 1.2.1. Growth (in %) 2.28% 4.79% 1.99% 5.29% 2.48% 1.3. GDP per worker (in USD) $2,125.44 $2,033.18 $2,227.17 $2,167.81 $2,237.80 $2,178.09 1.3.1. Growth (in %) -4.34% 4.79% 1.99% 5.29% 2.48% 2. Women's Contribution 2.1. in USD, bil. $1,084,390,549 $1,252,334,304 $1,278,234,205 $1,165,121,334 $1,298,483,448 $1,245,245,780 2.2. of GDP (in %) 26.78% 30.23% 30.12% 28.21% 30.45% 30.01% 2.2.1. Growth (in p.p. %) 3.46% 3.34% 1.43% 3.68% 3.23% 2.3. of Labor Output (in %) 39.04% 43.63% 43.01% 40.76% 43.40% 43.26% 2.3.1. Growth (in p.p. %) 4.60% 3.98% 1.72% 4.36% 4.23% 3. Sectoral Value Added 3.1. traditional, female (in USD) $647,951,018 $746,647,597 $813,208,892 $728,681,803 $512,652,499 $532,672,334 3.1.1. Growth (in %) 15.23% 25.50% 12.46% -20.88% -17.79% 3.2. modern, female (in USD) $436,439,531 $505,686,706 $465,025,312 $436,439,531 $785,830,948 $712,573,446 3.2.1. Growth (in %) 15.87% 6.55% 0.00% 80.05% 63.27% 3.3 traditional, male (in USD) $735,858,976 $809,219,818 $735,858,976 $735,858,976 $735,858,976 $764,595,353 3.3.1. Growth (in %) 9.97% 0.00% 0.00% 0.00% 3.91% 3.4. modern, male (in USD) $957,620,009 $808,615,124 $957,620,009 $957,620,009 $957,620,009 $868,347,818 3.4.1. Growth (in %) -15.56% 0.00% 0.00% 0.00% -9.32% Source: Authors’ simulation based on expanded version of CKW 2017 model. 38 Table 5.3. Multiple Convergences Estimates LFPR, Pay-Gap and Sectoral Complete Convergence 7.1 7.2 7.3 7.4 Baseline Closure Closure+ Closure Closure+ 1. GDP 1.1. GDP (in USD) $4,049,744,636 $4,514,124,396 $4,438,326,790 $5,007,472,629 $4,943,569,613 1.1.1. Growth (in %) 11.47% 9.60% 23.65% 22.07% 1.2. GDP per capita (in USD) $800.71 $892.53 $877.54 $990.07 $977.44 1.2.1. Growth (in %) 11.47% 9.60% 23.65% 22.07% 1.3. GDP per worker (in USD) $2,125.44 $2,215.82 $2,178.61 $2,428.51 $2,397.52 1.3.1. Growth (in %) 4.25% 2.50% 14.26% 12.80% 2. Women's Contribution 2.1. in USD, bil. $1,084,390,549 $1,624,414,351 $1,586,438,641 $1,873,707,498 $1,841,654,912 2.2. of GDP (in %) 26.78% 35.99% 35.74% 37.42% 37.25% 2.2.1. Growth (in p.p. %) 9.21% 8.97% 10.64% 10.48% 2.3. of Labor Output (in %) 39.04% 50.10% 50.10% 50.16% 50.16% 2.3.1. Growth (in p.p. %) 11.06% 11.06% 11.12% 11.12% 3. Sectoral Value Added 3.1. traditional, female (in USD) $647,951,018 $812,510,752 $833,158,308 $638,402,833 $657,565,149 3.1.1. Growth (in %) 25.40% 28.58% -1.47% 1.48% 3.2. modern, female (in USD) $436,439,531 $811,903,599 $753,280,333 $1,235,304,665 $1,184,089,763 3.2.1. Growth (in %) 86.03% 72.60% 183.04% 171.31% 3.3 traditional, male (in USD) $735,858,976 $809,219,818 $829,783,745 $634,376,428 $653,417,887 3.3.1. Growth (in %) 9.97% 12.76% -13.79% -11.20% 3.4. modern, male (in USD) $957,620,009 $808,615,124 $750,229,301 $1,227,513,601 $1,176,621,711 3.4.1. Growth (in %) -15.56% -21.66% 28.18% 22.87% Source: Authors’ simulation based on expanded version of CKW 2017 model. 39 Table 5.4. Non-Tradable Production (NTP) Output Complete Status-Quo Convergence Total NTP (in USD) $780,487,162 $880,138,323 As share of GDP, if added (in %) 16.16% 14.93% GDP, if NTP added (in USD) $4,830,231,798 $5,887,610,953 Growth (in %) 19.27% 45.38% Female NTP (in USD) $530,209,334 $597,905,484 Female share of NTP (in %) 67.93% 67.93% Male NTP (in USD) $250,277,828 $282,232,839 Male share of NTP (in %) 32.07% 32.07% Source: Authors’ simulation based on expanded version of CKW 2017 model. Notes: NTP stands for Non-Tradable Production. Complete Convergence refer to the value of the non- tradable production if the Model 7.3 would be applied. 40