PROTOCOL: The impacts of interventions for female economic empowerment at the community level on human development: a systematic review of the evidence in low- and middle-income countries

1.1 The problem

In many places of the world, women remain at a disadvantage to men. Women globally earn 24 per cent less than men for doing the same work and 75 per cent of women's work in developing countries is in the informal sector (World Economic Forum, 2015). Gender inequality in access to economic opportunities, a lack of voice and participation as well as limited opportunities to accumulate wealth could perpetuate vulnerability to poverty among women, limit human capital accumulation in the new generation and restrict economic growth.

With this view, the UN Millennium Development Goals (MDGs) and the Sustainable Development Goals (SDGs) have recognized that improving female economic empowerment is not only valuable in its own right but is also a key element to end poverty and boost economic prosperity (UN, 2000; UN, 2015). Based on this premise, international organizations, local governments, non-profit organizations and private companies have engaged in various initiatives that aim to decrease gender inequality and foster female empowerment. For instance, in 2016 the Bill and Melinda Gates Foundation contributed $80 million US dollars to help accelerate the progress of women and girls around the world.

Empowerment in a broader sense is defined as the expansion of people's (both men and women) ability to make strategic choices in a context where this ability was previously denied to them (Kabeer, 2012). To the extent that freedom of choice depends on economic, social, political and psychological aspects, empowerment is a multidimensional concept. Yet, one aspect that is considered to be crucial in increasing the ability of individuals to control their lives and exert influence in society is economic empowerment – defined as increased access to - and control over - resources and economic lives.

Given the importance of economic freedom to promote development and growth, this systematic review focuses on interventions that aim at fostering female economic empowerment. There is a wide range of interventions that have been developed to foster female economic empowerment. A useful classification is provided by Malhotra and Schuler (2005) who distinguish three dimensions of economic empowerment; namely, at the household level, community and national level. Household economic empowerment is related to the control of household income. Therefore, this dimension considers interventions that foster gender equity in rights to own and inherit land. At the community level, it refers to access to employment, credit, representation in trade associations and access to markets. It considers for example initiatives that give in-kind grants to women to establish an agricultural or micro-entrepreneurial activity, initiatives that give women access to labour markets by providing child or care for the elderly, or by opening short-term employment possibilities for women. At the national level, the focus is on female representation in high-paying jobs, the proportion of female CEOs and the representation of women's economic interests in macroeconomic policies and federal budgets. In this review, we focus on interventions that foster female economic empowerment at the community level and consider their impact on human development.

Human development is defined as the process of enlarging people's freedoms and opportunities and improving their well-being. Traditionally, this term has been associated with increases in per-capita income or the growth rate. Yet, recognizing that the quality of life not only depends on a monetary dimension, a more recent approach is the capabilities approach that considers the ability that an individual has to freely choose the life they want to live (Nussbaum and Sen, 1993). This approach recognizes that there are certain functional capabilities that allow a decent quality of life (Nussbaum, 2000). Following this conceptualization, and an approach similar to the UN human development reports, we consider two dimensions of human development. First, we consider a dimension of wealth and take into account household income-per-capita, assets and access to basic services. Second, we consider three capabilities: i) the capability to live a life of normal length with good physical and mental health, protected from violence and enjoying recreational activities; ii) the capability to think, reason and engage in critical reflection and iii) the capability to control one's environment by participating in decisions in the household and in the community.

While previous evidence suggests that diverse interventions have been successful in reducing gender inequality and fostering female empowerment (Dekker, 2013; Buvinic, et. al., 2013), there is relatively little evidence of their impact on human development. The evidence on the effects of female economic empowerment on development is relatively scarce (Morrison et al., 2010; Duflo, 2012 and Bandiera and Natraj, 2013). We build on previous work by doing a systematic review of studies from developing countries that consider a variety of interventions that aim at providing women access to income generation alternatives and their impact on human development (i.e. support for agricultural and micro-entrepreneurial activities, access to labour market and formal financial markets).

1.2 The Intervention

Programs that focus on female economic empowerment are very diverse both in terms of the intervention and the dimension of empowerment that they address (Buvinic, et. al., 2013). We adopt the conceptualization of Malhotra and Schuler (2005) and classify interventions according to the context in which they take place. Table 1 presents the indicators of female economic empowerment in each context and classifies the interventions according to whether they are covered or not in the present systematic review.

In this review, we only consider interventions that aim to increase female economic empowerment at the community level by providing alternatives for income generation. We consider interventions that: 1.

Foster female productivity in agricultural activities. We focus on interventions that a) give women access to inputs and new technologies (i.e. improved seeds, fertilizer, farmer field schools), b) support access to final markets (i.e. promote farmer associations, or provide information on prices) and c) provide subsidies to connect to existing infrastructure (i.e. electricity, irrigation).

In this review, we focus on interventions that attempt to directly reduce or eliminate the challenges that women face, which prevent them from participating in economic activities. Therefore, we do not consider interventions that could indirectly affect female labour force participation as conditional cash transfers (Leroy et al. 2009 and Yoong et al. 2012; Kabeer et al.2012); education, vocational and financial literacy, entrepreneurship training programs —including field schools (Miller et al. 2014; Tripney and Hombrados, 2013, Woodruff and McKenzie, 2014; Chinen et al. 2016); diverse types of information campaigns (i.e. Aker, 2010); interventions as certification to firms for good practice in hiring and retaining female employees, consulting and mentorship programs for women (Feigenberg, et al., 2013; Field, et al., 2016) or interventions that aim to reduce domestic violence against women (Day, 2009).

Finally, in the analysis, we do not consider interventions that aim to empower women at the national and multinational level by increasing representation among CEOs or decision-making positions in government, through political reservations or within the judiciary system (Dahlerup, 2013; Schmitt, 2015). We also do not consider interventions that aim to give women access to resources in the household. We, therefore, exclude interventions that give men and women equal rights under the law (i.e. land rights and land titling, inheritance laws and law on divorce settlements: marital property, child support, custody and alimony).

1.3 How the intervention might work

Improving the economic position of women by providing them with access to income generating opportunities at the community level is likely to reduce poverty and generate greater human development within the household (Morrison et al., 2010). Annex 1 presents the theory of change that provides a base for our analysis.

The main objective of female economic empowerment interventions is to increase the possibility for women to generate income. Income in the hand of women can affect human development by different channels. First, the most direct effect of female economic empowerment is that this will increase household's income allowing higher consumption which will be reflected in increased well-being. Second, it is expected economic opportunities for women will affect female's bargaining power within the household (Allendorf, 2007; Doss, 2013). This is expected to improve health outcomes of women (Beegle et al., 2001; Li and Wu, 2011), reduce intimate partner violence (Panda and Agarwal, 2005 and Rao, 1997) and have positive intergenerational effects. Increase female bargaining power has been shown to be associated with a larger share of budget used on food and on children's education and health (Quisumbing and Maluccio, 2003; Duflo and Udry, 2004; Doss, 2006; Robinson, 2012). Evidence also shows that income in the hands of women is associated with increased intergenerational welfare (Schultz, 2002; Rubalcava et al, 2009). Moreover, income in the hands of women has been shown to increase investment in girls, reducing future gender inequality (Duflo, 2003). The third indirect channel by which female economic empowerment could affect human development is related to fertility decisions. As the opportunity cost of time for women increases, it is expected that fertility rate drops and that income per capita, savings and asset accumulation increase (Schuler and Hashemi, 1994; Dupas and Robinson, 2009).

Interventions that provide financial access are expected to relax credit constraints that households face to borrow capital for their enterprises. This would be reflected in more investments in enterprises, adoption of new technologies and in higher productivity (Morrison et al., 2010). On the other hand, access to credit is expected to reduce the negative effect of unanticipated shocks reducing households’ vulnerability to poverty. Interventions that foster saving groups are also expected to lead to an increase in asset accumulation and have a positive impact on children's well-being (Pitt et al., 2003, Cheston, 2006, Dupas and Robinson, 2013).

While female economic empowerment could lead to a decrease in poverty and higher human development, it is not clear if this is the case. There are many contextual factors that may limit the possibility of an impact on human development. Even if women have access to resources this does not necessarily guarantee their control over their use. Social barriers might prevent women from controlling these resources. For example, social norms on the acceptable role of women in society might limit the mobility of women and the possibilities to participate in economic activities. Moreover, low initial bargaining power within the household might imply that men control the resources generated by women. For instance, Goldstein and Uldry (2008) show that males with higher status in Ghana were able to exploit this to their advantage using fallowing land. On the other hand, the systematic review by Vasseen et al. (2012) on the impacts of microfinance on a woman's control over spending, shows that the effect is relatively low and not significant, which is partly attributed to the demand of the social network including the husband.

Assuming that it is possible to enforce a woman's control over resources so that they make investment decisions, it is not clear if they would engage in more productive activities than men. Women might specialize in less profitable crops, such as food crops, foregoing the income generating opportunities stemming from cash crops (Croppenstedt et al. 2013). Alternatively, they could specialize in less productive enterprises that give them more flexibility to combine household responsibilities and working hours (Amin, 2010). Second, women might invest less in high-yield but riskier technologies. Udry (1996) shows that men are more likely to use agricultural inputs than women, which translates to differences in productivity. Similarly, experimental evidence shows that micro-loans directed at women were not associated with larger effects than those given to men (Karlan and Zinman, 2011). Different studies show that the return of in-kind grants is larger for male than female owned business (De Mel, et al., 2008 and Fafchamps et. al, 2011). Therefore, interventions that support female business have been criticized in that they enable less successful enterprises to survive longer (Woodruff and McKenzie, 2013). A critical aspect might be the lack of access to marketing opportunities.

Interventions that foster female labour force participation, through the provision of childcare or childcare subsidies, can also generate positive spill-over effects on children's nutrition and development (Ruel et al.,2006; Attanasio& Vera-Hernandez, 2004; Berlinski&Galiani, 2007). Another problem is that these interventions can increase employment, but at a lower salary (Todd, 2013) or generate negative spill-over effects on the jobs and wages of the nonbeneficiary population (Betcherman et al., 2004). A critical assumption of programs that rely on the promotion of female labour force participation is that they would not generate a negative effect on non-beneficiaries. Hence, it is important to try to capture general equilibrium effects. When available, we will extract information on the change in male employment that can be attributed to the intervention. Finally, the sustainability of these interventions is questioned as discrimination against women might continue to persist once the incentives have been removed (Groh et al 2012).

It is possible that female economic empowerment increases tensions within the household and the community. For example, women could be subject to the triple burden of work as they need to contribute income to the household and still be responsible for housework and childcare. This can lead to more stress and conflict within the household. (Ahmed 2005; Ahmed & Chowdhury 2001). As women gain access to income-generating alternatives, they may reduce their time with their children, resulting in worse performance of their children at school and hence an increased likelihood of their children dropping out of school, involvement in gangs or switching/transferring responsibilities of housework to younger daughters (Ruhm, 2008). As working mothers have less time available, they could switch to processed foods of high caloric value leading to nutrition problems (Cawley and Liu, 2012).

Increased access to labour market opportunities could also lead to an increase in domestic violence (Heath, 2014), though this correlation might disappear once endogeneity is controlled for (Lenze and Klasen, 2016). Participation in economic activities outside of the home could also make women more vulnerable to crime or could expose women to environmental hazards at work (Amaral et al. 2015).

In summary, the interventions for female economic empowerment that we consider in this review are expected to affect human development by the following three main channels. First, these interventions increase the opportunities for women to have economic independence by enabling them to generate their own income. Second, they generate a direct income effect in the form of higher household income. Third, they reduce the vulnerability of the households by reducing the variability of their income, either by the provision of insurance or by decreasing the covariance of a household's income.

However, various factors may prevent these effects from occurring. While it is difficult to capture these factors, we can use the proxies: region of analysis, religious beliefs and a measure of female participation in a particular sector before the intervention. To capture the effect of female control over resources, we use as an indicator for female bargaining power. Yet, this factor is considered in the analysis only if we have measures of human development.

A second factor that might restrict the success of the support programs for agriculture and entrepreneurship is financial access. Therefore, one needs to consider the financial situation of the beneficiaries before the program began, distance to financial institutions and the perceived financial access.

A third condition for the program to be successful is that the support programs lead to productive investments with good market access under reasonably good economic conditions. Hence, in the case of agricultural support programs, we need to consider if the beneficiaries invested in food or cash crops. Second, if they invested additional resources (i.e. inputs). Third, we need to consider how adverse effects might have affected the program (i.e. weather shocks, price shocks etc.) and lastly, consider if they work in a cooperative group or not because it may facilitate market access.

Finally, programs could have a positive effect reducing a household's vulnerability to poverty by providing more stable income. One channel by which this might occur is by decreasing the impact of negative shocks and the degree of covariance of a household's income. To the extent that this information is available, we would include it in the analysis. An intervention that is expected to last only a couple of months is expected to generate lower effects than long-term interventions. Therefore, we would need to consider the effect of the expected duration of the program in addition to the duration of exposure to an intervention.

1.4 Why it is important to do the review

Our motivation to undertake a systematic review is to enhance academic knowledge and provide useful insight for policy makers both at the country and international level. Many studies address the question of what works to promote female empowerment as well as female economic empowerment. For example, Brody et al. (2013) consider the impact of self-help groups; Woodruff and McKenzie (2013) business training programs; Cho and Honorati (2014) entrepreneurship programs; Bandiera et al. (2013) capital, training or a combination of both; Mehra et al (2013) financial services; Rodgers and Menon (2013) land rights; Doss et al (2013) and Knowles (2013) agricultural productivity; Todd (2013) employability and quality of work; Katz (2013) young employment; and Vaessan et al (2014) reviews the effects of microcredit. In our review, we pursue an additional step and assess which interventions of female empowerment have an impact on human development.

Previous studies have addressed the relation between gender equity and development (Duflo, 2012; Morrisson et. al, 2010) However, no study has used a compact measure of outcomes such as a human development indicator. In our analysis, we consider an aggregated measure of human development that has four dimensions: household income and assets, health outcomes, educational outcomes and attitudes towards gender equity. Our measure of outcome considers the aggregated effect at the household level.

Furthermore, we contribute to this line of research by focusing on Randomized Control Trials- RCTs and quasi-experiments that use relevant econometrics to explain the causal impact of an intervention. We pursue the use of a robust methodological approach which other reviews are unable to provide (with a few exceptions such as Brody et al., 2013; and Rodgers and Menon, 2013). For the search of interventions in developing countries, we follow a specific time frame and include studies that have been published between 1990 and 2017. In addition, the risk of selection bias is addressed by including different languages (English, French, German, Spanish), working papers as well as reports from particular programs in our search.

Our focus is to provide feasible policy relevant conclusions of the evaluated interventions. Female empowerment and gender equality is not only an important aim of the UNDP to achieve the Millennium Development goals (UNDP, 2014), but it has also been on the policy agenda of developing nations. Large scale programs, such as the Mahila Samakhiya in India, Business Development Schemes in Kenya, Uganda and Tanzania, are examples of the steps undertaken by various governments to improve female empowerment and development.

Through our review, we attempt to compare the relative impact of interventions across different interventions, studies and contexts. Depending on data availability, we will include information on the cost of an intervention, which could provide a base for comparison on the cost-effectiveness of the different interventions. In the absence of data, we will favour simple interventions that can be easily implemented, taken to scale and sustained even after the implementer has left. We build on the works of Buvinic et al. (2013): “Roadmap for Promoting Women's Economic Empowerment” and Dekker (2013): “Promoting gender equity and female empowerment.” While these studies consider the effectiveness of interventions on female economic empowerment (measured as increased productivity, wages, income and assets on the hand of women), in this review we consider the impact of these interventions on a very different of dimensions related to economic development.

3.1 Criteria for Including and Excluding Studies

Following Petticrew and Roberts (2006), we use the PICO model (Population, Intervention, Comparison and Outcome model) to define the inclusion and exclusion criteria.

3.2. Search strategy

3.3. Data Collection and Analysis

3.4. Statistical Procedures and Conventions

This section describes the methodology that we plan to follow to address the three research questions mentioned earlier. Specifically, for the first question, we aim to conduct meta-analysis and meta-regression to evaluate the effect of female empowerment on human development. The second question is basically to analyze the intermediate outcomes to identify the channels through which the intervention may affect human development. For this mediator analysis related to the second question, we will conduct meta-analysis and meta-regressions wherever possible considering data availability on mediator variables. If sufficient data is not available, then for both the first and second questions we will complement the quantitative synthesis with a narrative analysis of the effects. Our third question will be addressed through moderator analysis, sub-group analysis, and meta-regressions when possible. The following sub-sections elaborate on these methods further.