Version classiqueVersion mobile

Who Cares?

Aatif Somji

2. Literature Review

Texte intégral

Gender Gaps in Labour Market Outcomes

1The motivation for this research stems from my passion for women’s economic empowerment and my firm belief in its potential to redress wider, gender-based inequalities across society. This inspired me to explore the complex relationship between gender and labour market outcomes in greater detail, and better understand the factors underlying this relationship.

2Gender gaps – differences in outcomes between men and women – present themselves in a number of ways across labour markets, a pattern that transcends economic and geographic boundaries (ILO 2016). Globally, women are less likely than men to participate in the labour market: their participation rate of 48.5% stands at almost 27 percentage points below that of men (ILO 2018a). Among those who actively participate in the labour market, the rate of unemployment for women is higher than for men in almost all countries (ILO 2018a). Occupational segregation, meanwhile, sees those women who are employed overrepresented in the lowest paid jobs (ILO 2016). Finally, the global gender wage gap – the aggregate difference in mean wage between men and women – is estimated at 23 per cent, meaning that on average women earn only 77 per cent of what men earn (ILO 2016).

3There is a complexity of potential explanatory factors for these gender gaps in labour market outcomes, including: level of education, occupational choice, flexibility of working hours, gender-based preferences for work-life balance, and discriminatory social institutions (Ferrant et al. 2014). Many of these factors could be endogenous: decisions taken at the individual level (such as what level of education to pursue, what industry to work in, or how many hours to work) may already be a result of internalised expectations about gender gaps, which in turn reinforce themselves (Beblo et al. 2003). However, even when a large number of observable characteristics are controlled for, including many of those mentioned above, gender gaps still exist. Women earn less than men even when they are equally educated, graduated in the same field, have the same number of years’ experience and work in the same type of job (UN 2015). The puzzle as to what is driving this ‘unexplained’ gender gap remains to be fully understood.

4Microenterprises are a common feature across developing countries, in large part due to the limited employment opportunities within the formal sector (Leino 2009). Across Sub-Saharan Africa, for example, the proportion of active workers in informal employment stands at over 90 per cent – with the vast majority of these people self-employed as micro-entrepreneurs (ILO 2018b). An interesting avenue of research for those concerned with issues of gender, labour and development, therefore, is to explore the extent to which gender gaps exist between male- and female-owned microenterprises, and why they may occur. This intersection between gender and microenterprise forms the basis of this research.

5The literature review is divided into three sections. The first section examines the incidence of gender gaps within the microenterprise context, focusing on the economic literature that uses experimental techniques in an attempt to overcome the endogeneity problems described above. It details how various studies have attempted to measure gender gaps, the extent to which these gaps exist, and possible factors that could be causing them. The second section examines unpaid care and domestic work as a potential explanatory factor for gender gaps in labour market outcomes more broadly. It draws from a multidisciplinary, feminist literature to outline what unpaid care and domestic work is, examine its unequal distribution between women and men, explore how this could be contributing to gender gaps, and finally propose concrete steps that can be taken to address this. The final section combines these two analyses, applying the feminist lens of unpaid care and domestic work to the phenomenon of gender gaps in microenterprise. This forms the theoretical framework through which the research question will be addressed, providing an alternative approach to existing economic studies and making a small, original contribution to the wider literature on the topic.

Gender Gaps in Microenterprise

6A nascent economic literature explores gender gaps specifically among microenterprises. Much of this research uses experimental techniques in order to establish causal relationships that go beyond simple association. This section of the literature review will focus on the economic literature on this topic, which identify the presence of gender gaps and explore possible reasons why these gender gaps might exist.

7The research conducted by Suresh de Mel, David McKenzie and Christopher Woodruff (2009) can be seen as the seminal study on gender and microenterprise. It stems from their original work (2008), where they used an innovative approach to estimate returns to capital among Sri Lankan microenterprises. Previous studies exploring returns to capital in the microenterprise context tended to focus on microcredit clients, which generates potential selection biases both on the supply and demand side: microfinance institutions make decisions about who to lend to based on specific selection criteria; similarly, micro-entrepreneurs make the decision about whether to seek credit in the first instance, and if so whether to take up this credit at the given rate of interest. The authors thus overcome this bias by widening the target population to all microenterprises, not only those that apply for credit. They provided a randomly allocated positive capital shock of $100 or $200 to these businesses and observed how their profits changed as a result. The random allocation of the grants meant that there should be no correlation between receiving the treatment and other factors that could influence the profitability of these firms.

8On aggregate, they found that average real returns to capital are very high – roughly five per cent per month, or 60% per annum. More importantly, these returns are substantially higher than the market interest rates on loans charged by banks and microfinance institutions. These results suggest that the microenterprises studied are credit constrained: they are able to achieve marginal returns that are on average four times the market rate of interest. Economic theory thus suggests that if these businesses were able to access credit, they would do so until the marginal return of taking out the loan was equal to its interest rate. Therefore, a primary conclusion that can be drawn from the study is that micro-entrepreneurs simply lack the access to credit needed in order to develop their business.

9However, after disaggregating their data according to sex, the researchers found a stark contrast between male- and female-owned enterprises, casting doubt on this primary conclusion. Mean real returns to capital for men were estimated at 11 per cent per month, a finding that is statistically significant at the five per cent level. Meanwhile, the corresponding estimate for women was slightly negative and not statistically different from zero. This empirical result is puzzling for two reasons. Firstly, the fact that average returns are much lower for women than for men appears to go against the common assumption that women are more credit constrained – for instance due to their relatively limited access to economic and social mobility or lack of physical collateral (e.g. Khandker 1998). Secondly, it is unclear why female micro-entrepreneurs are generating zero returns from a positive capital shock.

10Several other economic studies have since explored the gendered effect that an increase in financial capital can have on the business outcomes of microenterprises, with broadly similar findings. Fafchamps et al. (2014) replicated the study of de Mel et al. (2008) in Ghana, randomly providing grants to male and female micro-entrepreneurs. For women running subsistence enterprises, they found that the grant had no effect on business profits. Berge et al. (2015) also found that providing grants to female micro-entrepreneurs in Tanzania had no effect on their business profits. Finally, Fiala (2018) randomly allocated subsidised loans to male and female micro-entrepreneurs in Uganda, and noted a strong, positive effect on business profits for the male group but no effect for the female group.

11Overall, the evidence from the economic literature appears to corroborate the two key findings of de Mel et al. (2009). Gender gaps consistently present themselves in returns to capital for microenterprises, and there is no statistically significant effect of positive capital shocks on the business outcomes of female-owned enterprises. This suggests that addressing only credit constraints is not enough to help poor women grow their business.

12Given that credit alone is not a sufficient condition for female microenterprise development, it is necessary to explore other possible constraints from across the literature. Alternative hypotheses for the persistent gender gap in microenterprise returns can be grouped into four categories: individual, enterprise, household and society.


13At the individual level, female micro-entrepreneurs may lack sufficient entrepreneurial ability to yield positive returns to capital. Prima facie, this seems unlikely as there is no reason to believe that women are intrinsically worse at doing business than men, or that they are being deprived of opportunities available to men to improve their business skills. To test the effect of human as well as financial capital on microenterprise development, Berge et al. (2015) and Fiala (2018) randomly allocated business training to micro-entrepreneurs. Both studies found that a combined intervention of human and financial capital had a large positive effect on the profits of male entrepreneurs but no effect on those of females, suggesting that entrepreneurial ability is unlikely to be driving the gender gap in microenterprise returns.

14Differences between men and women regarding attitudes towards risk and competition could also potentially explain the observed gender gap. De Mel et al. (2009) played a monetary incentivised lottery game with firm owners to elicit a measure of their risk aversion, and found no evidence that this is influencing the gender gap in returns. Berge et al. (2015) also used a game with monetary incentives to measure willingness to compete, finding that women are generally more competition averse than men. Moreover, their data indicates a positive correlation between willingness to compete and business profits, suggesting that competitiveness could be an important factor for entrepreneurial success, which women in general may lack.


15At the enterprise level, male and female micro-entrepreneurs may be self-selecting into very different industries, which in turn could explain the gender gap in returns. Suggestive evidence from de Mel et al. (2009), Berge et al. (2015) and Fiala (2018) all indicate occupational segregation along gender lines. For instance, Berge et al. (2015) show, at the baseline of their study, that there are statistically significant gender differences across sectors, with women more likely to be in the service sector and men more likely to work in manufacturing. De Mel et al. (2009) also investigated this possibility, and found that as the proportion of females in a sector increases, investment levels and returns to capital both decrease. Given that there does not appear to be a straightforward explanation as to why female-dominated sectors intrinsically yield lower returns, the authors explore how the proportion of females in a sector could be the proxy for other constraints to microenterprise development – most notably geography.

16Regarding geography, 74% of female-owned enterprises in the de Mel et al. (2009) sample are home-based, compared to 52% of those owned by males. Moreover, almost half of the female-owned businesses have all their customers within a one-kilometre radius of their business, with the corresponding figure for male-owned businesses estimated at 30%. The authors control for these various geographical constraints and conclude that returns to capital are still negatively associated with the proportion of females in the sector. Thus, while the mechanisms through which sectoral decisions affect returns to capital remain unclear, occupational segregation appears to partially explain the gender gap in returns to capital – though a gender difference remains even after accounting for this.


17At the household level, women may be channelling the positive capital shock they receive away from their business and towards the household. The Sri Lanka study (de Mel et al. 2009) finds that women do not invest any of the smaller treatment amount into their business. Similarly, Fafchamps et al. (2014) suggest that Ghanaian women running businesses with low initial profits, comparable to the entire sample of women in the Sri Lanka study, spend most of their grant on household expenditure. Finally, Berge et al. (2015) find that Tanzanian women who are randomly assigned a business grant receive less from their husband towards household expenditure, suggesting a crowding-out effect that the women may fill with their own business income.

18However, two factors suggest that household expenditure may not be driving the gender gap in microenterprise returns. Firstly, business outcomes are broadly similar for cash and in-kind treatments, despite the in-kind grants being more difficult to liquidate. Fafchamps et al. (2014) provide both modalities as part of their research and find no statistically significant effect of either of them on women with low-profit businesses. Secondly, de Mel et al. (2009) find that for women who receive the smaller treatment, which is seemingly not invested in the business, there is no statistically significant effect on monthly household expenditure. Moreover, women receiving the larger treatment amount in their experiment actually invest more in their business than men – but still appear to generate zero returns. Therefore, while women may be more likely to spend their business grant on household expenditure, this does not seem to provide a strong explanation for the gender gap.

19An alternative possibility from within the household is that of spousal capture. De Mel et al. (2009) suggest that fear of spousal capture could lead women to protect their grant by investing it in highly illiquid assets, irrespective of the returns these may generate. They explore this further by estimating how investment decisions and profits for women vary with empowerment – measured through a series of questions on decision-making power within the household. Focusing on those who invest the grant into their business, the authors found that empowerment increases investment in inventories – which are generally more liquid and therefore easier to capture than fixed capital. They also found a significant, positive effect of empowerment on profits. Berge et al. (2015) likewise explored the possibility of spousal capture influencing the gender gap in returns to microenterprises. They conducted an incentivised lottery experiment to test this, finding that greater fear of spousal capture is negatively associated with business profits for women. Together, these results provide suggestive evidence that spousal capture, or at least the fear of it, may be influencing women’s investment decisions and contributing to the gender gap in returns to microenterprise.

20The occupational composition of the household could also provide an explanation for the gender gap. Bernhardt et al. (2017) hypothesise that the low returns for female-owned microenterprises are due to the fact that male and female micro-entrepreneurs often belong to the same household. They propose an Enterprise Household Model, where multiple enterprise households rationally allocate capital towards the business with the higher returns – with women’s capital often invested into their husband’s business as a result. The authors test this model using data from the de Mel et al. (2008) study. They find that the positive capital shock – which had no impact on profits for the full sample of women – leads to a statistically significant seven per cent increase in profits among women who are the sole entrepreneur in their household. Meanwhile, they observe an increase in aggregate household income for the entire sample of female entrepreneurs receiving the positive capital shock, suggesting that women in multiple-enterprise households invest the extra capital in their husband’s business. While the empirical data appears to fit the Enterprise Household Model, the authors’ argument is flawed in that it is unable to explain the gender gap in returns to micro-entrepreneurs. They posit that this is driven by women in multiple-enterprise households rationally allocating capital to their husband’s business, due to the latter’s higher returns. But this does not explain why these male-owned enterprises are likely to have higher returns than those of their wife in the first place.

21An alternative explanation to the Enterprise Household Model is that being female and the sole entrepreneur in the household is a proxy for being a single woman (i.e. unmarried, separated, divorced, widowed). This is quite feasible given the high incidence of micro-entrepreneurs in most developing countries (ILO 2018b). On the basis of this assumption, there are two corollaries that could help to explain the gender gap in returns to microenterprises. First, there could be a difference in business strategy – sector choice, level of investment – based on whether or not a woman is the sole person responsible for providing for herself and her family. Second, being single would eliminate the possibility of spousal capture – established as a likely contributor to the gender gap. Without this, women may feel free to invest more efficiently in their business and therefore obtain greater returns. In sum, the Enterprise Household Model proposes an innovative way of approaching the issue of gender gaps in microenterprises but lacks explanatory power. Further research is therefore required to understand the specific mechanisms through which these observed effects occur.


22At the society level, social norms may strongly influence men and women in different ways, leading to significant heterogeneity in returns to capital. During qualitative interviews, women often express their strongly defined roles within the household and community, such as being responsible for childcare and household chores (e.g. Fiala 2018). Empirical evidence from Field, Jayachandran and Pande (2010) demonstrates the importance of social norms within the microenterprise context. Using an experimental approach, they explored the effect of traditional religious and caste institutions on entrepreneurship in India. They found that the most restricted social group did not respond to their business training intervention, despite positive effects among those with fewer restrictions, highlighting the importance of social constraints to enterprise development.

23While the concept of social norms is nebulous and therefore difficult to measure, it is plausible that these could be driving many of the possible reasons for the gender gap in returns to microenterprises. For instance, gendered social norms may dictate what is a socially acceptable sector to work in, which would explain the occupational segregation that potentially contributes to gender differentials in returns. Alternatively, a societal expectation of femininity being equated with submission to one’s husband could explain spousal capture, and why women who are more empowered therefore appear to generate significant profits. Similarly, gender norms around femininity could discourage women entrepreneurs from being as competitive as their male counterparts. Finally, social norms around the distribution of household labour between men and women could limit the ability of the latter to dedicate sufficient time – and by extension cognitive effort (Mani et al 2013) – to their business.

24To sum up, the economic literature on gender and microenterprise strongly indicate that women face multiple constraints to enterprise development, beyond access to credit, which operate at multiple levels. The aggregated findings above demonstrate the complexity of this issue. There is no single definitive explanation for the gender gap, rather a combination of many interlinked factors including attitudes towards competition, occupational segregation and spousal capture. Social norms around gender may provide the common thread for these possible explanations. The concept should therefore be unpacked further to potentially reveal additional insights into female-specific constraints to microenterprise development.

Unpaid Care and Domestic Work

25One avenue through which social norms may influence gender roles and constrain female microenterprise development is the distribution of unpaid care and domestic work. This can be understood as all unpaid services provided by individuals within the household and community for the benefit of its members, including care of persons, housework and voluntary community work (Elson 2000). Common examples include cooking, washing, cleaning, looking after children and caring for elderly, sick, or less able dependents. To deconstruct the phrase, unpaid care and domestic work is a form of work as it involves activities requiring time and effort, it is care as it helps to sustain or develop a decent standard of living, it is domestic as it is largely carried out within the home, and it is unpaid as those carrying out these activities are not remunerated (Elson 2000). It is clear from this definition that unpaid care and domestic work is essential for providing for individuals, families and communities, and can be regarded as the foundation upon which the market economy functions (Collas-Monsod 2011). It is for this reason that unpaid work is often understood as a crucial dimension of social reproduction (e.g. Benería 1979; Folbre 2014).

26There are strong underlying gender dimensions to unpaid care and domestic work, hereafter referred to simply as unpaid work. Across the world, women and girls carry out the majority of this work. According to global time-use data from the Organisation for Economic Co-operation and Development, women spend on average two to ten times more time on unpaid work than men (Ferrant et al. 2014). Complementary data from the United Nations shows similar findings. In developed countries, women spend on average 4 hours, 20 minutes per day on unpaid work while men spend 2 hours, 16 minutes. This inequality is even more acute in developing countries, where women spend on average 4 hours, 30 minutes per day on unpaid work and men spend just 1 hour, 20 minutes (UN 2015). At the macro level, then, poverty appears to be associated with a significant increase in both the absolute amount of time women spend on unpaid work and the relative proportion of unpaid work assumed by them. Finally, it is important to note that this gender imbalance in unpaid work starts early. Worldwide, girls aged five to nine spend 30 per cent more time helping around the house than boys, with this figure rising to 50 per cent for those aged ten to fourteen (UNICEF 2016).

27Gender inequality in unpaid work is associated with gender gaps in numerous labour market outcomes. Globally, gender inequality in the amount of time devoted to unpaid work is negatively correlated with gender inequality in labour force participation – even after controlling for many other variables including GDP per capita, fertility rate, urbanisation rate, maternity leave and gender inequality in unemployment and education (Ferrant et al. 2014). Similarly, gender gaps in unpaid work are linked to gender wage gaps. In countries where women spend disproportionately more time on unpaid work, the gender gap in hourly wage is also higher – despite controlling for female labour force participation and unemployment along with the previously mentioned variables (Ferrant et al. 2014). Overall, whilst unable to show a direction of causality, these findings demonstrate the clear relationship that exists between gender gaps in unpaid care work and gender gaps in labour market outcomes. Intuitively, unpaid work constrains the total amount of possible time that can be dedicated to market work. An interesting empirical finding from extrapolating the global data is that full gender equality in unpaid care work corresponds to a predicted female labour force participation of 50% of the total labour force (Ferrant et al. 2014). It appears that, to achieve equality in paid work, women also need to achieve equality in unpaid work.

28What can be done to redress the unequal distribution of unpaid work between women and men? Diane Elson (2017) summarises the strategies that can help to achieve this as recognising, reducing and redistributing unpaid work.

29Recognising unpaid work means understanding how this work underpins economies and valuing it accordingly. The first step towards this is to measure the extent of unpaid work through time-use surveys, which would help to make its contribution more visible (Benería et al. 2016). Next, the economic value of these contributions can be calculated by aggregating the total time spent on different activities and multiplying this by the cost of this time.

30Three methodological challenges render this strategy difficult to carry out in practice. Firstly, collecting time-use data is labour intensive, requiring significant effort on the part of the researcher and the respondent. Secondly, assigning value to non-market work can be ambiguous – for instance, it could be calculated using the market price of any output created, or instead by using a monetary value of the time taken to do this work, imputed either through the replacement or opportunity cost (i.e. the market wage of getting someone else to do this activity or the market wage of the person actually doing the activity) (Ferrant et al. 2014). Finally, the majority of unpaid work falls outside of the production boundary of the System of National Accounts, the internationally-agreed set of recommendations on measures of economic activity (Hirway 2015). Countries may therefore lack incentives to spend limited funds on collecting this data, despite its potential to inform policies aimed at promoting gender equality.

31Reducing unpaid work would free up time for caregivers to pursue other activities, including paid work. A reduction in unpaid work can be achieved through time-saving technology, physical infrastructure and social infrastructure (Elson 2017). Time-saving technology can help reduce the amount of time spent on unpaid work, for instance through fuel-efficient stoves which speed up the cooking process and minimise the need to collect fuel wood in developing countries (Hirway 2015). Investment in physical infrastructure such as access to a clean water supply, sanitation, electricity and public transport can significantly reduce unpaid work, while relevant social infrastructure includes the formal provision of care services for children and the elderly. Limited access to time-saving technology, physical and social infrastructure in poorer economies is likely to exacerbate the amount of unpaid work undertaken in these countries (ADB 2015), as alluded to previously in the time-use data (UN 2015).

32Three examples demonstrate how reducing unpaid work can potentially improve labour market outcomes. Ilahi and Grimard (2000) investigate how water infrastructure affects the time allocation of women in Pakistan. They find that improvements in the public provision of water are negatively associated with the time women spend collecting water and positively associated with time allocated to income-generating activities. Dinkelman (2011) analyses a rural electrification programme in South Africa to estimate its impact on employment growth. She finds positive effects on female labour supply on the extensive and intensive margin: female employment significantly rises by nine percentage points in the wake of electrification and women spend almost nine hours more per week in paid work. The Estancias Infantiles para Apoyar a Madres Trabajadoras (Child Crèches to Support Working Mothers) programme in Mexico was created with the specific aim of addressing labour market inequalities resulting from women’s unpaid care and domestic responsibilities. It provides childcare subsidies to mothers and single fathers who are working, seeking employment or studying (Holmes and Jones 2013). Angeles et al. (2014) conduct an impact evaluation of the programme, finding a statistically significant positive effect of the subsidised crèche facilities on the rate of female employment and number of hours spent by women in paid employment – similar to the previous results.

33While recognising and reducing unpaid work can be beneficial for those who undertake this work, gender equality requires that residual care duties are redistributed more equitably between men and women (Elson 2017). Government policies can promote this, for example through a more equal provision of paid maternity and paternity leave. This could encourage fathers to play a greater role in unpaid care and domestic responsibilities while at the same time reducing the disincentive for employers to hire women. Similarly, flexible working conditions could enable parents to better balance their paid and unpaid work (Ferrant et al. 2014). These potential solutions seem particularly suited to high-income countries. More broadly, redistributing unpaid work is likely to require changing social norms on masculinity and femininity – challenging the prevailing narrative of men as breadwinners and women as caregivers (e.g. Budlender 2010; Doyle et al. 2014).

Unpaid Work and Microenterprises: A Contribution to the Literature

34Returning the focus specifically to the microenterprise context, there are two key mechanisms through which the gendered distribution of unpaid work could be influencing gender gaps in microenterprise returns.

35A footnote in the de Mel et al. (2009) paper describes how women were more likely to report entering self-employment in order to have the flexibility to care for children or elderly parents. These women’s unpaid care and domestic responsibilities therefore appear to have driven their decision to become a micro-entrepreneur in the first place. Moreover, these responsibilities are likely to influence other crucial business decisions, including the choice of sector and geographical location of the enterprise. The economic literature review on microenterprise returns suggests that occupational segregation could be an important explanatory factor for gender gaps. The unpaid care and domestic responsibilities of women entrepreneurs could to some extent be constraining their full choice-set regarding sector choice and other business decisions, potentially contributing to gender gaps in microenterprise returns.

36Beyond its ability to constrain the decision-making abilities of women micro-entrepreneurs, unpaid care and domestic work takes time. A nuanced appreciation of these demands on micro-entrepreneurs, inspired by the feminist literature review, suggests that it easily diverts attention away from the business. This diverted time is extremely difficult to measure, as it tends to be short, sporadic, and often during working hours. The blurring of activities between paid and unpaid work is commonplace within the microenterprise context – such as women who tend to their children while selling their products (Folbre 2014). A significant critique of the economic literature therefore lies in how the time spent on paid work is calculated. Rudimentary self-reported measures, such as those used by de Mel et al. (2008) may fail to take into consideration that unpaid work is often done alongside paid work – especially so for women whose business is home-based or very close to the home. If this is the case, the self-reported figures on hours of paid work could be overstating the true amount of time dedicated specifically to this. As labour is an essential component of the production function, overstating this figure for women could in some way explain the gender gaps in returns to capital found in many studies (e.g. de Mel et al. 2009; Fafchamps et al. 2014; Fiala 2018).

37This research seeks to apply the feminist framework of unpaid care and domestic work to the microenterprise context. The central research question examined throughout this thesis is whether unpaid care and domestic work is a significant constraint to female microenterprise development. Establishing this requires accurately calculating how much time is being dedicated to unpaid care and domestic work by women micro-entrepreneurs, assessing whether this work is a key constraint to enterprise development, and investigating whether these responsibilities are specific to women. The focus on unpaid work provides an alternative explanatory approach to the existing economic literature on gender gaps in microenterprise. By situating this research at the intersection of economic and feminist literature, it is hoped that it can make a small, original contribution to the wider academic literature on gender and microenterprise. The next chapter explores the specific context in which this research question was addressed.


Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search