Introduction

For decades, women have been the focus of several development interventions since they were largely marginalized in society. In areas such as reproductive health, gender equality, gender-based violence, intimate partner violence, and even mainstream microfinance, emphasis was put on women. Increasingly, however, we are observing a shift towards engaging men in such interventions [See for example Subašić et al. (2018) for gender equality; Casey et al. (2018) for gender-based violence; Ruane-McAteer et al. (2019) for sexual and reproductive health; Flood (2011) for men’s violence against women]. There are increasing global efforts in this regard from major development actors. The United Nations for example through campaigns like the ‘HeForShe’ campaign recognizes the role men and boys must play to foster gender equality. The move towards male engagement in these different efforts stems from the realization that it may be difficult to achieve gender equality from development efforts if the focus is entirely on women and men are excluded. Since the marginalization of women is deeply rooted in discriminative gender norms inherent in several societies and men are normally the custodians of such norms, their role must be recognized if a change is to happen.

In this paper, we consider a large and growing development model - savings groups that are currently facing a similar trend of increased male focus. We investigate the influence of increasing male engagement on key aspects of the savings group’s operational model. Savings groups (SGs) are grassroots, community-managed financial institutions where members save collectively and borrow from pooled savings at an interest (Le Polain et al. 2018). Such institutions are flexible and robust to different environments (Mutebi et al. 2017; Ojong 2014) which makes them very popular among development actors like CARE International, Catholic Relief Services, etc. as a mode to reach largely excluded populations. Current estimates show that there are over 14 million members in savings groups globally (SEEP Network 2016). At inception, the model had a particular emphasis on women. For example, the first structured savings groups model, the Village Savings and Loan Associations, pioneered by CARE International in the 1990s started by reaching out to only women. Through the Mata Masu Dubara (meaning women on the move) groups, the goal was to drive the socio-economic empowerment of women (Grant and Allen 2002). Although women still make up around 80 percent of savings groups’ membership (Wheaton 2019), available data shows that male membership is increasing over time. In the sample used for this study, for example, the share of male members increased from about 9% in 2010 to over 27% in 2017. However, very little is known about the influence that this increasing male engagement has on the SG core operational model.

We shed light on the influence that male engagement has on savings groups. Specifically, we investigate the following research question: what impact does male membership have on the savings group’s profit-generating capacity? The SG model requires efficient handling of savings and borrowing activities to create wealth for the group members and for the sustainability of the groups. The group profit-generating capacity is the best way to measure whether the group is operating efficiently. Some scholars argue in favor of member homogeneity across certain elements within the group arguing that when people are similar, it is easier to enforce rules making transaction costs much lower in such groups (Cassidy and Fafchamps 2020; Slover 1991). On the contrary, heterogeneity may bring people of dissimilar interests together (Nagarajan et al. 1999) which can affect group functioning and performance in general. For example, men are usually portrayed as more self-interested whereas women usually tend towards collective interests (D'Espallier et al. 2013; Guérin 2011; Johnson 2004).

We use a novel data set covering 81,853 savings groups from 30 countries in the period 2010 to 2017. This data is obtained from the Savings Groups Information Exchange (SAVIX), the first platform of its kind to have comprehensive data on these informal financial associations. SG profit-generating capacity is measured using the return on savings that captures the wealth creation for the group and members’ financial outcome from group participation. The results indicate that as the percentage of male members in the group increases, the profit-generating capacity reduces. Additionally, we investigate whether the relation between gender composition and group profit-generating capacity is influenced by the level of gender equality in the country in which the group is located. Countries with low levels of gender equality are usually characterized by discriminatory gender norms that could curtail the ability of men and women to interact and collaborate in groups. For example, in certain societies, it is even prohibited for women to interact with men, especially in public spaces (Lata et al. 2020). Indeed, the results show that gender inequality in the countries where these groups are located worsens the negative relation between gender composition and SG’s profit-generating capacity.

This study is conceptually rooted in a larger body of work that investigates increasing male engagement in development interventions that previously focused on women. Several scholars have investigated male engagement albeit the effect of this has been mostly investigated on an individual and household level (Slegh et al. 2013; Stern et al. 2015). We shift the unit of analysis from the individual to the group level. By doing so, the research contributes to a broader body of knowledge and guides policy on engaging men in savings groups without harming the core aspects of the group.

Additionally, the paper adds to the evidence regarding savings group composition by investigating the gender composition question. Burlando and Canidio (2017) look at the socioeconomic status of the group members and how this influences the amount of savings collected and the loans disbursed. Cassidy and Fafchamps (2020) investigate how sorting of group members along certain characteristics like occupation may influence the financial intermediation capability of savings groups.

The rest of the paper is organized as follows: section "Relevant Literature and theoretical framework" presents the relevant literature; section "Method" describes the data and the empirical strategy used in the paper. In section "Results", the results are presented and discussed, and section "Discussion of Findings" concludes the paper.

Relevant Literature and Theoretical Framework

Background on Savings Groups

Savings groups are community-based financial institutions composed of 15-30 members that pool money in a common fund and borrow from the fund at an interest (Le Polain et al. 2018; Burlando and Canidio 2017). They are typically informal associations that are built on a foundation of cooperation (Cassidy and Fafchamps 2020). The savings group methodology is based on traditional financial groups schemes like the Rotating Savings and Credit Associations (ROSCAs) and Accumulating Savings and Credit Associations (ASCAs), which have been in existence for centuries (Beaman et al. 2014; Bouman 1995). The basic savings group model, the Village Savings & Loan Associations (VSLA), was pioneered by CARE International in Niger in 1990. Since then, different international Non-Government Organizations (NGOs) like Catholic Relief Services, Agha Khan Foundation, Oxfam, etc. have promoted variants of the VSLA model. The savings group methodology has gained popularity among NGOs owing to its capability to reach large populations cost-effectively (Le Polain et al. 2018; Karlan et al. 2017).

Figure 1 below summarizes the savings group operational model. Savings and loans are the core group activities. Members make savings by purchasing shares during the scheduled group meetings. These savings are then made available to interested members who can take loans from the group funds at a pre-determined interest rate. Normally, members can only borrow around three times their savings. Thus, collecting savings and giving out loans represents the main mechanism through which SGs create value and generate profits for the group members. At the end of the cycle, typically annually, the contents of the cash box (the savings plus the accumulated interest) are shared out among the group members. By the share-out date, all loans must be repaid. Hence, operating a SG effectively is not easy. At the beginning of a cycle, members must be motivated to bring in sufficient savings to make lending possible. When lending begins, the borrowers must be carefully selected to assure repayment. Fines and interests must be collected, and as the end of a cycle approaches all loans must be repaid. Taken together, a savings group is a complex business model to operate and the best proxy to measure whether a group is financially managed effectively and efficiently is the return on savings (ROS). This measurement summarizes a group’s ability to intermediate savings and create wealth for the group members.

Fig. 1
figure 1

The operational model of savings groups

Full size image

The first basic savings groups formed in Niger by CARE International under the Mata Masu Dubara (Women on the move) project focused on improving women’s socioeconomic conditions. Hence, all initial savings groups were comprised of only female members. Currently, there are still some facilitating agencies that exclusively focus on women. For example: in Sudan, Oxfam’s Saving for Change program targets only women (Stevens 2018). Though the savings groups are still dominated by females, recent developments have witnessed increasing male participation. Savings group promoters are encouraging the engagement of men to mitigate certain negative effects that may arise when men feel threatened by the increasing women empowerment. Such effects include increased intimate partner violence (Slegh et al. 2013; Gupta et al. 2013).

Inclusion of Men in Gendered Interventions

For decades, women have been on the receiving end of different forms of gender injustice, for example, gender-based violence, intimate partner violence, and exclusion from participation in political and economic activities. Most of these injustices are deeply rooted in discriminative social norms for which men are usually custodians hence branding men as perpetrators of gender injustice (Flood 2019). Several efforts have been undertaken to remedy such injustices including sensitization of women and equipping them with knowledge and skills, engaging and mobilizing communities, and influencing a change in policies and legislation (Flood 2019).

However, these have often been geared towards empowering only women. Lately, though, there has been a realization that to achieve desired results from these interventions, men and boys need to be involved in efforts toward ending gender injustice (Schwittay 2019). First, as earlier mentioned, men are in most cases the perpetrators of such injustices like gender-based violence. Second, the exclusion of men may lead to negative consequences like increased gender injustices and inequalities (Colpitts 2019; Slegh et al. 2013), and it, therefore, is important for men to understand that they too stand to gain from reduced gender injustices and empowered women.

Like in other gendered interventions, there are increasing efforts to engage men in savings groups mostly premised on two reasons (Wheaton 2019). Just like women, some men suffer financial exclusion. To remedy such male exclusion and move towards universal financial inclusion, actors are recognizing the need to refocus financial inclusion efforts such as savings groups and involve men. Additionally, there is increased recognition that access to finance alone is not sufficient for the holistic transformation of women in the face of prevailing gender norms (Vossengberg et al. 2018). Moreover, there is also an increased awareness that the focus on women may lead to unintended consequences (ICRW 2019; Edwards 2017) hence there is a need to ensure that savings groups cause no harm to female members.

Unlike male engagement in reproductive health, gender violence, and intimate partner violence, this study investigates male engagement in activities one can consider outside direct household structures. It will hence shed more light on how the engagement of men in such activities influences their operations.

Group Composition and Savings Groups’ Performance

To examine group composition and savings groups’ performance, two theoretical foundations are drawn on: the social identity theory (Tajfel 1978) and the Similarity-attraction paradigm (Byrne 1971). The social identity theory starts from the premise that belonging to a group confers social identity to a person. Tafjel (1972, p. 31) defines social identity as “the individual’s knowledge that he belongs to certain social groups together with some emotional and value significance to him of the group membership”. Further still, people seek to gain a positive social identity by belonging to a certain group which they derive from social comparisons with other relevant groups. It is through this process of social comparison that one acknowledges the distinctive differences between one’s group (the ingroup) and other groups (the outgroup) and develops “outgroup” bias. The social identity approach predicts that diversity creates subgroups within a group which increases “ingroup” favoritism and “outgroup” discrimination. Such behavior in turn reduces information exchange and collaboration between group members and increases conflict which leads to reduced productivity and satisfaction (Anthony 2005).

The similarity-attraction paradigm proposes that people prefer to associate with others whom they perceive to be similar to themselves and interact better with persons whom they share similar characteristics with (Mia 2022). This hence implies that homogeneous groups will exhibit stronger cohesion and better communication yielding better outcomes. The paradigm proposes that diversity reduces interpersonal attraction and liking within the group which reduces cohesiveness, trust, and efficient communication within the group ultimately leading to negative group performance.

In conventional microfinance, it is common to apply group lending methodologies. In such settings, several scholars have investigated the effect of the gender composition of credit groups. In a laboratory experiment setting of microfinance loan groups, Berge et al. (2016) find that female groups are better than all male and mixed groups in collaboration for problem-solving. Further, they find that female groups are more risk-taking. Studying loan repayment rates in credit groups, Anthony and Horne (2003) find a positive relationship between the percentage of women and individual loan repayments. Similarly, while investigating delinquency rates among microfinance loan groups in Bangladesh, Sharma and Zeller (1997) show that delinquency rate decreases as the percentage of females in the groups increases. Still, investigating repayment rates, D’Espallier et al. (2011) focus on microfinance institutions and investigate the gender question in terms of the percentage of female clients served by an MFI. Their findings suggest that an increase in the percentage of female clients results in a decrease in portfolio risk, write-offs, and loan loss provisions. In Guatemala, Kevane and Wydick (2001) highlight that female credit groups have the lowest instances of fund misuse with mixed-gender groups having the worst performance.

When it comes to gender composition and performance of savings groups, Eboh (2000) takes a descriptive approach when analyzing the sustainability of savings and credit associations in Nigeria. He shows that all-female savings and credit groups had higher average savings and loans per member compared to all-male groups. He further observes that absenteeism from meetings was lower for all-female savings groups compared to gender-mixed groups. Slover (1991), however, finds that groups with membership dominated by females have fewer funds mobilized than those comprised of predominantly males. He attributes this to the differences in risk-taking between men and women.

Why would male involvement as group members influence the savings group’s profit-generating capacity? One argument is that male involvement may increase conflicts and tensions in the group due to inherent differences in behavioral characteristics between men and women. There is a common perception that, unlike women, men are less cooperative and less socially oriented (D'Espallier et al. 2013; Anthony and Horne 2003). In other words, men are more individualistic (Johnson 2004). Further, literature proposes that men tend to lean more towards disagreement and are more competitive while women lean more towards agreement and are more conciliatory (Anthony and Horne 2003). In this paper, we argue that in line with the social identity theory and similarity attraction paradigm, such differences may affect group interactions, communication among group members, conflict resolution, and decision-making and consequently affecting group operations.

Additionally, men have alternative sources of credit and savings (D’Espallier et al. 2013). Due to this, they may join savings groups for other purposes and may not be interested in the group functions of savings and credit. For instance, women sometimes engage in informal financial associations like ROSCAs without the knowledge of their husbands and extended family because this is the only way they can have control over their money (Guérin 2011). Men may hence join such groups to monitor the actions of their wives and get more information about their savings and credit activities. This may affect the way women conduct their group activities and ultimately hurt the overall group performance.

With the view that men are poor at loan repayment (Enimu et al. 2017) and have lower savings rates (Ledgerwood 1999), trust issues may arise in the groups when men join which may also lead to hesitations in issuing loans to men. Moreover, there is a perception that men compared to women are not as influenced by social pressure (Johnson 2004) hence they may default on loan repayment without fear of any consequences or societal shaming (Shohel et al. 2022).

However, based on some socio-economic characteristics of men especially in the areas in which savings groups are prevalent, male engagement may enhance group performance in several ways. Men are usually involved in activities that require large lump sums of money and equally have responsibilities that require the same (Jonhson 2004). This may translate into an increased demand for loans which is beneficial for a group as a larger share of the savings is lent out. Also, men are generally more educated and more literate than women (Guérin 2011). This could bring advantages for the group as they could enhance the knowledge base in the group, devise more efficient ways of running the group, and growing the group fund. After all, as mentioned, operating a SG is complex. Moreover, the argument that men are more risk-seeking than women (Croson and Gneezy 2009; Jianakoplos and Bernasek 1998; Slover 1991) may positively influence the group’s profit-generating capacity. If this risk appetite is combined with an increase in interest rate to cover the additional risk brought on by men, it may lead to an increase in the group returns.

Does the relationship between gender and savings group’s profit-generating capacity vary with the level of gender equality in the different countries where the groups are located? One can argue that, indeed, the effect may be different in countries where men and women have more equal rights and access to equal opportunities compared to countries where there is high inequality between men and women. Gender norms inherent in different societies influence the way men and women interact in different spheres and have a bearing on the kind of opportunities that are available to the different genders (Shohel et al. 2021; Johnson 2004). For example, in some societies, the woman’s responsibilities are relegated only to childbearing and taking care of the household (Kabeer 2005). In such societies, women may not be allowed to freely participate in economic activities such as in financial associations like savings groups. Moreover, in several cases, even when women can participate in certain activities, they are limited from reaching their full potential owing to the discriminatory gender norms of such patriarchal cultures (Shohel et al. 2021; Kinkingninhoun-Mêdagbé et al. 2010). When the percentage of men in the group increases in such societies characterized by high levels of gender inequality, it will likely have a worse effect than in societies where gender inequality is low.

Method

Variable Definitions

Dependent Variables

The dependent variable, return on savings (ROS), is used to proxy the profit-generating capacity. Previous studies have used ROS to measure the performance of SGs (Gonzales Martinez 2020; Burlando et al. 2016; Burlando and Canidio 2017). ROS captures the profitability of the group and the efficiency with which a group manages the complex SG business model to generate profit for its members. It measures the value created for the group members when their savings are lent out at an interest, and it is obtained as the total profit over the average savings used to generate these profits.

Independent Variable

The independent variable, group gender composition is captured using two variables. One we call male members. This captures the percentage of male members in a group and is computed as the number of male members divided by the total number of members in the group. The second independent variable we call mixed gender is a dummy variable that takes on the value of one if the group is composed of both male and female members and zero if the group is composed of only female members. Whereas the dummy variable, mixed gender, covers the effect of the mere presence of male members in the group, the male members variable covers the difference in effects resulting from differing percentages of male members.

Moderating Variable

We conjecture that the relationship between gender composition and group profit-generating capacity is influenced by the level of gender equality in the country where the group is located. We investigate this by introducing a moderation term for gender equality. To measure gender equality, we use the global Gender Gap Index (GGI) obtained from the World Economic Forum. The GGI measures gender-based differences in relation to four dimensions i.e., economic participation, education attainment, health, and political empowerment. The index ranges from 0 to 1 with higher values meaning higher equality and vice versa. In this study, the index has been reverse-coded with higher values meaning higher inequality and vice versa for ease of interpretation.

Control Variables

We include as control variables several SG-specific and contextual variables. The SG-specific variables include the age of the group, total assets, group size, welfare fund, rural dummy, other development initiatives, and facilitating agency. Age is likely to influence the profit-generating capacity of the groups due to the learning effect that takes place over time. With subsequent cycles, members may have a better understanding and appreciation of the savings group’s methodology (Moret et al. 2020). We also control for total assets which represent the welfare of group members, i.e., their ability to bring in more savings to the group. Group size is also likely to influence the profit-generating capacity as larger groups are likely to have higher loan fund utilization rates due to a higher number of projects qualifying for loans which should translate into higher returns. However, larger group size may also expose the group to risks associated with free riding and may affect the group operations negatively. Alongside the core loan fund, some SGs keep an emergency fund referred to as a welfare fund to cater to member emergencies. The welfare fund is likely to influence the group’s profit-generating capacity (Gonzales Martinez 2020).

Additionally, we include a dummy variable that takes on the value of one if the group is in rural areas. The location of the group may influence the profit-generating capacity of the group because groups in urban areas may have access to more resources. SGs are usually used as channels through which other development initiatives are delivered to low-income populations. These other development initiatives are usually referred to as ‘add-ons’ or ‘plus activities’. They include health education, training on income-generating activities, etc. The presence of add-ons along with core group activities is likely to influence the SG profit generation capacity as illustrated by Gonzales Martinez (2020). SGs are normally formed and supported by international Non-Governmental Organizations referred to as facilitating agencies. These facilitating agencies usually have different objectives and modes of operation that may influence the profit-generating of the groups.

A savings group’s profit-generating capacity is influenced by the macroeconomic conditions of the countries in which they operate. To account for the different macroeconomic conditions under which the savings groups in our sample operate, we include three contextual control variables. These are the population density, gross domestic product (GDP) per capita, and the heritage index of economic freedom.

To minimize the effect of outliers, all the ratios are winsorized at the 1% and 99% levels. Table 2 presents all the variables and their definitions.

Data

We have access to a unique dataset on informal grassroots financial associations from the Savings Groups Information Exchange (SAVIX). The SAVIX is an online platform that contains standardized data on many savings groups globally. It was developed by the VSL Associates and funded by the Bill and Melinda Gates Foundation together with several facilitating agencies (CARE International, Catholic Relief Services, Oxfam International, and Plan International). Data is collected and uploaded to the SAVIX Management Information System (MIS) on a quarterly basis. The SAVIX captures several elements relating to the operations of the savings groups. The sample for this study consists of 81,853 savings groups from 30 African countries and covers the period from 2010 to 2017. Table 1 details the distribution of the data by country. Most savings groups in the sample are from Mali (19.72%), closely followed by Uganda (14.85%).

Table 1 Distribution of sample by country
Full size table

The SAVIX is the first and arguably the most comprehensive database there is on informal community-managed grassroots associations globally. The data is collected and uploaded by field officers visiting the groups on a quarterly basis. The SAVIX mainly contains data for savings groups that are supported by a facilitating agency, thus it does not cover several other groups that are formed spontaneously without the support of a facilitating agency which is a weakness of the dataset. Nonetheless, the SAVIX is the most representative dataset on savings groups as it covers different complementary metrics and contains several groups from different countries.

To supplement data from the SAVIX and consider the macroeconomic conditions of the countries where the SGs are located, we draw on several country indicators obtained from multiple sources. GGI data is obtained from the World Economic Forum global gender gap reports, population density is obtained from the world development indicators developed by the World Bank, Gross Domestic Product (GDP) is obtained from the world economic outlook database maintained by the International Monetary Fund and economic freedom index is obtained from heritage foundation.

Estimation Strategy

It should be noted that there could be some endogeneity concerns as the group profit-generating capacity and the percentage of male members may be simultaneously determined. If SGs formed completely or almost completely by women are successful for example in terms of producing high returns for group members, this will attract men to participate in the next cycle of the SG. Hence, the number of men in a group may be influenced by the profit-generating capacity of a group just as the number of men in a group may influence the profit-generating capacity. We address this simultaneity bias concern by analyzing only the data on the first cycle of group operations. In such a case, men are not basing their decision to join the group on the group’s previous cycle outcomes. Endogeneity concerns are also addressed under the further analyses section using a matched sample difference-in-differences approach.

We employ random effectsFootnote 1 regression analysis to investigate whether the profit-generating capacity is statistically related to the gender composition of the group. Specifically, we estimate the basic model below:

$${Y}_{it}={\beta }_{0}+{\beta }_{1}{Gender}_{it}+\delta {X}_{it}+{c}_{i}+{u}_{it}$$
(1)

where \({Y}_{it}\) captures the profit-generating capacity, measured in terms of ROS. \({Gender}_{it}\) is the gender composition of the ith savings group at time t. \({X}_{it}\) represents a vector of control variables. We also include time dummies in the different regressions to cater for time-fixed effects.Footnote 2\({c}_{i}\) is the group-specific unobserved effect and \({u}_{it}\) is the idiosyncratic error term. Standard errors are clustered at the group level.

Gender Composition of SGs Over Time

Although over 80% of savings groups’ membership globally is comprised of women (Wheaton 2019), Fig. 2 shows that the share of male members is growing rapidly over time. Male members have increased from around 9% in 2010 to over 27% in 2017. This shows the increasing significance of men in savings groups’ operations.

Fig. 2
figure 2

Share of male members over time

Full size image

Summary Statistics

Table 2 presents the summary statistics for the variables used in the study. From the table, we see that the average group in the dataset is composed of 17 percent male members. The mixed-gender dummy shows that about 55 percent of the groups have both male and female members. The other 45 percent is composed of only female members.

Table 2 Variable definitions and summary statistics
Full size table

Concerning the profit-generating capacity, the average group has a return on savings of 40.9%. A value of 40.9% means that at each end-of-cycle share-out, each member on average makes 40 cents on each dollar saved. This further shows that savings groups typically lend out at high rates and fetch higher returns for group members as compared to other savings mechanisms like formal financial institutions (Allen 2002).

Regarding the moderating variable, GGI, the average group is in a country with a score of 0.65 which indicates that most groups are located in countries with moderate levels of gender equality. The minimum value (0.57) and maximum value (0.82) show that SGs in the sample are in countries with diverse levels of gender equality i.e., from average levels to high levels of equality.

Moving on to the savings groups’ characteristics, we observe that the average group has been in existence for 351 days (corresponding to about one year since its first savings were initiated for the first cycle), has an asset size of 667.6 USD, and has 23 active members. Further still, 69.5% of the SGs keep a welfare fund and 59% are in rural areas. 37.6% of the SGs incorporate other development initiatives like health education, women empowerment, etc. This shows that savings groups, in addition to being mechanisms for local intermediation of savings, are also channels through which multiple services can be offered to the people at the bottom of the pyramid to tackle multi-dimensional poverty. The table further shows that the majority of the SGs in the sample (29.3%) are supported by Plan International, closely followed by CARE (24.2%), Oxfam (19%), and Catholic Relief Services (14.4%) with other facilitating agencies sharing the remaining 13 percent among them.

Looking at the macroeconomic conditions for the countries in our sample, the average country has a population density of 108.43 people per square kilometer, GDP per capita of 2531.6 USD, and a heritage economic freedom score of 57%.

Results

Gender and Savings Group’s Profit-Generating Capacity

Table 3 presents the random effects regression results for the effect of gender on the profit-generating capacity of the group measured in terms of the return on savings (ROS). The results reported in column 1 indicate that the proportion of male members is negatively related to the profit-generating capacity of the group. The estimated coefficient shows that a unit increase in the proportion of male members by say 100 percentage points leads to a 6.7 percentage point decrease in the ROS.

Table 3 Gender and SG profit generation capacity
Full size table

One potential argument for the observed effect is the perception that men are poor at loan repayment and hence have higher default rates than women (D’Espallier et al. 2011; Mutebi et al. 2017). The high default rates have a direct effect on the returns on savings as both the member savings and interest income are lost. Moreover, the potentially higher default rates could influence the fund utilization rate if loans to men are rationed to mitigate the effects of their poor repayment rates. In mixed gender, women may tend to limit the amount of money that they allow men to take as loans and reduce the overall lending in a group. A similar observation was made by Mutebi et al. (2017) where in some mixed-gender groups, there was a limit to how many men were allowed in the group owing to the high default rates of men. Additionally, there were more stringent terms imposed when men required loans, for example, they required that men have their spouses serve as their guarantors because women ensured that men fully paid back the acquired loan (Mutebi et al. 2017).

Another potential argument is that the nature of savings groups may not be suited to the socio-economic characteristics of men i.e., they more often require bigger loans that the group may not be able to offer. Johnson (2004) makes a similar observation with regard to ROSCAs being less suited to men’s financial responsibilities. Moreover, since men usually have alternative forms of savings and sources of credit unlike women, they may not care much for the savings group’s activities and may join for different purposes. Some men may join to monitor the activities of the women (Waller 2014). In such instances, this may affect the savings and borrowing behavior of women especially if they do not want the men to have knowledge of the same. For example, women may reduce how much they borrow because they do not want the men to know how much money they have under their control. Anderson and Baland (2002) observe that women keep their participation in informal savings groups secret from their husbands to hide their savings from misuse by the husbands so when men join such groups, it could infringe on the privacy that the women sought and influence their savings and borrowing behavior ultimately affecting the return on savings.

Looking at the control variables, some results are interesting. We observe a lower ROS for rural groups which hints at the poorer investment opportunities in rural areas hence the need to lend to members at lower interest rates or a lower share of the savings being converted into loans. Similarly, we observe that a lower return on savings is associated with SGs that deliver other development initiatives along the core group activities. This indicates that the other development initiatives may destruct the group from the core savings and borrowing activities, an area that should be researched further to find out how delivery of other initiatives influences savings groups operations and devise ways of constructively doing the same without harming the core group model. Further still, lower profit-generating capacity is associated with groups that keep a welfare fund for emergency purposes.

In column 2, gender composition is measured using the dummy variable, mixed gender. The results are consistent with those obtained when gender is measured using the proportion of male members. There is a negative relation between the mixed-gender dummy and the profit-generating capacity of the group. Specifically, mixed-gender savings groups have significantly lower ROS than savings groups composed of only female members.

Moderating Effect of Gender Equality

Does the above relation between gender composition and group profit-generating capacity vary with the level of gender equality in the country in which the group is located? To answer this, we include an interaction term between gender composition and the level of gender equality in the regression model. Gender equality is measured using the global gender gap index (GGI). Random effects regression results are presented in Table 4. As in Table 3 column 1 reports numbers related to the share of male members while column 2 reports results when gender is measured using the mixed gender dummy variable.

Table 4 Moderating role of gender equality
Full size table

As shown, contextual gender equality influences the effect of gender on group profit-generating capacity. Specifically, the negative relation between gender and ROS is worse as the level of gender inequality increases. The results shed light on the importance of understanding inherent gender norms for the functioning of savings groups.

Further Analyses

Alternative Estimation Method

To ascertain the robustness of the results and make an argument for a causal effect of gender composition on group profit-generating capacity, we employ the difference-in-differences (DID) approach (Chen et al. 2017). The DID approach further takes care of the endogeneity concerns of simultaneity as highlighted above. DID is used to estimate the effect of a treatment by comparing two similar groups, one with treatment and the other without treatment. The SAVIX dataset contains data that is collected on a quarterly basis. To form the sample for this analysis, we include groups that experience an increase in group membership. We include data on the group one quarter before and one quarter after the increase in group membership. To make the treatment group, the increase in group membership had to bring in male members hence the group moved from being all-female to mixed. This leaves 203 SGs making up the treatment group. In the control group, the increase in group membership still brings in female members and the group remains all-female. This leaves 1745 SGs making the control group.

Before performing the DID, we apply propensity score matching on the pre-treatment data to make the treatment and control groups as similar as possible. We use the nearest neighbor to select (Stuart and Rubin 2008), for each SG in the treatment group an SG from the control group with the closest propensity score based on several matching covariates. Specifically, each SG with mixed gender composition is matched to an SG with only female members that is as similar as possible to it along several covariates before male members joined the mixed SG. The control variables used in the previous regression models form the matching covariates that are adopted. The final dataset after matching contains data on 203 treated SGs and an equal number of matched control SGs.

To test the reliability of the matching, we conduct univariate t-tests for the matching covariates to compare the mixed-gender groups with their matched control SGs. Panel A of Table 5 shows that there is no statistically significant difference between both groups along the observable matching covariates. We then estimate the DID model below on the matched sample.

Table 5 Difference-in-differences estimation results
Full size table
$${ROS}_{it}=\alpha +{\beta }_{1}{Male\,joiners}_{it}+{\beta }_{2}{Post}_{it}+{\beta }_{3}{Male\,joiners}_{it}\times {Post}_{it}+ \delta {X}_{it}+{\varepsilon }_{it}$$
(2)

where \(Male\,joiners\) is an indicator variable equal to one if the SG is in the treatment group, and zero otherwise, \(Post\) is an indicator variable equal to one if the quarter is after membership increase. \({X}_{it}\) represents a vector of control variables as in Eq. 1. We include cycle, facilitating agency, and time-fixed effects. SGs may be in different cycles which could influence profit-generating capacity so cycle fixed effects take this into account.

Panel B of Table 5 presents the DID results. The coefficient on the \(\mathrm{Male joiners}\times \mathrm{Post}\) is negative and statistically significant at the 5% level which suggests that after male members joining the group, SGs have lower ROS than after female members join. The coefficient shows that SGs have ROS that is 10 percentage points lower the quarter after male members join than when female members join.

Alternative Definition of the Gender Variable

We recode the gender variable into four categories to assess if the manner of the composition has any effect on the profit-generating capacity. The categories are; all female (takes the value of 1 if all members in the group are female and zero otherwise), minority men (assumes the value of 1 if the group is composed of fewer men than women and zero otherwise), balanced (assumes the value of 1 if the group is composed of an equal number of men and women, and zero otherwise) and majority men (assumes the value of 1 if the group is composed of more men than women and zero otherwise).

The results of this analysis are presented in Table 6 and the ‘all female’ is the reference category. Compared to groups that are composed of only female members, groups that have some men in them, have a lower ROS further confirming the previous results. This is evidenced by the negative significant results in all categories i.e., minority men, balanced, and majority men. In fact, the coefficient for the majority being men is the highest, illustrating again that the more ‘male’ a group becomes the worse the group performs.

Table 6 Robustness analysis—redefined gender categorization
Full size table

Does It Matter How Many Male Members a Group Has?

In further analysis, we decompose the gender composition as follows: We compare groups with (a) no male members i.e., all-female groups with (b) groups with one male member, (c) groups with two male members, and (d) groups with three or more male members. We run random effects regressions with the reference category as the groups with no male members.

We find that generally, groups with two or more male members have a lower ROS than groups with no male members or with one male member (Table 7). Specifically, having one male member has no effect on ROS. However, having two male members significantly reduces group ROS and having three members further reduces the ROS.

Table 7 Does it matter how many male members a group has?
Full size table

Discussion of Findings

Drawing on the social identity theory and similarity-attraction paradigm as theoretical foundations, this paper examined the effect of gender composition on the performance of savings groups. We specifically use data on 81,853 savings groups in 30 African countries to investigate the effects of gender composition on SG profit-generating capacity. We further investigated how this relationship changes depending on the gender equality of the country in which the group is located.

Confirming the premises of the two adopted theories, the findings showed a negative relationship between gender composition and group profit-generating capacity. Specifically, the findings show that the profit-generating capacity of the group decreases as the proportion of male members in the group increases. The finding is in line with Berge et al. (2016) who find higher collaboration in female microfinance groups compared to all male and mixed groups. The findings also corroborate Anthony and Horne (2003) who find that loan repayment in credit groups increases with the percentage of women.

One potential argument for the observed effect in this study is the perception that men are poor at loan repayment and hence have higher default rates than women (D’Espallier et al. 2011; Mutebi et al. 2017). The high default rates have a direct effect on the returns on savings as both the member savings and interest income are lost. Moreover, the potentially higher default rates could influence the fund utilization rate if loans to men are rationed to mitigate the effects of their poor repayment rates. In mixed gender, women may tend to limit the amount of money that they allow men to take as loans and reduce the overall lending in a group. A similar observation was made by Mutebi et al. (2017) where in some mixed-gender groups, there was a limit to how many men were allowed in the group owing to the high default rates of men. Additionally, there were more stringent terms imposed when men required loans, for example, they required that men have their spouses serve as their guarantors because women ensured that men fully paid back the acquired loan (Mutebi et al. 2017).

Another potential argument is that the nature of savings groups may not be suited to the socio-economic characteristics of men i.e., they more often require bigger loans that the group may not be able to offer. Johnson (2004) makes a similar observation with regard to ROSCAs being less suited to men’s financial responsibilities. Moreover, since men usually have alternative forms of savings and sources of credit unlike women, they may not care much for the savings group’s activities and may join for different purposes. Some men may join to monitor the activities of the women (Waller 2014). In such instances, this may affect the savings and borrowing behavior of women especially if they do not want the men to have knowledge of the same. For example, women may reduce how much they borrow because they do not want the men to know how much money they have under their control. Anderson and Baland (2002) observe that women keep their participation in informal savings groups secret from their husbands to hide their savings from misuse by the husbands so when men join such groups, it could infringe on the privacy that the women sought and influence their savings and borrowing behavior ultimately affecting the return on savings.

Further, the findings also showed that the gender equality of the country where the group is located exacerbates the observed effect of the gender composition on the group’s profit-generating capacity. The findings hence highlight the fact that gender discrimination affects women’s ability to participate in economic and social activities. Moreover, the findings also highlight the importance of contextual factors such as culture and norms in achieving results for different interventions (Zhang and Posso 2017).

Conclusion and Suggestions for Future Research

While there are increasing efforts to engage men in gender-focused development projects, little is known about how this influences the effectiveness of the projects. This is particularly true for savings groups which have become a ‘darling’ for development agencies that are interested in financial inclusion and empowerment through group efforts for people living at the bottom of the pyramid. In this paper, we investigate the relationship between gender composition and the profit-generating capacity of savings groups, i.e., the financial efficiency of groups. We further probe whether this relationship is moderated by the gender equality situation in the country where the savings groups are located. Gender composition is measured through the proportion of male members, and a dummy variable that shows whether the group is composed of both male and female members or only female members. Profit-generating capacity is measured through the return on savings which measures the wealth created from member savings. Using a large sample of data on 81,853 savings groups in 30 countries, we employ random effects regression analysis to investigate the effects of gender composition on SG profit-generating capacity. We address endogeneity concerns due to simultaneity bias by doing the empirical analysis on observations from only the first cycle of group operations, i.e., the gender composition of the group is decided ex-ante the performance of the group.

Results indicate that there is a negative relationship between gender composition and group profit-generating capacity suggesting that SGs with a higher proportion of male members have a lower return on savings. The results for the dummy variable for the mixed gender group are consistent showing that SGs with both male and female members have a lower return on savings as compared to those that have only female members. Further, results for the interaction term between the gender composition and gender equality variable show that the negative relationship between gender composition and profit-generating capacity is generally made worse in contexts with high gender inequality. These findings are robust to alternative estimation methods and alternative definitions of the gender composition variable. Results from the differences-in-differences robustness estimation further minimize endogeneity concerns.

A couple of recommendations emerge from the study. First, to avoid harming the savings group operational model, practitioners should be aware of the downsides that come with male engagement in SGs and should derive ways of solving them. Male engagement can be an important component in the empowerment of women. However, as seen from the results, having both males and females in the same group has a negative effect on some core aspects of the savings group model. Thus, SG promoters should be wary of this and come up with necessary balancing efforts like for instance gender awareness training and increased monitoring of gender-mixed groups.

Second, development actors should exercise increased contextual caution when implementing programs. Underlying gender norms of the different program areas should be taken into consideration before implementing programs in different areas. This recommendation stems from the observation that the negative relationship between gender composition and SG profit-generating capacity varies with the gender equality of the different countries in which the SGs are located.

Savings groups are often used by development actors to deliver other development initiatives apart from savings and credit. Gender equality dialogue and training is one such initiative that is often delivered through these groups. Further research can show how the effect of gender on group operations is moderated by gender equality training as an add-on. Could such trainings provide a potential solution to the observed negative effect that mixed-gender groups have a lower ROS than groups that are all females?

Future studies are also needed to capture the effect on women which cannot be captured by the current quantitative investigation. Are women changing their behavior for example in terms of expression during group meetings when men join the groups? The lower ROS for mixed gender groups could signal a change in saving and borrowing behavior among female members when men join the groups. Understanding how women react to men joining the groups will expose the behavioral biases that women may have towards men and with this, solutions can be devised on how the two genders can collaborate for the overall good of the group and individuals in the group.

Moreover, on a more general level, a qualitative investigation may help decipher the reasons why men join SGs and particularly why they would like to be in the same group as women. Do they trust dealing with women more than fellow men? Do they join such groups to monitor activities of the women as suggested by Waller (2014) and have an idea of the financial position of the women? As gender norms paint women as a generally weaker gender in societies with high gender inequality, do the men join female groups to take advantage of the position of women? Do they join such SGs because they are lured by the development agencies, who are interested in driving the gender equality agenda? Are development agencies more interested in the social function of the group as compared to the financial aspect?