The relationship between financial
sustainability and outreach for sustainable and
unsustainable microfinance institutions
Twan Goossen
University of Groningen
Faculty of Economics and Business
June 2020
Supervisors:
Prof. Dr. B.W. Lensink
Abstract
The drive towards financial sustainability in the microfinance industry has raised concerns regarding the effect it has on the social outreach of the institutions. This study contributes to the trade-off discussion in the microfinance industry in two ways. First, it explores the relationship between both depth and breadth of outreach and financial sustainability, while dealing with the often neglected endogeneity concerns of simultaneity between outreach and financial sustainability. Secondly, the relationship is tested using a more in-depth view. The microfinance institutions (MFI’s) are assigned to groups based on financial sustainability. This creates three groups: financially unsustainable MFI’s, MFI’s focusing on financial sustainability financially sustainable MFI’s. The evidence shows complementary effects between the depth of outreach and financial sustainability, which are strongest for sustainable MFI’s. We also find diminishing complementary effects between the breadth of outreach and financial sustainability as MFI’s become sustainable.
2
1. Introduction
Microfinance as a way to reduce poverty has existed for a long time. It focuses on providing products and services, especially in the form of loans, to people from low-income countries, who otherwise do not have access to financial services1 (Yunus, 2008). In the literature,
providing financial services to people who do otherwise not have access to financial services is described as outreach and comprises two dimensions: depth of outreach and breadth of outreach. Depth of outreach focuses on reaching the poorest people, while the breadth of outreach focuses on the scale to which the financial services are provided (Conning, 1999). Currently, the microfinance industry is characterised by trying to accomplish two important components: financial sustainability2 and outreach. However, one should ask the question if
these two components can be achieved simultaneously or if microfinance institutions experience a trade-off between the two. This specific question has led to a long-lasting and widespread discussion in the microfinance literature. On the one hand, there are the institutionalists who believe financial sustainability is a necessity to fight poverty in the long run and believe that the two components go hand in hand. On the other hand, there are the welfarists who believe sustainability will shift focus from serving the poor to financial performance, making it impossible to improve outreach and financial sustainability simultaneously. The current literature available on this discussion, however, copes withsome shortcomings. First, they do not consider endogeneity issues between financial sustainability and outreach. This causes the coefficients to be unable to picture the correct relationship between financial sustainability and outreach. And secondly, the trade-off is analysed in too broad of a setting and should take a closer look at the trade-off between microfinance institutions based on different stages of financial sustainability. This study will contribute to this debate in two ways. First, by analysing the trade-off between financial sustainability and outreach considering previously ignored endogeneity issues. And secondly, by examining the trade-off between financial sustainability and outreach considering different stages of financial sustainability.
Most studies assume either that outreach is the dependent variable, and financial sustainability the independent variable (see Hermes et al., 2011; Cull et al., 2007; Kar, 2013), or that financial sustainability is the dependent variable and outreach is the independent variable (see Ayayi and Sene, 2010; Bassem, 2012). However, in these studies, they rarely consider both financial sustainability and outreach as endogenous variables. A few studies do address endogeneity and simultaneity and control for this by implementing instruments. They use a lagged variable as an instrument to determine the explanatory regressors (see Kar, 2013; Mersland and Strom, 2010). However, substituting an endogenous explanatory variable
1 These financial services include, but are not limited to: loans, savings, insurance, but also non-credit
services, such as training.
3
with its lagged value does not avoid the simultaneity bias, and the returned coefficients are still inconsistent (Reed, 2015). To properly use a lagged variable, one has to implement a lagged variable of the dependent variable using an instrumental variable estimation. Mersland and Strom (2010) try to use a set of country variables as additional instruments but found them to be relatively weak. Studies which properly addressed simultaneity issues are very limited and differ widely in instruments and results (see Churchill, 2019; Nurmakhanova, 2015; Quayes, 2012). This study will contribute to the limited literature available by considering both financial sustainability and different dimensions of outreach as endogenous variables. The study will use a new set of instruments to control for endogeneity concerns to analyse the trade-off between financial sustainability and outreach.
Additionally, in the on-going discussion, previous research has tried to find if different microfinance institutions experience different relationships. Cull et al. (2007) study if the trade-off holds for different groups of MFI's, by distinguishing MFI’s that focus on individual loans, group loans and village banks. Another group of researchers made a distinction between for-profit MFI’s and non-profit MFI's to reflect the difference between financially sustainable MFI's and financial unsustainable MFI’s (see Churchill, 2019; Hartarska, 2007; Cull et al., 2007). This follows the reasoning that for-profit MFI’s focus on their financial performance and sustainability, while non-profit MFI’s mainly focus on outreach. However, the distinction between for-profit and non-profit is insufficient to test the difference between financially sustainable MFI’s and financially unsustainable MFI’s, because also non-profit MFI’s are focusing on becoming financially sustainable (Conning, 1999). To the best of my knowledge, nobody has studied the relationship between financial sustainability and outreach considering microfinance institutions in different stages of sustainability. This study follows the reasoning from the financial systems theory which argues that financially sustainable firms can better perform in the outreach dimensions as they can expand their outreach by reaching more clients due to them not being reliant on subsidies (Robinson, 2001). Therefore this study will make a distinction between microfinance institutions in different stages of sustainability: financially unsustainable MFI’s and the financially sustainable MFI’s, while also distinguishing MFI's who are focusing on becoming financially sustainable, based on their current sustainability and how their sustainability improved compared to the previous year. This distinction is important because it will explore if effort spent by microfinance institutions to become financially sustainable will pay off in the long run by being able to target more and poorer clients.
4
2. Literature review
Financial sustainability has become an increasingly important issue for MFI’s since the rise of commercialization of microfinance. Instead of being subsidized institutions providing affordable finance for people who otherwise do not have access to finance, MFI’s now experience a shift towards becoming sustainable commercial organizations (Robinson, 2001). This results from several developments in the field of microfinance, mainly the entrance of competition and the impact of commercially oriented institutions (Rhyne and Otero, 2006). The increasing competition among MFI's and commercial entities trying to maximize their potential pressures MFI's to increase their efficiency and revise their way of doing business. Christen (2000) studies the commercialization in Latin America and shows high levels of profitability and increased competition in the region. With the new commercialization also came the question about the effect it had on the mission of the microfinance institutions, serving the poor. Many MFI’s have tried to combine focusing on providing finance to the poor and becoming financially sustainable institutions. If the two can be achieved simultaneously has been widely studied by researchers, but has so far not yielded clear results. On the one hand, focusing on outreach can lead to reduced administrative and monitoring costs, as smaller loans are often disbursed in the form of group loans or similar loan structures (Quayes, 2012). Due to the implementation of group loan structures, the microfinance institutions can work more efficiently through which they increase their financial performance. Additionally, focusing on financial performance attracts investors, which increases the funding for microfinance institutions allowing MFI’s to expand their outreach by increasing the supply of finance (Lensink, 2011). Finally, financial performance promotes innovations, through which the costs can be reduced or the benefit for clients increases (Schreiner, 2002), for example through lower interest rates or better loan terms. On the other hand, focusing on outreach can induce higher transaction costs. This is because many small loans are more expensive than one large loan transaction for a richer borrower (Armendáriz and Morduch, 2010). This, in turn, results to lower financial performance due to higher costs. To hold on to the financial performance, but at the same time keep the serving the poorest borrowers, MFI's must charge higher interest rates or incur a greater cost of disbursing loans (Conning, 1999). This can result in lower loan disbursement, as many poor borrowers will not be able to afford the increased interest rates. Additionally, focusing on financial performance itself can decrease outreach by targeting clients who are better off, as they are generally less expensive (Churchill, 2019). It can as well lead to an increasing interest rate to target maximized profits.
2.1 Empirical evidence: trade-off financial sustainability and outreach
5
the two, where focusing on financial sustainability will decrease the outreach and focusing on outreach will lead to lower financial sustainability. Hermes et al. (2011) focus on the trade-off between the depth of outreach and efficiency. Their findings show that MFI’s with a lower average loan size are less efficient. They also show that MFI’s who borrow more to female borrowers are less efficient3. This means that with the current trend for
commercialization, MFI’s will become more efficient, and thus have a lower depth of outreach. These findings are confirmed by Cull et al. (2007). While he finds that smaller loans aren’t necessarily less profitable, he does find that larger loan sizes are associated with lower average costs. If an MFI were to strive for profit maximization, it would benefit from larger loan sizes and thus targeting less poor clients.
Makame and Murinde (2006) analyse the trade-off between financial sustainability and both depth of outreach and breadth of outreach. They use various measures of depth of outreach based on the average loan size and measured financial sustainability through efficiency. Using a small panel dataset of 33 MFI's based in Eastern-Africa, they find a trade-off for both measures of outreach. They find that the operating expenses over gross loan portfolio increase as MFI's target more and poorer borrowers, confirming the findings of a previous study based in Latin-America by Olivares-Polanco (2005). Additionally, Quayes (2012) finds results both rejecting and accepting the trade-off between outreach and financial performance. His study uses a dummy variable for financial sustainability, with 0 meaning the MFI is not sustainable and 1 for a sustainable MFI. Using over 700 MFI’s across 80 countries, they find that a 1% decrease in the average loan size (and thus an increase in the depth of outreach) would increase the probability of being financially self-sustainable with 3.68%. However, a 1% increase in the number of active borrowers4 decreases the probability
of being financially self-sustainable with 5.59%. Quayes (2012) also splits the findings into low data disclosure MFI's and high data disclosure MFI's. He argues that the results of high-disclosure MFI's are more reliable because they are audited more often. The results of high disclosure MFI's are in line with his previous findings, only the effect of the average loan size increases the probability of financial sustainability from 3.68% to 6.36% and the number of active borrowers with similar probabilities. These findings indicate that while the depth of outreach increases as MFI's become more financially sustainable, the breadth of outreach decreases.
2.2 Empirical evidence: No trade-off financial sustainability and outreach
Another group of researchers shows that financial sustainability and outreach do not experience a trade-off. Ayayi and Sene (2010) study the drivers of financial sustainability for
3 Another generally accepted measure of the depth of outreach, following the reasoning that poorer
borrowers are often female.
4 This is a widely accepted measure of the breadth of outreach. An increase in active borrowers would
6
MFI’s. They find that the average loan size has a positive impact on the financial sustainability of an MFI. However, the coefficient is very low, implying a minimal effect of this variable. Additionally, they did not find any significant effect of the percentage of women on financial sustainability. This minimal to no effect is confirmed by Cull et al. (2007) and Bassem (2012) who also study the effects of outreach on profitability. They find no significant effect of the depth of outreach on the financial performance of MFI’s.
Some studies also show evidence of financial sustainability and outreach being complementary. Mersland and Strom (2010) study the effect of the average profits on the average loan size5. They find that as the average profits of an MFI increase, the average loan
size becomes larger as well. This indicates that MFI’s who strive for higher profits would focus on borrowers who are better off, driving out the poorest borrowers. However, they find that the average loan size increases more from an increase in the average cost. If an MFI increases its efficiency and therefore decreases its average costs, it can increase the service to poorer borrowers. These results would mean that an MFI can increase its profits and can become more efficient without reducing the impact of loans to the poorest borrowers.
Another study looks at the effect of profitability of MFI's on the depth of outreach. Similar to Mersland and Strom (2010), Kar (2013) uses lagged variables of the explanatory variables to overcome the endogeneity bias. The study finds that an increase in return on assets (ROA) leads to a decrease in the average loan size and an increase in the percentage of female borrowers for both individual-based lenders and solidarity-based lenders. However, they do not find statistically significant evidence for the effect of financial self-sustainability on the depth of outreach. Cull et al. (2007) also study outreach and profitability in terms of financial self-sustainability. They disaggregate the effect by lending type6. They find a significant
negative coefficient for financial self-sustainability on the average loan size and a significant positive coefficient for the percentage of woman borrowers when looking at individual-based lenders. This indicates that for this sort of lending type, self-sustainability and outreach can go hand in hand. No evidence is found for the village banks and solidarity-based lenders. However, they do find that the relationship is different from the individual-based lenders.
2.3 Empirical evidence of simultaneous equation modeling
One major concern with the preceding research is that they measure financial performance and outreach independent of each other. In reality, outreach can influence financial performance, while at the same time financial performance influences outreach. The simultaneous influences between the two are neglected in most research (Nurmakhanova et al., 2015). Because decisions on outreach and financial performance are made
5. An often-used proxy to measure the depth of outreach, in which a higher average loan size indicates less
outreach because the MFI is expected to have fewer loans to poor clients (Hermes et al., 2011)
7
simultaneously, both variables should be measured simultaneously. In doing so, Nurmakhanova et al. (2015) find no evidence of a trade-off between financial sustainability and outreach. On the contrary, he finds that the percentage of female borrowers and the number of active borrowers tend to go up as MFI's become more financially sustainable. Quayes (2012) adds on these results, and he finds complementary effects between financial self-sustainability and the depth of outreach. This study, however, fails to implement strong instruments to deal with endogeneity issues. They also observe data from one year only, and therefore the results should be interpreted with caution. Similar to Nurmakhanova et al. (2015), Churchill (2019) uses a simultaneous equation model to deal with simultaneity and endogeneity issues. In their study, they measure both the effect of financial sustainability on outreach, as well as the effect of outreach on financial sustainability simultaneously. To measure outreach, they create an index measuring the depth and breadth of outreach. They find that the depth of outreach has a significant negative impact on the financial sustainability of an MFI. Additionally, their results show that financial sustainability has an even larger negative impact on the depth of outreach, confirming findings from previous studies (Mersland and Strom, 2010; Cull et al., 2007). The difference in negative impact suggests that focusing on outreach might negatively impact financial performance, but a focus on financial performance has even more negative consequences for the depth of outreach. The study finds a reversed effect when looking at the breadth of outreach. When an MFI increases in its number of active borrowers, they also perform better in terms of outreach and vice versa. These findings are in line with both the financial systems approach, as well as one of the major concerns of this approach, namely that while financial sustainable MFI’s can deal with the demand of loans, they tend to focus less on the poorest borrowers.
3. Hypothesis development
8
cope with the demand because they have more resources at their disposal (Robinson, 2001) and as a result, can target more poor borrowers. Based on this reasoning, the following hypothesis is proposed:
Hypothesis 1: As firms become more financially self-sustainable they deviate from the
poorest borrowers, thus there is a trade-off between financial sustainability and depth of outreach. However, the breadth of outreach and financial sustainability are complementary. More interesting is the relationship for non-sustainable and sustainable microfinance institutions. With the rise of commercialization and the entrance of commercially oriented institutions, the microfinance world has been split into two camps. On the one hand, there are the non-profit MFI's, while on the other hand there are the for-profit MFI's. The financial systems theory argues that self-sustainability is a necessity for microfinance institutions to expand their outreach in the long term. This is based on the reasoning that self-sustaining and profitable MFI's are more likely to find financial resources which help to focus on the goal, battling poverty (Daher and Le Saout, 2013). Additionally, the pursuit of profits can expand the drive for innovations, which can benefit both outreach and performance. An MFI driven by donations, on the other hand, is less incentivised to focus on innovating (Schreiner, 2002). Contrary, the poverty lending approach believes that the pursuit of profits is not beneficial for reaching the highest outreach. This approach argues that focusing on profits will cause a drift to the less poor, as the poorest people are too expensive to make profits. By focusing mainly on the poor, they allow interest rates below market rates as the poorest people cannot afford high interest rates (Hermes and Lensink, 2011). With the ongoing debate about the trade-offs between financial sustainability, one can argue that non-profit and for-profit MFI’s do not experience the same relationship between outreach and financial sustainability. Sustainable and profitable MFI’s are more likely to find financial resources to help focus on outreach (Daher and Le Saout, 2013). Because self-sustaining MFI’s can gather more financial resources, and do not suffer from donor constraints (Morduch, 2000), they could reach higher levels of outreach than MFI’s who have to cope with these constraints. As for-profit MFI’s are characterized by focussing on financial performance, such as efficiency and profit margins, it is more likely that they are financially sustainable (Churchill, 2019). This is formally investigated by Tchakoute-Tchuigoua (2010). His findings show that for-profit MFI’s are indeed more sustainable than non-profit MFI’s. Alternatively, while for-profit MFI's are less limited by donors, they are more constrained by shareholders, whose main objective might not be to maximize outreach but to maximize profits (Tchakoute-Tchuigoua, 2010). Non-profit MFI’s do not have to deal with maximizing shareholder values, giving them the freedom to pursue maximizing outreach.
9
sustainability and outreach, considering if an MFI is identified as for-profit and non-profit. In terms of depth of outreach, he finds that both for-profit and non-profit MFI's cope with a trade-off between financial sustainability and outreach. When a for-profit MFI increases its sustainability, the negative impact on the depth of outreach is much larger than the impact on sustainability when the MFI increases its outreach. Surprisingly, the result for non-profit MFI's is the opposite. A focus on outreach has a much more negative impact on sustainability than the other way around. In terms of breadth of outreach, there is a clear distinction between for-profit and non-profit MFI's. The breadth of outreach and sustainability are complementary when looking at for-profit MFI's, while there is a trade-off for non-profit MFI's. The difference can be explained by the sustainability of for-profit MFI's. They are expected to earn profits from additional clients, exceeding the costs of these clients, while non-profit MFI's earn less than the costs of additional clients7. Other research has attempted
to find additional results showing a difference between for-profit and non-profit MFI’s (Numakhanova, 2015; Hartarska, 2007; Cull et al., 2007). However, they were unable to find significant results.
One explanation for the limited findings is that previous research makes a distinction between for-profit and non-profit, but it is no longer only for-profit MFI's who strive towards sustainability. Also, non-profit MFI's are pressured by donors to focus on sustainability (Conning, 1999). Because of this, the literature should no longer focus on the difference between for-profit and non-profit, but the distinction should be made between sustainable and non-sustainable MFI's. In addition to this, another dimension is introduced: MFI's who are in the transition of becoming financially sustainable. These three dimensions capture the different stages in which an MFI can be located, respectively, financially unsustainable MFI's, MFI's in the transition of becoming financially sustainable and sustainable MFI's. Unsustainable MFI's are expected to primarily focus on reaching the poorest borrowers (Churchill, 2019), and as a result, experience a trade-off for both depth and breadth of outreach. This formulates the following hypothesis:
Hypothesis 2: There is a trade-off between both the depth and breadth of outreach and
financial sustainability for unsustainable MFI’s.
MFI's who are in the transition of becoming financially sustainable are expected to focus on financial performance. As showed by Christen (2000) in Latin America, in becoming financially sustainable, MFI's focused primarily on increasing profitability. As a result, they are expected to focus less on the poorest borrowers to increase the profitability (Mersland and Strom, 2010), but in turn, increase the number of borrowers they can serve.
7. Concerning the depth of outreach, this can be explained by for-profit MFI's focusing on less poor
10 Hypothesis 3: There is a trade-off between depth of outreach and financial sustainability,
but the breadth of outreach and sustainability are complementary for MFI's in transition. Finally, MFI’s who have become financially sustainable can be described as both for-profit as well as non-profit. Following the reasoning from the financial systems approach, MFI’s in this stage can increase the supply of outreach (Robinson, 2001), meaning a complementary relationship between the breadth of outreach and sustainability. Additionally, MFI’s will be able to serve poorer borrowers as a result of higher efficiency and a decrease in cost (Mersland and Strom, 2010) while remaining financially sustainable.
Hypothesis 4: Both the breadth of outreach and the depth of outreach experience a
complementary relationship with financial sustainability when the MFI is financially sustainable.
4. Methodology
This study will use panel data, containing observations of multiple MFI's over multiple time periods, to test our hypotheses: the existence of a trade-off between outreach and financial sustainability. The model used is specified as:
𝑂𝑆𝑆𝑖𝑐𝑡 = 𝛼 + 𝛿𝑂𝑈𝑇𝑖𝑡+ 𝛽1𝐼𝑆𝑖𝑡 + 𝛽2𝐶𝑆𝑐𝑡+ 𝑈𝑖 + 𝑉𝑡+ 𝜇𝑖𝑡 (1)
𝑂𝑈𝑇𝑖𝑐𝑡 = 𝛼 + 𝛾𝑂𝑆𝑆𝑖𝑡 + 𝛽1𝐼𝑆𝑖𝑡+ 𝛽2𝐶𝑆𝑐𝑡+ 𝑈𝑖 + 𝑉𝑡+ 𝜀𝑖𝑡 (2)
11
As mentioned before, because outreach and financial sustainability are simultaneously determined, it is inappropriate to use a standard OLS regression. Therefore, the two equations mentioned above ((1) and (2)) need to be estimated simultaneously. Simultaneity occurs when two variables on both sides of the equation, influence each other at the same time (Brooks, 2019). Ignoring this issue would translate into simultaneity bias and create biased estimates. This can, in turn, lead to incorrect conclusions. To be able to simultaneously estimate the model, we need to identify the equations by adding instruments for both measures of outreach and the measure of financial sustainability. Adding these instruments gives us the following two models, which can be measured simultaneously:
𝑂𝑆𝑆𝑖𝑐𝑡 = 𝛼 + 𝛿𝑂𝑈𝑇𝑖𝑡+ 𝛽1𝐼𝑆𝑖𝑡 + 𝛽2𝐶𝑆𝑐𝑡+ 𝛽3𝐼𝑛𝑠𝑡𝑟. 𝑂𝑆𝑆𝑖𝑡+ 𝑈𝑖 + 𝑉𝑡+ 𝜇𝑖𝑡 (3)
𝑂𝑈𝑇𝑖𝑐𝑡 = 𝛼 + 𝛾𝑂𝑆𝑆𝑖𝑡 + 𝛽1𝐼𝑆𝑖𝑡+ 𝛽2𝐶𝑆𝑐𝑡+ 𝛽3𝐼𝑛𝑠𝑡𝑟. 𝑂𝑈𝑇𝑖𝑡+ 𝑈𝑖 + 𝑉𝑡 + 𝜀𝑖𝑡 (4)
Instr.OSS and Instr.OUT are two vectors containing instruments for operational self-sustainability and outreach. For an instrumental variable to be effective, it is required that it is independent of the error term. It should be not be related to the dependent variable in any other way than through the explanatory variable which is being instrumented while being related to the explanatory variable. Previous research has often implemented the lagged explanatory variable as an instrument (Mersland and Strom, 2010; Kar, 2013; Churchill, 2019). Mersland and Strom (2010) additionally added country-specific variables as instruments, such as GDP per capita, GDP growth and inflation, but find these instruments to be generally weak. This research will use a new set of country-specific variables. Anderson et al. (2011) introduced the use of higher-level averages as an instrument for firm-level endogenous variables. Following this approach, this study computes two types of instruments for the sustainability and outreach variables.
12
reasoning can be held for using the average outreach of a country as an instrument for outreach. While country-specific characteristics can influence the average depth and breadth of outreach, such as the number of people in a country living below the poverty rate (Xu et al., 2016). If a country has a high number of poor inhabitants, MFI's are expected to have higher outreach. An individual MFI, however, will have limited power to influence the country averages. The second instrument is the averages of sustainability and outreach based on the legal status and size of institutions per year. This variable is unlikely to have a relationship with the dependent variables, nor is it expected to be prone to influences of individual MFI’s. The relevant use of this instrument follows two lines of reasoning. First, MFI’s with the same legal status have multiple similarities. Estapé-Dubreuil and Torreguitart-Mirada (2015) show that governance mechanisms are significantly different among MFI's with different legal status. Previous studies have shown that governance mechanisms and legal status have an impact on the outreach and sustainability of companies (Harstarksa, 2005; Gutierrez-Nieto et al, 2007). Secondly, it has been shown in multiple studies that size affects both financial sustainability and outreach (Kar, 2013; Nurmakhanova, 2015; Churchill, 2019). MFI's with similar legal status might differ a lot in sustainability and outreach because they are different in size. Therefore MFI's are grouped into ten quadrants based on size and split into different legal status. While an average based on the combination between size and legal status can be a good indicator of the explanatory variables, it is unlikely that individual MFI’s can influence these averages. Using the two instruments as described above, are believed to be good measures to solve the simultaneity problems and independent of the error term. Because equation (3) and (4) are identified, it is possible to measure them simultaneously using a simultaneous equation model. This study will measure the equations using a two-stage least square approach (2SLS)8. Using the 2SLS option of
stata (for both equations) allows us to use fixed effects and clustered robust standard errors9.
To test the hypotheses about the effect of financial sustainability on outreach for MFI’s in different stages of sustainability, we rewrite equation (4) by implementing dummy variables in the regression, as well as interaction dummies combined with the measure of financial sustainability. We estimate the following model using 2SLS:
8 While the three-stage least square (3SLS) is generally considered being more efficient (Belsley, 1988),
2SLS can handle violations of i.i.d.8, for example, heteroskedasticity, serial correlation and clustered
robust standard errors, it also can easily control for fixed effects.
9 It is important to note that we do not use the simultaneous equation model option, but we do measure
13
𝑂𝑈𝑇𝑖𝑐𝑡 = 𝛼 + ∑ 𝛿𝑘(𝑂𝑆𝑆𝑖𝑡∗ 𝑆𝑇𝐴𝐺𝐸𝑖𝑡) + ∑ 𝛽𝑘𝑆𝑇𝐴𝐺𝐸𝑖𝑡 + 𝛽1𝐼𝑆𝑖𝑡+ 𝛽2𝐶𝑆𝑐𝑡+
𝛽3𝐼𝑛𝑠𝑡𝑟. 𝑂𝑈𝑇𝑖𝑡+ 𝑈𝑖 + 𝑉𝑡+ 𝜀𝑖𝑡
(5) Similar to previous equations (3) and (4) we use instruments to identify the equations. However, because we are only interested in the effect of financial sustainability on outreach for MFI’s in different stages of financial sustainability, we test only the rewritten equation (4). In addition to equations (4), STAGE represents dummy variables of the three different stages as described earlier. The first stage contains MFI’s who are unsustainable. The second stage contains MFI’s who are growing in sustainability and the last stage contains MFI’s who are sustainable. OSS*STAGE represents the interaction of the different stages with the explanatory variable of interest.
5. Data and descriptive statistics
This research contains institutional and country-specific data. The institutional specific data used in this research is obtained from MIX Market, a data source in the databank of the World Bank. It contains microfinance information for over 3000 MFI's from 1999 till 2019. The data available is self-reported by MFI's which might cause some bias (Chan, 2009). However, the data is heavily monitored by the Mix Market and therefore widely seen as accurate and reliable (Galema et al., 2011; Quayes, 2012). Because MFI's are not necessarily public institutions, it can be difficult to obtain large scale data regarding institutional specific information elsewhere, the MIX Market is therefore the most reliable data source available. The country-specific data is obtained from the World Bank database, except for the human development index10. Because the MIX Market dataset moved to the World Bank in 2019
(Hadi and Cull, 2020), minimal data was lost in merging the data from institutional and country-specific data, because they were already compatible. The data used in this research is an unbalanced panel dataset containing data of 2600 MFI's over a time period from 1999 to 2016. The timespan has been chosen because of data availability and to cover as many observations as possible. To preserve the panel dimension of the dataset, MFI's with less than three years of observations were removed from the analysis. Additionally, any country with less than 5 MFI's is removed to maintain the validity of using country average instruments. This leaves us with a final dataset containing 1726 MFI's from 1999 to 2016. This reflects roughly 66.5% of the total available data. When analyzing the remainder of the data, some outliers were discovered (see appendix 1). To reflect as much of the data as possible, the data has been winsorized11 with one percent to prevent the loss of valid data.
14 Figure 1: Amount of MFI’s over time
In figure 1 the increase of the MFI's over time is displayed. The amount of MFI's increased especially after 2002. Table 1 gives an overview of the number of observations of MFI's for each year, split into different legal status. The most observations are for non-bank financial institutions (NBFI) and non-governmental organizations (NGO's) with respectively 35.1% and 34.4% of all observations. They are followed by 13.8% of credit unions or cooperatives and 10.6% banks. Rural banks and other legal forms are the last occurring with a combined 6.0% of all observations. 0 500 1000 1500 2000 2500 Total MFI's 2000 2005 2010 2015 Years
Table 1: Observations per year per legal status
Years Bank Credit Union / Cooperative NBFI NGO Other Rural Bank Total
1999 26 7 33 37 103 2000 36 15 56 58 4 169 2001 38 44 83 81 1 3 250 2002 48 65 119 122 3 10 367 2003 57 97 171 225 4 47 601 2004 72 106 221 289 6 56 750 2005 86 129 272 356 6 75 924 2006 94 161 331 379 6 87 1058 2007 96 155 337 358 11 76 1033 2008 103 169 359 344 10 67 1052 2009 104 140 352 347 10 52 1005 2010 108 136 359 361 14 40 1018 2011 95 141 351 366 21 31 1005 2012 86 112 330 278 18 10 834 2013 89 100 323 232 17 13 774 2014 94 101 338 259 19 17 828 2015 86 83 315 260 16 18 778 2016 83 57 271 185 11 15 622 Total 1401 1818 4621 4537 173 621 13171
15 5.1 Endogenous variables
This research contains two endogenous variables, which are expected to influence the other simultaneously: Outreach and Financial sustainability. Both variables will be used as a dependent variable, with the other being an independent variable. Outreach can be separated into two forms of outreach, depth and breadth of outreach. Outreach, in general, is defined as the extension of loans and other financial services to the wider and poorer of the poor (Conning, 1999). In this definition, depth of outreach is focused on loans to the poorer of the poor. It measures how far clients are below the poverty line (Adhikary and Papachristou, 2014). Because the depth of outreach is not measurable itself a proxy is placed to measure its effect. The most widely used proxy for the depth of outreach is the average loan size (ALS) (Hermes et al, 2011). It follows the reasoning that the poorest people have a lower average loan size because they cannot afford it otherwise. As the average loan size increases, the depth of outreach decreases. Because this study compares MFI's across countries, the average loan size is taken as a fraction of the gross net income (GNI) to make it comparable across countries. Operational self-sustainability and the average loan size are expected to be positively related. As the OSS increases, the ALS increases as well, thus, lowering the depth of outreach. The second measure of outreach is the breadth of outreach. It captures to which extent microfinance institutions promote financial inclusion (Churchill, 2019). This is often reflected in how many clients are served (See Churchill, 2019; Adhikary and Papachristou, 2014; Quayes, 2012; Von Pischke, 1996). Following this approach, the breadth of outreach will be measured using the number of active borrowers (NAB). A borrower is defined as active if there is currently an outstanding loan balance with the institution or is primarily responsible for repaying any portion of the gross loan portfolio. A borrower will only be counted once, even if the borrower has multiple outstanding loans. An increase in the number of active borrowers is considered an increase in the breadth of outreach. The operational self-sustainability and the number of active borrowers are expected to be positively related. An increase in financial sustainability would lead to an increase in the breadth of outreach and vice versa.
16
use. OSS is measured as the operating revenues divided by expenses12 and multiplied by 100.
Therefore an MFI is considered unsustainable if OSS < 100.
This study will examine the relationship between sustainability and outreach for three groups of MFI's. These MFI's will be grouped based on the sustainability and the growth of sustainability from the previous year. It will be measured using the OSS and the growth rate of the OSS. The first stage contains MFI's who are not sustainable and experience limited growth in sustainability (thus, no sign of transitioning towards being sustainability or focusing on sustainability). An MFI is experiencing limited growth if the growth rate is below 15%, an estimation of the average growth rate of profitability in the microfinance sector during the sample period (the period 2000-2010 experienced a growth of 20% while the period after experienced a growth of 7% (Microfinance Barometer, 2019)). MFI's in the first stage has an OSS below 100 and have a growth rate below 15%. The second stage contains MFI's who are growing in sustainability. The MFI's in this stage are believed to focus on sustainability either because they focus on performance, or because they want to become sustainable. The MFI's in the second stage has a growth rate of OSS above 15%. The last stage contains MFI's who are already sustainable and is based on an OSS > 100 and a growth rate below 15%. In this stage, MFI's are believed to have reached their goal of sustainability, and will now continue to focus on increasing their outreach, while remaining financially sustainable.
Table 2 shows the summary statistics for the endogenous variables. Stage 1 has the lowest average loan size (ALS) when looking at the average and the median, and thus has a higher depth of outreach compared to other stages; stage 3 has the highest average loan size and thus performs worst in terms of depth of outreach. The breadth of outreach (NAB) seems to increase as MFI’s are placed in higher stages, which would carefully hint that the breadth of outreach is higher for sustainable MFI’s. Additionally, stage 3 has by far the most observations, indicating that considerably many observations display sustainability. When considering the mean and standard deviation of the variables, it shows extreme standard deviations for the outreach variables. This is due to the variables being very rightly skewed, indicating that a few cases on the far right side push the standard deviation up. Not solving this could wrongfully impact the regression. Therefore, the outreach variables will be transformed into natural logarithms13. Additionally, operational self-sustainability (OSS) will
also be transformed into a natural logarithm. This allows us to interpret the results as elasticity14.
12 OSS: Operating revenue / (Operating expense + Financial expense + Impairment losses on loans) * 100 13 Previous research has taken a similar approach (see Hermes et al., 2011; Kar, 2013).
14 Elasticity refers to ΔVariable1 = (ΔVariable2)^x, where x equals the coefficient. If x > 1 the change in
17 Table 2: Summary statistics for endogenous variables
Entire sample
Variable Observations Mean Median Std.Dev. Min Max
OSS 14766 114.11 111.54 40.30 15.80 287.09
ALS 15053 68.97 29.48 120.36 1.76 837.67
NAB 15263 74.11 8.22 411.09 0.00 8166.29
Stage 1
Variable Observations Mean Median Std.Dev. Min Max
OSS 3188 71.76 78.87 24.89 15.80 99.99
ALS 2878 61.31 24.90 110.87 1.76 837.67
NAB 2891 49.91 5.69 394.34 0.00 8166.29
Stage 2
Variable Observations Mean Median Std.Dev. Min Max
OSS 2363 121.32 115.57 50.81 18.57 287.09
ALS 2252 67.00 27.81 114.86 1.76 837.67
NAB 2261 53.41 6.54 332.14 0.00 7100.00
Stage 3
Variable Observations Mean Median Std.Dev. Min Max
OSS 9215 126.92 118.07 30.46 100.01 287.09
ALS 9923 71.63 31.38 124.07 1.76 837.67
NAB 10111 85.65 9.82 430.82 0.00 7290.00
OSS, operational self-sustainability; ALS, average loan size; NAB, number of active borrowers (x1000). Stage 1 contains MFI’s with OSS < 100 and a growth of OSS < 0.15. Stage 2 contains MFI’s with a growth of OSS > 0.15. Stage 3 contains MFI’s with OSS > 100 and a growth of OSS < 0.15. 5.2 Exogenous variables
18
Additionally, a set of dummy variables is added to denote the legal status of an MFI: Bank, Credit union / Cooperative, Non-bank financial institution (NBFI), non-governmental organisation (NGO), rural bank and other legal status. Previous research showed that NGO’s do better in terms of outreach (Hartarska, 2005), while financial institutions do better in terms of sustainability (Churchill, 2019).
19
are expected to have a higher impact on outreach. While MFI's operating in high social and economic developed countries are expected to do better in terms of financial sustainability.
Table 3: Summary statistics for exogenous variables
Institution specific variables
Variable Observations Mean Std.Dev. Min Max
DON 11951 94467 291481 0 2089070 TA 13177 15.638 2.078 10.601 20.749 NBFI 13213 .35 .477 0 1 Bank 13213 .106 .308 0 1 NGO 13213 .343 .475 0 1 RBank 13213 .047 .212 0 1 COOP 13213 .138 .344 0 1 Other 13213 .013 .114 0 1
Country specific variables
Variable Observations Mean Std.Dev. Min Max
REG 12823 39.773 17.078 2.913 92.718
PCGDP 12977 30.556 21.055 .35 149.196
FLABOUR 13115 39.075 9.379 8.014 55.658
HDI 11680 .621 .118 .263 .843
DON, donations; TA, total assets; NBFI, non-bank financial institution; NGO, non-governmental organisation; COOP, credit union/cooperative; REG, regulatory quality; PCGDP, private credit to gross domestic product; FLABOUR, female labour participation rate; HDI, human development index.
6. Results
First, we will measure equation (1) and (2) using an ordinary least squares regression (OLS). The results of these regressions should be interpreted with caution, as they are merely used to compare the results of not dealing with endogeneity and dealing with endogeneity. Table 4 displays the results. This regression contains the relationship between financial sustainability and the depth of outreach in Panel A and the relationship between financial sustainability and the breadth of outreach in Panel B.
20
with 0.22 percent. Again, these results should be interpreted with caution, due to possible endogeneity bias. Measuring the relationship between financial sustainability and outreach, considering endogeneity issues as done in equation (3) and (4) is shown in table 5.
Table 4: Financial sustainability and outreach (OLS regression)
VARIABLES PANEL A PANEL B
Depth of outreach Breadth of outreach
OSS ALS OSS NAB
Average Loan Size (ALS) 0.006
(0.014)
Operational self-sustainability (OSS) 0.012 0.230***
(0.029) (0.039)
Number of active borrowers (NAB) 0.090***
(0.015) Donations -0.010*** -0.000 -0.010*** 0.006 (0.003) (0.002) (0.003) (0.004) Total assets 0.117*** 0.186*** 0.048*** 0.742*** (0.011) (0.019) (0.014) (0.025) Regulatory quality 0.000 -0.003* 0.001 -0.000 (0.001) (0.002) (0.001) (0.002) Private credit to GDP -0.004*** 0.001 -0.004*** -0.003** (0.001) (0.002) (0.001) (0.002)
Female labour participation rate 0.006 0.036*** 0.009 -0.030***
(0.006) (0.009) (0.006) (0.010)
Human development index 1.675*** 2.076* 1.953*** -3.434***
(0.636) (1.127) (0.641) (1.202)
Constant 1.567*** -2.553*** 1.543*** 0.004
(0.504) (0.867) (0.505) (0.930)
Observations 10,463 10,463 10,487 10,487
Number of MFI’s 1,729 1,729 1,729 1,729
Entity FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Legal status Yes Yes Yes Yes
Standard errors are in parentheses. The legal status denotes the 6 legal statuses of MFI’s: NBFI, NGO, Bank, Rural Bank, Credit unions/Cooperatives and Other. Significance is denoted with: *** p<0.01, ** p<0.05, * p<0.1.
over-21
identification15, we can use the Hansen J test statistic to test if the instruments are
uncorrelated with the error term. We are unable to reject the null hypothesis (instruments are valid and uncorrelated with the error term) and therefore accept the instruments. Additionally, we test the strength of the instruments using an F-test of excluded instruments and reject the hypothesis at the 1 percent level that the excluded instruments together are equal to zero, also confirming the strength of the instruments.
When focusing on the relationship between financial sustainability and the depth of outreach (Panel A), we find that not controlling for endogeneity pushes the coefficients more towards a trade-off relationship. Controlling for endogeneity, we find the coefficients to be negative and significant. We find evidence that the depth of outreach and financial sustainability can be achieved simultaneously, rejecting part of the first hypothesis. As MFI’s become more financially sustainable, they are better able to reach the poorest borrowers. A 1 percent increase in average loan size decreases the operational self-sustainability with 0.08 percent. Similarly, an increase of operational self-sustainability with 1 percent decreases the average loan size with 0.09 percent. Because the average loan size decreases, the depth of outreach increases. The results show that when MFI’s shift to wealthier clients, they generally do worse in terms of financial sustainability, suggesting that these wealthier clients are more expensive or less profitable, possibly because they require more extensive monitoring or require a lower interest rate as their marginal returns are lower (Armendáriz and Morduch, 2010). Additionally, when MFI's become more financially sustainable, they do so by increasing profits or reducing costs, by focusing on poorer clients. While this evidence does not show a large impact it does confirm that microfinance institutions can focus on financial sustainability and simultaneously expand their outreach. These findings are in line with previous findings in the literature (Ayayi and Sene, 2010; Kar, 2013; Quayes; 2012). However, Quayes (2012) found the relationship to be much stronger but did implement clustered standard errors and fixed effects as he used a three-stage least squares method. The results partly support the findings of Nurmakhanova (2014) who found no evidence of a trade-off but no complementary effects either and reject the findings of Churchill (2019) who found a trade-off between depth of outreach and financial sustainability.
The results of Panel B display the relationship between financial sustainability and the breadth of outreach. The results do not show a clear relationship between financial sustainability and the breadth of outreach. An increase in the number of active borrowers has a negative effect on operational self-sustainability. An increase in active borrowers of 1 percent leads to a decrease in operational self-sustainability of 0.08 percent. These results also contradict the first hypothesis, which states that a complementary relationship was
15 Over identification happens when the number of instruments (L) exceeds the number of regressors (K):
22
expected between financial sustainability and the breadth of outreach. This indicates that the returns obtained per borrower are not enough to cover the costs of borrowing, as increasing the number of borrowers and thus, increasing the breadth of outreach lowers financial sustainability.
Table 5: Financial sustainability and outreach (2SLS regression)
VARIABLES PANEL A PANEL B
Depth of outreach Breadth of outreach
OSS ALS OSS NAB
Average Loan Size (ALS) -0.087***
(0.029)
Operational self-sustainability (OSS) -0.096* 0.187***
(0.056) (0.067)
Number of active borrowers (NAB) -0.082*
(0.048) Donations -0.009*** -0.003 -0.009*** 0.004 (0.002) (0.002) (0.002) (0.004) Total assets 0.092*** 0.153*** 0.136*** 0.644*** (0.012) (0.018) (0.039) (0.032) Regulatory quality -0.000 -0.002 -0.000 -0.000 (0.001) (0.002) (0.001) (0.002) Private credit to GDP -0.002** -0.001 -0.002** -0.002 (0.001) (0.001) (0.001) (0.002)
Female labour participation rate 0.013** 0.030*** 0.008 -0.025***
(0.006) (0.008) (0.006) (0.010)
Human development index 0.353 0.240 -0.088 -3.587***
(0.548) (0.923) (0.600) (1.123) Legal averages 0.448*** 0.130*** 0.456*** 0.098*** (0.041) (0.025) (0.043) (0.019) Country averages 0.742*** 0.531*** 0.746*** 0.184*** (0.041) (0.034) (0.044) (0.022) Constant -2.733*** -2.267*** -2.520*** -1.140 (0.473) (0.712) (0.495) (0.879) Observations 10,463 10,463 10,487 10,487 Number of MFI’s 1,727 1,727 1,729 1,729
Hansen J-test, p-value 0.133 0.687 0.657 0.975
F-test of excluded instruments, p-value 0.000 0.000 0.000 0.000
Entity FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Legal status Yes Yes Yes Yes
23
These results also contradict the findings of previous research (see Adhikary and Papachristou, 2014; Schäfer and Fukasawa, 2011). Surprisingly, an increase in financial sustainability shows a positive effect on the number of active borrowers. An increase in 1 percent of operational self-sustainability leads to an increase in active borrowers of 0.18 percent. This would confirm part of our first hypothesis, when MFI's focus more on financial sustainability, they also reach more borrowers. These results are in line with the findings of Churchill (2019). However, because we find opposing results when considering the number of active borrowers and the operational self-sustainability as the explanatory variable, we cannot confirm a complementary effect.
Additionally, we found some results for our control variables. Because these variables (except for total assets) are not transformed to logarithms, we need to interpret them as follows: a unit increase in the explanatory variable leads to a percentage increase in the dependent variable, where the percentage increase is calculated as 𝑒𝛽 with β being the
coefficient of the explanatory variable. For the institutional specific variables, we find donations to be negatively related to financial sustainability as expected. However, the donations do not explain the level of outreach, while one would expect that MFI's with higher donations would be able to reach higher outreach. The size of MFI’s is significantly positive for all dependent variables. This is in line with findings of previous research (see Cull et al., 2007; Nurmakhanova, 2014). As the size of MFI's increase, they generally do better in terms of sustainability and outreach. For the country-specific variables, we find that regulatory quality does not explain either outreach or financial sustainability. This contradicts the findings of Mueller and Uhde (2010) and Halouani and Boujelbène (2015). Private credit to GDP captures the financial development of a country (Ahlin et al., 2011). Contradictory to previous results, we find that as the financial development of a country increases, the financial sustainability of MFI’s in that country decrease. The female labour participation rate is significantly positive at the 1 percent level for the depth of outreach. This is in line with the reasoning that a high female participation rate reflects the poverty of a country (Verick, 2014). The female labour participation rate is, however, significantly negative for the breadth of outreach. If there is a lot of poverty in a country, MFI’s generally do worse in terms of reaching borrowers. This is confirmed by the negative coefficient for the human development index. As a country is less socially and economically developed, MFI’s in this country do worse in terms of the breadth of outreach.
To test the trade-off between sustainability and outreach between different stages of financial sustainability (see equations (5)), we run both an OLS regression as well as a 2SLS fixed effects regression again. The results for the 2SLS regression are displayed in Table 616. The
OLS regression can be found in Appendix 4. The coefficients should be interpreted as
24
follows: OSS shows the baseline coefficient, representing the relationship in the first stage. The coefficients in stage 2 and 3 show the difference from the baseline. The coefficients for the second and third stage are obtained by adding the displayed coefficient to the baseline: ALS + Stage 2 * ALS. A full overview of these coefficients can be found in table 7,
including differences between the second and third stage.
Table 6: Financial sustainability and outreach using stage interactions
VARIABLES ALS NAB
Operational self-sustainability (OSS) -0.024 0.334***
(0.075) (0.106)
Stage 2 * Operational self-sustainability (OSS) -0.016 -0.199**
(0.065) (0.085)
Stage 3 * Operational self-sustainability (OSS) -0.490** -0.503*
(0.229) (0.288) Stage 2 0.061 0.871** (0.288) (0.368) Stage 3 2.347** 2.347* (1.081) (1.358) Legal averages 0.127*** 0.101*** (0.025) (0.019) Country averages 0.538*** 0.184*** (0.034) (0.022) Constant -2.600*** -1.787* (0.776) (0.992) Observations 10,487 10,487 Number of MFI’s 1,729 1,729
Hansen J-test, p-value 0.953 0.938
F-test of excluded instruments, p-value 0.000 0.000
Entity FE Yes Yes
Year FE Yes Yes
Legal status Yes Yes
Control variables Yes Yes
Standard errors are in parentheses. The country averages and legal averages denote the averages of the country and legal status for the dependent variable: ALS and NAB respectively. They are used as instruments to identify the equation. The Hansen (1982) J-test tests if the instruments are independent of the error term, and thus dealing with the endogeneity. A low p-value (< 0.1) rejects the independence of the instruments. The Wald F-test tests the hypothesis that all coefficients together are equal to zero, testing the validity and strength of the instruments (Greene, 2003). Significance is denoted with: *** p<0.01, ** p<0.05, * p<0.1.
25
and sustainability has disappeared for the second stage. The coefficients are significantly different from the first stage, but not significantly different from zero, indicating that there is neither a trade-off nor a complementing relationship between the breadth of outreach and sustainability. For the third stage, we find significant evidence of a trade-off between financial sustainability and the breadth of outreach.
More importantly however, are the results shown in table 6, as this regression controls for endogeneity concerns. Similar results are found for the relationship between the breadth of outreach and financial sustainability; however, the results drastically differ for the relationship between the depth of outreach and financial sustainability. Unsustainable MFI's, as displayed in the first stage, do not experience a trade-off between the depth of outreach and financial sustainability. The same is found for MFI’s that focus on their financial performance, stage 2: we find no evidence of a trade-off. This indicates that MFI’s can focus on either outreach or focus on financial performance, without risking the other. In the third stage, where MFI’s are sustainable, we find that the depth of outreach and financial sustainability have become complementary. We find that for sustainable MFI’s an increase in financial sustainability with 1 percent, leads to a decrease in the average loan size, and thus an increase in the depth of outreach of 0.46 percent, which is a significant and important practical finding. These results can imply that sustainable MFI's can utilize their funds more efficiently and effectively than unsustainable MFI's, resulting in the ability to target poorer borrowers, while at the same time benefit financially.
26 Table 7: F-tests for 2SLS regressions
VARIABLES ALS NAB
Stage 2 * Operational self-sustainability (OSS) -0.039 0.135*
Stage 3 * Operational self-sustainability (OSS) -0.513** -0.169
Stage 2 OSS = Stage 3 OSS -0.474** -0.304
Table 7 displays the results from F-tests with the assumption that the coefficients are significantly different from zero. Stage 2 = Stage 3 tests if the coefficients of stage 2 and stage 3 are significantly different from each other. Significance is denoted with: *** p<0.01, ** p<0.05, * p<0.1.
Overall, we reject our second hypothesis. We find evidence that unsustainable microfinance institutions do not experience a trade-off in the depth of outreach and financial sustainability, and that they experience a complementary relationship with the breadth of outreach and financial sustainability. These findings reject the previous findings that MFI's cope with a trade-off between sustainability and reaching the poorest borrowers (see, Hermes et al., 2011; Churchill, 2019). We partly accept our third hypothesis, as we do find evidence of a complementary relationship between financial sustainability and the breadth of outreach. However, unlike what we expected, this relationship developed for the worse compared to the first stage. Finally, we partly accept our final hypothesis. While the complementary effect between the breadth of outreach and financial sustainability has faded away, we find that sustainable MFI's do experience complementary effects between outreach and financial sustainability. This leads to an important conclusion, as unsustainable MFI's can focus on sustainability through which they can increase their breadth of outreach, and once sustainable, they can target poorer borrowers.
7. Conclusion
27
relationship between the breadth of outreach and financial sustainability. This indicates the importance of controlling for endogeneity. Finding complementary effects between depth of outreach and financial sustainability is not a new phenomenon. Existing literature has found evidence of this (see Quayes, 2012; Mersland and Strom, 2010). However, they failed to appropriately address endogeneity concerns. Studies which addressed endogeneity by implementing different instruments found mixed results (see Churchill, 2019; Numakhanova et al., 2015). This study supplements the current literature using higher-level averages as instruments and uses exogenous control variables. We find evidence of the existence of complementary effects between financial sustainability and outreach and show that the drive towards sustainability can improve reaching the poorest borrowers, the group for which microfinance is most important. The relationship between the breadth of outreach and financial sustainability has received indubitably less attention than its counterpart. The number of borrowers has often been used in research examining determinants of sustainability (see Ayayi and Sene, 2010; Schäfer and Fukasawa, 2011) and was found to complement sustainability. The evidence of this study, however, does not find a clear relationship between the two and we are therefore unable to conclude the effect of increasing sustainability on the breadth of outreach.
28
sustainable, they can increase their depth of outreach simultaneously with financial sustainability.
The difference between the findings for the relationship between financial sustainability and the breadth of outreach forces to ask what causes these contradicting results. One possible explanation can be that the instruments used to determine the number of active borrowers did not correctly explain the actual number of active borrowers. This could result in a different outcome when using the number of active borrowers as an independent variable. Since the breadth of outreach has received by far less attention than the depth of outreach, we advise future research to focus more on determining the breadth of outreach.
Additionally, the results provided in this study, reveal promising evidence which can open the door to further research, specifically focusing on using another set of instruments to analyse different stages of microfinance institutions. An important concern to mention is the endogeneity concerns that arise because of the different stages. Similar to the financial sustainability and the outreach being determined simultaneously, the different stages are based on financial sustainability and are therefore also determined simultaneously with financial sustainability and outreach. Further research should concentrate on defining the stages while dealing with arising endogeneity concerns. Furthermore, as the evidence suggests: the relationship between financial sustainability and the breadth of outreach may differ as MFI's become sustainable, implying the possibility for a non-linear relationship. Additional research can focus on a non-linear relationship, which can help to determine the turning point in the relationship between financial sustainability and outreach.
29
8. Appendix
Appendix 1: Outliers
30 Appendix 2: Variable definitions
Variable Definition Source
Operational self-sufficiency (OSS)
OSS = Operating revenue / (Operating expense + Financial expense + Impairment losses on loans). Measures how well the MFI can cover its costs with its operating revenue
MIX Market
Average Loan Size (ALS) Average loan size / GNI per capita (%) MIX
Market Number of active borrowers
(NAB)
The number of individuals (x1000) who currently have an outstanding loan balance with the MFI
MIX Market
Donations (DON) The number of donations an MFI receives (x1000) MIX
Market
Size The logarithm of the total assets of an MFI MIX
Market
NBFI A dummy variable taking the value of 1 if the legal status is an
NBFI and 0 otherwise
MIX Market
Bank A dummy variable taking the value of 1 if the legal status is a
Bank and 0 otherwise
MIX Market
NGO A dummy variable taking the value of 1 if the legal status is an
NGO and 0 otherwise
MIX Market
RBank A dummy variable taking the value of 1 if the legal status is a
Rural bank and 0 otherwise
MIX Market
COOP A dummy variable taking the value of 1 if the legal status is a
credit union/cooperative and 0 otherwise
MIX Market
Other A dummy variable taking the value of 1 if the legal status is
another legal form and 0 otherwise
MIX Market Regulatory quality (REG)
The ability of the government to formulate and implement policies and regulations that permit and promote private sector development
World Bank Private credit to GDP
(PCGDP)
The domestic credit to the private sector to measure the financial development of a country
World Bank Female labour participation
rate (FLABOUR) The percentage of female labour to the entire labour force
World Bank Human development index
(HDI)
An index to measure the social and economic development of a country
31 Appendix 3: Financial sustainability and outreach (2SLS regression, full)
VARIABLES PANEL A PANEL B
Depth of outreach Breadth of outreach
OSS ALS OSS NAB
Average Loan Size (ALS) 0.073**
(0.028) [0.201]
Operational self-sustainability (OSS) 0.244*** -0.599
(0.090) (0.407)
[0.088] [-0.064]
Number of active borrowers (NAB) -0.018**
(0.009) [-0.167] Donations -0.007*** -0.000 -0.008*** -0.036** (0.003) (0.002) (0.003) (0.018) [-0.052] [-0.001] [-0.055] [-0.027] Total assets 7.793*** 6.839*** 9.808*** 80.226*** (0.933) (1.855) (1.196) (19.482) [0.414] [0.131] [0.520] [0.455]
Non-bank Financial Institution (NBFI) -43.378 10.984 -42.784 3.706
(28.927) (10.093) (28.923) (34.038)
[-0.544] [0.050] [-0.536] [0.005]
Bank 10.346*** 2.062 12.135*** 98.719***
(3.073) (6.510) (3.343) (37.329)
[0.079] [0.006] [0.093] [0.081]
Non-governmental Organisation (NGO) 19.782* -3.317 20.235** -34.177
(11.058) (3.596) (9.810) (50.458) [0.250] [-0.015] [0.255] [-0.046] o. Rural Bank - - - - COOP 8.523** -9.953*** 9.172** 73.087* (3.649) (3.186) (4.277) (43.095) [0.081] [-0.034] [0.087] [0.074] o. Other - - - - Regulatory quality 0.065 -0.404** 0.020 -1.124* (0.086) (0.186) (0.083) (0.619) [0.028] [-0.064] [0.009] [-0.053] Private credit to GDP -0.381*** 0.474* -0.355*** -0.586 (0.092) (0.246) (0.090) (0.541) [-0.204] [0.092] [-0.190] [-0.034]
Female labour participation rate 0.361 1.181 0.131 -18.638*
(0.543) (0.989) (0.556) (10.277)
32 Appendix 3: continued.
Human development index 163.524*** -21.582 173.674*** 856.605
(57.935) (141.998) (60.505) (616.188) [0.513] [-0.024] [0.544] [0.287] Constant -123.930** -117.971 -147.525*** -865.028** (49.172) (111.830) (51.462) (406.180) Observations 10,463 10,463 10,487 10,487 Number of MFI’s 1,729 1,729 1,729 1,729
Hansen J-test p-value 0.679 0.218 0.442 0.262
F-test of excluded instruments, p-value 0.000 0.000 0.000 0.000
Entity FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Standard errors are in parentheses and standardized coefficients in brackets. The Hansen (1982) J-test tests if the instruments are independent of the error term, and thus dealing with the endogeneity. A
low p-value rejects the independence of the instruments. The F-test tests the hypothesis that all coefficients together are equal to zero, testing the validity and strength of the instruments (Greene, 2003).Variables starting with o. were omitted as a result of Multicollinearity. Significance is denoted
33 Appendix 4: Financial sustainability and outreach using stage interactions (OLS)
VARIABLES ALS NAB
Operational self-sustainability (OSS) 0.011 0.357***
(0.042) (0.064)
Stage 2 * Operational self-sustainability (OSS) -0.003 -0.182***
(0.038) (0.055)
Stage 3 * Operational self-sustainability (OSS) 0.044 -0.349***
(0.073) (0.086) Stage 2 0.009 0.786*** (0.171) (0.241) Stage 3 -0.215 1.604*** (0.338) (0.395) Constant -2.546*** -0.559 (0.884) (0.959) Observations 10,463 10,463 Number of MFI’s 1,729 1,729
Hansen J-test, p-value
F-test of excluded instruments, p-value
Entity FE Yes Yes
Year FE Yes Yes
Legal status Yes Yes
Control variables Yes Yes
VARIABLES F-tests
ALS NAB
Stage 2 * Operational self-sustainability (OSS) 0.008 0.175***
Stage 3 * Operational self-sustainability (OSS) 0.055 0.008
Stage 2 OSS = Stage 3 OSS 0.048 -0.167***
34 Appendix 5: Financial sustainability and outreach using stage interaction (full)
VARIABLES PANEL A PANEL B
OLS Regression 2SLS Regression
ALS NAB ALS NAB
Operational self-sustainability (OSS) 0.011 0.357*** -0.024 0.334***
(0.042) (0.064) (0.075) (0.106)
Stage 2 * Operational self-sustainability
(OSS) (0.038) -0.003 -0.182*** (0.055) (0.065) -0.016 -0.199** (0.085)
Stage 3 * Operational self-sustainability
(OSS) (0.073) 0.044 -0.349*** (0.086) -0.490** (0.229) -0.503* (0.288) Stage 2 0.009 0.786*** 0.061 0.871** (0.171) (0.241) (0.288) (0.368) Stage 3 -0.215 1.604*** 2.347** 2.347* (0.338) (0.395) (1.081) (1.358) Donations -0.000 0.000 -0.000 0.000 (0.000) (0.000) (0.000) (0.000) Total assets 0.186*** 0.737*** 0.147*** 0.635*** (0.019) (0.026) (0.018) (0.032)
Non-bank Financial Institution (NBFI) -0.062 0.194*** 0.142 0.210***
(0.054) (0.074) (0.134) (0.047)
Bank 0.248*** -0.294*** 0.451*** -0.339***
(0.062) (0.067) (0.054) (0.066)
Non-governmental Organisation (NGO) -0.018 -0.060 -0.225 0.072
(0.100) (0.118) (0.243) (0.144) o. Rural Bank - - - - COOP -0.117*** 0.113 0.015 0.128 (0.036) (0.100) (0.052) (0.110) o. Other - - - - Regulatory quality -0.003* -0.000 -0.002 -0.000 (0.002) (0.002) (0.002) (0.002) Private credit to GDP 0.001 -0.003** -0.001 -0.002 (0.002) (0.002) (0.001) (0.002)
Female labour participation rate 0.036*** -0.030*** 0.031*** -0.025**
(0.009) (0.010) (0.008) (0.010)
Human development index 2.072* -3.314*** 0.310 -3.446***
(1.130) (1.204) (0.930) (1.135)
Legal average ALS 0.127***
(0.025)
Country average ALS 0.538***
