THE PROFIT ORIENTATION OF MICROFINANCE INSTITUTIONS. Peter W. Roberts * Goizueta Business School. Emory University.

Transcription

1 THE PROFIT ORIENTATION OF MICROFINANCE INSTITUTIONS AND EFFECTIVE INTEREST RATES Peter W. Roberts * Goizueta Business School Emory University 1300 Clifton Road, Atlanta, GA, peter_roberts@bus.emory.edu * The author is grateful for the helpful comments provided by David Kyle, Anand Swaminathan and Peter Thompson. 1

2 THE PROFIT ORIENTATION OF MICROFINANCE INSTITUTIONS AND EFFECTIVE INTEREST RATES Since the arrival of for profit microfinance institutions (MFIs), commentators have been asking whether the sector benefits by MFIs adopting a stronger profit orientation. We address this question by analyzing their adoption of the for profit legal form, appointing private sector representation and banking acumen to MFI boards, and participation in more extensive for profit networks. The results consistently indicate that a stronger for profit orientation corresponds with higher effective interest rates for MFI clients. However, these effects do not lead to greater profitability and therefore sustainability because these variables are also associated with increases in the major elements of an MFI s costs. Key Words: microfinance, global, interest rates, nonprofit 2

3 Some Fear Profit Motive to Trump Poverty Efforts in Microfinance New York Times headline, August 28, INTRODUCTION Microfinance institutions (MFIs) are banking organizations whose primary purpose is that of providing financial services to poor and otherwise marginalized clients (Mersland & Strøm, 2010). Collectively, the microfinance sector is lauded for modifying standard banking practices in order to effectively extend credit to the poor and in doing so helping to elevate their standards of living. More specifically, the innovations that led to the modern microfinance movement overcame two problems previously thought to prohibit lending to the poor: small loan sizes and little or no collateral (Armendariz & Morduch, 2005). The recent evolution of the microfinance sector can be viewed in terms of the rapid growth in the number of active MFIs, increases in the volume of business they conduct, a broader range of financial services on offer, and changes in the types and motivations of MFIs. In this latter respect, an important marker in the sector s evolution is the arrival of profit oriented MFIs. Although lauded by many as critically important for the maturation of the sector, these MFIs also ushered in debates about whether it is possible to effectively blend nonprofit ideals and for profit orientations and practices (Morduch, 2000). More practically, these debates are rooted in questions about whether MFIs with stronger profit orientations are better able to sustainably address the needs of poor borrowers. Some commentators clearly place emphasis on the anti poverty orientation of MFIs: the first goal of MFIs is to reach more clients in the poorer strata of the population, and the second goal is financial sustainability (Mersland & Strøm, 2008a, pg. 663). However, it is also believed that a large number of (especially nonprofit) MFIs are not earning sufficient income to cover their full costs of operation and expansion and must therefore rely on subsidies in the form of grants and donations to sustain themselves. Concerns about the reliability of these funding sources makes the financial 3

4 viability of MFIs a major concern for sector participants (Mersland & Strøm, 2010). In this respect, profit oriented MFIs are part of the movement toward a more businesslike microfinance sector. With heightened business acumen and a stronger market orientation, profitoriented MFIs are supposed to set more appropriate loan prices and deliver greater efficiencies, and thus have an easier time attracting needed investment into the sector. This should, in turn, allow the social impacts that they generate to be more sustainable (Hermes & Lensink, 2007). The arrival of profit oriented MFIs also raises concerns about MFIs trading off social for financial performance. In contrast, a greater focus on profitability might push concerns about the well being of poor clients to the back burner. While these concerns clearly apply to for profit MFIs, they also apply to nonprofit MFIs where attention to sustainability are leading some to emphasize the generation of financial surplus, even if that surplus is never distributed to outside shareholders. In both cases, many worry that we will see MFIs abandoning the poorest clients in search of more reliable profit streams something commentators call mission drift (Copestake, 2007). Even MFIs that remain focused on the poorest clients might alter the mix of costs and benefits that they offer as they strive for enhanced profitability (Yunus, 2011). We address this latter question by examining the impact of MFIs having a stronger profit orientation on the effective interest rates charged to clients. While this is not the only variable that matters to microfinance clients (Cull, Demirguc Kunt, & Morduch, 2009), it does capture the effective price of credit access. We examine differences between the effective interest rates charged by nonprofit versus for profit MFIs. However, we also recognize that the profit orientation of an MFI extends beyond its decision to operate as a for profit organization. It is also manifested in the more subtle choices and commitments that an MFI makes. In particular, we examine the composition of MFI governing boards to ascertain whether they contain private sector representation and/or individuals with banking acumen. We also examine the extent to which the network support organizations (Cook & Isern, 2004, pp., pg. 3) that an MFI participates in are themselves populated by numerous other for 4

5 profit MFIs. In the end, our analyses paint a somewhat sobering picture of the influence of a stronger profit orientation on the effective interest rates charged to MFI clients. Each of these three variables is associated with higher effective interest rates. However, these increases do not manifest in higher MFI profitability and therefore sustainability because the MFIs that are more profit oriented also tend to have higher costs. 2. PROFIT ORIENTATION OF MICROFINANCE INSTITUTIONS The collective push to see a more profit oriented microfinance sector is most evident in the relatively high incidence of for profit MFIs around the world. In 2009, 490 of the 1,169 MFIs (42%) in the MIX Market database were for profit MFIs. They collectively controlled roughly two thirds of the more than $65 billion in assets deployed in that year. Clearly, adopting a for profit legal form suggests a stronger profit orientation. However, the decision to operate as a for profit organization is not the only choice that indicates the profit orientation of MFIs. In fact, Mersland & Strøm (2008b) recently concluded that MFI ownership (e.g., shareholder versus NGO) is not particularly relevant in determining its social or commercial orientation. Rather, as Cull et al. (2009) suggest, earning profits does not imply being a for profit bank. Nonprofit MFIs can and do earn positive profits that are simply not distributed to shareholders but are re invested in activities that further service their clients. Therefore, we also look at several other organizational choices that plausibly correspond with an MFI s profit orientation. Appointing individuals to the board of directors represents an important strategic decision for MFIs. Advice from and oversight by these board members have consequences for decisions taken within an MFI. Those interested in adopting best practices from the for profit world and from the traditional banking sector might therefore tend to appoint individuals to their boards who bring experience from these domains. More specifically, appointing individuals from the private sector, as opposed to the government or NGO sectors, indicates a desire to be influenced in this direction. The 5

6 same can be said about appointing individuals with banking acumen. Thus, having private sector representation and banking acumen on the board suggests a stronger profit orientation. Another way that MFIs deepen their commitment to a profit orientation is by participating in networks comprised of other for profits MFIs. Sociological and managerial research on networks indicates that the behavior and performance of organizations is influenced by the networks in which they participate (Brass, Galaskiewicz, Greve, & Tsai, 2004). From an organizational learning perspective, the ties that make up organizational networks are conduits through which knowledge and ideas flows. This effect is evident in a statement taken from the website of one prominent microfinance network: MicroFinance Network (MFN) is an international association of leading microfinance institutions. Through the MFN, 31 members from 27 countries share ideas, experiences, and innovative solutions to the challenges they face in search of continuous growth and progress. MFN members seek to be models of what is possible in the industry ( The latter aspiration points to a second control aspect of networks. The predominant participants in an organizational network tend to define the norms and practices in that network (Owen Smith & Powell, 2004). If other MFIs that populate microfinance networks are largely for profit MFIs, then the orientations and ideas that tend to flow through those networks will support and reinforce a stronger profit orientation among network participants. This suggests that participating in networks comprised of more for profit MFIs suggests a stronger profit orientation. Assuming these variables indicate the profit orientation of an MFI, the question becomes how this orientation influences behavior and performance. The optimistic view is that social impacts will be improved by incorporating more market discipline and commercial acumen into the traditionally non profit microfinance sector. In other words, the lure of profit, economists assume, motivates all entrepreneurs and managers and fosters efficient decision making by private firms (Weisbrod, 1998, p. 70). The pessimistic view suggests that this orientation is either distracting of detracting from the pursuit of poverty reduction as an organizational goal. After all, a nonprofit 6

7 organization has little incentive to skimp on quality of output or otherwise take advantage of poorly informed customers (Weisbrod, 1998, p. 70). In the former case, a focus on profits tends to lead MFIs away from the commitment that improves their ability to effectively lend to the poor while the latter suggests that a desire to improve profitability will lead to deliberate choices to cut services or raise interest rates to meet the demands of higher profitability. The middle ground between these two positions (and thus the null hypothesis in all that follows) suggests that debates about the implications of profit seeking are quite irrelevant. Mersland and Strom s (2008a) analysis found that the impact of an MFI s adopting a for profit orientation on its performance is effectively null. This centrist position is based on the belief that nonprofit and profit oriented MFIs can be equally concerned with both alleviating poverty and financial sustainability: A priori, one would consider that SHFs are more profit oriented than NGOs. Similarly, that NGOs should care more about reaching the poorest clients than SHFs However, an alternative hypothesis may be that SHFs and NGOs do not perform differently, because they may use the same business model to compete and serve customers in the microfinance market (Mersland & Strøm, 2008b). Given the unsettled debates and the relative paucity of systematic empirical analysis (Hermes & Lensink, 2007, pg. F2), the following sections provide a comprehensive analysis of the implications for effective interest rates of an MFI having a stronger profit orientation. 3. DATA AND SAMPLE Our analysis combines two different MIX data sources ( their archive of MFI financial information and their more recent Social Performance Reports. These latter reports capture detailed information about specific choices and configurations that pertain to an MFI s social orientation. The 358 MFIs in our sample all completed Social Performance Reports in 2008 or 2009 and had corresponding financial data for To assess the representativeness of this 7

8 sample, we compare it to the broader sample of all MFIs in the 2009 MIX financial information file and find it to be broadly comparable (see table 1). Table 1 about here Our analysis focuses on one variable that has direct implications for the clients of MFIs the effective interest rate charged on the funds that they borrow. Following accepted practice (Gonzalez, 2010), our effective interest rate variable is real total earned interest income and fees divided by the average gross loan portfolio. Among the sampled MFIs, the average effective interest rate is 28.06%. The MIX data also report the profit status of each MFI. We use this information to create a For profit MFI dummy variable set to one for MFIs that operate as for profit organizations. According to table 1, 35% of the sampled MFIs are for profit organizations. The MIX Social Performance Reports ask respondents several questions about the composition of their boards. Two questions are salient for this analysis. The first asks whether an MFI s board has private sector representation and the second asks whether banking acumen is present on the board. We use this information to create two additional dummy variables: Private sector on board and Banking acumen on board. Finally, the MIX website reports on the composition of roughly 100 network support organizations that work with MFIs around the world. These networks facilitate and provide support to organizations that are committed to delivering financial services to the poor, and can be national, regional or international in orientation (Cook & Isern, 2004, pp., pg. 3). They range in size from four to more than 150 members and provide a range of services to their members, including financial and technical services, knowledge management, research and development and policy advocacy. After recording the names of all MFIs participating in each network, we count the total number of for profit MFIs that a focal MFI is tied to by shared network affiliation. The maximum number of for profit ties is 56 and so we divide the observed number of ties for each sampled MFI 8

9 by 56 to obtain a normalized variable that ranges from zero (no ties to for profit MFIs) to one (the maximum number of observed ties). Differing social and economic conditions across countries and regions have implications for the supply of and demand for micro lending (Ahlin, Lin, & Maio, 2011). We control for locating across the six regions isolated in the MIX databases Africa, East Asia and the Pacific, Eastern Europe and Central Asia, Latin America and The Caribbean, Middle East and North Africa and South Asia with a series of region fixed effects. At the country level, we account for local economic and political problems by taking the average of the six dimensions of the World Bank s Worldwide Governance Indicator ( voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law and control of corruption. Because total population influences the demand for microcredit, we also control for the natural log of each country s total population in Prevailing interest rates should also be influenced by the degree of microfinance sector competition. Following a large body of organizational ecology research (Hannan & Freeman, 1989), we proxy for the degree of competition by counting the number of MFIs active in each country in 2009 (as reported in the MIX database). The overall count ranges from a low of one (in Tunisia) to a high of 163 (in India). Given the debates about the differential competitiveness of for profit and nonprofit MFIs, we decompose this variable into for profit and nonprofit density measures. In doing so, we see that some countries, like India, have large numbers of both for profits (62) and nonprofits (95). Mexico has many for profits (55) but relatively few nonprofits (ten). On the other hand, Bangladesh has many nonprofits (70) but relatively few for profits (3). Due to economies of scale and experience effects, larger and older MFIs should be more efficient (Gonzalez, 2007). We therefore account for the size (natural log of assets) and age (natural log of the years since an MFI started its microfinance operations) of each MFI. There is also interest in the role played by regulation in shaping the behavior and performance of MFIs (Cull, Demirguc 9

10 Kunt, & Morduch, 2011). We include a dummy variable set to one for banks that reported being under the influence of a regulatory authority in MFIs also vary in the complexity and scope of their offerings. In this regard, two variables that might influence the cost and performance of MFIs are the extent to which they also engage in deposit taking activities and their outreach levels. To capture the former effect, we include a variable that measures the savings deposits to assets ratio for each MFI. We proxy for the degree of outreach (in terms of number of individuals touched) by including another dummy variable set to one for MFIs that are reported in the MIX database as having either medium or large outreach levels. Commentators also note that costs and therefore interest rates can be influenced by several variables associated with the degree of difficulty associated with providing micro loans across different client segments (Mersland & Strøm, 2010, pp., pg. 35). Absent direct measures of the extent to which an MFI targets marginalized clients, accepted proxies include average loan size, targeting women borrowers and targeting individuals in rural areas (Cull, et al., 2009). Our average loan size variable is the median loan size as percentage of country gross national income per capita (Gonzalez, 2010). The fraction of women borrowers is reported in MIX as the share of outstanding loans held by women borrowers. The Social Performance Reports ask respondents whether their MFI targets rural clients. We use this information to create another dummy variable. Finally, published mission statements provide some indication of the extent to which poverty alleviation is a central concern for an MFI. We examine the mission statement of each MFI and isolate those that explicitly mention poor clients or poverty alleviation. Our final control variable relates to the lending model adopted by each MFI. Following Hermes and Lensink s (2007) discussion of group (or joint liability) lending, we include a variable that indicates whether an MFI employs individual, as opposed to group or village lending practices. Table 2 reports descriptive statistics for, and pair wise correlations among all of the 10

11 variables in our analysis. Table 2 about here 4. ANALYSIS AND RESULTS We begin by entering the control variables into an ordinary least squares regression model (see model 1 in table 3). The significant coefficients reveal an interesting asymmetry between the estimated effects of for profit versus nonprofit competition. The latter negative effect is consistent with expectations. Greater competition from larger numbers of nonprofit MFIs drives down effective interest rates. However, the estimated effect is opposite for the for profit competition variable. We revisit these two results which are robust across the various models that we estimate in the concluding section of the paper. The MFI size variable has the expected negative effect on interest rates. As expected, MFIs that offer more cost effective larger loans also charge lower interest rates. On the other hand, interest rates are significantly higher among the MFIs that have medium or large outreach levels, those that target women clients, and those that emphasize poverty reduction in their mission statements. Table 3 about here We are primarily interested in the organizational choices made by MFIs that indicate a stronger commitment to profitability: adopting the for profit status, having private sector representation and banking acumen on boards, and having more extensive ties to other for profit MFIs. Model 2 introduces these variables into the effective interest rate model. 1 The for profit MFI variable is positive but not statistically significant at conventional levels (p=.0.11). Moreover, each of the three board and network variables inflates effective interest rates. Having private sector 1 Given concerns about multicollinearity, we checked the variance inflation factors (VIFs) and found the highest (5.02) to be within the acceptable range. We also estimated four separate models that entered each of the profit orientations variables individually and obtained the same results. Finally, we estimated an (unreported) model that removed outlier observations (i.e., those for which the residual is more than two standard deviations away from zero) and obtained the same pattern of results. 11

12 representation on the board corresponds with an estimated increase in the effective interest rate of 3.53 percentage points. Ensuring that the board has banking acumen corresponds with a larger 4.06 percentage point increase. Finally, increasing ties to other for profit MFIs from zero to the maximum level corresponds to a 6.97 percentage point increase in effective interest rates. 2 Given the similar magnitudes of the estimated board and network effects, we performed an F test to assess the null hypothesis that their coefficient estimates are equal. This test can not reject the null hypothesis of equal effects (F=0.57; p=0.57). We therefore sum the three variables into a single index that reflects the overall profit orientation of each MFI. This variable ranges from a low of zero (in eight observations) to a high of 2.96, and averages Closer inspection shows that this profit orientation variable is significantly higher among the for profit MFIs; averaging 1.71 compared to 1.29 for the nonprofit MFIs (t=5.33; p=0.00). When the overall profit orientation index is substituted into model 2, its effect is positive and significant; a unit increase corresponding to a 4.05 percentage point increase in effective interest rates (see model 3). Some researchers express concerns about data quality in models evaluating MFI performance (Mersland & Strøm, 2010). We therefore estimated an (unreported) variant of model 3 based on the 310 MFIs that received four or five stars for data quality from MIX. The results are virtually identical to those reported in model 3. Two thirds of the nonprofit MFIs in this sample are NGOs, with credit unions/cooperatives and non bank financial institutions comprising the remaining third. Roughly 80% of the for profit MFIs are non bank financial institutions. The remaining for profit MFIs are either banks or rural banks. 3 To ensure that our results are not an artifact of the legal form adopted by the sampled MFIs, we ran another (unreported) model that 2 To assess whether the network ties effect simply reflects the overall degree of connectedness, we created a second network variable that ranges from zero to one as an MFI moves from having no ties to other nonprofit MFIs to one when it has the maximum number of observed ties (131). In an unreported model, the coefficient on this new variable is negative and insignificant ( = 1.55; p=0.68) while the magnitude and significance of the for profit tie variable remains positive and marginally significant ( =7.89; p=0.03). 3 This distribution of legal forms in this sample is qualitatively similar to that reported by Cull et al (2009). 12

13 includes dummy variables for each legal form. Again, the results reported in model 3 are replicated. Finally, to ensure that the estimated effects of the profit orientation variables are robust, we estimated a variant of model 3 that replaces all region and country control variables with a set of country fixed effects (see model 4). Because we only include observations from countries that have at least three sampled MFIs, this reduces the sample to 339 MFIs. Again, the results from model 3 are replicated. In this model a unit increase in the profit orientation variable corresponds with a 3.60 percentage point increase in effective interest rates charged to MFI clients. One of our basic premises is that the organizational choices that MFIs make to incorporate a stronger profit orientation can have implications above and beyond that of adopting the for profit form. Given pressures on nonprofit MFIs to act more like their market oriented for profit counterparts, these choices should also influence their behavior and performance. Model 5 (in table 4) re estimates model 3 using the sub sample of nonprofit MFIs and shows a similar pattern of effects. This time, a unit increase in the profit orientation variable is associated with a more substantial 5.00 percentage point increase in effective interest rates. In the sub sample of 127 forprofit MFIs, the effect of a stronger profit orientation is still positive, although no longer statistically significant. It seems that the more damaging effects of a stronger profit orientation (in terms of higher interest rates charged to clients) are confined to nonprofit MFIs, although it is instructive to observe that among the for profit MFIs, the variables that ought to correlate with improved business and banking acumen do not help to lower the interest rate variable. Table 4 about here Many of the claims about the importance of a stronger profit orientation relates to the ability of MFIs to encourage more investment to flow into the sector. In support of this claim, table 2 shows that for profit MFIs tend to be larger than their nonprofit counterparts and tend to operate in more populous countries. Extending this line of thought, it is plausible that the stronger profit orientation is more suited to larger MFI size. We might therefore expect to see the effects of the 13

14 profit orientation variables disappear or reverse in the sub sample of MFIs that are above the median sample size. In model 7, we do see a reduction in the estimated effect size of the profit orientation variable. Relative to the sub sample of smaller MFIs, the adverse effect of a stronger profit orientation is roughly halved; producing a 2.23 percentage point increase compared to a 5.44 percentage point increase in the smaller MFI sub sample. However, although the adverse effect of the profit orientation variable on interest rates is less pronounced it is still significant. (a) Costs and Sustainability Effective interest rates are the sum of profits earned plus three major components of an MFI s costs: operating expenses, financial expenses and losses due to loan impairment. We continue our analysis by looking at how the variables in model 3 also influence these three cost components. Operating expenses are largely a function of salaries and staff productivity (Gonzalez, 2010). Thus, the operating expense variable that we analyze is the sum of these non financial expenses (plus depreciation and amortization and other administrative expenses) divided by the average gross loan portfolio (Mersland & Strøm, 2008a). The variable that captures financial expenses is calculated as the total financial expenses relative to the average gross loan portfolio. Finally, the loan losses variable is the similar ratio of the value of loans written off divided by the average gross loan portfolio. Given the obvious interdependence among these three cost variables, we estimate the effects of our covariates in a seemingly unrelated regression framework (Zellner, 1962). The results from this system of equations presented as model 8 in table 5 suggest that most of the systematic variance in MFI costs pertains to operating expenses. Having more nonprofit competition in a country corresponds with lower operating expense ratios. The for profit MFI competition variable has no discernible effect. The MFI size variable has a negative and significant effect on operating expenses, corroborating expectations about economies of scale in the MFI 14

15 sector. There is also evidence of the documented trade off between focusing on poverty alleviation and cost efficiency (Hermes, Lensink, & Meesters, 2011). The estimated effects of targeting women borrowers and emphasizing poverty reduction in the mission statement are positive and significant in the operating expenses equation. On the other hand, the decision to target rural clients has a marginally negative effect, perhaps due to lower land and labor costs in less developed areas. Finally, taking in more deposits, and thereby increasing organizational complexity, corresponds with higher average operating expenses. Table 5 about here Turning to the variables of interest, the profit orientation variable is associated with significantly higher operating expenses ( =5.77) and higher loan impairment expenses ( =0.67). Its effect on financial expenses is also positive but not significant. The estimated effect of the for profit MFI variable is also positive in all three equations, albeit only marginally significant in the financial expenses equation. Thus, there is no evidence of the expected efficiency benefits of adopting a stronger profit orientation. These findings are consistent with Hudon and Traça (2011), who find that MFIs that receive higher levels of subsidies which are probably the less profit oriented MFIs in this sample are actually more efficient. More generally, the pattern reinforces a core finding of Mersland and Strom (2008b), who find no evidence that shareholder owned MFIs are systematically more cost effective than their NGO counterparts. One justification for the higher interest rates charged by MFIs who demonstrate a stronger commitment to profitability relates to the need for MFIs to be financially sustainable. In theory, by providing more business acumen and market discipline, a stronger profit orientation reduces the need for MFIs to rely on subsidies. Instead, higher interest rates and/or lower costs allow them to earn the surpluses that allow them to sustain operations on their own terms. An MFI s financial self sufficiency ratio equals its net income divided by its total costs. Values greater than one indicate that an MFI is able to cover its costs and therefore sustain itself 15

16 over time. We create a dummy variable set to one for MFIs with financial self sufficiency ratios greater than one and analyze its covariates in a logistic regression model. Arguably, these are the MFIs that generated enough profit in 2009 to sustain themselves over time. The significant coefficients in model 10 (in table 6) suggest that greater competition from other for profit MFIs increases the probability than an MFI is financially self sustaining. Whether or not a variable leads to significant financial sustainability differences depends on the extent to which the estimated interest rate and cost effects offset one another. In this respect, the fact that competition from other for profit MFIs increases the likelihood that an MFI will be financially self sustaining is explained by the fact that its estimated impact on interest rates ( =0.27 in model 3) is greater that the corresponding effects on costs (consistently null across the three equations in model 9). On the other hand, emphasizing women in lending portfolios and poverty alleviation in mission statements both reduce the likelihood that an MFI is financially self sustaining. MFIs that place greater emphasis on women tend to have higher operating costs ( =18.90 in the operating expenses equation in model 9) that are not fully accounted for by the relatively higher interest rates that they charge their clients ( =9.11 in model 3). This leads to the overall negative effect on the probability of being financially self sustaining. A similar set of observations applies to MFIs that emphasize poverty in their mission statements. Table 6 about here The profit orientation variable exerts no significant effect on sustainability. This overall null effect is explained by the fact that the positive effect on interest rates charged ( =4.05 in model 3) is more than fully offset by its adverse effects on operating expenses and loan impairment expenses ( =5.77 and =0.23 respectively in model 9). Similarly (although estimated with less provision) forprofit MFIs tend to charge higher interest rates but also operate with higher expenses, especially financial expenses. 16

17 5. DISCUSSION AND CONCLUSION A review of microfinance policy reports reveals that most of them highlight the strengths of SHFs and the weaknesses of NGOs. In particular, they emphasis that NGOs are less commercial and professional because they lack owners with the pecuniary incentive to monitor management (Mersland & Strøm, 2008b). As the microfinance sector matures, more questions will be asked about whether it is evolving in a way that advantages the poorest people on the planet. In particular, we expect that the role of fully commercial, profit seeking institutions in providing such microfinance loans [will continue to be] controversial (Cull, et al., 2009). Profit oriented MFIs are expected to be more efficient but then distribute more of their earned surplus to outside shareholders. Nonprofits are expected to be less concerned about generating surplus for owners but also less operationally efficient. The baseline expectations about the effects of a stronger profit orientation which have been challenged elsewhere are not at all supported in this analysis. The variables that suggest a stronger profit orientation do not lower any of the major components of an MFI s cost. Nor do they significantly improve MFI sustainability. The only thing that we can conclude is that the effective interest rates charged by MFIs with stronger profit orientations are significantly higher on average. In light of the persistent commentary regarding the need for a more market based orientation in the sector, this offers a somewhat sobering account of the implications of MFIs having stronger forprofit orientations. The effects revealed in this analysis suggests that advisory inputs from individuals with private sector backgrounds, with traditional banking acumen or experience running for profit MFIs do not help MFIs navigate the trade offs between efficient service delivery on the one hand and organizational sustainability on the other. They also reinforce an observation made by Armendariz and Morduch (2005), who note how pioneering models grew out of experimentation in lowincome countries rather than from adaptations of standard banking models in richer countries. 17

18 It seems that the insights that are required to achieve poverty alleviation along with financial sustainability will similarly not come from the importation of advice from those familiar with standard banking models. The variables that indicate a stronger profit orientation never seem to produce the expected benefits for MFI clients. So, we conclude the paper by joining others in stressing that rather than concentrating on an MFIs commercialization, attention should be focused on how to reduce costs per client (Mersland & Strøm, 2010, pg. 35). Given the strong correspondence between variables that systematically influence both MFI costs and effective interest rates, it seems that discussions of how to stimulate stronger profit orientations should be replaced with those which more directly address MFI efficiency. For instance, consideration might be given to how one might induce more nonprofit competition, or to effectively scaling efficient and effective MFIs both nonprofit and forprofit as both of these variables seem to reduce operating costs and lower effective interest rates (the latter with increased profitability). Other findings from tables 3 and 4 warrant further scrutiny. Consider, for example, the asymmetric effects of the levels of for profit and nonprofit competition. The results that pertain to nonprofit competition are consistent with our understanding of how markets are supposed to operate. Greater numbers of suppliers force MFIs to become more cost efficient in order to attract clients. The combination of competition and induced efficiencies drives down effective interest rates. This is the specific dynamic that commentators want to see within the sector. However, the corresponding effects of increased numbers of for profit MFIs are counter intuitive. Here, larger numbers tend to correspond with significantly higher effective interest rates (for all MFIs and for the sub sample of nonprofit MFIs) and higher MFI profitability. This result is so robust in these data that it seems odd to equate the increased numbers of for profit MFIs with increased MFI competition. These asymmetric competitive effects are clearly worthy of further analysis. In the meantime, we must question the net benefit of inducing greater participation of MFIs with stronger 18

19 profit orientations. In addition to their adverse direct effects on interest rates, their proliferation in a country leads to even further interest rate increases. In closing, we propose that this kind of research allows practitioners and commentators to look beneath broad generalizations about the direction of the microfinance sector and appreciate the diversity in observed outcomes that are attributable to the more specific choices that both forprofit and nonprofit MFIs make. In this respect, it may not be that important to determine whether, on average, the MFI sector is experience what some are calling mission drift. It may not even be important to ascertain whether, on average, for profit and nonprofit MFIs are making different choices and trade offs. What is important is the knowledge of how the specific decisions taken by MFIs are able to more or less effectively meet the twin challenges of addressing poverty while sustaining and scaling these impacts. That said, we must also stress the need for an appropriate interpretation of these crosssectional results. All we can say for sure is that in 2009, MFIs that displayed stronger profit orientations tended to charge higher effective rates while operating at higher cost. These findings are germane to those that seek to offer advice on what kinds of MFIs tend to produce what kinds of performance outcomes and social impacts. They are also important to those responsible for directing funds toward more impactful and sustainable MFIs. Here, one might stress the need for those with available funds even funds that seek market returns look past the for profit versus nonprofit distinction and look to lend to MFIs that are operating efficiently and pricing competitively. However, our results are silent on the causal effects of an individual MFI moving toward (or away from) a stronger for profit orientation. Addressing this kind of question, which is clearly part of the policy questions that pertain to how existing MFIs might better serve their clients, requires a longitudinal analysis of MFI s that switch to become for profit organizations, that change the composition of their boards, or that change the extent to which their networks are dominated by for profit MFIs. Given the benefits associated with isolating such exogenous impacts 19

20 on MFI behavior and performance, this kind of data and analysis will be most useful, and will surely add rigor to the debates addressed in this paper. 20

21 TABLE 1. CURRENT SAMPLE All MIX Market Sample (2009) N 358 1,169 Share of for profit MFIs 35% 42% Average assets (log) Average effective interest rate

22 TABLE 2. DESCRIPTIVES AND CORRELATIONS Mean (0) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (0) Effective interest rate (1) Country problems (2) Country population (log) (3) For Profit MFIs in country (4) Nonprofit MFIs in country (8) Assets (log) (9) Age (log) (5) Regulated (6) Deposits to Assets (7) Medium/Large Outreach (10) Average loan size (log) (11) Fraction women borrowers (12) Poor/poverty in mission (13) Target rural clients (14) Lend to individuals (15) For profit MFI (16) Private sector on board (17) Banking acumen on board (18) Ties to for profit MFIs (norm)

23 TABLE 3. EFFECTIVE INTEREST RATES CHARGED BY MICROFINANCE INSTITUTIONS Model 1 (Controls) Country problems 0.01 (0.05) Country population (log) 0.76 (0.70) For Profit MFIs in country 0.33 ** (0.05) Nonprofit MFIs in country 0.26 ** (0.04) Assets (log) 1.96 ** (0.59) Age (log) 0.21 (1.01) Regulated 0.74 (1.48) Deposits to Assets 3.39 (3.10) Medium/Large Outreach 5.02 ** (1.92) Average loan size (log) 8.61 ** (2.50) Fraction women borrowers ** (3.01) Poor/poverty in mission 3.86 ** (1.31) Target rural clients 1.86 (1. 43) Lend to individuals 0.13 (2.02) Model 2 (Profit Orientation) 0.01 (0.05) 1.08 (0.69) 0.25 ** (0.05) 0.23 ** (0.04) 2.52 ** (0.59) 0.46 (1.03) 2.26 (1.53) 1.20 (3.04) 3.97 * (1.89) 8.47 ** (2.45) 8.85 * (2.93) 4.14 ** (1.27) 0.65 (1. 41) 0.86 (1.97) Model 3 (Single Index) 0.01 (0.05) 1.02 (0.68) 0.27 ** (0.05) 0.24 ** (0.04) 2.40 ** (0.58) 0.61 (1.02) 2.13 (1.52) 1.31 (3.03) 3.92 * (1.88) 8.76 ** (2.43) 9.11 * (2.96) 4.14 ** (1.27) 0.67 (1. 41) 0.83 (1.96) Model 4 (Country Fixed Effects) 1.21 * (0.55) 0.98 (0.97) 2.92 (1.57) 2.67 (3.10) 0.44 (1.73) 8.65 ** (2.60) 9.91 ** (2.97) 2.72 * (1.21) 1.18 (1. 34) 0.52 (1.80) For profit MFI 2.49 (1.55) 2.46 (1.55) 2.42 (1.55) Profit orientation 4.05 ** (0.88) 3.60 ** (0.86) Private sector on board 3.53 ** (1.31) Banking acumen on board 4.06 * (1.62) Ties to for profit MFIs (norm) 6.97 * (2.93) Fixed region effects (yes) (yes) (yes) (no) Fixed country effects (no) (no) (no) (yes) N Adjusted R ** p<0.01; * p<0.05; p<

24 TABLE 4. MODERATING FACTORS: FOR PROFIT STATUS AND SIZE Model 5 (Nonprofit MFIs) Country problems 0.04 (0.06) Country population (log) 1.16 (0.82) For Profit MFIs in country 0.12 (0.07) Nonprofit MFIs in country 0.17 ** (0.06) Assets (log) 2.39 ** (0.82) Age (log) 0.62 (1.36) Regulated 0.55 (1.93) Deposits to Assets 2.76 (4.31) Medium/Large Outreach 3.89 (2.45) Average loan size (log) ** (4.15) Fraction women borrowers 7.44 * (3.76) Poor/poverty in mission 2.26 (1.60) Target rural clients 2.02 (1.88) Lend to individuals 2.31 (2.27) Model 6 (For Profit MFIs) 0.11 (0.09) 0.92 (1.33) 0.40 ** (0.09) 0.26 ** (0.06) 2.50 ** (0.85) 1.46 (1.69) 5.19 (3.25) 1.54 (4.69) 2.99 (3.43) 8.04 ** (3.16) 4.46 (5.99) 6.23 ** (2.23) 2.09 (2.20) 4.15 (4.07) Model 7 (Larger MFIs) 0.01 (0.06) 0.37 (0.85) 0.37 ** (0.06) 0.27 ** (0.04) 1.59 (0.81) 1.76 (1.25) 2.88 (1.93) 2.52 (3.48) 1.58 (2.73) 9.12 ** (2.61) 8.40 * (3.84) 3.28 * (1.50) 1.93 (1.71) 2.88 (2.61) Model 8 (Smaller MFIs) 0.04 (0.07) 2.14 * (1.05) 0.10 (0.09) 0.19 ** (0.07) 2.44 * (1.18) 1.02 (1.60) 3.13 (2.46) * (5.60) 5.84 * (2.94) ** (4.78) 9.92 * (4.79) 4.55 * (2.06) 2.67 (2.30) 0.83 (2.97) For profit MFI 2.75 (1.85) Profit orientation 5.00 ** * (1.14) (1.48) (1.09) 3.29 (2.65) 5.44 ** (1.40) Fixed region effects (yes) (yes) (yes) (yes) N Adjusted R ** p<0.01; * p<0.05; p<

25 TABLE 5. THREE ELEMENTS OF MFI COSTS Model 9 a Operating Expenses Country problems 0.24 * (0.11) Country population (log) 3.04 * (1.52) For Profit MFIs in country 0.03 (0.12) Nonprofit MFIs in country 0.35 ** (0.08) Assets (log) 6.38 ** (1.30) Age (log) 2.75 (2.27) Regulated 2.38 (3.41) Deposits to Assets ** (6.78) Medium/Large Outreach 3.02 (4.21) Average loan size (log) 4.05 (5.42) Fraction women borrowers ** (6.62) Poor/poverty in mission ** (2.84) Target rural clients 5.22 (3.16) Lend to individuals 5.05 (4.38) For profit MFI 3.47 (3.46) Profit orientation 5.77 ** (1.97) Financial Expenses 0.06 (0.03) 0.32 (0.42) 0.01 (0.03) 0.03 (0.02) 0.12 (0.36) 0.02 (0.63) 0.48 (0.95) 1.49 (1.88) 1.12 (1.17) 0.70 (1.50) 1.60 (1.84) 0.07 (0.79) 1.29 (0.87) 0.85 (1.22) 1.86 (0.96) 0.55 (0.55) Loan Impairment 0.03 (0.02) 0.06 (0.26) 0.02 (0.02) 0.01 (0.01) 0.01 (0.22) 0.33 (0.39) 0.00 (0.58) 2.32 * (1.15) 0.12 (0.71) 0.59 (0.92) 0.90 (1.12) 0.09 (0.48) 0.61 (0.54) 1.20 (0.74) 0.23 (0.59) 0.67 * (0.33) Fixed region effects (yes) (yes) (yes) N 358 R ** p<0.01; * p<0.05; p<0.10 a Seemingly unrelated regression model 25

26 TABLE 6. SELF SUFFICIENCY (SUSTAINABILITY) Model 10 a Sustainability Pr(F.S.S.>1) Country problems 0.01 (0.01) Country population (log) 0.01 (0.16) For Profit MFIs in country 0.04 ** (0.01) Nonprofit MFIs in country 0.00 (0.01) Assets (log) 0.21 (0.15) Age (log) 0.50 * (0.25) Regulated 0.01 (0.37) Deposits to Assets 1.85 * (0.95) Medium/Large Outreach 0.20 (0.46) Average loan size (log) 0.49 (0.59) Fraction women borrowers 1.46 * (0.74) Poor/poverty in mission 0.66 * (0.31) Target rural clients 0.34 (0.34) Lend to individuals 0.64 (0.48) For profit MFI 0.06 (0.37) Profit orientation 0.14 (0.21) Fixed region effects (yes) N 358 Log Likelihood ** p<0.01; * p<0.05; p<0.10 a Logistic regression model 26

27 REFERENCES Ahlin, C., Lin, J., & Maio, M. (2011). Where does microfinance flourish? Microfinance institution performance in macroeconomic context. Journal of Developmental Economics, 95, Armendariz, B., & Morduch, J. (2005). The Economics of MicroFinance (Second ed.). Cambridge, MA: MIT Press. Brass, D. J., Galaskiewicz, J., Greve, H. R., & Tsai, W. (2004). Taking stock of networks and organizations: A multilevel perspective. Academy of Management Journal, 47, Cook, T., & Isern, J. (2004). What Is a Network? The Diversity of Networks in Microfinance Today. Copestake, J. (2007). Mainstreaming microfinance: Social performance management or mission drift? World Development, 35, Cull, R., Demirguc Kunt, A., & Morduch, J. (2009). Microfinance meets the market. Journal of Economic Perspectives, 23, Cull, R., Demirguc Kunt, A., & Morduch, J. (2011). Does Regulatory Supervision Curtail Microfinance Profitability and Outreach?. World Development, 39, Gonzalez, A. (2007). Efficiency drivers of microfinance institutions (MFIs): The case of operating costs.unpublished manuscript. Gonzalez, A. (2010). Analyzing microcredit interest rates: A review of the methodology proposed by Mohammed Yunus.Unpublished manuscript. Hannan, M. T., & Freeman, J. (1989). Organizational Ecology. Cambridge, MA: Harvard University Press. Hermes, N., & Lensink, R. (2007). The empirics of microfinance: what do we know? The Economic Journal, 117, F1 F10. Hermes, N., Lensink, R., & Meesters, A. (2011). Outreach and efficiency of microfinance institutions. World Development, 39, Hudon, M., & Traça, D. (2011). On the efficiency effects of subsidies in microfinance: An empirical inquiry. World Development, 39, Mersland, R., & Strøm, R. Ø. (2008a). Performance and governance in microfinance institutions. Journal of Banking and Finance, 33, Mersland, R., & Strøm, R. Ø. (2008b). Performance and trade offs in microfinance organisations Does ownership matter? Journal of International Development, 20, Mersland, R., & Strøm, R. Ø. (2010). Microfinance mission drift? World Development, 38, Morduch, J. (2000). The microfinance schism. World Development, 28, Owen Smith, J., & Powell, W. W. (2004). Knowledge networks as channels and conduits: The effects of spillovers in the Boston biotechnology community. Organization Science, 15, Weisbrod, B. A. (1998). Institutional form and organizational behavior. In W. W. Powell & E. S. Clemen (Eds.), Private Action and the Public Good (pp ). New Haven, CT: Yale University Press. Yunus, M. (2011, January 14, 2011). Sacrificing Microcredit for Megaprofits. New York Times, Zellner, A. (1962). An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association, 57,