What drives microfinance institution's financial sustainability

Ayi Gavriel Ayayi Maty Sene

The Journal of Developing Areas, Volume 44, Number 1, Fall 2010, pp. 303-324 (Article) Published by Tennessee State University College of Business DOI: 10.1353/jda.0.0093

For additional information about this article

http://muse.jhu.edu/journals/jda/summary/v044/44.1.ayayi.html

Access Provided by Indian Institute of Technology, Bhubaneswar at 08/26/12 2:06PM GMT

WHAT DRIVES MICROFINANCE INSTITUTIONS FINANCIAL SUSTAINABILITY

Ayi Gavriel Ayayi Universit du Qubec at Trois Rivires, Canada Maty Sene Audencia & Universit de Nantes, France
ABSTRACT Microfinance promises to trim down poverty. To achieve this noble objective microfinance institutions (MFIs) have to become steady profitable because donor constancy is not a given. Thus important question is: what factors drive the financial sustainability of MFIs? Using data on 217 MFIs in 101 countries distributed by region and type of MFIs over the period of 1998-2006, we report three important findings. First, we show that a high quality credit portfolio, coupled with the application of sufficiently high interest rates that allow a reasonable profit and sound management are instrumental to the financial sustainability of MFIs. Second, we show that the percentage of women among the clientele has a weak statistically non-significant negative effect on financial sustainability of MFIs. Third, we find that the client outreach of microfinance programs and the age of MFIs have a positive but lesser impact on attainment of financial sustainability. The policy implication is that MFIs have to emulate profit-making banking practices by implementing a sound financial management and good managerial governance to assure their financial sustainability. Keywords: Microfinance, financial sustainability, portfolio-at-risk, interest rates, client outreach, Legal status, Credit methodology JEL Classification: G21, I30, R00 Corresponding Authors Email: [email protected] or [email protected]

INTRODUCTION The economic and financial crises of the 1980s, coupled with mismanagement of banks and financial institutions, have propelled the liquidation of development banks and specialized financing structures dedicated to reducing poverty. This situation has accentuated the exit of small economic operators and the rural world from classic financing structures. To avoid reproducing past errors, microfinance has emerged as a solution for small economic operators in the countryside and towns that were evicted from classical financing structures, together with all disadvantaged persons that wished to develop an income-generating economic activity to increase their well being and that of their immediate family. Microfinance subsequently proliferated in countries with a paucity of bank infrastructures, such as most of Africa, Asia, Latin America and Eastern Europe. In some of these countries, less than 20% of the population has a bank account. Drawing lessons from the failures of the previous development structures, emphasis was placed on the perennial financial sustainability of MFIs. However, to ensure that microfinance does not deviate from its original mission of facilitating access to financial services for the poor, particular attention has been paid to the capacity of MFIs to

304
sustainably facilitate access to financial services for excluded individuals in order to improve their well being, and to evaluating this capacity by impact measures. The objective of this article is to identify the determining factors of financial sustainability of MFIs. Financial sustainability is crucial because donor constancy is not a given. As the proverb states, it is better to teach someone to fish than to give this person a fish. Moreover, sustainability implies increased access to financial services for the poor and for everyone excluded from the classic financial sector. It consequently has a significant impact on poverty reduction. MFIs can thus stimulate economic and social development by providing quality financial services to disadvantaged populations. Moreover, for microfinance to be a development tool, it must not be dependent on aid, because most aid policies around the world have failed. To test our hypotheses, we analyzed a sample of 217 MFIs with 5 diamond ratings from MIX Market database over a period of 9 years, namely from 1998 to 2006. Our results show that a high quality credit portfolio is the most determining component of financial sustainability of MFIs, followed by the application of adequate interest rates and effective management to control personnel expenses. Further, our results show that the client outreach of microfinance programs and the age of MFIs, whose coefficients are positive and statistically significant, have a lesser influence on financial sustainability of MFIs. In addition, we demonstrate that the percentage of women borrowers has a weak statistically nonsignificant negative effect on financial sustainability. In general, the results of our regressions by geographical region, legal status and credit methodology are consistent with the overall results of the general model. The remainder of the article is structured as follows. Section II provides the literature review. Section III covers the data and the methodology. Section IV presents the results of the determinants of financial sustainability. Section V reports the analysis by geographic region, legal status and credit methodology, and Section VI concludes the paper. LITERATURE REVIEW Financial Sustainability of MFIs The financial sustainability of an MFI can be defined as its capacity to cover all of its expenses by its revenue and to generate a margin to finance its growth. In other words, it is the capacity of an MFI to carry out its activities without the need for subsidies in the form of concessional loans or donations.1 Nonetheless, a study by CGAP (1995)2 concludes that microfinance cannot be financially sustainable.3 It is generally believed that small loans are too costly to provide, and the resulting income is insufficient to ensure profitable operations. Brau and Woller (2004) corroborate this hypothesis by showing that unlike financial institutions in the formal sector; most MFIs are not financially sustainable. They add that many MFIs could not function without the subsidies that they receive from governments and other funders. 4 This view is shared by Hermes and Lensink (2007), who contend that because of the high cost of providing microfinance products and services, most MFIs are not financially sustainable and are thus reliant on donor subsidies. Nonetheless, Littlefield and Rosenberg (2004) argue that MFIs could cover all of their expenses if they would offer their services at adequate interest margins, constantly

305
strive for efficiency and apply an aggressive loan recovery policy. Morduch (2000) underlines the need to continue to serve the poor through financially sustainable MFIs, by developing the institutional capacity and management of these MFIs and by encouraging new innovations and experimentation in the field of microfinance. Gibbons and Meehan (2000) point out that MFIs must adopt commercial for-profit strategies because as MFIs begin to free themselves from dependence on subsidies and abide by serious banking standards, they will become obliged to contain their costs. This will not only enhance the quality of services for borrowers but will also facilitate attainment of financial sustainability. In the same vein, Von Pischke (1996) notes that a compromise between financial sustainability, client outreach and social impact on the poor is possible. Navajas et al. (2000) emphasized that MFIs that are truly financial sustainable tend to target clients that are either slightly above and slightly below the poverty line, because they can benefit from economies of scale by extending their loans to marginally poor and non-poor clients. Moreover, Hollis and Sweetman (2001) show that microcredit institutions in Ireland in the 18th and 19th centuries were financially sustainable for more than a century because they adapted to their economic and financial environment. Patten and al (2001) compared the performance of the Indonesian MFI Bank Rakyat and the formal banking sector during the financial crisis in East Asia. They showed that the Rakyat MFI exhibited superior performance in terms of repayment rates and other portfolio ratios. Financial Sustainability and Interest Rates There are three conflicting schools of thought regarding interest rate levels in microfinance. The first school, endorsed in particular by western NGOs and their partners in developing countries, maintains that the poor cannot pay interest rates at market prices and consequently they must be loaned funds at very low interest rates (between 1% and 3% regardless of inflation). Such a model can function only with subsidies. The second school supported by mutual societies and savings and credit unions, advocates interest rates slightly below those of commercial banks. This approach is feasible because these institutions pay out little or none of the funds collected, or benefit from subsidized lines of credit or international aid. The third school is espoused by organizations that want to manage microcredit by covering the effective rates to attain financial sustainability and to ensure the continuity of their actions. Their primary objective is to eliminate the bank exclusion of the poor even if they must bear high interest rates. Accordingly, Morduch (2000) notes that the high interest rate results from the fact that the demand for credit is not very elastic among populations for whom credit were previously rationed. It is important to avoid confusing moral and economic judgments: to make credit accessible to a large number of players, one must follow a financial logic. This includes covering ones costs and setting rates accordingly. As the loan amounts are modest and their duration short, this does not pose a problem for micro entrepreneurs, especially in developing countries where microcredit allows individuals to circumvent unscrupulous loan sharks that impose prohibitive interest rates in the 100% to 400% range. It is this financial logic that has led MFIs around the world to charge high interest rates. Acclassato (2006) notes that in West Africa, financially sustainable MFIs have had to apply interest rates of 84%, and that in Indonesia the Bri Unit Desa charges rates of

306
between 35% and 60%. Similarly, in Bolivia, the Banco Sol. Corposol applied an effective rate of 52% to attain a real return of 4.9% on its total assets (CGAP, 1995). Contrary to the commercial banks, the compensation structure of credit granted by MFIs must include the interest rate, i.e. the rent on the money, the training committee, the monitoring/advisory committee and the collection committee. According to CGAP (2004), all of these elements justify interest rates that are generally higher than those of the banking sector, but below those of informal borrowers such as loan sharks. To attain financial sustainability, MFIs must apply high but not exorbitant interest rates. This was confirmed by Cull, Kunt and Morduch (2007), who show that for MFIs that grant individual loans, when the interest rates applied surpass 60%, the MFIs are no longer profitable because the demand for credit decreases. Moreover, MFIs that apply the highest interest rates are the best performers. They are also most efficient and most financially sustainable. This assertion was confirmed by empirical studies (Robinson (1996) and Conning (1999)) that show that only the MFIs that generate profits are able to accumulate equity capital, a source of financing, to ensure their continuity and increase their reach among the poor. As MFIs capital grows, so does their financing capacity, either in terms of debt or by accepting larger deposits from their clientele, which in turn broadens the reach of their actions, which reinforces financial sustainability. It follows that to attain financial sustainability; an MFI must set sufficiently high interest rates to cover its expenses because too low a rate would cause financial distress that may lead to the bankruptcy. In contrast, a needlessly high interest rate would penalize the clients, that the MFI could lose, which would then undermine its social mission. Financial Sustainability and Quality of the Credit Portfolio Micro-enterprises, the main clients of MFIs, do not generally supply material guarantees or reliable financial and accounting information when they seek microcredit. Thus, discriminating between micro-entrepreneurs without resorting to adverse selection is a crucial problem that MFIs must face to attain financial sustainability, as are very strong constraints on repayment. To identify micro-entrepreneurs with higher potential, MFIs can apply sequential loans (Chowdhury, (2007)). Fuelled by borrowers hopes of obtaining subsequent credit to sustain the growth of their micro-enterprise, and by extension improve their social well being, this mechanism serves not only to separate micro-entrepreneurs with dubious repayment possibilities from reliable ones, but also to improve the quality of the portfolio-at-risk. This mechanism of discrimination of questionable micro-entrepreneurs from good ones can avert the phenomenon of perverse contagion that spreads among low risk borrowers that find that it is pointless to repay their credit when others are not doing so. This attitude consequently accelerates the deterioration of the portfolio quality and contributes to the erosion of the financial sustainability of MFIs. Aside from dynamic loans, several risk management methods are used. Examples include credit scoring5 (Schreiner, 2003) and pre-default6 (Lanha, 2004), which is based on the possibilities of staggering the repayment of microcredit. Gibbons and Meehan (2000) maintain that the portfolio-at-risk rather than the loan repayment rate must be controlled to improve the quality of MFIs portfolios because the chronological classification of

307
payment arrears allows loans in the portfolio to be ranked according to different risk levels for calculation of provisions for losses on MFI loans. Such information allows the managers to make fully enlightened decisions, particularly concerning the weaknesses of their institution, which may lead to a crisis situation if necessary corrections are not made. Moreover, group loans with joint surety are often used as a mechanism to minimize the risk of default in order to improve the MFI portfolio return and thus attain financial sustainability (see Pitt and Khandker (1998), Ghatak (1999), Armendariz de Aghion and Morduch (2000), Laffont and Nguessam (2000), Casar and al (2007)). Hermes, Lensink and Mehrteab (2006) note that monitoring by the group leader, who possesses social ties and local knowledge, plays a determining role in improving repayment rates. Ahlin and Townsend (2007) and Casar et al (2007) note that social sanctions are another means of preventing repayment default stemming from ill will by some members of the group, which contributes to the financial sustainability of MFIs. In the same vein, Cull, Kunt and Morduch (2007) note that group loans have a positive impact on the quality of the credit portfolio of MFIs. They show that for most MFIs that grant individual loans, the portfolio-at-risk increases according to the interest rates applied, and that beyond the threshold of 60% an increase in these rates is not associated with long-term profits, which is not the case for MFI that rely on a joint liability loan methodology. These results reinforce the idea that monitoring and control by peers to minimize moral hazard,7 together with the extent of social relations among the group members and the MFI employees (Woolcock, 2001), contribute to the sustainability of MFIs. Financial Sustainability and Management The research on financial sustainability dictates that MFIs should have modern and efficient employee management and should integrate new information technologies in their management that allow them to control operating costs and by extension their personnel costs, which are currently exorbitant. The integration of new information technologies in management can allow MFIs to monitor, analyze and control operations. An information system that generates precise data in a timely fashion allows managers to continually assess performance, better predict the need for liquidities, anticipate needs and quickly respond to crises. By using new information technologies adapted to the specific needs of MFIs, managers can offer their clients financial services at reduced costs. For example, an information system enabled the Indian MFI Spandana to compile reliable data and control the performance of its entire network of 45 branches (Mishra, 1994). With the same objective, 62 banks and MFIs mentioned that they integrated POS terminals and portable telephones in their services in order to reduce their credit issuance costs (CGAP, 2005). The use of new information technologies can not only reduce credit granting costs but can also increase the productivity of loan officers, allowing them to render high quality services to the poorest populations and contribute to increasing the financial sustainability of MFIs. In addition to technological management, good employee management through continuous training plays a vital role in equipping MFIs with proven technical expertise in the field of microfinance. Well trained loan officers motivated by financial incentives that include a well adapted bonus system are indispensable to attain financial

308
sustainability. Thus, MFIs that peg a portion of the wages to financial performance report that they are satisfied with the results obtained (CGAP, 2005). A staff with proven technical expertise in the field of microfinance is another prerequisite for the effective choice and monitoring of projects to finance. In addition, in the quest for efficiency, cost reduction and optimization of means to attain financial sustainability, MFIs should increasingly adopt good management structures and incorporate existing efficient banking practices in their management, which notably entails emphasizing client relations and loyalty (Churchill, 2000, Schreiner, 2000 and Norell, 2001). These practices include creation of a marketing department and of audit and risk identification committees via an efficient internal control system. Management analysis reveals that good technological management combined with efficient human resources management should translate into a significant reduction in personnel costs. DATA AND METHODOLOGY Sources of Data The data set for this study originates from the MIX market (www.mixmarket.org), a database that compiles information on MFIs. Created in June 2002 as a private non-profit organization, the MIX (Microfinance Information eXchange) promotes the exchange of information within the microfinance sector. The mission of the MIX is to help create a microfinance market by offering data collection services, performance tracking tools, sector comparisons and specialized information services. To procure this information, the MIX adopted a quality control system that awards diamonds on a scale of 1 to 5. An MFI with a 5-diamond rating has financial statements audited over several years and data adjusted by a third party (ratings/evaluation, due diligence and other benchmarking assessment reports or studies). 8 Choice of Sample For the purposes of this research we formed a sample of 233 MFIs that obtained 5 diamonds. We excluded MFIs for which information was unavailable concerning the variables examined in our study. Thus, on August 6, 2007 we had a sample of 217 MFIs in 101 countries over a 9-year period (1998-2006). Our sample is distributed by region and type of institution as follows: The chart shows that 5-diamond MFIs are found predominantly in the Latin America-Caribbean region (41.63%), followed by Eastern Europe-Central Asia (20.60%), Sub-Saharan Africa (15.87%), Southeast Asia/Pacific (15.45%) and the Middle East/North Africa (6.45%). This situation can be explained by the fact that the Latin-America Caribbean region has the highest level of institutionalization of microfinance, and that many MFIs have adopted the operating rules of banks. In contrast, the low level of transparency observed in the Middle East/North Africa region is explained by the nascent state of microfinance in this region compared with Latin-America. Moreover, NGOs make up the largest segment of MFIs with five diamonds ratings (39.91%), followed by NFBIs (38.62%), banks (9.87%), savings and credit cooperatives (7.72%), other (3%) and lastly rural banks (0.88%). This phenomenon is

309
attributable to the fact that the NGOs that were the precursors of microfinance imposed rigorous management rules to enhance their credibility among fund providers (donors, government agencies etc.) and endeavored in particular to show that the money is indeed used to reduce poverty. Increased management transparency was a corollary of these objectives. FIGURE 1. DISTRIBUTION OF MFIS IN SAMPLE
Figure1:Distribution of MFIs in sample
60

50

40
Number of MFIs

30

20

Bank Savings and credit cooperative Non-bank financial intermediary Non-governmental organization Other Rural bank

10

0 1 Sub-Saharian Africa 2 Southeast Asia and Pacific 3 Latin America/ Caribbean 4 Eastern Europe/ Central Asia 5 Middle East/ North Africa

Description of Variables Dependent variable For our analyses we will use financial sustainability (Finsus) as the dependent variable. We can thus evaluate the financial performance of MFI over time because when the financial revenue continuously exceeds the total expenses (Finsus >100), we can deduce that the MFI can forgo subsidies and turn to financial markets if necessary. MFIs obtain financial revenue from loans and other financial services in the form of interest, penalties and commissions. The financial revenue also includes income from other financial assets such as investment income. MFIs expenses include operating expenses, cost of credit and provisions for probable losses on loan default. In contrast with Cull, Deminrg-Kunt and Morduch (2007), who, in addition to financial sustainability, used operational selfsufficiency (OSS) and adjusted return on assets (ROA), we concentrate uniquely on financial sustainability because our objective is to examine economic profit in the sense

310
of creation of value, rather than accounting profit, which is more similar to operational self-sufficiency.9 Independent variables The first independent variable is the Portfolio-at-risk at 30 days (Par>30). This risk and liquidity ratio allows us to measure the impact of the portfolio-at-risk measured at a specific period on credit granted in the following period. In effect, the functioning of MFIs is characterized by revolving credit logic, i.e. that lendable funds are derived almost entirely from credits. Under these conditions we anticipate that the portfolio-at-risk will have a negative influence on MFI performance. In other words, MFIs that have high Par>30 on a period would have fewer funds to lend and would therefore be obliged to ration credit, which would have a negative influence on their financial performance. We will also consider the risk hedging ratio (Riskhdgrat), another risk and liquidity ratio, to see which of the two ratios (Par>30 and Riskhdgrat) contributes most significantly to financial sustainability. Concerning our second independent variable--interest rate applied by MFIs--we used the financial revenue ratio (Finrevrat) and the accounting rate of return of the portfolio (Accretport). These two ratios that result from operating revenue enable us to take into account the interest rates MFIs charge their clients because the financial revenue of MFIs consists mainly of interest charged on loans and other financial services offered to clients. Moreover, given that good management can contribute to the improvement in and enduring growth of MFIs financial sustainability, we have chosen as our third independent variable the operating expense ratio (Opexprat) and the ratio of personnel costs (Percosrat), both of which can capture good management. Conceivably, a well managed MFI that applies best banking practices can effectively control its operating expenses. Moreover, the personnel cost ratio can also illustrate human resources management, specifically with respect to training, customer relations and motivation. These factors can play an important role in the financial performance of MFIs. Given that good management is an element that can boost productivity and therefore improve financial sustainability, we have chosen as the fourth independent variable the three productivity ratios, i.e. number of loans per staff member (Loanstaff), number of borrowers per loan officer (Borlooff) and the number of loans compared with the number of loan officers (Lolooff) to see which of the three contributes most significantly to financial sustainability. Lastly, to assess the impact of institutional indicators and the client outreach on financial sustainability of MFIs, we retained as the remaining independent variables a set of measures of client outreach, namely: the percentage of women borrowers (Percwomen) and the average loan balance (Avloanbal), and as institutional indicators the staff members (Staffmeme) and age of MFIs (Agemfi). We then refined the data to retain only the pertinent variables that allow us to obtain the least biased results. Refinement and quality of data Examining the profile of our sample prior to our regressions enables us to determine whether there are anomalies that could bias the validity of our estimates. Accordingly, we

311
analyzed each variable in terms of dispersion and normality. Concerning normality, we evaluated the skewness10 and kurtosis11 of each variable. The variables indicated by a star in Table 1 are those that simultaneously meet the two criteria of normality. In general, we can say that the distribution of each of these variables is symmetrical with respect to dispersion. Further, the normality test of Kolmogorov Smirnov done with STATA9 software, the results of which are not presented in the article, confirm the choice of variables that we retain based on skewness and kurtosis. For all variables retained we also analyzed other descriptive statistics. For the Finsus variable, the mean of 106.12, i.e. greater than 100, shows that our global sample is made up of financially sustainable MFIs. Nonetheless, the minimum of 68 shows that there are MFIs that, although transparent, cannot be considered financially sustainable. Regarding the variable Par>30, the higher its value, the TABLE 1 DESCRIPTIVE STATISTICS.
Variable Finsus* Par>30* Finrevrat* Opexprat Percosrat* Agemfi* Staffmem Avloanbal* Percwomen* Profmargin Lostaffmem Borlooff* Lolooff Accretport Riskhedrat # obs. 1930 1930 1930 1930 1930 1930 1899 1760 1930 1892 1929 1929 1929 1922 1899 Mean 106.12 2.46 27.50 18.03 10.25 9.64 159.17 385.16 69.19 16.09 127.79 227.55 231.87 34.40 1.13 Standard deviation 17.6474 1.74 7.18 6.08 3. 38 3.32 162.69 280.16 16.60 10.55 44.15 66.84 71.69 8.07 0.57 Min 68 0.1 13 10 5.1 5 21 71 39.1 -0.87 45 81 81 21 0.5 Max 144 7.2 44.8 33.8 18.7 18 933 993 100 86 256 394 453 562 3.4 Skew - 0. 38 0.61 0.53 1.05 0 .57 0.5 2.70 0. 55 0.23 2.14 0.35 -0.11 0.17 0.81 1.93 Kurtosis 2.41 2. 54 2.76 3.33 2.41 2.31 11.83 2.01 2.00 11.70 3.12 2.54 3.14 3.09 7.30

riskier the credit portfolio, which can have a negative influence on the financial sustainability of the MFI. For our sample, the mean of 2.50 and the maximum of 7.2 imply that the credit portfolio of some MFIs in the sample is fairly risky. The higher the variable Finrevrat is, the more the MFI earns from the products and services it offers its clients. For the sample, the mean is 27.42 with a maximum of 44.8, which is 3.8 times the minimum. This clearly illustrates the disparity of rates offered by MFIs. Regarding the variable Percosrat, the higher its value, the greater the personnel expenses, which could have a negative impact on financial performance of MFIs if there is no improvement in productivity. Within our sample, control of personnel expenses varies considerably, from a maximum of 18.70 to a minimum of 5.10, with a mean of 10.19. The variable Agemfi demonstrates that the MFIs in our sample are on average 10 years old. This variable also illustrates a disparity in the experience of MFIs in the sector, which ranges from a minimum of 5 years to a maximum of 18 years. Moreover, the variable Avloanbal exhibits extreme values, from a minimum of 71 to a maximum of 993, for a mean of 387.18. The Percwomen variable, with a mean of 69.21; a minimum

312
of 39.1 and a maximum of 100, shows that the MFIs in our sample have more women borrowers than men. Concerning the Borlooff variable, values obtained illustrate diversity in the productivity of MFIs in our sample. Nonetheless, before we began our regressions, we analyzed the mean and the variance of the dependent variable according to legal status, geographical region and credit methodology to see whether it is sufficiently stable, and thus eliminate heterogeneity bias. Analysis of mean and variance for dependent variable For this analysis we divide our sample according to three criteria: legal status, geographical region and credit methodology of MFIs. Table II shows that overall the dependent variable is fairly stable. TABLE 2. MEAN OF DEPENDENT VARIABLE BY LEGAL STATUS, CREDIT METHODOLOGY AND GEOGRAPHICAL REGION OF MFIS.
Legal status Bank Savings and credit cooperative NFBI NGO Rural bank Credit Methodology Individual Individual /Joint Joint group Village bank Geographical region Africa Asia ECA ALC
MENA

# obs 217 24 14 80 85 14 217 68 111 18 20 217 35 33 45 90

14

Mean 110,4 112 101 110 112 117 108,75 116 109 105 105 112 91 116 117 111
125

St dev 5,85662 2.98710 5.35487 2.85472 1.95645 3.08947 5,18812 3.21854 2.81560 5.72451 5.01973 12,76715 5.96470 5.85626 5.96580 5.02873
5.72630

Min 101 107 98 105 110 111 105 113 105 95 100 91 81 110 113 110
114

Max 117 115 110 112 113 120 116 123 111 110 113 125 98 126 130 123
130

If we consider the three groups, the mean is consistently above 100. In addition, standard deviations are relatively low compared with the means, which confirms the stability of this variable. Moreover, Table II shows that the stability of financial sustainability does not vary greatly according to the legal status of MFIs (the range is (117 101) =16). We note that rural banks seem to maintain financial sustainability more effectively than savings and credit cooperatives. Concerning distribution of MFIs by credit methodology, we observe less variance (range of (116-105) = 11) in financial sustainability than relative to legal status. Here, financial sustainability is more prevalent in MFIs that grant individual credit exclusively than by those that offer other credit methodologies. This can be explained by the fact that individually there is more motivation to reimburse microcredit, whereas within groups some members may harbor ill will. Guarantees required for individual loans can also contribute to financial sustainability. Lastly, we observe that

313
financial sustainability is less stable relative to geographical region than to legal status or credit methodology. MFIs with the best financial sustainability are found in the Middle East, whereas those situated in Africa are by far the worst, with a mean below 100. Methodology Hypotheses At the starting point of our analysis, we hypothesize that a MFI that is financial sustainable is better posited to provide quality financial services to the poor that will have a profound impact on poverty alleviation. To make this working hypothesis testable, we now develop more specific hypotheses. Basically they say that good financial and managerial governance are the epitome of MFIs financial sustainability. Following our in-depth literature review we paid particular attention the following testable hypotheses. Hypothesis 1: The application of interest rates that generate a profit margin should significantly contribute to attainment of MFIs financial sustainability. We posit a direct relationship between the interest rate and financial sustainability. Hypothesis 2: Good control over the portfolio-at-risk, i.e. significant reduction in the global MFI portfolio, increases MFIs profits and consequently has a positive impact on financial sustainability of MFIs. We predict an inverse relationship between the portfolioat-risk and financial sustainability. Hypothesis 3: Good management through investment in human and technological resources, together with financial incentive mechanisms, should increase the quality and profitability of loan officers, lower personnel costs and heighten loan officer productivity. Consequently, we anticipate an inverse relationship between the ratio of personnel expenses and financial sustainability, along with a direct relationship between the number of borrowers per loan officer 12 and financial sustainability. In addition, given that the majority of MFI clients are women, and that as they gain experience, MFI can better refine their client selection strategy and portfolio-at-risk management, we decided to assess the impact of these elements on financial sustainability of MFIs, as stated in the following hypothesis: Hypothesis 4: the age of the institution, the client outreach of its actions and the percentage of female clients should have a positive impact on the financial sustainability of MFIs. Correlation coefficient matrix Before we examine the links between financial sustainability and the independents variables, we verify that the links between all the variables retained are not too strong or unbalanced (for example, two variables are very strongly correlated while others are nearly independent). Table III presents the correlation coefficient matrix.

314
TABLE 3. CORRELATION COEFFICIENTS BETWEEN THE VARIABLES.
Finsus Finsus Par>30 Finrevrat Percosrat Agemfi Avloanbal 1.000 -0.51*** 1.000 0.19*** -0.18*** 1.000 1.000 -0.35*** -0.02 0.34*** -0.47*** 1.000 0.08** 0.35* 0.41*** 1.000 -0.48*** -0.12*** 1.000 -0.10*** 1.000 Par>30 Finrevrat Percosrat Agemfi Avloanbal Percwom en Borlooff

-0.17*** -0.25*** 0.68*** 0.12*** 0.26*** 0.42*** 0.006

-0.12*** 0.11*** -0.25*** 0.34*** 0.35*** -0.16***

Percwomen 0.03 Borlooff 0.14***

*** Indicates level of statistical significance at 1%; ** Indicates level of statistical significance at 5%*; * Indicates level of statistical significance at 10%

Financial sustainability is positively correlated with the financial revenue ratio, at a statistically significant level. This correlation is anticipated in that the application of interest rates that integrate real costs and a profit margin that reflects the cost of services rendered invariably have a positive impact on the financial sustainability of MFIs . Financial sustainability and the portfolio-at-risk at 30 days are negatively correlated and statistically significant. This finding is compatible with the argument whereby the gradual increase in volume of unpaid loans is detrimental to the financial sustainability of MFIs. Financial sustainability is negatively correlated with the personnel expense ratio but positively correlated with the age of MFIs and the number of borrowers per loan officer. Moreover, the percentage of women borrowers is positively correlated with financial sustainability but is not statistically significant. Although women are by far the predominant consumers of microcredit, the nature of the activities that they undertake (mainly intended to satisfy primary needs on the small scale) does not generate additional profits that would allow MFIs to apply interest rates that could contribute significantly to their financial sustainability. The positive and statistically significant correlation of the average loan balance, a variable of client outreach, with financial sustainability appears to corroborate the theory that the massification of microcredit reinforces the financial sustainability of MFIs (Ayayi, 2007). The financial revenue ratio is positively correlated with the personnel expense ratio and is statistically significant. This corresponds with the concept of incentive wage in that the increase in financial revenue reflects the increase in employee productivity, which should translate by an increase in wages, which in turn is manifested by an increase in personnel expenses. Further, the positive and negative correlations between independent variables analyzed in pairs are weak. Accordingly, we do not observe collinearity between the explanatory variables.

315
Regression model The model retained is intended to estimate the impact of the portfolio-at-risk, the financial revenue ratio, the personnel expense ratio, the age of MFIs, the average loan balance, the percentage of women borrowers and the number of borrowers per loan officer on the financial sustainability of MFIs. The empirical model to estimate these impacts can be expressed as: Finsus = 0+ 1*Par>30 + 2*Finrevrat +3*Percosrat + 4*Agemfi+ 5*Avloanbal + 6*Percwomen + 7*Loanstaff + 8*Borlooff + where 0 8 are the coefficients of the variables and is the random error term. The results obtained after the first regression of the model imply that our model is well specified. However, to be certain, we refine the analysis by examining the residuals to ascertain that there are no problems of endogeneity and heteroskedasticity. Given that the results of the endogeneity13 test show that there is no significant correlation between the error term and the explanatory variables, we ascertained that there are no problems of endogeneity. In contrast, the Breusch Pagen14 test reveals heteroskedasticity. To correct this heteroskedasticity we used the White-Eicker method. These conclusive tests enabled us to determine that the coefficients obtained from our regressions are not biased. Table IV presents the results of the regression with heteroskedasticityconsistent t-statistics in parentheses in the second line of the table. Table 4 reveals that the model is well specified with non-biased coefficients. Fischers F = 529.33 with the prob >F = 0.00 signifies that the model has good global significance. This result is confirmed by the adjusted R2 of 58.40%, which signifies that 58.40% of the variance of the dependent variable is explained by the independent variables. TABLE 4. RESULTS OF TESTS OF MODEL
Finsus Cons Par>30 Finrevrat Percosrat Agemfi Avloanbal
Percwomen

Coef. 99.29 -7.02 1.20 -2.67 0.842 0.005 -0.041 0.046

t-statistic (26.09) *** (-26.56) *** 21.84) *** (19.22) *** (7.25) *** (3.15) ** (-1.24) (7.59) ***

# obs

F (7, 1728)

Prob > F

Adjusted R2

Root MSE

1736

529.33

0.000

0.5840

11.523

Borlooff

*** Indicates level of statistical significance at 1%; ** Indicates level of statistical significance at 5%; * Indicates level of statistical significance at 10%. RESULTS Portfolio-at-risk at 30 Days (Par>30) The coefficient of the portfolio-at-risk at 30 days is negative, as expected, and statistically significant. This confirms our second hypothesis, namely that a significant

316
reduction in the portfolio-at-risk at 30 days in the global portfolio should have a positive impact on the MFIs financial sustainability. In other words, a high portfolio-at-risk would limit the revenue derived from microcredit operations and therefore decrease the amount of lendable funds. This would lead to the rationing of credit and ultimately the inability to sustainably supply quality services to the clientele, and have a negative impact on MFIs financial results and hence their financial sustainability. The negative value of the coefficient of -7.02 of the portfolio-at-risk, which is statistically significant, clearly illustrates this problem. Another adverse consequence of the inverse relation between financial selfsustainability and the portfolio-at-risk is that a decrease in financial sustainability that leads to a rationing of credit hinders the MFIs ability to increase its client outreach, which results from the massification of microcredit intended to substantially reduce poverty. Thus, adjustment measures must be taken because beyond this worrisome situation, persistent financial distress could lead to bankruptcy and consequently the cessation of the MFIs activities. The threshold of Par>30 of 5% stipulated by CGAP (1999) seems to be much higher than the percentage that would enable MFIs to ensure their sustainability. As part of a study conducted in Senegal in December 2007, the managers of the MFI ACEP informed us that they would require a Par>30 of 2% to ensure the financial sustainability of their institution. This assertion was corroborated by the administrators of the Egyptian MFI Lead Foundation that we met at the Global Microfinance Investment Congress in Paris in January 2008. Moreover, the coefficient of the portfolio-at-risk at 30 days is the highest regression coefficient in absolute value. It follows that the portfolio-at-risk (Par>30) is the most determining indicator of the financial sustainability of MFIs. This finding reinforces our second hypothesis, which states that a high quality credit portfolio is instrumental to the financial sustainability of MFIs. In other words, to be financially sustainable MFIs must undertake preventive risk management to build a quality credit portfolio. Accordingly, to increase their financial sustainability, MFIs must engage in risk management planning by assessing their capacity to anticipate problems of recovering outstanding credit to prevent risks of perverse contagion among the borrowers. In a context of under-mobilization of savings and shrinkage of external financing possibilities, only a weak Par>30 underlies the sustainability of MFI. This inevitably leads to a professionalization of MFIs as they must exhibit effective management of their credit risk. To face all these difficulties, and as part of good credit risk management, MFIs can group together to create a credit bureau to foster dialogue and improve the flow of information on borrower solvency. When MFIs begin to compete and to dispute over the clientele, the default repayment rate rapidly rises if they do not have access to a database containing the relevant elements of their clients behavior. Moreover, in the absence of a credit bureau, increased MFI control over individual loans and group loans with joint surety (see Conning (1999), Laffont and Nguessam (2000), Ahlin and Townsend (2007) and Casar and al (2007)) should play a preponderant role in environments marked by solidarity of members, such as Africa and Asia. Joint surety is both an ex ante and ex post guarantee that mitigates the risk of default (see Stiglitz, (1990) Gatak, (1999) and Morduch, (1999)). Thus, sound credit risk management that promotes a satisfactory loan repayment rate would enable MFIs to increase their average

317
credit portfolio, at identical expense levels. They could then gradually lower their effective interest rates, thus improving their competitiveness. Other Independent Variables The coefficient of the financial revenue ratio (Finrevrat) is positive and statistically significant. This confirms hypothesis 1 and hence the argument that the higher the interest rates, the higher the chances of MFIs attaining financial sustainability. Further, the fact that the coefficient is statistically significant indicates that MFIs should apply sufficiently high interest rates to generate financial revenue that lets them cover all their expenses while generating a profit to ensure their financial sustainability. The management and productivity variables have the expected signs and are statistically significant. The coefficient of the personnel cost ratio (Percosrat) is negative, which is compatible with the argument that ceteris paribus, the increase in personnel costs causes a decrease in financial sustainability. Accordingly, to attain sustainability, MFIs must significantly decrease personnel expenses. This can only be achieved by emphasizing employee training and financial motivations to augment productivity. In addition, they must put in place and judiciously use new technologies and efficient information systems to facilitate the attainment of financial sustainability. The coefficient of the number of borrowers per loan officer (Borlooff) is positive. This result is consistent with the argument that higher employee productivity is associated with a stronger likelihood of an MFIs attaining financial sustainability. Nonetheless, the low value of the coefficient implies that gains in productivity are not proportionate with the increase in the number of loan officers, or simply that the increase in the number of borrowers is more proportionate with the increase in the number of loan officers. This result also indicates that beyond financial sustainability, MFIs are heavily inclined to grant microcredit to a large number of poor people to significantly increase the client outreach of their actions. Concerning the other variables, the age coefficient of MFIs (Agemfi) is positive and significant. This implies that as MFIs mature, and thus acquire experience in their sector; they increase their likelihood of attaining financial sustainability. This can be explained by the fact that MFIs gradually improve their control over all operations related to issuance of microcredit. In other words, MFIs that have considerable experience in the microfinance sector have diligently applied credit risk management and general efficient management techniques to attain financial sustainability. The coefficient of average loan balance (Avloanbal), an indicator of client outreach, is positive and statistically significant. Nonetheless, the very low value of this coefficient (0.005) shows that the client outreach of the action of MFIs is not a major variable for financial sustainability, as we have explained with respect to the increase in the number of loan officers, which is less than proportionate with the increase in the number of borrowers. The coefficient of the percentage of women (Percwomen) is negative and statistically non-significant, and therefore refutes our hypothesis. This result leads us to reject the commonly held notion that women are more industrious, more reliable and are better repayers of microcredit than men. This belief is grounded in the argument that the majority of women are in the most disadvantaged margin of MFI operations, and that they consequently must be helped to launch activities that generate subsistence income

318
for their families in order to directly reduce poverty. Moreover, given that womens activities tend to be small scale and intended to generate income to attain food selfsufficiency, and therefore are not conducive to the accumulation of wealth in the form of savings, it is quite understandable that the percentage of women borrowers has a negative impact on financial sustainability owing to possible repeated bankruptcies of these microactivities. ANALYSIS ACCORDING TO GEOGRAPHICAL REGION AND LEGAL STATUS AND CREDIT METHODOLOGY After presenting and analyzing all the results of our general model, we will now examine financial sustainability relative to geographical region, legal status and credit issuance method to illustrate similarities and differences among MFIs. Analysis by Geographical Region Table 5 presents the results of the regression with heteroskedasticity-consistent t-statistics in parentheses below the coefficients for the geographical region. Similar to the results for the general model, the results of our regressions show that the quality of the portfolio, the financial revenue ratio and personnel expense ratio are by far the three variables that affect financial sustainability most significantly when the region of implementation of MFIs is examined. Moreover, the age coefficient of MFIs and that of borrowers per loan officer are positive and statistically significant for all geographic regions except for Sub-Saharan Africa, where they are positive but statistically non-significant. In the Latin America-Caribbean region, where there are the most MFIs, the coefficient of the average loan balances is positive and statistically significant. In the other geographical regions, the coefficients of the average loan balance are partly positive but very low compared with that of the Latin American-Caribbean region, and are not statistically significant. Lastly, for all geographical regions, the coefficients of the percentage of women borrowers are negative and statistically non-significant. This result confirms the results of the regressions of the general model, namely that the percentage of women among the MFI clientele does not explain MFIs financial sustainability.
TABLE 5. RESULTS OF TESTS OF MODEL BY GEOGRAPHICAL REGION Robust Finsus Par>30 Finrevrat Percosrat Agemfi Avloanbal AFRICA -6.96805 (-11.67) *** 1.30025 (4.19) *** -2.70322 (-3.72) *** 0.7032254 (1.08) 0.005143 ASIA -7.05952 (-12.58) *** 1.22609 (10.55) *** -2.65428 (-9.37) *** 0.82277 (3.34) *** 0.005123 ECA LAC MENA -6.957981 (-13.26) *** 1.247146 (11.11) *** -2.66959 (-9.75) *** 0.801180 (3.43) *** 0.005383

-7.11530 -7.03688 (-8.15) *** (-18.50) *** 1.26275 1.21140 (7.12) *** (15.27) *** -2.76035 -2.61084 (-6.39) *** (-13.48) *** 0.804742 0.00538 (2.10) ** (4.96) *** 0.004464 0.829936

319
Percwomen Borlooff Constant # obs F (7, 1728) Prob > F (0.54) -0.032992 (-0.18) 0.0438528 (1.40) 99.8733 (4.97) *** 315 17.52 0.0000 (1.38) -0.04689 (-0.67) 0.04427 (3.51) *** 100.567 (12.46) *** 297 120.72 0.0000 (0.77) -0.577181 (-0.52) 0.041630 (2.18) ** 102.600 (8.16) *** 405 52.94 0.0000 (2.14) ** -0.042822 (-0.90) 0.04534 (5.25) *** 99.7449 (18.23) *** 810 256.84 0.0000 (1.54) -0.040008 (-0.61) 0.0445126 (3.76) *** 99.44594 (13.23) *** 126 131.81 0.0000

R-squared 0.5897 0.5858 0.5888 0.5847 0.5876 Root MSE 12.069 11.538 11.575 11.522 11.497 *** Indicates level of statistical significance at 1%; ** Indicates level of statistical significance at 5%; * Indicates level of statistical significance at 10%

Analysis by Legal Status Table 6 presents the results of the regression with heteroskedasticity-consistent t-statistics in parentheses below the coefficients related to legal status. Regressions based on legal status confirm the results obtained above. For all types of MFIs, the highest coefficient in absolute value is that of the portfolio-at-risk, which confirms the fact that a quality credit portfolio is a determining factor in the financial sustainability of MFIs. Moreover, this coefficient is statistically significant regardless of the legal status of the MFI. The same principle applies to coefficients of the variables Finrevrat, Percosrat, Agemfi and Borlooff, which indicates that the application of adequate interest rates, good management that entails control of personnel expenses, the experience of the MFI and productivity of loan officers significantly influence MFIs attainment of financial sustainability. These results also show that the client outreach of MFI programs and the percentage of women among the clientele are not major variables in the financial sustainability of these MFIs. Analysis According to Credit Methodology Table 7 presents the results of the regression with heteroscedasticity-consistent t-statistics in parentheses below the coefficients based on credit methodology. TABLE 6. RESULTS OF TESTS OF MODEL ACCORDING TO LEGAL STATUS
Robust Finsus Par>30 Finrevrat Percosrat Agemfi Bank -7.1758 (-5.90) *** 1.256339 (5.58) *** -2.77998 (-4.99) *** .9787173 (2.00) ** Savings and credit union -7.11530 (-8.15) *** 1.262753 (7.12) *** -2.760358 (-6.39) *** .8047423 (2.10) ** NFBI -7.0441 (-15.82) *** 1.216078 (13.12) *** -2.62473 (-11.61) *** .8276614 (4.23) *** NGO -6.9958 (-21.21) *** 1.19973 (17.15) *** -2.5829 (-15.03) *** .8280633 (5.68) *** Rural bank -7.16501 (-9.54) *** 1.144308 (7.47) *** -2.47989 (-6.43) *** .9428494 (2.91) **

320
.004692 .004464 .005303 .0056773 .0049433 (0.61) (0.77) (1.80) * (2.61) * (1.00) Percwomen -.073236 -.057181 -.0441165 -.0367499 -.0591254 (-0.49) (-0.52) (-0.79) (-0.89) (-0.63) Borlooff .0385927 .0416308 .045001 .0458448 .042422 (1.51) (2.18) ** (4.46) *** (5.99) *** (0.014) *** Constant 103.4727 102.6004 100.0071 98.99158 101.5486 (5.74) *** (8.16) *** (15.62) *** (20.85) *** (9.42) *** # obs 216 126 720 765 126 F 83.63 52.94 188.76 229.98 67.37 Prob > F 0.0000 0.0000 0.0000 0.0000 0.0000 R-squared 0.5902 0.5888 0.5850 0.5843 0.5896 Root MSE 11.868 11.575 11.527 11.547 11.552 *** Indicates level of statistical significance at 1% ; ** Indicates level of statistical significance at 5%; * Indicates level of statistical significance at 10% Avloanbal

TABLE 7. RESULTS OF TESTS OF THE MODEL BASED ON CREDIT METHODOLOGY

Robust Finsus Par>30 Individual Individual/ Joint Group joint Village bank

-7.008421 -7.036885 -7.115304 -7.004999 (-12.23) *** (-18.50) *** (-8.15) *** (-6.16) *** Finrevrat 1.25319 1.211402 1.262753 1.306477 (10.15) *** (15.27) *** (7.12) *** (5.01) *** Percosrat -2.725016 -2.610846 -2.760358 -3.031484 (-9.12) *** (-13.48) *** (-6.39) *** (-5.01) *** Agemfi .7479151 .8299364 .8047423 .5997264 (2.88) ** (4.96) *** (2.10) ** (1.04) Avloanbal .0051816 .0053871 .004464 .0088761 (1.35) (2.14) ** (0.77) (1.20) Percwomen -.0385724 -.0428227 -.057181 .0172565 (-0.54) (-0.90) (-0.52) (0.12) Borlooff .0454753 .0453434 .0416308 .051886 (3.52) *** (5.25) *** (2.18) (2.07) ** Constant 100.1745 99.74491 102.6004 96.2362 (12.29) *** (18.23) *** (8.16) *** (6.35) *** # obs 612 999 162 180 F 108.14 256.84 52.94 62.53 Prob > F 0.0000 0.0000 0.0000 0.0000 R-squared 0.5883 0.5847 0.5888 0.5842 Root MSE 11.446 11.522 11.575 11.901 *** Indicates level of statistical significance at 1% ; ** Indicates level of statistical significance at 5%; * Indicates level of statistical significance at 10%

Here again the results show that the highest coefficient in absolute value is that of the portfolio-at-risk, which is statistically significant. The same applies for the variables Finrevrat, Percosrat and Agemfi, with the exception of village banks, for which the age coefficient of MFIs is positive but statistically non-significant. For village banks, the lack of training of managers, who are generally from the same village, may explain the fact the experience acquired by the MFI does not significantly influence its financial sustainability. Similarly, we note that the variables Percwomen and Avloanbal do not

321
have significant coefficients. Here again, a quality portfolio, efficient management and the application of appropriate interest rates to cover the expenses related to issuance of microcredit are the most determining factors in the attainment of financial sustainability. CONCLUSION Microfinance is often portrayed in the literature as a tool that allows individuals excluded from the formal financial system to gain access to sources of financing, i.e. as a means of combating exclusion and poverty. This research presents an original approach to studying management of MFIs based on financial sustainability. Determining the most pertinent indicators of financial sustainability of MFIs was the main objective of this article. To attain this objective we began by reviewing the literature, and then identified indicators of financial sustainability that could apply to the empirical data. After collecting these data on the MIX Market site, we formed a basic sample of 217 MFIs from 101 countries over a 9-year period (1998-2006). Subsequently, we processed and analyzed the data gathered to test our model and clarify the determinants of financial sustainability of MFIs. The results show that the quality of the credit portfolio resulting from good credit risk management is the determining factor in the financial sustainability of MFIs. In all our regressions, the variable Par>30 had the highest coefficient in absolute value, thus confirming our hypothesis that credit risk management that ensures a quality portfolio is the determining factor of financial sustainability of MFIs. Our results also show that the application of sufficiently high interest rates that generate a profit, together with quality management that integrates good expense control, the application of effective banking practices and the implementation of an adequate information system are indispensable for financial sustainability of MFIs. Further, our results show that the client outreach of microfinance programs, the age of MFIs and the percentage of women among the clientele do not significantly influence the MFIs attainment of financial sustainability. Note that the coefficient of the percentage of women variable is not significant in any case. Lastly, the results pertaining to geographical region, credit method and legal status of MFIs corroborate those of the general model. ENDNOTES
Definition by MicroBanking Bulletin (2005) This study shows that in the past, the use of subsidized or targeted credits led to flagrant failures that spawned a generation of skeptics. There are three reasons for these failures. First, MFIs were not financially autonomous, which caused their decapitalization. Second, the funds did not reach the target populations. Lastly, these MFIs distorted the financial markets and thus hindered the effective financing of important sectors of the economy. 3 Morduch (1999) reports diverging positions regarding the role of subsidies in the continuity of MFIs stemming from the avowed failures of State-subsidized credit policies intended to foster the emergence of financially sustainable institutions. 4 They explain that Morduch (2000) estimated the number of MFIs that are presently financially sustainable at 1% and predicts that 5% would be sustainable in the near future.
2 1

322
5

Credit scoring uses quantitative performance measures and characteristics of preceding loans to predict performance of future loans with similar characteristics. Credit scoring serves to neither approve nor reject a loan application; rather it predicts the probability of occurrence of bad performance (default) as defined by the borrower. Credit scoring is a means of reducing information asymmetries and transaction costs. 6 Pre-default allows timely detection of a borrowers difficulties to prevent definitive default (Lanha, 2004) 7 One of the adverse consequences of moral hazard is the deterioration of portfolio quality. 8 For MFIs that obtain a one-diamond rating, only general information is available. 9 Moreover, the fact that the correlations between these three variables are positive and significant but not perfect (Cull, Deminrg-Kunt and Morduch (2007)) is an additional reason that led us to choose our dependent variable judiciously. 10 The distribution is normal if the value of skewness is less than one and does not exceed twice the standard error, i.e. n Distribution is normal if the absolute value of kurtosis is below 3. 12 The ratio of the number of borrowers per loan officer is a variable that illustrates the productivity of the loan officer. Consequently, an increase between periods indicates positive growth in the productivity of the loan officer. 13 The results of the endogeneity test are not presented in this paper but can be supplied to the editor upon request, for verification 14 The Breusch Pagen test is also not presented in the paper but can be supplied to the editor upon request, for verification.
11

REFERENCES Acclassato D H (2006), Taux dintrt effectif, viabilit financire et rduction de la pauvret par les institutions de micro finance au Bnin.Cahiers de recherches du laboratoire dconomie dOrlans, 18P Ayayi A (2009), Microfinance: Debt or Equity? What are the Implications for Profitability and Social Welfare, Global Economic and Finance Journal, Vol. 2, No.2 PP6480 Ahlin C and Townsend R (2007), Using Repayment Data to Test Across Models of Joint Liability Lending , Economic Journal, Vol. n 117, pp. F11-F51. Armendariz de Aghion B and Morduch J. (2000), Microfinance Beyond Group Lending , Economics of Transition, Vol. n8, pp. 401-420. Brau J. and Woller G M (2004), Microfinance: A Comprehensive review of the Existing Literature. Journal of Entrepreneurial Finance and Business Venture, Vol. 9, no.1, PP.1-26 Casar A, Crowley L and Wydick B (2007), The Effect of Social Capital on Group Loan evidence From Fields Experiments, Economic Journal, Vol. n 117, pp. F85-F106. CGAP (1995),Comment maximiser ltendue du financement des micro entreprises : Ce que nous enseignent les institutions performantes . Focus n 5 .October 1995. CGAP (1999), Mesurer les taux dimpay en micro finance, les ratios peuvent tre dangereux pour votre sant , Etude Spciale n 3, Washington.

323
CGAP (2004),Explication raisonne des taux dintrt utiliss pour le microcrdit, Note sur la microfinance, (6), 2p. CGAP (2005), Etude sur le transfert dargent des migrs au Sngal et les services de transfert en microfinance, Publi par : Unit des finances sociales de l'Organisation Internationale du Travail, 50p. CGAP (2006), La micro finance et les banques: Le down scaling: une opportunit pour les banques, Etude spciale n 5, Washington Chowdhury P R (2007), Group-Lending with sequential financing Contingent renewal and social capital, Journal of Development Economics, Vol. n 84, pp. 487-506 Churchill G. (2000), Banking on Customer Loyalty, Journal of Microfinance, vol. n 2, pp.1-21. Conning J. (1999), Outreach, Sustainability and Leverage in Monitored and Peermonitoring Lending, Journal of Development Economic, Vol. 60, pp. 229-248. Cull R, Demirg-Kunt A and Morduch J (2007), Financial Performance and Outreach: A Global Analysis of Leading MicroBank, Economic Journal, Vol. 117, pp. F107-F133. Gibbons D S and Meehan J W. (2000), Le dficit du micro crdit: comment obtenir lautonomie financire institutionnelle tout en prservant lengagement datteindre les familles les plus pauvres, Abidjan. Ghatak, M., Group Lending, Local Information, and Peer Selection, Journal of Development Economics, 1999, Vol. 60, pp. 229-248 Hermes N, Lensink R and Mehrteab HT, Does The Group Leader Matter: The Impact of Monitoring Activities and Social Ties of Group Leaders on The Repayment Performance of Group-Based Lending in Eritrea, African development review, 2006, Vol. n 18, pp. 72-97. Hermes N, Lensink R, Introduction: The empirics of microfinance: what do we know?, Economic Journal, 2007, Vol. 117, pp. F1-F10 Hollis A. and Sweetman A , The Life-cycle of a Microfinance Institution: The Irish Loan Funds, Journal of Economic Behavior& Organization, 2001, Vol. 46, pp. 291-311. Laffont J J and NGuessam, T. , Group Lending with Adverse Selection, European Economic Review, 2000, Vol. n 44, PP. 773-784. Lanha M., Efficacit relative du prt de groupe avec caution solidaire, Cahier de Recherches, Laboratoire d Economie dOrlans, 2004. Littlefield E and Rosenberg, R., La microfinance et les pauvres: La diffrence entre et secteur formel sestompe, Finance et Dveloppement, 2004,Vol. 41, n 2, pp. 38-40. Mishra P. , Crop Insurance and Crop Credit: Impact of the Comprehensive Scheme on cooperative Credit in Gujarat, Journal of International Development, 1994,Vol. n 6, pp. 529-568. Morduch J.,The microfinance promise, Journal of Economic Literature, 1999, Vol.37, pp.1569-1614. Morduch J., the microfinance schism, World Development, 2000, Vol. 38, pp. 617-629. Navajas S Schreiner M., Meyer, R. L., Gonzalez-Vega C. and Rodriguez-Meza J.,, Microcredit and The Poorest: Theory and Evidence from Bolivia, World Development, 2000, Vol. 28, and pp.333-346. Norell, Dan, How to reduce arrears in microfinance institutions, Journal of Microfinance, 2001,Vol .3, pp.115-130.

324
Patten R H, Rosengard J.K. and Johnston Jr. D. E. , Microfinance Success Amidst Macroeconomic Failure: The experience of Bank Rakyat Indonesia during the East Asian crisis, World Development, 2001,Vol. 29, p. 1057-1069. Pitt M M and Khandker R. , The Impact of group-based credit programs on poor households in Bangladesh: Does the gender of participants matter? The Journal of Political Economy, 1998, Vol.6, pp 958-996. Robinson, M.S, Addressing some key questions on finance and poverty, Journal of International Development, 1996, Vol. 8, pp153-163 Schreiner M, Credit scoring for microfinance: Can it work?, Journal of Microfinance, 2000, Vol. 2, pp 105-118. Schreiner M., Les vertus et les faiblesses de lvaluation statistique credit scoring en microfinance, Microfinance Risk Management and Center for Social Development, Washington University in St Louis, 2003 Stiglitz, J. E., Peer monitoring and credit markets. World Bank Economic Review, 1990, Vol. 3, pp. 351-366. Von Pischke J., Measuring the trade-off between outreach and sustainability of microentreprise lenders, Journal of International Development, 1996, Vol. 8, pp. 225-239. Woolcock M. (2001), Microenterprise and Social Capital: A Framework for Theory, research, and Policy, Journal of Socio-Economics, 2001, Vol.30, pp. 193-198.