measuring the efficiency of sub-saharan africa's microfinance institutions and its drivers

24
Annals of Public and Cooperative Economics 84:4 2013 pp. 399–422 MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS AND ITS DRIVERS by KEMONOU Richard Senami Segun and M. ANJUGAM Tamil Nadu Agricultural University, India ABSTRACT: This paper uses a large panel data set covering 70 MFIs in 25 Sub- Saharan African countries to analyze the efficiency of MFIs. This is important, given that MFIs have to operate efficiently to fulfil its dual mission of serving the poor and being sustainable. The results reveal that MFIs are inefficient in meeting the goals of either providing microfinance related services to their clients or intermediating funds between borrowers and depositors. The MFIs lack ability to reach efficient sizes of their performing loan portfolio at the same time they reach an efficient number of clients served. Keywords: microfinance institutions, efficiency, data envelopment analysis, Tobit regression, Sub-Saharan Africa. Medida de la eficiencia de las instituciones de microfinanzas en el ´ Africa Subsahariana Este art´ıculo se basa en una gran muestra de datos de panel relativos a 70 instituciones de micro- finanzas (IMF) en 25 pa´ ıses del ´ Africa Subsahariana, con la finalidad de analizar su eficiencia. Esto es importante en la medida en que las IMF deben operar con eficiencia para cumplir su doble misi´ on de prestar servicio a los pobres y de sostenibilidad en el tiempo. Los resultados ponen de manifiesto que las IMF son ineficientes tanto para proporcionar servicios de microfinanzas a sus clientes como para servir de intermediarios de fondos entre prestatarios y depositantes. Las IMF carecen de aptitudes para alcanzar dimensiones eficientes para sus carteras de pr´ estamos, siendo rentables en t´ erminos del n ´ umero de clientes atendidos. Messung der Effizienz von Mikrofinanzinstitutionen und ihrer F ¨ uhrungen in afrikanischen L ¨ andern s ¨ udlich der Sahara In diesem Beitrag wird ein großes, 70 Mikrofinanzinstitutionen (MFIs) in 25 afrikanischen L ¨ andern udlich der Sahara umfassendes Paneldataset benutzt, um die Effizienz von MFIs zu analysieren. Dies ist wichtig, wenn man bedenkt, dass MFIs effizient arbeiten m ¨ ussen, um ihre duale Mission zu erf ¨ ullen, n ¨ amlich den Armen zu dienen und nachhaltig zu sein. Die Ergebnisse zeigen, dass E-mail: [email protected]; [email protected] © 2013 The Authors Annals of Public and Cooperative Economics © 2013 CIRIEC. Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Upload: m

Post on 16-Feb-2017

214 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

Annals of Public and Cooperative Economics 84:4 2013 pp. 399–422

MEASURING THE EFFICIENCY OF SUB-SAHARANAFRICA’S MICROFINANCE INSTITUTIONS

AND ITS DRIVERS

by

KEMONOU Richard Senami Segun∗ and M. ANJUGAMTamil Nadu Agricultural University, India

ABSTRACT: This paper uses a large panel data set covering 70 MFIs in 25 Sub-Saharan African countries to analyze the efficiency of MFIs. This is important, giventhat MFIs have to operate efficiently to fulfil its dual mission of serving the poor andbeing sustainable. The results reveal that MFIs are inefficient in meeting the goals ofeither providing microfinance related services to their clients or intermediating fundsbetween borrowers and depositors. The MFIs lack ability to reach efficient sizes of theirperforming loan portfolio at the same time they reach an efficient number of clientsserved.

Keywords: microfinance institutions, efficiency, data envelopment analysis, Tobit regression,Sub-Saharan Africa.

Medida de la eficiencia de las instituciones de microfinanzas en el AfricaSubsahariana

Este artıculo se basa en una gran muestra de datos de panel relativos a 70 instituciones de micro-finanzas (IMF) en 25 paıses del Africa Subsahariana, con la finalidad de analizar su eficiencia.Esto es importante en la medida en que las IMF deben operar con eficiencia para cumplir su doblemision de prestar servicio a los pobres y de sostenibilidad en el tiempo. Los resultados ponen demanifiesto que las IMF son ineficientes tanto para proporcionar servicios de microfinanzas a susclientes como para servir de intermediarios de fondos entre prestatarios y depositantes. Las IMFcarecen de aptitudes para alcanzar dimensiones eficientes para sus carteras de prestamos, siendorentables en terminos del numero de clientes atendidos.

Messung der Effizienz von Mikrofinanzinstitutionen und ihrer Fuhrungenin afrikanischen Landern sudlich der Sahara

In diesem Beitrag wird ein großes, 70 Mikrofinanzinstitutionen (MFIs) in 25 afrikanischen Landernsudlich der Sahara umfassendes Paneldataset benutzt, um die Effizienz von MFIs zu analysieren.Dies ist wichtig, wenn man bedenkt, dass MFIs effizient arbeiten mussen, um ihre duale Missionzu erfullen, namlich den Armen zu dienen und nachhaltig zu sein. Die Ergebnisse zeigen, dass

∗ E-mail: [email protected]; [email protected]

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC. Published by John Wiley & Sons Ltd, 9600 Garsington Road, OxfordOX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Page 2: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

400 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

MFIs ineffizient hinsichtlich beider Ziele sind: der Bereitstellung von Dienstleistungen im BereichMikrofinanz an ihre Kunden wie der Tatigkeit als Finanzmittler zwischen Kreditnehmern undKreditgebern. Den MFIs fehlt die Fahigkeit, effiziente Großenordnungen ihrer laufenden Kredit-portfolios zu erreichen und zugleich eine effiziente Anzahl von Kunden zu gewinnen.

Mesure de l’efficacite des institutions de microfinance en AfriqueSub-Saharienne et de ses determinants

Cet article se base sur un grand echantillon de donnees en panel portant sur 70 institutions demicrofinance (IMF) dans 25 pays d’Afrique Sub-Saharienne afin d’analyser leur efficacite. C’estimportant dans la mesure ou les IMF doivent operer avec efficacite pour remplir leur double missionde service aux pauvres et de durabilite. Les resultats revelent que les IMF sont inefficaces tant pourfournir des services de microfinance a leurs clients que pour servir d’intermediaires de fonds entreemprunteurs et depositaires. Les IMF manquent d’aptitude a atteindre des tailles efficaces pourleurs portefeuilles de prets tout en etant rentables en termes de nombre de clients servis.

1 Introduction

One of the most stylized facts of developing economies is that formal financial institu-tions leave the poorest population tightly constrained in their access to financial services.It is widely recognized that economic progress relies largely on access to financial ser-vices such as savings, insurance and credit. It has been argued for long that formalbanking institutions have not met the credit needs of financially challenged people whoare not able to offer collaterals but who have feasible and promising investment ideasthat can turn into profitable initiatives (Hollis and Sweetman 1998). The incidence ofsuch exclusion has also been highlighted by Littlefield et al. (2003) stating that thecommercial banking sector does not consider the poor bankable owing to their inabilityto meet the eligibility criteria including collateral. This has limited the accessibility andprovision of timely and adequate credit from formal financial institutions for the poor-est section of the society. The World Bank Rural Finance Access Survey (World Bank2003) indicates that rural banks serve primarily the needs of richer rural borrowersin developing countries. According to the survey, 66 per cent of large farmers have de-posit accounts, 44 per cent have access to credit. Meanwhile, the rural poor face severedifficulties accessing savings and credit from the formal sector. It is also revealed that70 per cent of the poorest households (marginal/landless farmers) do not have a bankaccount, 87 per cent have no access to credit. Only 20 per cent of African householdshave access to formal financial services and the unmet demand for financial services inSub-Saharan Africa is vast (African Union 2009). This led to a search for alternativepolicies, systems and procedures, savings and loan products, other complementary ser-vices and new delivery mechanisms that would fulfil the requirements of the poor. Inthis context, microfinance has emerged as the most suitable and practical alternative tothe conventional banking in reaching the hitherto unreached poor population.

This paper analyzes the efficiency of microfinance institutions (MFIs) in Sub-Saharan Africa. As microfinance has evolved as an accepted institutional frameworkto provide financial services to poor and microfinance institutions are considered as

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 3: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 401

the vehicle for advancement of microcredit to them, success of microfinance programmedepends on the successful functioning of these institutions. Therefore, it becomes im-perative to have an insight into the efficiency of their operations. The purpose of thepaper is to find out whether microfinance institutions are performing efficiently in theiroperations. The efficiency of microfinance institutions in Sub-Saharan Africa is veryimportant to highlight because of two main reasons. First, there has been little studyconducted on efficiency of microfinance institutions in the region. Secondly, efficientmicrofinance institutions are needed to avoid mission drift (Mersland and Strøm 2008,Freixas and Rochet 2008), but the question is how these institutions can become effi-cient? This paper is contributing by identifying and studying the factors determining theefficiency of microfinance institutions and their effects through which decision makersmay acquire a better knowledge of where to put in effort to improve the efficiency ofmicrofinance institutions.

Efficiency has been defined and measured in multiple ways in the microfinance lit-erature. Farrell (1957) first proposes an approach to estimate the efficiency of observedunits and decomposes efficiency into two elements viz., (a) technical efficiency whichmeasures the firm’s success in producing maximal output with a given set of inputs and(b) allocative (price) efficiency which estimates the firm’s success in choosing an opti-mum combination of inputs, given their respective prices. These two elements are thencombined to provide the total economic efficiency. Efficiency measures include Data En-velopment Analysis and Stochastic Frontiers which involve mathematical programmingand econometric methods, respectively. This paper is concerned with the measurementof efficiency through Data Envelopment Analysis, using a panel data set collected fromthe Microfinance Information Exchange1 (MIX) spanning 70 MFIs in 25 countries overthree years. The factors determining the efficiency of MFIs are studied through regres-sion analysis. We take into account the specific methodological problems associated withthis type of estimation by (a) using a Tobit regression and (b) right censoring the efficientMFIs.

The outline of the paper is the following: institutional details, selection issues,description of the data and empirical analysis are in Section 2. Section 3 has the resultsand discussion. Section 4 presents the study’s conclusions and possible implications.

2 Context

(a) Institutional background

According to various researchers (African Development Bank 2006, Ouedraogoand Gentil 2008), microfinance has existed for centuries in Sub-Saharan Africa. Thereare many examples in the history of microfinance ranging from small-scale, rotatingsavings-and-loans clubs ‘tontines’ in South Africa, Cote d’Ivoire and Ethiopia duringthe 18th century to Savings and Credit Cooperatives in Kenya in the 19th century. InNigeria, microfinance goes back to the 15th century. The mainstreaming, formalizationand recognition of microfinance as part of the formal financial sector in Sub-SaharanAfrica begins to gain momentum in the late 1990s.

1 See www.mixmarket.org

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 4: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

402 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Microfinance undergoes four distinct phases of evolution in Sub-Saharan Africa.The first phase of microfinance begins in the 1950s and comprises of directed, subsidizedcredit often targeting individuals who did not have the means to repay loans. Theseschemes assume that the lack of money is the main obstacle to eliminate poverty. Thesecond phase starts in 1970s consists of microcredit offered mostly through NGOs,beginning with the Grameen Bank in Bangladesh and followed in Sub-Saharan Africaby a majority of NGOs which provide microcredit to the poor. However, financial self-sufficiency is not important for such NGOs as they receive resources from developmentagencies. NGOs are acting as intermediaries, largely functioning as income transferagents for social purposes rather than financial intermediaries. The third phase is theformalization of microfinance institutions in most Sub-Saharan African countries in1990s. In response to the demand, MFIs begin offering more financial services such assavings and insurance products. Microfinance also demonstrates that it can improve thesocioeconomic well-being of its clients and their families. The fourth phase is happenedto be the period of mainstreaming of microfinance institutions into the formal financialsector and then microfinance institutions become part of the formal financial sectorwithout losing its focus on serving the poor. As on 31 December 2010, 22,900 MFIs arefunctioning in Sub-Saharan Africa with an outreach of 4.5 million active borrowers andgross loan portfolio of US$14.9 billion (MIX and CGAP 2012).

The data used in this paper include information on income statement and balancesheet of the reporting MFIs to the MIX, which is the leading business informationprovider dedicated to strengthening the microfinance sector through database creation.The organization’s core focus is to provide objective data and analysis on microfinanceproviders. MIX promotes financial transparency in the industry and helps to build theinformation infrastructure in developing countries. The present study takes the averagedata for the three consecutive years 2008, 2009 and 2010 for its analysis.

(b) Selection of sample microfinance institutions

We adopt a multistage purposive sampling technique for the selection of sampleMFIs. In the first stage, 147 Sub-Saharan African MFIs are selected purposively out of508 MFIs reporting data to the MIX based on the information available for all MFIs forthe three consecutive years 2008, 2009 and 2010. In the second stage, MFIs reportedwith diamond2 levels 4 and 5 are purposively selected since the data available from theseMFIs will have a high level of reliability. The MFIs of diamond level 4 provide generalinformation, data on outreach, financial indicators and audited financial statementswhereas, the MFIs of diamonds level 5 are rated by rating agencies in addition to thecharacteristics of MFIs of level 4. We assume that there is no significant qualitativedifference between the MFIs of diamonds levels 4 and 5. Hence, out of 147 MFIs selectedin the first stage, 111 MFIs comprising of 71 MFIs of diamond level 4 and 40 MFIs ofdiamond level 5 are purposively considered.

2 MIX uses a scale of diamond ranging from 1 to 5 to assess the transparency and quality ofinformation provided by MFIs. The highest level of diamonds indicates that the MFI has auditedfinancial statements.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 5: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 403

Since the study intends to collect data for the three consecutive years 2008, 2009and 2010 on the MFIs, missing data in the panel setting are found. The way to handlethe problem will be either to impute the missing data or to limit the study with the MFIshaving complete information throughout the panel. In this paper, the latter approach isfollowed i.e., a sample of 70 MFIs which are having full information on required datafor the three years 2008, 2009, 2010 are purposively selected. The list of these MFIs isfurnished in the Appendix.

(c) Data Envelopment Analysis

This paper employs Data Envelopment Analysis (DEA) to analyze the efficiencyof MFIs i.e., a unit’s ability to obtain maximum output from a given set of inputs of theselected MFIs. We focus on assessing only the technical efficiency of MFIs because dataavailability hinders the examination of alternative categories of efficiency. In particular,prices are needed to examine both allocative and economic efficiency. There is not enoughinformation from the available data to estimate these prices. Moreover, many of theMFIs did not obtain their inputs at market terms. In general, MFIs behave like otherproduction firms because they acquire most of their inputs in the market, their linkageswith other firms are through the market and their output is supplied in the market (Leon2001). However, in Sub-Saharan Africa, many of MFIs inputs (funds in particular) areobtained from non-market sources. Other subsidized inputs are heterogeneous and ofdiverse quality that it will be difficult to obtain a single shadow price for them. The twoorientations of the DEA technical efficiency measure viz., Output-oriented and Input-oriented Measures are discussed in the following sections.

An Output-oriented Model (OOM) implies that technical efficiency is estimated bythe output of firms (MFIs) relative to the best-practice level of output for a given level ofinputs. In order to specify the mathematical formulation of the OOM, let us assume thatwe have K decision making units (DMU) using N inputs to produce M outputs. Inputsare denoted by xjk (j = 1, . . . ,n) and the outputs are represented by yik (i = 1, . . . ,m) foreach MFI k (k = 1, . . . ,K). The efficiency of the DMU can be measured as (Coelli 1998):

TEk =m∑

i=1

(uiyik)/ n∑

j=1

(vjxjk)

where yik is the quantity of the i-th output produced by the k-th DMU, xjk is the quantityof j-th input used by the k-th DMU and ui and vj are the output and input weightsrespectively. The DMU maximizes the technical efficiency TEk, subject to:

TEk =m∑

i=1

(uiyik)/ n∑

j=1

(vjxjk

) ≤ 1 where, ui and vj ≥ 0

The above equation indicates that the technical efficiency measure of a DMUcannot exceed 1 and the input and output weights are positive. The weights are selectedin such a way that the DMU maximizes its own technical efficiency. To select optimalweights the following linear programming (Output-oriented) is specified (Coelli 1998):

MaxTEk

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 6: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

404 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Subject tom∑

i=1

(uiyik − xjk + w) ≤ 0 where, k = 1, . . . . . , k

vjxjk −n∑

j=1

(ujxjk) ≥ 0 and ui and vj ≥ 0

The Input-oriented Model (IOM) is used in order to obtain the given level of outputby input minimization. Therefore, the following mathematical programming model isspecified (Coelli 1998):

MinTEk

Subject tom∑

i=1

(uiyik − yjk + w) ≥ 0 where, k = 1, . . . . . , K

xjk −n∑

j=1

(ujxjk

) ≥ 0 and ui and vj ≥ 0

The Input-oriented Model identifies technical efficiency as a proportional reduc-tion in input usage whereas the Output-oriented Model identifies technical efficiencyas a proportional increase in output production. According to Coelli (1998), the Input-and Output-oriented Models estimate exactly the same frontier and therefore, identifythe same set of DMUs as being efficient. It is only the efficiency measures associatedwith the inefficient DMUs that may differ between the two models. Hence, the choiceof an appropriate orientation is not crucial. It will have only minor influences uponthe results obtained (Coelli and Perelman 1996). Many researchers have used only theInput-oriented Model because the DMUs have particular orders to fill e.g. electricitygeneration, total payroll and hence the inputs appear to be the primary decision vari-ables. It is the case in microfinance industry where MFIs managers pay most attentionto the total personnel expenses, the number of credit officers, the cost incurs per bor-rower, etc. We consider each of the sample MFI as a Decision Making Unit and usethe Input-oriented Model of the DEA technical efficiency methodology to analyze theefficiency of MFIs operating in Sub-Saharan Africa.

The above Input-oriented Model shows technical efficiency (TE) under ConstantReturns to Scale (CRS) assumption if w = 0 and it changes into Variable Returns to Scale(VRS) if w is used unconstrained. In the first case, it leads to technical efficiency andin the second case pure technical efficiency (PTE) is estimated. As the CRS assumptionholds good only when all DMUs are operating at an optimum scale while imperfectcompetition, accessibility to fund etc., may not allow all DMUs to operate at optimal scale(Coelli 1998), considering the suggestion by Banker et al. (1984), the CRS DEA technicalefficiency (TECRS) have been decomposed into two components, technical efficiency underVRS assumption (TEVRS) and scale efficiency (SE) where, TECRS = TEVRS x SE. This canbe alternatively stated as TE = PTE x SE (Coelli 1998).

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 7: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 405

The value of scale efficiency does not indicate the nature of the scale efficiencies i.e.,whether the DMU is operating in an area of increasing or decreasing returns to scale,this may be determined by running an additional DEA problem with non-increasingreturns to scale (NIRS) imposed. The nature of the scale efficiencies for a particularDMU can be identified by observing whether the TENIRS score is equal to TEVRS score.While the unequal scores indicate the existence of increasing returns to scale for theparticular DMU, the equal scores indicate decreasing returns to scale. In this paper,the data are analyzed using Data Envelopment Analysis Computer Program (DEAP)Version 2.1.

To analyze the relative technical efficiency of any sort of financial institution, it isnecessary to define the model that will be adopted in order to measure the flow of servicesprovided by the institution. Different philosophical approaches as to what a financialinstitution does and what is meant by efficiency leads to different models (Berger andMester 1997). Two basic models are prevalent in the literature: intermediation and pro-duction approach (Athanassoupoulos 1997). Under the production approach, financialinstitutions are considered as the producers of services for clients. In this case, outputis measured by the number and type of transactions (or by the number of participants)process over a given period of time. Nevertheless, as information on the flow of transac-tions is not usually available, data on stock of the number of deposit or loan accountsor the number of insurance policies outstanding as of a given date are used as a proxyvariable (Fried et al. 2008).

The intermediation approach considers financial institutions primarily as firmsthat intermediate funds between savers (depositors) and investors (borrowers). Thepurpose of intermediation is the transfer of purchasing power from surplus to deficitunits, which improves the allocation of resources in the economy (Gonzalez-Vega 1986).Meanwhile, the degree of financial deepening reflects the volume of funds transferred.Formally, the term financial deepening describes a process of expansion of financialtransactions through markets at a pace faster than the growth of non-financial activities(Gonzalez-Vega 2003). The level of intermediation can be measured as a flow or as astock. The magnitude of the flows depends on the maturity of contracts and the featuresof lending technologies. Stocks are a better measure in this case. Usually, flow data onintermediation services are less readily available. Thus, outputs are commonly assumedto be proportional to the stock of financial value in the accounts, such as the outstandingvalue of loans, deposits or insurance in force as of a given period (Fried et al. 2008).

Some of the operating MFIs in Sub-Saharan Africa do not mobilize deposits fromtheir clients, but all of them issue loans. It has been wrongly assumed that credit is theonly financial service demanded by microenterprises (Adams et al. 1984). Particularlyin Sub-Saharan Africa, loans have been privileged among financial services supply byMFIs, implying that MFIs are incomplete intermediaries and that clients do not haveaccess to deposit facilities as a valuable service. Moreover, available data on both thevalue of the outstanding loan portfolio and the number of outstanding loans offeredby the MFIs allow the paper for the implementation of both the production and theintermediation approaches to measure the MFIs efficiency.

Since it is difficult to define inputs and outputs in the banking industry, Mlimaand Hjalmarsson (2002) defined inputs and outputs in terms of labour, machines andmaterials under the production approach. The number of outstanding loans at the end

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 8: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

406 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

of each financial year is defined as a measure of output in terms of outreach. The largerthe number of outstanding loans with poor people that each MFI reaches, the more itcontributes to the development goals pursue by the authorities. Therefore, we considerthe average number of outstanding loans over the three financial years 2008, 2009, 2010as a measure of output under the production approach.

The most important input in microfinance is labour i.e., the efforts of loan officers.The efficiency with which labour is deployed is influenced by the support services thatloan officers use in their tasks such as gasoline and other transportation costs, computerservices, materials, as well as office supplies, security and utilities (e.g. telephone andelectricity). Therefore, the two most important inputs used by any financial firm toproduce its outputs are labour and services (capital) in support of operations. Labourand services for operations are non-financial resources that MFIs employ to perform theiractivities. Provision of microfinance services is very labour intensive. MFIs usually lendto low-income clients, who have very few assets and lacks security. As a result, MFIsneed to find collateral substitutes such as group joint liability and to spend much effortin screening and monitoring borrowers (Joshi 2005). Because microfinance is character-based lending, frequent visits to the home and business office by the loan officer arerequired while the risk of public embarrassment rather than the risk of legal actionare used to create incentives to repay and enforce contracts. Loan officers are critical inassessing an applicant’s creditworthiness, the latter is identified through frequent visitsto the clients, which help both in developing mutual respect (trust and implicit contracts)between the clients and the loan officers and in reaching the very subjective decisionon the suitability of the amount of loan, term to maturity, frequency of payments, etc(Ledgerwood 1999). In this paper, the total number of employees at the end of the threeconsecutive years 2008, 2009, 2010 is considered as a proxy for labour inputs. The secondinput is ‘Services in support of operations’. These services are more relevant than anyother input because in order to produce loans in a particular period, the MFIs do notexactly need to own a building or a computer. Rather, they need the services of a locale,computation and office supplies. Also, loan officers need support to their operations,especially human capital support such as credit bureau reports, appraisals or securityservices may be provided by firms outside the institution. These services offer the staffthe logistic support needs to accomplish their tasks and increase the productivity of theirefforts. In this paper, ‘total personnel expenses’ of MFIs is used as a proxy for Servicesin support of operations.

As credit remains the most important financial service that MFIs provide to theircustomers, loan or portfolio outstanding during the period of study can be an indicatorfor the level of outreach under the intermediation approach. In this paper, three yearaverage portfolio outstanding for financial years ending 2008, 2009 and 2010 has beentaken as output variable. Further, Norman and Stocker (1991) reveal that the primaryinputs required to produce loans are labour and expenditure. They consider two inputsin the intermediation approach viz., the number of credit officers involve in the MFIs asa proxy for labour and the cost per borrower as a proxy for expenditure.

Microrate and Bank (2003) define credit officers as personnel whose main activityis the direct management of a portion of the loan portfolio. Credit officers are thosehaving direct relationship with the clients. They identify clients, screen them and givefollow-up and monitoring. As credit officers actively engage with the development ofloan portfolio as well as maintenance of its quality, it is taken as a proxy of labour. Cost

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 9: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 407

Table 1 – Summary statistics of inputs and outputs (all variables are averaged over thethree financial years 2008, 2009, and 2010)

Approach Variables Minimum Maximum Mean Standard Deviation

Intermediation Output Gross loan portfolio (US$) 34832.7 871068039 39171267 137500515Inputs Credit officers (Nos.) 2 2636.3 166.7 367.7

Cost per borrower (US$) 10.33 797 186.3548 160.1Production Output Outstanding loans (Nos.) 765.33 625926.3 46170.2 96102.7

Inputs Employees (Nos.) 4.33 4384.7 379.8 760.4Personnel expenses (US$) 69.3 601087.7 35141.4 94976.5

per borrower or Cost per client indicates the average cost of providing an active creditclient (Microrate and Bank 2003). The cost per borrower in this paper indicates theoperating expenses i.e., expenses related to operations including all personnel expenses,depreciation, amortisation and administrative expense per active borrower. Table 1reports the summary statistics of outputs and inputs consider under both productionand intermediation approaches.

The mean of gross loan portfolio is US$39,171,267, the mean of credit officers is167 and the mean of cost per borrower is US$186 under the intermediation approach.Under the production approach, the mean of number of outstanding loans, number ofemployees and personnel expenses are 46,170, 380 and US$35,141, respectively. Thestandard deviation of all variables is found to be high, indicating that the variation ofthe amount of gross loan portfolio, number of credit officers, cost per borrower and thenumber of outstanding loans, employees and personnel expenses from the mean is high.This is reflected from the large difference between the minimum and maximum valuesof all variables.

(d) Tobit analysis

There are many factors that can affect the efficiency of MFIs in performing mi-crofinance activities. In this paper, an attempt is made to analyze the determinants ofefficiency of MFIs. Given that DEA efficiency scores fall between the interval 0 and 1,the dependent variable is a limited dependent variable. It is inferred from the previ-ous studies that computation of a censored regression specifically a Tobit can handlethe characteristics of the distribution of efficiency measures and thus, provide resultsthat can guide policies to improve efficiency. The Tobit regression is suggested as anappropriate multivariate statistical model in the second step in order to consider thecharacteristics of the distribution of efficiency measure (Grosskopf 1996).

In the Tobit regression, there is an asymmetry between observations having posi-tive dependent variable and those with negative dependent variable. The standard Tobitcan be defined as follows:

Yt = α + βXt + ut if Yt > 0 or ut> − α − βXt

Yt = 0 if Yt ≤ 0 or ut ≤ −α − βXt

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 10: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

408 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

where:Yt = Efficiency scores of the t-th MFIXt = Vector of Determinants of Efficiency of the t-th MFI

The basic assumption behind the Tobit is that there exists an index function Yt =α + β Xt + ut for each MFI being studied. For Yt ≤ 0, the value of the dependent variableis set to be zero, while for Yt > 0, the value of the dependent variable is set to be It (Itvarying between zero and one). Under the assumption of normal distribution of ut withmean zero and variance σ 2, the standard normal random variable is denoted by Z =ut/σ . The probability density function of the standard normal variable Z is denoted byf(z) and its cumulative density function by F(z) i.e., P[Z ≤ z]. Hence, the joint probabilitydensity function for the MFIs with positive efficiency scores is given by the followingexpression:

P1 =i=m∏i=1

f[

Yi − α − βXiσ

]

Where � denotes the product and m is the number of MFIs in the sub-sample for whichefficiency scores of MFIs are positive.

In the second sub-sample of MFIs of size n for which the efficiency scores arenegative i.e., Yt = 0, the joint probability density function for the random variableut ≤ −α − β Xt is:

P2 =j=n∏j=1

F[−α − βXj

σ

]

The Maximum Likelihood for the entire sample of MFIs is therefore given by:

L = P1P2

which is nonlinear in α and β due to potential unbiasedness. The procedure for obtainingestimates of α and β is to maximize L with respect to its parameters.

With the efficiency of MFIs estimated through Data Envelopment Analysis, weuse Tobit regression to analyze the factors that influence the efficiency of MFIs. Thedependent variable i.e., efficiency scores are right censored at their maximum valuebecause all efficient MFIs are considered equal. The Tobit regression is performed twotimes, one for the efficiency scores obtained from the frontier built with the intermedia-tion approach and another one for the efficiency scores obtained from the frontier builtwith the production approach.

The determinants of efficiency of MFIs may include characteristics of the operatingenvironment and characteristics of the manager such as human capital endowments(Fried et al. 2008). In this paper, care has been taken to group the variables expected toinfluence the efficiency of MFIs under three wide categories viz., governance, presenceand outreach and financial management and performance.

Rock et al. (1998) define governance of MFI as the process by which the board ofdirectors through management provide guidance to an institution to meet its missionand to protect its assets without diluting the quality. According to Leon (2001), gover-nance refers to those rules and agreements set by the owners of a firm with internaland external agents, in order to meet their primary economic goals. Specific governance

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 11: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 409

rules emerge from the three main types of legal status found in the sample of MFIs:Anonymous Stocks Corporation (S.A.), Civil Association (A.C.) and Civil Society (S.C.).The S.A. is governed by an assembly of shareholders who vote according to the value oftheir shares. Thus, voting power is correlated with the shareholder’s stake in the cor-poration, which creates compatible incentives in pursuing efficiency. The shareholdersappoint one or more managers, who implement the day-to-day direction of the corpora-tion. Their authority can be removed at any time, which creates compatible incentive fortheir behaviour. When there are two or more managers, they will set up a managementboard. To operate legally, half of the members shall attend the meetings of the board andthe resolutions will be valid when the majority endorses them. In case of a tie, the pres-ident of the board will break it. The A.C. is regulated by the civil code of each country inSub-Saharan Africa. In most countries, the direction of these institutions is bestowed onthe general assembly, usually, the rule of one person one vote is adopted, which makes iteasier for a small share of the membership to take over the direction of the association.In this case, the probability that particular interests prevail over the broader pursuit ofefficiency is higher. The management of an S.C. is conferred to one or more members.In case the management is not limited to some of the members, every member willhave the right to lead and manage the common business and decisions will be takenbased on the majority of votes. In this paper, a dummy variable indicating whether theMFI is a for-profit (S.A. or S.C.) or not-for-profit (A.C.) organization is included as aproxy for governance. Since the MFIs are concerned with the livelihood promotion bymeeting the unmet credit demand of the economically challenged section of the society,borrower per staff or credit officer productivity is considered as a proxy for the coverage.It is expected that borrower per staff has a positive effect on the MFIs overall effi-ciency. The degree of asset quality is captured by portfolio at risk 30 days of individualMFI.

Under the second category of presence and outreach, are considered the age ofmicrofinance operation and size of the MFIs. According to Hartarska et al. (2006), MFIsbecome more efficient over time. Thus, we consider the years of operation of each MFIfrom the year of its legal establishment until 31 December 2010. It is expected that asthe age of MFIs increases, their overall efficiency also increases. To capture the effectof the size of MFIs, the average value of assets for financial years ending 2008, 2009and 2010 is considered. It is hypothesized that large MFIs with more number of yearsin the sector perform better than newer entrants and with relatively small size. Size ofthe MFIs is measured in terms of total assets.

Financial and economic viability of the MFIs are important to keep the microfi-nance operation viable from an organizational point of view (Hulme and Mosley 1996).Financial indicators such as Debt-Equity Ratio, Return on Assets, Return on Equity,Operational Self-Sufficiency, Yield on Gross Portfolio and Financial Expenses per Assetare included in the study under the category of financial management and performance.It is expected that higher Debt-Equity Ratio reduces MFIs’ efficiency as it reflects thehigher financial dependence on outside sources of borrowing. On the other hand, boththe Return on Assets and Return on Equity are expected to have a positive effect onthe efficiency of MFIs. Operational self-sufficiency represents the ability of MFIs tomeet their operating costs from their income. It indicates whether enough revenue isearned to cover the MFIs’ cost which includes financial expenses, operating expensesas well as impairment loss. The study hypothesized that an increasing Operational

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 12: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

410 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Table 2 – Summary statistics of variables used in Tobit regression (all variables areaveraged over the three financial years 2008, 2009, and 2010)

Variable Minimum Maximum Mean Standard Deviation Skewness Kurtosis

Non-Financial Services (D1) 0 1 0.5 0.5 −0.02 −2.1Not-For-Profit MFIs (D2) 0 1 0.66 0.48 −0.68 −1.59Borrower per Staff (BPS) 22 704.6 143.4 126.4 2.52 7.56Age of MFIs (AGE) 3 45 12.75 8.10 1.51 3.34Portfolio at Risk 30 days (PAR) 0 0.52 0.08 0.08 3.12 13.3Financial Expenses / Assets (FEPA) 0 0.17 0.06 0.03 1.83 4.96Total Asset 5.01 9.12 6.94 0.81 0.24 0.53Debt Equity Ratio (DER) −1.70 1.84 0.37 0.52 −0.54 3.68Return on Asset (ROA) −0.39 0.21 −0.014 0.097 −1.73 4.80Return on Equity (ROE) −354.43 1.32 −5.1 42.36 −8.36 69.98Operational Self Sufficiency (OSS) −0.47 0.39 0.006 0.13 −0.75 3.04Yield on Gross Portfolio (YOGP) 0.11 1.05 0.42 0.21 0.55 −0.31

Self-Sufficiency (OSS) indicates the financial viability of the MFIs. Yield on Gross Port-folio and Financial Expenses per Asset are taken as proxies for interest rates chargedby the MFIs to its borrowers and the cost of borrowing for the respective MFIs. As thetotal assets primarily include loan portfolio, Financial Expenses per Asset explains thefinancial expenses incur to build a unit of asset. As a higher speed i.e., higher the differ-ence between the income and expense of the MFIs, greater the possibility of becomingfinancially self-sustainable, it is hypothesized that while Yield on Gross Portfolio is pos-itively related to efficiency, Financial Expenses per Asset holds a negative relation withefficiency.

Because output has been defined exclusively in terms of financial services (loans),resources distracted into the production of other types of services will reduce effi-ciency. These services will contribute to the efficiency from this perspective, only ifthey strengthen loan recovery. Most MFIs are specialized in only providing financialservices, while others also provide non-financial services such as social intermediation,enterprise development, health, nutrition, education and literacy training (Ledgerwood1999). In order to identify whether the MFIs provide non-financial services or not, adummy variable is employed. This variable has the value of 1 if either the MFIs providetraining to microenterprises or education workshops or health care training or its helpin the development of commercial networks or its provide insurance services and 0 oth-erwise. It is hypothesized that provision of non-financial services has a negative effecton the MFIs overall efficiency.

Table 2 reports the descriptive statistics of the variables employ in the Tobitregression. The results reveal that the standard deviation of all variables except Ageof MFIs, Borrowers per staff and Return on Equity are found to be low, indicating thatthe variation of these variables from the mean is small. This is reflected from the smalldifference between the minimum and maximum values of these variables.

The skewness and kurtosis of Non-Financial Services, Not-For-Profit MFIs, TotalAsset and Yield on Gross Portfolio are found to be around zero, which mean thesevariables are normally distributed. However, kurtosis of Borrower per Staff, Age ofMFIs, Financial Expenses/ Assets, Debt Equity Ratio, Return on Asset, OperationalSelf Sufficiency are far from zero, but since the skewness are found to be around zero

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 13: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 411

therefore these can be categorized as normally distributed. The skewness and kurtosisof Portfolio at Risk 30 days and Return on Equity indicate that both variables are notnormally distributed. It implies that Portfolio at Risk and Return on Equity will affectthe results of the Tobit regression.

3 Results and discussion

(a) Technical efficiency of MFIs under the intermediation approach

The DEA technical efficiency of 70 MFIs are calculated by considering the in-termediation approach under both the Constant Returns to Scale (CRS) and VariableReturns to Scale (VRS) technology assumption. Table 3 reports the empirical resultsof technical efficiency, pure technical efficiency and scale efficiency of each MFI basedon the DEA methodology. Looking at the table, we observe that only two MFIs are onthe technical efficiency frontier when constant returns to scale is assumed, while sixMFIs are on the efficiency frontier when variable returns to scale is assumed. UnderVRS assumption, we find higher efficiency scores for all MFIs and more efficient MFIsbecause the MFIs operating efficiently under CRS are accompanied by new efficientMFIs that might be operated under either increasing returns to scale or decreasingreturns to scale. The MFIs that remain technically efficient under both CRS and VRSassumption are Alliance de Credit et d’Epargne pour la Production (ACEP Senegal) andEquity Bank. ACEP Senegal is a Credit union/ Co-operative located in Senegal which hasan outreach of 30,503 active borrowers with a gross loan portfolio of US$60.8 million.ACEP Senegal is fully engaged in microfinance related activities and provides exclu-sively financial services. Equity Bank is a bank located in Kenya and has an outreachof 524,902 active borrowers with a gross loan portfolio of US$925 million. Equity Bankis fully engaged in microfinance related activities and provides non-financial services aswell.

The mean technical efficiency, pure technical efficiency and scale efficiency arefound to be 20.9 per cent, 39.5 per cent and 52.4 per cent, respectively. It impliesthat MFIs under CRS assumption could decrease 79.1 per cent of their inputs withoutaffecting the existing output level i.e., gross loan portfolio. The mean VRS efficiency scoreimplies that 60.5 per cent of inputs could be decreased without affecting the existinggross loan portfolio. The mean scale efficiency score indicates that MFIs are operatingbelow the optimal scale.

Examining the returns to scale, we observe that the majority of the MFIs experi-ence economies of scale i.e., 89 per cent of MFIs are operating at the phase of increasingreturns to scale. It indicates that MFIs managers still have to utilize the inputs ef-ficiently to increase their overall efficiency in providing and distributing loans to theSub-Saharan African population.

(b) Technical efficiency of MFIs under the production approach

The DEA technical efficiency of 70 MFIs are calculated by considering the produc-tion approach under both the Constant Returns to Scale (CRS) and Variable Returns to

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 14: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

412 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Table 3 – Single output-two inputs DEA technical efficiency of MFIs under intermediationapproach

Technical Pure Technical Scale ReturnsS.I. No. Name of MFI Efficiency Efficiency Efficiency to Scale

1 ACEP Cameroon 0.663 0.704 0.942 irs2 ACEP Senegal 1.000 1.000 1.000 –3 AfricaWorks 0.052 0.218 0.241 irs4 Akiba 0.340 0.35 0.972 irs5 AlidA C© 0.123 0.401 0.306 irs6 AMfB 0.082 0.117 0.705 irs7 APED 0.031 0.171 0.183 irs8 BIMAS 0.074 0.203 0.362 irs9 BOM 0.037 0.100 0.372 irs10 BRAC – UGA 0.062 0.280 0.220 irs11 Capitec Bank 0.932 1.000 0.932 drs12 CAPPED 0.129 0.248 0.519 irs13 CAURIE Micro Finance 0.222 0.393 0.565 irs14 CDS 0.803 0.953 0.843 irs15 CEDEF 0.012 1.000 0.012 irs16 CFF 0.071 0.91 0.078 irs17 CMMB 0.148 0.588 0.252 irs18 CMS 0.839 0.893 0.940 drs19 CRG 0.084 0.228 0.370 irs20 CUMO 0.018 0.305 0.060 irs21 Faulu – KEN 0.093 0.149 0.624 irs22 FDM 0.042 0.228 0.184 irs23 FINCA – DRC 0.077 0.110 0.697 irs24 FINCA – MWI 0.035 0.098 0.352 irs25 FINCA – UGA 0.082 0.116 0.703 irs26 FINCA – ZMB 0.031 0.105 0.296 irs27 FUCEC Togo 0.424 0.446 0.951 irs28 CVECA Kita 0.024 0.405 0.059 irs29 ECLOF-KEN 0.075 0.169 0.445 irs30 Hluvuku 0.140 0.362 0.388 irs31 GGEM Micro S. 0.036 0.557 0.064 irs32 K-Rep 0.557 0.573 0.973 irs33 Kafo Jiginew 0.381 0.429 0.889 irs34 KixiCredito 0.094 0.122 0.770 irs35 KSF 0.176 1.000 0.176 irs36 KWFT 0.363 0.465 0.781 irs37 Equity Bank 1.000 1.000 1.000 –38 ID-Ghana 0.027 0.296 0.090 irs39 LAPO 0.140 0.303 0.460 irs40 Madfa SACCO 0.025 1.000 0.025 irs41 MED-Net 0.050 0.159 0.317 irs42 MECREF 0.274 0.521 0.525 irs43 MEC FEPRODES 0.201 0.66 0.305 irs44 MGPCC DEKAWOWO 0.300 0.941 0.319 irs45 Micro Africa 0.126 0.223 0.564 irs46 MicroCred – MDG 0.127 0.174 0.730 irs47 MUL 0.220 0.492 0.448 irs48 NovoBanco – MOZ 0.166 0.182 0.911 irs49 OI – TZA 0.022 0.107 0.204 irs50 OIBM 0.267 0.298 0.897 irs

Continued

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 15: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 413

Table 3 – Continued

Technical Pure Technical Scale ReturnsS.I. No. Name of MFI Efficiency Efficiency Efficiency to Scale

51 OISL 0.089 0.116 0.770 irs52 Opportunity Finance 0.247 0.353 0.698 irs53 Otiv Diana 0.085 0.215 0.398 irs54 PAIDEK 0.099 0.366 0.272 irs55 PAPME 0.183 0.225 0.814 irs56 ProCredit – GHA 0.166 0.176 0.946 irs57 RCPB 0.721 0.738 0.976 irs58 SEAP 0.202 0.733 0.276 irs59 SAT 0.155 0.203 0.764 irs60 SEDA 0.052 0.155 0.332 irs61 Soro Yiriwaso 0.112 0.555 0.201 irs62 Reliance 0.097 0.159 0.612 irs63 RML 0.106 0.218 0.488 irs64 TIAVO 0.04 0.095 0.423 irs65 U-Trust 0.143 0.186 0.769 irs66 U-IMCEC 0.271 0.383 0.708 irs67 UCEC/MK 0.149 0.249 0.597 irs68 UNION DES COOPECs 0.084 0.313 0.269 irs69 UOB 0.073 0.137 0.53 irs70 WAGES 0.268 0.326 0.822 irs

Mean 0.209 0.395 0.524

Scale (VRS) technology assumption. Table 4 reports the empirical results of technicalefficiency, pure technical efficiency and scale efficiency of each MFI based on the DEAmethodology. Looking at the table, we observe that only one MFI is on the technicalefficiency frontier when constant returns to scale is assumed, while eight MFIs are onthe efficiency frontier when variable returns to scale is assumed. Under VRS assump-tion, we find higher efficiency scores for all MFIs and more efficient MFIs because theMFI operating efficiently under CRS is accompanied by new efficient MFIs that mightbe operated under either increasing returns to scale or decreasing returns to scale. TheMFI that remains technically efficient under both CRS and VRS assumption is KrabanSupport Foundation (KSF) which is an NGO located in Ghana and has an outreach of8,017 active borrowers with a gross loan portfolio of US$1.2 million. KSF is fully engagedin microfinance related activities and provides non-financial services as well.

The mean technical efficiency, pure technical efficiency and scale efficiency arefound to be 23.2 per cent, 36.8 per cent and 69.6 per cent, respectively. It implies thatMFIs under CRS assumption could decrease 76.8 per cent of inputs without affecting theexisting output level i.e., number of outstanding loans. The mean VRS efficiency scoreimplies that 63.2 per cent of inputs could be decreased without affecting the existingnumber of outstanding loans. The mean scale efficiency score indicates that MFIs areoperating below the optimal scale.

Examining the returns to scale, we observe that majority of MFIs experiencediseconomies of scale i.e., 61 per cent of MFIs are operating at the phase of decreasingreturns to scale. It indicates that MFIs could improve their efficiency by decreasingtheir inputs to improve their overall efficiency in providing and distributing loans to theSub-Saharan African population.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 16: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

414 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Table 4 – Single output-two inputs DEA technical efficiency of MFIs under productionapproach

Technical Pure Technical Scale ReturnsS.I. No. Name of MFI Efficiency Efficiency Efficiency to Scale

1 ACEP Cameroon 0.106 0.110 0.968 irs2 ACEP Senegal 0.215 0.298 0.721 drs3 AfricaWorks 0.104 0.133 0.785 irs4 Akiba 0.082 0.103 0.797 drs5 AlidA C© 0.313 0.366 0.856 drs6 AMfB 0.069 0.070 0.991 drs7 APED 0.259 0.308 0.841 drs8 BIMAS 0.261 0.306 0.853 drs9 BOM 0.075 0.076 0.991 irs10 BRAC – UGA 0.188 0.47 0.400 drs11 Capitec Bank 0.121 0.301 0.403 drs12 CAPPED 0.069 0.101 0.679 irs13 CAURIE Micro Finance 0.652 0.904 0.722 drs14 CDS 0.317 0.445 0.713 drs15 CEDEF 0.516 1.000 0.516 irs16 CFF 0.594 0.789 0.753 irs17 CMMB 0.177 0.294 0.604 irs18 CMS 0.196 0.414 0.474 drs19 CRG 0.474 0.944 0.503 drs20 CUMO 0.389 0.546 0.711 drs21 Faulu – KEN 0.194 0.446 0.434 drs22 FDM 0.130 0.203 0.641 irs23 FINCA – DRC 0.206 0.296 0.697 drs24 FINCA – MWI 0.161 0.201 0.798 drs25 FINCA – UGA 0.154 0.220 0.701 drs26 FINCA – ZMB 0.116 0.134 0.867 drs27 FUCEC Togo 0.134 0.291 0.462 drs28 CVECA Kita 0.918 0.931 0.986 irs29 ECLOF-KEN 0.219 0.277 0.790 drs30 Hluvuku 0.155 0.198 0.782 irs31 GGEM Microfinance 0.262 0.367 0.712 irs32 K-Rep 0.196 0.356 0.550 drs33 Kafo Jiginew 0.161 0.301 0.534 drs34 KixiCredito 0.081 0.084 0.972 drs35 KSF 1.000 1.000 1.000 –36 KWFT 0.403 1.000 0.403 drs37 Equity Bank 0.219 1.000 0.219 drs38 ID-Ghana 0.136 0.178 0.766 irs39 LAPO 0.199 1.000 0.199 drs40 Madfa SACCO 0.417 1.000 0.417 irs41 MED-Net 0.143 0.146 0.981 irs42 MECREF 0.177 0.261 0.679 irs43 MEC FEPRODES 0.243 0.312 0.78 irs44 MGPCC DEKAWOWO 0.111 0.205 0.543 irs45 Micro Africa 0.160 0.208 0.767 irs46 MicroCred – MDG 0.093 0.112 0.833 drs47 MUL 0.185 0.318 0.583 irs48 NovoBanco – MOZ 0.064 0.087 0.735 drs49 OI – TZA 0.088 0.110 0.800 irs50 OIBM 0.181 0.257 0.702 drs

Continued

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 17: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 415

Table 4 – Continued

Technical Pure Technical Scale ReturnsS.I. No. Name of MFI Efficiency Efficiency Efficiency to Scale

51 OISL 0.120 0.172 0.698 drs52 Opportunity Finance 0.160 0.202 0.793 irs53 Otiv Diana 0.060 0.090 0.673 irs54 PAIDEK 0.469 0.531 0.883 drs55 PAPME 0.092 0.109 0.840 drs56 ProCredit – GHA 0.044 0.055 0.797 drs57 RCPB 0.171 0.399 0.427 drs58 SEAP 0.568 1.000 0.568 drs59 SAT 0.464 0.918 0.506 drs60 SEDA 0.176 0.212 0.830 drs61 Soro Yiriwaso 0.696 1.000 0.696 drs62 Reliance 0.035 0.052 0.676 irs63 RML 0.108 0.186 0.580 irs64 TIAVO 0.064 0.064 0.995 –65 U-Trust 0.107 0.140 0.765 drs66 U-IMCEC 0.230 0.280 0.820 drs67 UCEC/MK 0.145 0.239 0.607 drs68 UNION DES COOPECs 0.102 0.203 0.502 irs69 UOB 0.236 0.329 0.718 drs70 WAGES 0.076 0.105 0.719 drs

Mean 0.232 0.368 0.696

(c) Determinants of efficiency of MFIs under intermediation approach

Table 5 reports the results of determinants of the efficiency of MFIs by consideringthe intermediation approach. Looking at the table, we observe that dummy variableindicating MFIs providing non-financial services, dummy variable indicating not-for-profit MFIs, Portfolio at Risk 30 days, Total Asset, Return on Asset, Operational SelfSufficiency and Yield on Gross Portfolio are found to be statistically significant on theefficiency of MFIs.

The dummy variable indicating MFIs providing non-financial services is found tobe significant at 5 per cent level and has negative influence on the technical efficiency ofMFIs as expected. It is clearly evident from the fact that MFIs which are providing non-financial services besides financial services reduce their technical efficiency. Portfolio atRisk 30 days is found to be significant at 10 per cent level and has positive influenceon the technical efficiency of MFIs against the expectation. We observe that an increaseof 1 per cent in the proportion of Portfolio at Risk 30 days would increase the technicalefficiency of MFIs by 0.50 per cent. Total Asset is found to be significant at 1 per centlevel and has positive influence on the technical efficiency of MFIs. We observe that anincrease of 1 per cent in the proportion of Total Asset would increase the technical effi-ciency of MFIs by 0.22 per cent. Return on Asset is found to be significant at 10 per centlevel and has negative influence on the technical efficiency of MFIs against the expecta-tion that a higher return on asset would lead towards the long run sustainability of themicrocredit operation by reinvesting the surplus into the operation. We observe that anincrease of 1 per cent in the proportion of Return on Asset would decrease the technicalefficiency of MFIs by 1.10 per cent. Operational Self Sufficiency is found to be significant

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 18: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

416 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Table 5 – Determinants of efficiency of MFIs under intermediation approach

Technical Efficiency Pure Technical Efficiency Scale Efficiency

Variable Coefficient t-ratio Coefficient t-ratio Coefficient t-ratio

Constant − 1.15∗ − 5.11 0.41 1.02 − 1.19∗ − 5.93Non-Financial Services − 0.08∗∗ − 2.18 − 0.09 − 1.40 − 0.04 − 1.09Not-For-Profit MFIs 0.04 0.85 0.06 0.79 − 0.10∗∗ − 2.52Borrower per Staff 0.00 − 0.35 0.00 1.97 0.00 − 3.77Age of MFIs 0.00 − 0.37 0.00 − 0.51 0.00 − 0.75Portfolio at Risk 30 days 0.5∗∗∗ 1.73 0.42 0.78 0.31 1.13Financial Expenses /Assets 0.42 0.58 0.36 0.28 − 0.28 − 0.36Total Asset 0.22∗ 7.10 0.02 0.44 0.29∗ 10.65Debt Equity Ratio − 0.07 − 1.55 − 0.05 − 0.57 − 0.01 − 0.35Return on Asset − 1.1∗∗∗ − 1.77 − 0.02 − 0.02 0.35 0.61Return on Equity 0.00 − 0.63 0.00 − 0.67 0.00 − 0.49Operational Self Sufficiency 1.00∗∗ 2.08 0.38 0.45 − 0.11 − 0.25Yield on Gross Portfolio − 0.36∗ − 3.04 − 0.59∗ − 2.82 − 0.24∗∗ − 2.33Sigma 0.14 0.25 0.13Log likelihood 32.89 −10.34 43.38LR chi2(12) 76.14 27.11 122.61Prob > chi2 0.00 0.01 0.00

Notes: Size of the MFI is measured by natural logarithm of total assets; we have also taken the natural logarithmof Debt Equity Ratio and Operational Self Sufficiency as suggested by the literature.∗Denote Statistical Significant at 1 per cent level.∗∗Denote Statistical Significant at 5 per cent level.∗∗∗Denote Statistical Significant at 10 percent level.

at 5 per cent level and has positive influence on the technical efficiency of MFIs. Weobserve that an increase of 1 per cent in the proportion of Operational Self Sufficiencywould increase the technical efficiency of MFIs by 1 per cent. Yield on Gross Portfoliois found to be significant at 1 per cent level and has negative influence on the technicalefficiency of MFIs against the expectation that a higher difference between the incomeand expense of an MFI lead towards the long run financial self-sustainability of themicrocredit operation. We observe that an increase of 1 per cent in the proportion ofYield on Gross Portfolio would decrease the technical efficiency of MFIs by 0.36 per cent.

Yield on Gross Portfolio is found to be statistically significant at 1 per cent level andhas negative influence on the Pure Technical Efficiency of MFIs against the expectation.We observe that an increase of 1 per cent in the proportion of Yield on Gross Portfoliowould decrease the pure technical efficiency of MFIs by 0.6 per cent.

The dummy variable indicating not-for-profit MFIs is found to be significant at5 per cent level and has negative influence on scale efficiency of MFIs against theexpectation. It reveals that not-for-profit MFIs reduce their scale efficiency. Total Assetis found to be significant at 1 per cent level and has positive influence on the scaleefficiency of MFIs. We observe that an increase of 1 per cent in the proportion of TotalAsset would increase the scale efficiency of MFIs by 0.29 per cent. Yield on Gross Portfoliois found to be significant at 5 per cent level and has negative influence on the scaleefficiency of MFIs against the expectation. We observe that an increase of 1 per cent inthe proportion of Yield on Gross Portfolio would decrease the scale efficiency of MFIs by0.24 per cent.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 19: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 417

Table 6 – Determinants of efficiency of MFIs under production approach

Technical Efficiency Pure Technical Efficiency Scale Efficiency

Variable Coefficient t-ratio Coefficient t-ratio Coefficient t-ratio

Constant 0.34∗ 4.51 0.09 0.26 1.19∗ 4.8Non-Financial Services 0.00 − 0.35 0.02 0.33 − 0.03 − 0.72Not-For-Profit − 0.01 − 0.42 0.02 0.32 − 0.09∗∗∗ − 1.69Borrower per Staff 0.00 24.41 0.00 7.54 0.00 1.75Age of MFIs 0.00 0.88 0.01 1.33 0.00 − 1.48Portfolio at Risk 30 days − 0.07 − 0.69 − 0.28 − 0.61 0.34 0.1Financial Expenses /Assets − 0.34 − 1.4 − 0.08 − 0.07 − 1.51∗∗∗ − 1.89Total Asset − 0.04∗ − 3.98 − 0.01 − 0.23 − 0.06∗∗∗ − 1.90Debt Equity Ratio 0.02 1.30 0.04 0.63 0.02 0.42Return on Asset 0.51∗∗ 2.45 1.12 1.21 − 0.84 − 1.23Return on Equity 0.00 − 0.15 0.00 − 0.2 0.00 1.04Operational Self Sufficiency − 0.35∗∗ − 2.21 − 0.61 − 0.86 0.42 0.80Yield on Gross Portfolio − 0.05 − 1.37 0.03 0.17 0.07 0.56Sigma 0.05 0.21 0.16Log likelihood 111.8 −1.02 28.65LR chi2(12) 199.28 62.16 22.30Prob > chi2 0.00 0.00 0.034

Notes: Size of the MFI is measured by natural logarithm of total assets; we have also taken the natural logarithmof Debt Equity Ratio and Operational Self Sufficiency as suggested by the literature.∗Denote Statistical Significant at 1 per cent level.∗∗Denote Statistical Significant at 5 per cent level.∗∗∗Denote Statistical Significant at 10 percent level.

(d) Determinants of efficiency of MFIs under production approach

Table 6 reports the results of determinants of the efficiency of MFIs by consider-ing the production approach. We observe that dummy variable indicating not-for-profitMFIs, Financial Expenses/Assets, Total Asset, Return on Asset and Operational SelfSufficiency are found to be statistically significant on the efficiency of MFIs.

Total Asset is found to be significant at 1 per cent level and has negative influenceon the technical efficiency of MFIs. We observe that an increase of 1 per cent in theproportion of Total Asset would decrease the technical efficiency of MFIs by 0.04 per cent.Return on Asset is found to be significant at 5 per cent level and has positive influenceon the technical efficiency of MFIs as expected. We observe that an increase of 1 per centin the proportion of Return on Asset would increase the technical efficiency of MFIsby 0.51 per cent. Operational Self Sufficiency is found to be significant at 5 per centlevel and has negative influence on the technical efficiency of MFIs. We observe that anincrease of 1 per cent in the proportion of Operational Self Sufficiency would decreasethe technical efficiency of MFIs by 0.35 per cent.

The dummy variable indicating not-for-profit MFIs is found to be significant at10 per cent level and has negative influence on scale efficiency of MFIs. We observe thatnot-for-profit MFIs reduce their scale efficiency. Financial Expenses/Assets is found tobe significant at 10 per cent level and has negative influence on the scale efficiencyof MFIs as expected. We observe that an increase of 1 per cent in the proportion ofFinancial Expenses/Assets would decrease the scale efficiency of MFIs by 1.51 per cent.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 20: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

418 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

Total Asset is found to be significant at 10 per cent level and has negative influence onthe scale efficiency of MFIs. We observe that an increase of 1 per cent in the proportionof Total Asset would decrease the scale efficiency of MFIs by 0.06 per cent.

4 Conclusion

This paper uses a large panel data set covering 70 MFIs in 25 Sub-Saharan Africancountries to analyze the efficiency of MFIs and to test for the efficiency drivers of theseMFIs. This is important, given that MFIs have to operate efficiently to fulfil their dualmission of serving the poor and also being financially sustainable, but the question ishow the MFIs can operate more efficiently? By studying the factors determining theefficiency of MFIs and their effect, decision makers can acquire a better knowledge ofwhere to put in effort to increase the MFIs efficiency, hence it makes this paper relevant.As far as we know, no rigorous empirical study has been devoted to this issue in Sub-Saharan Africa. To analyze the efficiency of MFIs, Data Envelopment Analysis (DEA)is employed using both intermediation and production approaches. The determinants ofthe efficiency of MFIs are analyzed using Tobit Regression.

Comparing the results of both approaches, it is evident that the MFIs in Sub-Saharan Africa are not efficient in meeting the goals of either providing microfinancerelated services to their clients or intermediating funds between borrowers and deposi-tors. But, most of the MFIs under intermediation approach are operating at the phaseof increasing returns to scale. Hence, these MFIs could improve their overall efficiencyby increasing their inputs in terms of labour or capital. However, under production ap-proach, most of the MFIs are operating at the stage of decreasing returns to scale. Hence,these MFIs could improve their overall efficiency by decreasing the level of inputs. Thefact that MFIs are operating at the stage of decreasing returns to scale under the produc-tion approach and at the stage of increasing returns to scale under the intermediationapproach obviously suggests that they are not efficient in meeting either goal. Yet, sincemany MFIs in Sub-Saharan Africa pursue those two goals simultaneously, it might wellbe that they are operating at their optimal scale. Advocating more efficiency for theseMFIs necessarily means reducing the weight of one goal and increasing the weight ofthe other. That is why we recommend to the regulatory authority in Sub-Saharan Africato define the structure of the market they want to pursue i.e., intermediation goal orproduction goal or both and make it clear to the MFIs.

The analysis on determinants of efficiency of MFIs under intermediation approachreveals that the dummy variable indicating MFIs providing non-financial services,dummy variable indicating not-for-profit MFIs, Portfolio at Risk 30 days, Total Asset,Return on Asset, Operational Self Sufficiency and Yield on Gross Portfolio are identifiedas the most significant factors determining the overall efficiency of MFIs.

The analysis on determinants of efficiency of MFIs under production approachreveals that Total Asset, Return on Asset, Operational Self Sufficiency, the dummyvariable indicating not-for-profit MFIs and Financial Expenses/Assets are identified asthe most significant factors determining the overall efficiency of MFIs.

The results have important implications, for both the microfinance industry andpolicy makers. Since most of the MFIs in Sub-Saharan Africa are found not to be effi-cient in their operations, the managerial skills of the employees/credit officers may be

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 21: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 419

improved to increase their output in terms of gross loan portfolio or the number of loansprovided to their clients in order to increase their overall efficiency.

Since ACEP Senegal and Equity Bank are found to be efficient under intermedi-ation approach, MFIs willing to perform the role of financial intermediary may followthese two MFIs models to become efficient by reaching medium level of outreach, bor-rowers and depositors, providing credit access to at least 40 per cent of women, havingless that 10 per cent of gross loan portfolio at risk and being regulated in the countrythey operate.

Since KSF is found to be efficient under production approach, MFIs willing toperform the role of provider of microfinance related services may follow its model tobecome efficient by reaching at least smaller outreach, borrowers and depositors, pro-viding credit access to at least 80 per cent of women, having less that 10 per cent of grossloan portfolio at risk and being regulated in the country they operate.

On a wide scale, future research is needed to analyze in greater details the effi-ciency drivers of MFIs in Sub-Saharan Africa. As the present paper deals with averagedata of three consecutive years for a single output and two inputs to measure theefficiency of MFIs, there is further scope for research on comparing the efficiency ofindividual MFIs as well as the group as a whole on a longer panel data set which wouldshow the direction of efficiency of MFIs in the Sub-Saharan African context. The non-availability of the borrower’s side secondary data as a utilization of the borrowing fund,microeconomic activities undertaken, average duration of loans and loan instalmentsmay have limited the Tobit regression results. Further, the MFIs may be sub-dividedamong their legal status as a Bank, Credit union/ Co-operative, Non-GovernmentalOrganization and Non-Banking Financial Institution.

REFERENCES

ADAMS D. W., GRAHAM D. H. and VON PISCHKE J. D., 1984, Undermining RuralDevelopment with Cheap Credit. Boulder, CO: Westview Press.

AFRICAN DEVELOPMENT BANK, 2006, Microfinance Policy and Strategy for the BankGroup. Operation Policies and Review Department. ADB.

AFRICAN UNION, 2009, Advancing the African Microfinance Sector, Follow-up to theAfrican Microfinance Roadmap. Paper presented at the Extraordinary Conference ofAfrican Ministers of Economy and Finance, CAMEF, Addis Ababa, Ethiopia, Decem-ber 2009.

ATHANASSOPOULOS A., 1997, ‘Service quality and operating efficiency synergies formanagement control in the provision of financial services: evidence from Greek bankbranches’, European Journal of Operational Research, 98(2), 300–313.

BANKER R. D., CHARNES A. and COOPER W. W., 1984, ‘Some models for estimat-ing technical and scale inefficiencies in Data Envelopment Analysis’ ManagementScience, 30, 1078–1092.

BERGER A. N. and MESTER L. J., 1997, ‘Inside the black box: what explains differencesin the efficiencies of financial institutions?’ Journal of Banking and Finance, 21, 895–947.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 22: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

420 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

COELLI T. J. and PERELMAN S., 1996, ‘A comparison of parametric and non-parametric distance functions: with application to European railways’, CREPP Dis-cussion Paper, University of Liege, Liege.

COELLI T., 1998, ‘A guide to DEAP Version 2.1: a Data Envelopment Analysis (com-puter) program’, CEPA Working Paper 96/08, University of New England, Australia.

FARRELL M. J., 1957, ‘The measurement of productive efficiency’. Journal of the RoyalStatistical Society, A CXX, Part 3: 253–290.

FREIXAS X. and ROCHET J. C., 2008, Microeconomics of Banking, Cambridge, MA:MIT Press, 2nd ed.

FRIED H. O., KNOX LOVELL C. A. and SCHMIDT S. S., 2008, The Measurement ofProductive Efficiency and Productivity Change, New York: Oxford University Press,Doi: Oxford Scholarship Online. 3–91.

GONZALEZ-VEGA C., 1986, Mercados Financieros y Desarrollo. Santo Domingo, R. D.Centro de Estudios Monetarios y Bancarios.

GONZALEZ-VEGA C., 2003, ‘Deepening rural financial markets: macroeconomic, policyand political dimensions’, Paving the Way Forward for Rural Finance: An Interna-tional Conference on Best Practices, Washington, D.C.

GROSSKOPF S. 1996, ‘Statistical inference and nonparametric efficiency analysis, inFried Lovel C. A. K. and Schmidt S. S., 2008, The Measurement of Productive Efficiencyand Application. New York, Oxford University Press.

HARTARSKA V., CAUDILL S. B. and GROPPER D. M., 2006, ‘The cost structure ofmicrofinance institutions in Eastern Europe and Central Asia. Williamson DavidsonInstitute Working Paper 809.

HOLLIS A. and SWEETMAN A., 1998, ‘Microcredit: what can we learn from the past?’World Development, 26(10).

HULME D. and MOSLEY P., 1996, Finance against Poverty. London: Routledge.

JOSHI M., 2005, ‘Access to credit by hawkers: what is missing? Theory and evidence fromIndia’, Unpublished Ph.D Thesis submitted to The Ohio State University, Columbus,Ohio, USA.

LEDGERWOOD J., 1999, Microfinance Handbook: An Institutional and Financial Per-spective. The World Bank/Sustainable Banking with the Poor (SBP), Washington,D.C.

LEON J. V., 2001, ‘Cost frontier analysis of efficiency: an application to the Peruvianmunicipal banks, unpublished Ph.D Thesis submitted to The Ohio State University,Columbus, Ohio, USA.

LITTLEFIELD E., MURDUCH J. and HASHEMI S., 2003, ‘Is microfinance an effectivestrategy to reach the millennium development goals?, CGAP Retrieved, November2008.

MERSLAND R. and STRØM R., 2008, ‘Microfinance mission drift’, Working Paper,Kristiansand 2008.

MICRORATE and BANK I. A. D., 2003, Performance Indicators for Microfinance Insti-tutions. Technical Guide, 3rd ed., Washington, DC.

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 23: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

SUB-SAHARAN AFRICA’S MICROFINANCE INSTITUTIONS 421

MIX and CGAP, 2012, 2011 Sub-Saharan Africa Regional Snapshot. February 2012.

MLIMA P. A. and HJALMARSSON L., 2002, ‘Measurement of inputs and outputs inthe banking industry’, Tanzanet Journal, 3(1), 12–22.

NORMAN M. and STOCKER B., 1991, Data Envelopment Analysis: the Assessment ofPerformance. John Wiley and Sons, New York, 9–16.

OUEDRAOGO A. and GENTIL D., 2008, La Microfinance en Afrique de l’Ouest – His-toires et Innovations. KARTHALA Editions, Juin 2008.

ROCK R., OTERO M. and SALTZMAN S., 1998, ‘Principles and practices of microfinancegovernance’, ACCION International, Retrieved 2 December 2008.

WORLD BANK, 2003, Microfinance in India: Issues, Challenges and Policy Options. TheWorld Bank, Washington D.C.

Appendix: List of the Selected MFIs for the Study

S.I. No. Name of MFIs Country

1 ACEP Cameroon Cameroon2 CAPPED Congo, Republic of the3 CDS Cameroon4 CRG Guinea5 FINCA – DRC Congo, Democratic Republic of the6 PAIDEK Congo, Democratic Republic of the7 UCEC/MK Chad8 Akiba Tanzania9 BIMAS Kenya10 BRAC – UGA Uganda11 Faulu – KEN Kenya12 FINCA – UGA Uganda13 ECLOF-KEN Kenya14 K-Rep Kenya15 KWFT Kenya16 Equity Bank Kenya17 Madfa SACCO Uganda18 MED-Net Uganda19 Micro Africa Kenya20 MUL Uganda21 OI – TZA Tanzania22 SEDA Tanzania23 RML Rwanda24 U-Trust Uganda25 UNION DES COOPECs U. Rwanda26 UOB Rwanda27 AfricaWorks Mozambique28 BOM Mozambique29 Capitec Bank South Africa30 CUMO Malawi31 FDM Mozambique32 FINCA – MWI Malawi33 FINCA – ZMB Zambia34 Hluvuku Mozambique35 KixiCredito Angola

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC

Page 24: MEASURING THE EFFICIENCY OF SUB-SAHARAN AFRICA'S MICROFINANCE INSTITUTIONS AND ITS DRIVERS

422 KEMONOU RICHARD SENAMI SEGUN AND M. ANJUGAM

S.I. No. Name of MFIs Country

36 MicroCred – MDG Madagascar37 NovoBanco – MOZ Mozambique38 OIBM Malawi39 Opportunity Finance South Africa40 Otiv Diana Madagascar41 TIAVO Madagascar42 ACEP Senegal Senegal43 AlidA C© Benin44 AMfB Nigeria45 APED Ghana46 CAURIE Micro Finance Senegal47 CEDEF Ghana48 CFF Ghana49 CMMB Benin50 CMS Senegal51 FUCEC Togo Togo52 CVECA Kita/BafoulabA C© Mali53 GGEM Microfinance Services Ltd. Sierra Leone54 Kafo Jiginew Mali55 KSF Ghana56 ID-Ghana Ghana57 LAPO Nigeria58 MECREF Niger59 MEC FEPRODES Senegal60 MGPCC DEKAWOWO Togo61 OISL Ghana62 PAPME Benin63 ProCredit – GHA Ghana64 RCPB Burkina Faso65 SEAP Nigeria66 SAT Ghana67 Soro Yiriwaso Mali68 Reliance Gambia, The69 U-IMCEC Senegal70 WAGES Togo

© 2013 The AuthorsAnnals of Public and Cooperative Economics © 2013 CIRIEC