Financial viability and social utility of microfinance institutions in Cameroon: a multi criteria and DEA combined analysis. By Joseph NZONGANG1, Faculty of Economics and Management, University of Dschang, Cameroon Abstract This paper measures the social performance and the financial performance of the institutions of microfinance within a network. It provides a practical tool to the network's manager for decision making. The DEA approach is used jointly with the PROMETHEEGAIA approach. The results show that the IMF located in Cameroon member of the MC² network fulfil the objective of providing financial and social objectives with a waste of resources. Furthermore, the existence of a linkage between the efficiency of the IMF and the level of their assets is highlighted. Keywords: Microfinance –Efficiency- MC2 – DEA- PROMETHEE GAIA. 1 E-mail : jnzongang@yahoo.fr Microfinance like many other instruments of development provokes debates on its capacity to alleviate poverty. It is defined as the provision (supplying) of financial services to the fringe of population excluded from the formal system. According to many professionals, microfinance should go together with other actions in order to be able to effectively improve the living conditions of the beneficiaries. Some institutions of microfinance (IMF) positioning themselves by the side of the poorer have promoted the idea of “complete” microfinance services (Hickson, 1999). According to this approach, the IMF should provide to their beneficiaries at the same time financial services and non financial services. These services aimed at strengthening the beneficiaries’ capacity to develop activities that produce durable incomes. Other practitioners have challenged this approach. Several arguments have been put forward, like the lack of competence or the absence of mandate of the IMF concerning the provision of non financial services. Such services can divert them from their main objective which is the provision of financial services and by so doing give contradictory messages to the clients, in particular if these non financial services are free. Since the non financial services lead to some important costs which are added to those of the microfinance operations, the non financial services can also affect the capacity of the IMF to get to financial self-sufficiency. Microfinance has the particularity of trying to accomplish a social mission while functioning as a classical financial institution. The consequence of this dualism is that a financial success tends to lead to a social failure and vice-versa. This gives rise to a debate as to the determination of the aspect which would have priority. The divergence of views becomes more pronounced due to the fact that the different participants of the microfinancial sector have different objectives. The donors are interested in the social impact of the institutions that they finance while the investors care about the financial durability (sustainability). However, whatever may be the orientation, many questions remain. Have the resources put at the disposal of the institutions of microfinance (IMF) been used in an optimal manner or have they been wasted? Has the institution been managed in an efficient way? In developing countries, microfinance is presented as a mean of poverty alleviation. The objective of IMFs in these countries is to reach the best possible performance (efficiency); this could be achieved when they are able to reconcile two requirements: social performance (SP) by reducing poverty and financial performance (FP) by ensuring a sustainable profitability. However, these two requirements provoke a debate between two conflicting schools of thought: the school of welfare highlights the social requirement of the targeting of the poor and their living conditions; the institutionalist approach stands up for the economic requirement of profitability and of the institutions’ viability. The IMFs of the “Mutuelles Communautaires de Croissance “ network (MC2) in Cameroon provide a pertinent illustration of this debate and the analysis of their activities permits to answer the following question: is there any compatibility between social performance and financial performance? The search for the complementarity between SP and FP becomes inevitable inasmuch as it conditions the durability (sustainability) of the institution, gives confidence to the potential partners and permits to build fame. However, the question of SP versus FP in the IMF has not attracted much attention in the literature (Navajas et al., 1998; Paxton, 2000; Woller and Shreiner, 2002; Gutiérrez-Niéto et al., 2005 ; Polanco, 2005 ; Cul, Demirguc-Kunt and Morduch, 2006 ; Cornée, 2006). These works measure the degree of efficiency of the structures studied. But, the issue of the efficiency does not only concern the investors and the donors. It is equally of interest for the managers, the operators of the IMF on a daily basis. In the Cameroonian context, most of the IMF try to gather together into networks2. To improve the performance of a network, the manager needs an instrument panel in order to improve his/her management. It is precisely this problem that we try to solve in the scope of this study. Our study tries to reconcile the SP and the FP of the IMF gathered within a network and to provide some concrete elements of assistance to the managers’ decision taking. For that, we are going to combine two methodologies (DEA3 and PROMETHEE-GAIA4). In order to achieve this objective, the work has been structured into three parts. The first part reviews the related literature on the different contrasting points of view of the two schools of thought on the social performance and the financial performance of the microfinance institutions. The second part brings out the concept of the efficiency frontier as well as the different steps of the empirical analysis. Finally, the third part presents the results and proposes some recommendations in order to ameliorate the performance of the IMFs studied. 1. Social performance versus financial performance of the IMF Microfinance is a mean of poverty alleviation in developing countries through the financing of the activities productive of incomes for poor households. However, the best way of helping the poor to have access to the financial services contrasts the welfare approach with the approach of the institutionalists. Though they all share the objective of poverty alleviation, these two approaches put microfinance at the crossroads. 1.1. The welfare approach This school of thought has been identified as a school of poverty measurement (Asselin and Anyck, 2000). According to this trend, an individual is considered as poor when he/she is situated below a minimum economic well-being. The advocates of this school of thought are based on the theory of social responsibility vis-à-vis the clientele in order to meet their expectations (Caroll, 1979). They assess the performance of an IMF in the point of view of the client through the social scope « Outreach » (Lafourcade; Isern; Mwangi and Brown, 2005) and the assessment impact (Harper; Hill; Horn; Salib and Walen, 2005). It targets the poorest people whose incomes are at 50% inferior to the poverty threshold ($1 per day) and aims at improving their living conditions. It is essentially made up of solidarity institutions – NGOs or cooperatives – which consider microfinance as an important means of the fight against poverty for the poor (Hamed, 2004). Even though it insists on the rational management of the resources and does not exclude the fact that the IMF could run a profitable activity at the end of 5 to 12 years, this school of thought advocates a supply of financial services at relatively low rates of interest and a large resort to grants (Olszyma-Marzys, 2006). This welfare approach has however created the rates of settlement below 50% as well as very high functioning costs leading to the failure and to the disappearance of some IMF though it is based on the logic of grants and the dependence of the beneficiaries. As a matter of fact, these IMFs are faced with obstacles (the viability and durability problem) which are limits to their development and to their capacity to contribute to the development of the person they help and to a poor performance. In this manner, the welfare approach has been the 2 3 4 In Cameroon, more than 75% of the IMF are affiliated to a network (Creusor, 2006). Data Envelopment Analysis. PROMETHEE = Preference Ranking Organization Method for Enrichment Evaluation; GAIA = Geometrical Analysis for Interactive Assistance. subject of many criticisms on the grounds of its subjectivity, its costs and the methodological difficulties that it brings about (De Briey, 2005). A revival of the financial and economic thought was necessary in order to re-assess the conditions of success of the IMF where the interest shown by the economists and the practitioners in the study of the IMF efficiency in poverty alleviation paves more and more the way for an efficiency analysis in financial and accounting terms. 1.2. The institutionalist approach Supported by international organisations such as the United Nations and the World Bank, the institutionalist approach came to birth (Woller, Dunford and Woodworth, 1999). It is based on the theory of contracts which considers that the non-fulfilment of contracts can lead to some opportunist behaviours of those who need loans (Ghatak and Guinane, 1999). Its protagonists consider that the unique manner to reach the great majority of the poor without access to the financial services is to increase the activity of the microfinance through its integration into the formal financial system. In this way, they try to put the IMF within the logic of market by insisting on the will of the setting up of a sustainable microfinance system as well as the will of rendering loans within the reach of the majority of the population (De Briey, 2005). Each IMF should aim at financial sustainability by maximising its efficiency and its productivity. They have conceived a set of “best banking practices” in order to increase the efficiency of the management systems, whose adoption is an essential step to get at financial self-sufficiency on the industrial scale and to have access to financial market and to reach the maximum number of poor clients (Morduch, 2000). Consequently, the sustainability necessarily passes through the access to the financial autonomy which is a criterion that better fulfills the social mission (Cornée, 2006). The measurement of the social impact passes through a proxy, profitability, while they appreciate success through the selfsufficiency of the programme (Otero and Ryhne, 1994). The interest for self-sufficiency emerged from the recognition of the scarcity of funds. As a matter of fact, the institutionalists believe in the necessity of large scale intervention which demands financial resources above what the bankers can provide. They are afraid of the changing nature of these national and international sponsors because an IMF which wants to operate in the long run, by becoming structurally dependent on grants, would risk being a short-lived programme. But the only way to get the financial resources which they need is to resort to private sources (saving, commercial debts, capital outlays and venture capital). This approach has not only been adopted by most of the literature edited in the scope of microfinance but it can also be observed at the moment through two great trends. We find the upgrading process where certain regulated IMFs are now created in countries offering a regulation process of the specialised institutions in microfinance on the one hand. These IMF are NGOs which give rise to regulated financial institutions having the status of limited companies and which come clearly within the scope of profitability the logic (De Briey, 2005). We find the downgrading process where certain traditional commercial banks which are looking for new markets have most recently entered into the microfinance sector. These banks have not only been convinced of the potentialities of micro credit, but they also have easy access to the funds and to better tools of marketing. They can directly grant loans to the micro entrepreneurs or can acquire shares in the IMF on the other hand. However, this institutionalist approach has been the object of many criticisms. At the level of the targeted population, it has as a preferred clientele the micro entrepreneurs very close to the poverty line, geographically concentrated, having highly profitable activities with short production cycle. Moreover, it requires from its clients very high interest rates in order to ensure financial autonomy. Finally, this approach recommends the attainment of the institutional and financial viability for long term microfinance programmes. The welfare and the institutionalist approaches have been the object of few criticisms. The first approach is confronted with the problem of viability and that of sustainability induced by the grants, the low rates of reimbursement and the increase of the functioning costs; the second approach has as preferred clientele the micro entrepreneurs very close to the poverty line on which are applied very high interest rates in order to ensure the financial autonomy of the IMF. This contrast leads to trade-offs between the two paradigms. 2. The methodology of the efficiency frontiers The complementarity between SP and FP can be analysed through the notion of efficiency. Efficiency is a relative notion which permits to estimate the performance as well as the development potential of a company and to place it in comparison with its competitors. It implies that the resources of a company should be used at best according to its objectives. For an IMF, it is a matter of getting a social and financial result without wasting the given micro financings. The efficiency methods, deriving from the production theory, permit to characterise the performances of the firms (Kopp and Diewert, 1982; Kumbhakar, 1988). The companies are considered as technically efficient when they can supply their products from a minimum level of resources without changing their combination of factors (Atkinson et Cornwell, 1994). Two methods can be used to calculate the efficiency scores (Berger and Humphrey, 1997). The parametric methods specify the structural relations between the variables with the help of a function which characterises the production function and resorts to econometrics to estimate it. The nonparametric methods, to which belongs the Data Envelopment Analysis (DEA) approach, do not a priori set hypotheses on the relations which exist between the variables but directly construct a frontier from the observations thanks to linear programming. The measure of efficiency is obtained by comparing the results of a unit of production to those it would have obtained if it adopted the practices of the best which are located at the efficiency frontier. Its efficiency is measured by calculating the distance which separates it from the frontier. This distance is expressed with the help of an efficiency score. In the present work, we have adopted the DEA nonparametric approach for it permits to “reveal” the inquired information (the efficiency) from the observed data by specifying a few hypotheses on the structure of the efficiency frontier. The DEA and the PROMETHEE-GAIA methods are first presented. The data used as well as the selected variables are then described. 2.1. DEA and PROMETHEE-GAIA : tools of estimation of efficiency and of aid to decision 2.1.1. The DEA method The DEA method constructs an efficiency frontier which envelopes the observations of the sample. In this way, the information located on this frontier corresponds to the 100% efficient entities. The entities located outside this frontier are considered as not being totally efficient. The score of efficiency corresponds to the measure of the gap existing between the observations and the frontier. When we want to reduce wasting of resources, this score gives us some information on the possibility of reduction of all the inputs without any modification of the initial combination of factors and of the level of outputs supplied. The DEA method is adapted for the study of the efficiency of small samples of observations. It analyses each unit separately and measures its efficiency in comparison with the set of units of the sample. Its important adjustment to the available data represents an indisputable asset in the domain of analysis of the IMF efficiency. However, this advantage is also a limit for it is impossible to draw conclusions as to the absolute values of the efficiency analysed. On the other hand, it is not easy to know the contribution of each variable in the realisation of efficiency. The DEA approach does not estimate the coefficients to describe the function characterising the frontier of efficiency5. In the absence of such information, a certain degree of precision is lost; this can reduce the possibilities of operational use of the results. It is to circumvent this limit that we combine the PROMETHEE-GAIA method with the DEA approach. 2.1.2. The PROMETHEE-GAIA method PROMETHEE-GAIA is a method of aid to the multi criteria decision (Brans and Mareschal, 1994, 2002). It permits to treat multi criteria alternatives in view of an arrangement of variables and observations. It is applied according to a process which comprises four steps: - comparisons of pairs of alternatives from variables in order to determine an index of aggregate preferences; - calculations of streams of out classification; - the arrangement of the alternatives from the streams of out classification; - the visualisation of the alternatives in a plane called GAIA. 2.1.3. Justification of the combination: DEA - PROMETHEE-GAIA In the DEA model, the researchers decide a priori on the model to use in selecting the variables which will be used in measuring the efficiency of the sample of observations (Pedraja, 1999). Serrano et al. (2001, p.3) draw attention on the fact that a variable can be included in the model when it does not contribute to the calculation of efficiency. Inversely, it is possible that a variable not included in the model could be very important for the analysis. Serrano et al. suggest circumventing this problem by calculating some DEA scores of all the possible combinations of the variables. The use of the Principal Component Analysis (PCA) permits afterwards to analyse the results (Gutiérrez et al., 2007 ; Cornée, 2006). We want to bring in new concrete elements of help to the decision making of the manager of the IMF network in order that those evolving below their potentialities should be identified and pushed towards the efficiency. We can not then make a combination of different DEA models since each model proposes its own corrective measures. The man ager would find himself/herself with as many propositions as models, without any selection criterion for retaining the most adapted to the IMF network that he/she manages. The solution is then to stick at one model and to analyse the results obtained in using a method of help to the multi criteria decision. 5 To get specific scores per input and output, two approaches are proposed in the literature on the measure of the efficiency. The first one is based on some ex post calculations combining the efficiency scores and the gap variables of the linear programme (Torgensen et al., 1996). The second one uses a non radical efficiency measure in the sense where a score is obtained for each output. The total measure describing the firm efficiency is then the average of the set of the specific scores calculated (Färe and Lovell, 1978). One of the advantages of the combination of the DEA and the PROMETHEE-GAIA models is to allow making some comparisons between the different IMF and the variables chosen. This permits us to know what IMF is “good” or “bad” on such or such input or output. 2.2. Data and variables selected 2.2.1. Sources of data Launched in 1992, the activity of the “MC2” aims at endowing the village communities with rural development micro banks created and managed by their members, in the respect of the sociocultural values. The “MC2” propose to the populations adapted solutions in order to overcome their problems of access to the financial services and permit them to improve their living conditions in a sustainable manner. It is a question of an endogenous approach of development which permits the underprivileged populations to create wealth. As any microfinance institution, the “MC2” have a twofold objective. An economic objective which concerns their financial viability and a social objective which is that of reaching the poorest levels of the populations by financing small and micro activities. The “MC2” are institutions of first category6 sponsored by Afriland First Bank which plays at the same time the role of a commercial bank and provides the technical assistance in partnership with the NGO “ADAF” (Appropriate Development for Africa Foundation). On the 31st December 2007, there were 66 operational “MC2”. On this same date, the network deals directly with 82 280 individuals, 9 844 groups and associations and indirectly with about 574 480 persons. The total amount of deposits is 11, 87 billions of CFA Francs, the capital raised in the “MC2” amounts to 2, 36 billions of CFA Francs. A total amount of 25, 43 billions of CFA Francs has been granted in a form of loans since 1992 (ADAF, 2008). The flexibility of the “MC2” as well as its adaptability to each sociocultural context permits its fast introduction in the different milieu which experience poverty problems and which the populations have chosen to become members in order to emerge from poverty. On the basis of the data at our disposal, we have retained the 20 “MC2”of the network having more than 10 years of experience. This selection guarantees a financial sustainability of the IMF as well as an effective social settling. The encoded data are stemming from the financial statements of the 2006 financial year. 2.2.2. Selection of the variables The role of a financial institution can be described according to the production and the intermediation models. In the first case, the institution uses some factors (capital, labour) to proceed to the financial transactions (saving and credit). On the other hand, if we consider the role of intermediation, the financial institution collects the deposits and grants loans, the deposits are considered as inputs while the loans are considered as outputs. To carry out our study, we have retained three inputs (the total asset, the number of the employees, the exploitation expenses) and three outputs (the outstanding loans, the exploitation revenues, the number of women membership). The choice of the first two inputs (personnel and assets) has been done according to the production approach. The variables “exploitation expenses” and “exploitation revenues” synthesize the information on the operational and the financial self-sufficiency of the “MC2”. The outstanding loans measure 6 The regulation n°01/02/CEMAC/UMAC/COBAC relative to the exercise of the microfinance activities in Central Africa states in its section 7 that the IMF of the first category treat only with their members. the accessibility of the IMF. Furthermore, to represent the interaction between the financial and social objective of the “MC2”, the number of women members is introduced as an output of the DEA model. Women are the targets having priority of the IMF (Montalieu, 2002). 3. The results and the recommendations We are now going to interpret the results stemming from the combination of the DEA and PROMETHEE-GAIA models and we shall give some recommendations in order to improve the efficiency of the “MC2”. 3.1. Interpretation of the results of the DEA model7 Table1: Efficiency and pair of reference of the MC2 MC² Babouantou Bafia_Makenen e Bafou Bafoussam_Rur al Baham Bambalang_Ba munka Bamendjou Bandja Bandjoun Bangangté Bangouu Batoufam Bayangam Doumbouo Manjoo Melonggg Muyuka Ngaoundal Penka_Michel Batié Source: author Bab BMk Baf B_r Bah B_B Bam Bja Bjn Bté Bng Bat Bay Dou Man Mel Muy Nga Pen TPD Efficiency 100.00% 100.00% 85.83% 100.00% 93.28% 60.79% 100.00% 100.00% 100.00% 100.00% 100.00% 87.81% 100.00% 100.00% 82.80% 97.94% 100.00% 78.42% 94.31% 62.00% The ‘’MC2’’ of reference Muy Dou Bjn Bjn Bay Bng Bab Muy Bam Bjn Bng Bab Bjn Bté Muy B_r Dou Bab Bam Dou Muy Bté Muy Dou Bng Bté Muy BMk Dou Table 1 above represents the efficiency scores obtained by every one of the “MC2” retained in the sample of this study. A score of 100% means that the “MC2” is efficient. It uses its resources in an optimal manner to supply the financial outputs (loans and products) and the social output (membership of women). These “MC2” define the efficiency frontier. The “MC2” having a score inferior to 100% presents a technical inefficiency. They waste productive resources in the process of supplying the outputs retained in the model. They can refer to the practices of the “MC2” of reference which are associated to them in order to become more efficient in the management of their inputs. Proximity of the efficiency scores emerges from table 1. This can be the result of two elements. These “MC2” have a certain mastery of their activity due to their length of service 7 For the calculation of the scores of the efficiency, we have used the “DEA Frontier” software of Joe Zhu. and they also have a neighbouring geographical localisation (14 are located in the West Region), and this leads to some similarities in the habits and customs of the local populations. Table 2: Excesses of inputs, lacks of outputs of every “MC2” Excesses of inputs Lacks of outputs Act Em Ch Cr Pro Fem Bab 0 0 0 0 0 0 BMk 0 0 0 0 0 0 Baf 0 1 2 591 923 3 744 268 0 0 B_r 0 0 0 0 0 0 Bah 53 908 741 0 0 0 0 0 B_B 0 2 0 0 0 0 Bam 0 0 0 0 0 0 Bja 0 0 0 0 0 0 Bjn 0 0 0 0 0 0 Bté 0 0 0 0 0 0 Bng 0 0 0 0 0 0 Bat 0 3 0 16 617 727 0 1 Bay 0 0 0 0 0 0 Dou 0 0 0 0 0 0 Man 0 1 0 24 117 892 0 0 Mel 0 0 2 623 586 0 0 0 Muy 0 0 0 0 0 0 Nga 0 3 2 890 797 0 0 19 Pen 76 245 493 0 0 975 565 0 0 TPD 13 185 386 1 0 0 0 0 Act : stands for assets Ch : stands for charges Fem : stands for women Cr : stands for credits Em : stands for employees Pro : stands for products Source: author MC2 Table 2 illustrates for each “MC2” the existing possibilities of reduction on certain inputs or of possible increase of certain outputs after the correction of the efficiency. These values indicate the situations where the “MC2” is not efficient in the sense of Pareto. There still remains some unused resources or inversely some outputs not produced which would permit to the concerned “MC2” to go farther in the improvement of its performance8. As an example, the “MC2” of Bafou has an efficiency score of 85, 83%. In comparison with its “MC2” of reference, it can produce the same level of outputs by reducing its consumption of inputs by 14, 17%. However, in addition to the proportional reduction of all its inputs, it is still possible to reduce some of them. In this way, the costs can still be more reduced by 2 591 923 CFA Francs. Similarly, the number of employees can be reduced by one unit. Originally, Bafou had 5 employees. After the correction by the score of efficiency, the number of employees can be reduced by one person. After taking into account the information given by the model and relative to the surpluses of inputs of each “MC2”, there subsists one employee too many. Hence the optimal number of the employees is 3. As to the 8 On a technical plane, these values describe the non saturation of the constraints of the linear programme from which the DEA efficiency frontier is constructed. outputs, this “MC2” has at its disposal an outstanding loans that 3 744 268 CFA Francs less than what it could manage with the means it has. In this manner, thanks to the information relative to the efficiency scores and to the different surpluses of inputs and lacks of outputs, the manager of the network has an idea of the objective to achieve in order to make every “MC2” more efficient. The table 3 below gives the target values to achieve for each IMF in order to render them completely efficient. These values can be used by the manager as a reference instrument in the management of each “MC2” Table 3: Aims in term of inputs and of outputs of each ‘’MC²’’ Aims in term of inputs Act Em Ch Bab 375 502 748 5 20 200 287 BMk 1 071 805 827 8 77 096 702 Baf 117 390 471 3 10 423 811 B_r 457 158 397 6 32 415 904 Bah 635 248 630 6 39 892 770 B_B 38 006 503 1 4 064 586 Bam 349 085 862 5 29 197 031 Bja 259 717 413 5 15 099 086 Bjn 600 347 781 6 38 692 016 Bté 795 791 217 5 53 871 251 Bng 105 977 137 4 12 981 132 Bat 145 677 594 2 8 460 042 Bay 443 629 053 5 23 634 979 Dou 272 508 929 7 24 229 070 Man 162 837 309 3 12 950 687 Mel 688 230 016 7 47 185 232 Muy 511 028 672 8 44 753 666 Nga 56 383 497 2 6 112 931 Pen 336 993 985 6 22 902 824 TPD 68 315 719 4 11 729 846 Source :author MC2 Aims in term of outputs Cr Pro 133 464 101 34 455 607 356 571 255 80 284 044 47 483 443 8 604 928 170 369 043 42 233 800 184 317 763 44 901 834 21 533 690 2 322 565 138 583 833 33 025 642 105 878 847 15 004 364 290 395 356 45 123 366 142 538 493 63 318 936 64 942 423 5 504 700 53 459 124 13 597 319 132 493 808 22 864 692 109 147 135 19 180 835 63 547 199 14 777 842 278 229 980 54 974 556 228 199 671 53 265 644 30 773 402 4 117 056 117 597 415 27 856 699 70 172 568 10 321 838 Fem 333 869 295 543 617 65 467 319 571 664 209 127 520 728 235 662 425 66 509 252 The DEA provides elements for identifying the inefficient and efficient “MC2”, the target values for the construction of dashboards for every “MC2” and the identification of the IMF of reference in the management of inputs. However, it reveals itself insufficient in a managerial perspective of the IMF. As a matter of fact, it is very important to know the behaviour of every “MC2” with respect to the different variables of the model in order to facilitate the taking of corrective measures. The combination of the DEA and the PROMETHEE-GAIA approaches will permit to characterise the relation existing between the “MC2” and the different variables at our disposal. 3.2. The combination of the DEA and the PROMETHEE-GAIA approaches The most significant element in the DEA analysis is the relation which exists between the “MC2” and the variables of the DEA model. This information can be obtained in analysing the corresponding GAIA plane. 3.2.1. The results of the PROMETHEE-GAIA approach9 The links which exist between the “MC2” and the variables of the DEA model are synthesized in the GAIA plane represented in Figure 1 below. The different “MC2” are represented by the triangles and the different variables by the squares. The horizontal axis contrasts the information relative to the inputs (green squares) and to the outputs (red squares) of the DEA model. The vertical axis contrasts the social objective with the financial objective (the women membership) of the considered IMF. Figure 1 : Typology of the ‘’MC²’’ in the GAIA plane Source :author On figure 1 above, the criteria of PROMETHEE decision is represented by the circles. They permit to identify four groups of “MC2”. The “MC2” located in the each group are similar. In this manner, we observe that the groups 1 and 3 are made up of very different IMF as to their behaviour relative to the different variables of the model. The group 2 is presented as a median position between the two extremes that are the groups 1 and 3. On the other hand, the “MC2” of Muyuka and Doumbouo isolate themselves from groups 1 and 2 and define a larger group (group 4) comprising the ellipses 1 and 2 and contrasting with group 3. The interpretation of the variables is based on the measure of the angle made by vectors which link them to the origin. In this way, the figure 1 shows that there exists an angle – which is almost 90 degrees – between the variables “women membership” and the other outputs of the model (loans and exploitation revenues). This means that the social output is independent from the financial outputs. The fact that an “MC2” enjoys some important exploitation revenues does not influence the number of women membership. On the other hand, there exists an angle with about 180 degrees between the variables exploitation revenues (outputs) and the assets (inputs). This means that they are contrasting. If one “MC2” is successful on one of these two variables, then it is inefficient on the other one. The group 3 is next to the criterion “active”. It gathers the “MC2” which have modest assets. It contrasts with group 4, which “MC2” have more important assets. For example, the “MC2” 9 For the calculations of the PROMETHEE-GAIA method, we have used the “Decision Lab” software. of Bafia-Makenene (group 4) has outstanding loans of more than one billion while the “MC2” of Bambalang-Bamunka (group 3) has a level of assets which is less than 63 millions of CFA Francs. The group 4 represents institutions which have important assets. However, most of these “MC2” have average performance in the maximisation of the exploitation revenues and the number of women membership, that is to say, in the achievement of the social and financial objective of the “MC2”. In this group, the “MC2” which have the highest assets are in group 1 while those having average sizes are in group 2. Within the group 4, two “MC2” distinguish themselves from the other. Muyuka is more interested in the financial objectives than Doumbouo which favours the social objective. Generally, the positioning of these “MC2” which have been studied does not result from a decision of their companies’ managers, but rather from the facts. Consequently, the knowledge of the situation of the IMF with respect of the social and financial objectives can help the network managers identify the measures to be taken in order to ameliorate the IMF results about the neglected aspect. 3.2.2. Combined interpretation DEA- GAIA Table 4 below presents the efficiency scores deriving from the DEA approach when we keep as “MC2” of reference Doumbouo and Muyuka. All the “MC2” are then evaluated with respect to these two IMF in order to provide an efficiency score which situates each institution of micro credits in relation to two objectives (social and financial) recognised from their activity. The results are presented by ranking the “MC2” in a decreasing order of efficiency. Table 4 : Scores et benchmarks of the MC² of the reduced DEA model MC2 DEA scores Benchmarks Doumbouo 100.00% - - Muyuka 100.00% Bamendjou 99.48% Dou Bafoussam-rural 95.57% Dou Muy Babouantou 90.41% Dou Muy Bafou 85.83% Dou Muy Manjo 81.26% Dou Muy Batoufam 80.35% Dou Muy Melong 79.95% Dou Muy Bangangté 79.41% Dou Muy Bandjoun 77.52% Dou Muy Bafia-makenene 75.14% Dou Muy Muy Bangou 73.82% Dou Muy Penka-michel 71.06% Dou Muy Bandja 67.24% Dou Muy Baham 63.70% Dou Muy Ngaoundal 62.03% Dou Muy Bayangam 61.16% Dou Muy Batié 47.13% Dou Muy 47.08% Dou Muy Bambalang-bamunka Source :author Using the scores given in table 4 and the GAIA plane, it is possible to identify some general trends. The most successful institutions have a score which is situated around 90% and belong to group 2; the median group on figure 1. These institutions have assets comprised between 200 and 400 millions of CFA Francs. The group 1 is constituted of institutions whose efficiency varies between 75% and 80%, apart from the “MC2” of Bafoussam rural (score of 95%) and of Baham (score of 63%). The value of the assets of that group amounts to 200 millions. The group 3 is the most disparate. It groups together “MC2” having good scores like Bafou (86%) and Manjo (81%) and “MC2” having the poorest scores: Batié and BambalangBamunka (47%). The value of the assets for this group exceeds 400 millions of CFA Francs. These results show that there exists a link between the level of efficiency of a “MC 2” and the value of its assets. It seems that a “MC2” is “mature” in terms of the management of its resources when it passes the 200 millions CFA Francs. Below this threshold, it falters out and can get all the levels of efficiency. However, when its assets are superior to 400 millions CFA Francs, its efficiency declines; the most important level of resources becoming probably more difficult to manage. The GAIA plane has then permitted us to deepen the understanding of the efficiency scores. It has appeared judicious not to limit ourselves to only the results given by the DEA approach. The exploitation of the GAIA plane has permitted to understand the “MC2” scores. Not understanding the DEA results could lead to some hesitations and experimentations that a manager should not allow. This is even more important if we should run a whole network. 3.3. Recommendations for the optimisation of the management of the “MC2” network The efficiency scores have allowed us rank the IMF according to their level of performance. The target values given by the DEA approach can be used to lead the “MC2” towards a higher level of performance, serving as objective to the “MC2”. Our first recommendation is to carry out the classification of the IMF according to their level of performance in a regular way. The evolution of the scores obtained each year could give an idea on the progress achieved by each IMF and globally by the network and could allow the updating of the objectives to be reached. In the Cameroonian context where there is a great cultural diversity, it is interesting for the network manager to proceed to a deeper analysis. From a more important sample, it could be possible to carry out a similar study per region. As a matter of fact, we have noticed that among the “MC2” used, that of Ngaoundal which is located in Muslim region is the IMF of our sample which produces the lowest social output (women membership). This level of analysis goes beyond the managerial scope to integrate the sociocultural context of the “MC2”. Concerning the characterisation of the “MC2” of reference, we suggest a twofold approach to be able to capitalise the obtained information. At first, it is a question of creating an exchange between the inefficient “MC2” and the persons in charge of “ADAF” in order to identify the weak points of this “MC2” and to agree on the elements coming from the “MC2” of reference which could serve as examples and then work in collaboration with the managers of the efficient “MC2” for the setting up of more efficient management methods. Secondly, it is conceivable to favour exchanges between the efficient “MC2”, the inefficient “MC2” and “ADAF” by organising, for instance, seminars during which the different participants could learn from one another. Conclusion Microfinance is characterised by a consensus on its objective which is poverty alleviation. However, there exists a divergence on the manner to help the poor through the access to financial services. Two trends emerge. The client approach considers microfinance as a means for reducing the poverty of the poorest people. It stresses on the amelioration of the welfare of the targeted populations, that is to say, on the social impact of microfinance. The institutionalist approach considers that the only way to reach the great majority of the poor is to increase microfinance activity through a policy favouring the granting of loans to the masses, the search for the financial autonomy and the integration of the IMF into the formal financial system. The aim of this study was to measure the performance of the IMF of the “MC2” network. As a matter of fact, for the cohesion of the network, their managers have all interest to make sure that the units which constitute it should be efficient in their management in order that the resources of the whole network should not be wasted. The data analysis has allowed getting the efficiency scores for each “MC2”. They constitute a synthetic measure of their relative performance, that is to say, of their capacity to limit wastage in the provision of social and financial services. The DEA approach used in this study has led to the identification of the inefficient and efficient “MC2” as well as to the definition of the target values permitting to the network administrator to define dashboards for each “MC2”. Afterwards, the use of the PROMETHEE-GAIA approach has permitted a better understanding of the behaviour of each IMF with regards to the different inputs and outputs of the DEA model. The additional information can be used by the administrator and the concerned “MC2” to define the process of amelioration of its activity towards a higher social and financial level of performance. References ACCLASSATO D. (2006), « Taux d’intérêt effectif, viabilité financière et réduction de la pauvreté par les institutions de microfinance au Bénin », Document de rechercheLaboratoire d’Economie d’Orléans -UMR CNRS 6221. ADAMS D. W, VON PISCHKE J.D (1992), « Microenterprise Credit Programs: Déjà Vu », World Development, vol 20, n°10, pp. 1463-1470. ATKINSON E.S., CORNWELL C., « Estimation of out put and input Technical efficiency using a flexible functional form and panel data », International Economy Review, n° 35, 1994, pp. 245-255. BASSEM B.S., « Efficiency of Microfinance Institutions: in the Mediterranean: An Application of DEA », Transition Studies Review, volume 15, n° 2, 2008, pp. 343-354. BERGER A.N., HUMPHREY D.B., « Efficiency of financial institutions: international survey and directions for future research », European Journal of Operational Research, volume 98, n° 2, 1997, pp.175–212. BRANS J.P., MARESCHAL B., « How to differentiate hard from soft multicriterial problems », accepted for publication in Discrete Mathematics 1994. BRANS J.P., MARESCHAL B., « Prométhée-GAIA : une méthodologie d'aide à la décision en présence de critères multiples », Editions de l'Université de Bruxelles, Éditions Ellipses, Paris, 2002, 192 pages. CGAP (1998) ; « Stratégies de mobilisation de l’épargne : leçons tirées de l’expérience de quatre institutions », Focus n°13. CGAP (2003) ; « L’impact de la microfinance », Note sur la microfinance, n°13 CGAP (1997) ; « Pérennité financière, ciblage des plus pauvres et impact sur le revenu : quels compromis pour les institutions de microfinancement ? », Focus N°5. CHESTON S. et al., « Comment mesurer la transformation : une évaluation et amélioration de l’impact du microcrédit », Réunion des Assemblées du Sommet du Microcrédit, 24-26 juin. CORNEE S., « Microfinance : entre marché et solidarité. Analyse de la convergence entre performances financières et performances sociales : application de la méthode Data Envolpment Analysis sur 18 institutions de microfinance péruviennes », Mémoire de Master en Sciences de Gestion, IGR-IAE, Université de Rennes 1, 2006, 101 pages. CREUSOT A.C., « L'état des lieux de la microfinance au Cameroun », BIM, n° 09 mai, 2006, 5 pages. CULL R, DEMIRGUC-KUNT A, MORDUSH J. (2006); «Financial performance and outreach: aglobal analysis of leading microbanks », Document de travail de recherche sur les politiques, WPS3827. DE BRIEY V. (2005), « Pleins feux sur la microfinance en 2005 », Regards Economiques, n°28, Mars, pp. 1-14. FARE R., LOVELL C.A.K., « Measuring the technical efficiency of production », Journal of Economic Theory, volume 19, n° 1, 1978, pp. 150-162. GUERIN I. (1999) : « Microfinance dans les pays du sud : quelle compatibilité entre solidarité et pérennité ? », Revue d’Economie Financière, pp145-163. GUTIERREZ-NIETO B., SERRANO-CINCA C., MAR MOLINERO C., « Microfinance institutions and efficiency », The International Journal of Management Science, volume 35, n° 2, 2007, pp. 131-142. HAMED Y. (2004); « Microcrédit et financement de la microentreprise au Maghreb », Thèse de Doctorat, Sciences Economiques, Université Paris 12. HICKSON, R. (1999), Reaching extreme poverty: financial services for the very poor, Working Paper, Office for Development Studies, UN Development Programme. KOPP R.J., DIEWERT W.E., « The decomposition of frontier cost deviation into measure of technical and allocative efficiency» Journal of Econometrics, n° 9, 1982, pp. 319-322. KUMBHAKAR S.C., « Estimation of input specific technical and allocative inefficiency in stochastic frontier models », Oxford Economic Papers, n° 40, 1988, pp. 535-549. LUZZI F. G., WEBER S. (2006): « Measuring the performance of microfinance institutions», Microfinance and Public Policy, pp 153-169. MARGUERITE S. R. (2001); “The Microfinance Revolution: Sustainable Finance for the Poor”, World Bank. MONTALIEU T. (2002), « Les institutions de micro-crédit : entre promesses et doutes. Quelles pratiques bancaires pour quels effets ? », Monde en développement, Tome 30, PP. 119-121. MORDUCH J. (1999); “The microfinance promise", Journal of Economic Litterature, vol XXXII, pp 1569-1614. MORDUCH J. (2000); “The microfinance schism”, World Development, vol 28, n°4, pp 617629. NAVAJAS S., SCHREINER M., MEYER R.C., GOZALEZ-VEGA C. et. RODRIGUEZMEZA J. (2000); “Microcredit and the Poorest of the Poor: Theory and Evidence from Bolivia”, World Development, vol. 28, n° 2, pp 333-346. OLSZYNA-MARZYS R. (2006): “Microfinance institutions: profitability at the service of outreach? A study of the microfinance industry in the ECA region”, College of Europe Bruges campus, European economic studies department. PAXTON J. (2002): “Depth of outreach and its relation to the sustainability of microfinance institutions”, Savings and Development, Giordano Dell'Amore Foundation, vol. 26, n°1, pp 69-85. PEDRAJA-CHAPARRO F., SALINAS-JIMENEZ J., SMITH P., « On the quality of the DEA model », The Journal of the Operational Research Society, volume 50, n° 6, 1999, pp. 636-644. POLANCO OLIVARES- F. (2005): “Commercializing microfinance and Deeping outreach?: Empirical Evidence from Latin America”, Journal of microfinance, vol. 7, n° 2, pp 47-69. PRESTON L.E., O’BANNON D.P. (1997); “the corporate social-financial performance relationship: A typology and analysis”, Business &Society, Vol 36, N°4, December, pp 419429. SERRANO CINCA C., MAR MOLINERO C., « Selecting DEA specifications and ranking units via PCA », Discussion Papers in Management, n° M01-3, University of Southampton, 2001, 23 pages. SOULAMA S. (2005); “Microfinance, pauvreté et développement”, Agence Universitaire de la Francophonie, Archives contemporaines (Paris). SUFIAN F., « The efficiency of non-bank financial institutions: empirical evidence from Malaysia ». International Research Journal of Finance and Economics, n° 6, 2006, pp. 49-65. TORGENSEN A.M., FORSUND F.R., KILLELSEN S.A.C., « Slack-adjusted efficiency measures and ranking of efficient units », Journal of Productivity Analysis, volume 7, n°4, 1996, pp. 379-398. WADDOCK S.A., Graves Samuel B. (1997); “The corporate Social performance- Financial performance Link”, Academy of Management Review, vol 10, n°4, pp 303-319. Annex : Data relative to each “MC2” MC2 Assets Bab 375 502 748 BMk 1 071 805 827 Baf 136 774 713 B_r 457 158 397 Bah 738 843 766 B_B 62 519 267 Bam 349 985 862 Bja 259 717 413 Bjn 600 347 781 Bté 795 791 217 Bng 105 977 137 Bat 165 903 426 Bay 443 629 053 Dou 272 508 929 Man 196 660 112 Mel 702 699 016 Muy 511 028 672 Nga 71 901 702 Pen 438 172 022 TPD 200 169 585 Source : ADAF (2006) Inputs Emp 5 8 5 6 6 5 5 5 6 5 4 6 5 7 4 7 8 6 6 5 Costs 20 200 287 77 096 702 15 164 973 32 415 904 42 768 932 6 686 091 29 197 031 15 099 086 38 692 016 53 871 251 12 981 132 9 634 632 23 634 979 24 229 070 15 640 663 50 855 974 44 753 666 11 481 788 24 284 652 11 729 846 Credits 133 464 101 356 571 255 43 739 175 170 369 043 184 317 763 21 533 690 138 583 833 105 878 847 290 395 356 142 538 493 64 942 423 36 841 397 132 493 808 109 147 135 39 429 307 278 229 980 228 199 671 30 773 402 116 621 850 70 172 568 Outputs Products 34 455 607 80 284 044 8 604 928 42 233 800 44 901 834 2 322 565 33 025 642 15 004 364 45 123 366 63 318 936 5 504 700 13 597 319 22 864 692 19 180 835 14 777 842 54 974 556 53 265 644 4 117 056 27 856 699 5 504 700 Wom 333 869 295 543 617 65 467 319 571 664 209 127 520 728 235 662 425 66 509 252