Social and financial efficiency of microfinance institutions in Pakistan.
Khan, Zahoor ; Sulaiman, Jamalludin
Targeting financially marginalised communities in an efficient way
is the most desirable and possibly the most effective strategy for
microfmance institutions (MFIs) to reduce incidences of absolute
poverty, create self-employment opportunities, and improve socioeconomic
wellbeing of the poor communities. This paper attempts to investigate
social and financial efficiency of Pakistani microfinance institutions
to bring forth optimal strategies for financing non-bankable poor in an
efficient and self-sustainable way. The investigation of efficiency of
the MFIs in Pakistan can help the major stakeholders of the industry in
understanding the current scenario and to design optimal policy agenda
for the future. The sample size of this study consists of all MFIs in
Pakistan for the year 2013. The data about the MFIs has been taken from
'Mixmarket' database. After specifying 19 different DEA
models, with the help of three input and four output variables,
representing various dimensions of MFIs such as cost structure,
financial structure and organisational characteristics, the study
reveals that MFIs efficiency is sensitive towards the selection of input
and output variables, the choice of CCR and BCC models and the number of
input and output variables in the model. The study further reveals that
there is no single way to efficiency, however, it mainly depends on the
scale, age and types of MFIs. Microfinance banks perhaps are not
appropriate financial institutions to extend microcredit to poorer
community members and to achieve the goal of women empowerment through
the extension of credit to women. Our rough estimates suggest that
inefficient MFIs should focus on the optimal use of Asset (which is
common among the socially efficient MFIs irrespective of their types and
size) followed by reducing operating costs and improving the quality of
loan officers.
JEL Classification: 043, L25
Keywords: Efficiency, Microfmance Institutions, DEA
1. INTRODUCTION
Financial efficiency and profitability of 'for profit'
institutions have been traditionally measured with the help of financial
ratios [Hassan and Sanchez (2009)]. However, financial ratios are
inappropriate to investigate the sources of inefficiency, estimate
financial or social efficiency with multiple inputs and outputs, and to
decompose the sources of efficiency or inefficiency into technical,
technological and scale efficiencies or inefficiencies respectively
[Hassan and Sanchez (2009)]. Microfinance Institutions (MFIs) are
special institutions, which simultaneously consider their social role to
uplift the marginalised community members along with their commercial
objective to secure self-sustainability. In standard literature this
phenomenon is coined MFIs as being 'double bottom line"
institutions. [Gutierrez-Nieto, Serrano-Cinca, and Mar Molinero (2007);
Gutierrez-Nieto, Serrano-Cinca, and Molinero (2007)]. This simultaneity
differentiates MFIs from conventional financial institutions. The
achievement of socioeconomic efficiency is indispensable for MFIs to
operate independently and on a wider scale. Thus investigation of
socioeconomic efficiency of MFIs is important for monitoring and optimal
policy implications.
Efficiency assessment techniques are broadly divided into
parametric; such as Stochastic Frontier Analysis (SFA), Thick Frontier
Analysis (THA) etc., and nonparametric techniques such as Data
Envelopment Analysis (DEA) [Berger and Humphrey (1997); Gutierrez-Nieto,
et al. (2007)]. According to Berger and Humphrey (1997) the popular
efficiency assessment technique is DEA. This technique does not assume
any prior specific shape of distribution and is also free from specific
functional form. In spite of the mentioned strengths of DEA, it also has
some demerits as well. Before investigating efficiency of Pakistani
MFIs, it is important to consider the limitation of DEA. Otherwise it
may yield misleading results. For example, an inefficient DMU may become
efficient and vice versa because of inappropriate specification of the
model or irrelevant input or output variables [Gutierrez-Nieto, et al.
(2007)]. How to avoid or minimise the biasedness of this technique is a
question of central importance for researchers and policy institutions?
The DEA technique identifies an efficient MFI based on extreme
information therefore, it is unsafe to conclude that an MFI is efficient
or otherwise based on a single input-output specification. To deal with
this issue, this paper attempts to identify an efficient MFI based on
all possible and theoretically important combinations of input and
output variables. This idea was developed by Cinca and Molinero (2004).
There are different statistical techniques such as "Factor
Analysis" to identify factor inputs or outputs, which are more
important than other combinations of input and output variables for
model specifications.
Pakistan initiated microfinance programmes in 1980s. The Agha Khan
Rural Support Programme (AKRSP) and the Orangi Pilot Project were the
first microcredit programmes initiated in Pakistan. Today microfinance
sector in Pakistan consists of; Microfinance banks, Rural Support
Programmes, NGOs, Islamic microfinance NGOs and specialised MFIs. Major
changes have been observed in the Microfinance sector in Pakistan.
First, there were no practices of the provision of other financial
services like micro-insurance, deposit; micro-pensions etc., except
microcredit but in recent days MFIs provide a set of financial products
and policies. Second, microcredit programmes were imitated to help the
poor and marginalised people without any commercial objectives but
todays' most of the MFIs have changed their intentions and now they
are looking for both: commercial gains and social success. Third,
microfinance programmes in Pakistan were multidimensional in nature but
todays' microfinance programmes are more specific and specialised.
Pakistan is one of the developing countries, which recognised the
importance of microfinance as a strong tool for socioeconomic uplifting
of the poor and financially marginalised segments since the early
80's. Although, the country has initiated the efforts for the last
30 years against poverty and gender disparity; however, the desired
outcome has not been achieved. Under the "Microfinance Strategy
2007", the state bank of Pakistan set a target to reach 3 million
borrowers until the end of 2010. Further, the target is expected to grow
from 3 million to 10 million by the end of 2015 [SBP(2011)]. (1)
Flowever, until the end of fiscal year 2012-13, around 2.43 million poor
have only been reached by microfinance institutions [Mixmarket (2012)].
This paper aims to gauge financial and social efficiency of
Pakistani MFIs across the country to know the underlying factors which
make a particular DMU efficient or otherwise. These factors have been
investigated in different dimensions such as organisational
characteristics, cost and financial structure of MFIs, the ability of
MFIs to generate maximum profit, disburse maximum loans, and targeting
'poorer and financially marginalised' community members.
2. THEORETICAL FRAMEWORK OF THE STUDY
The roots of micro-financing, to facilitate the poor by providing
small loans for productive utilisation and self-employment, can be
traced back to philosophical concern of conceptualising poverty as
lacking of access to financial capital [Engberg-Pedersen and Munk
Ravnborg (2010); Hulme and Shepherd (2003)]. According to this concept
poor are assumed to be productive, capable of running their own small
businesses and creditworthy to payback their loans. This idea initiated
the extension of microcredit to the poor at different formal and
informal levels [Ledgerwood (1999)]. Informal credit has remained a more
dominant source for the poor who were not able to produce physical
collateral for conventional financial institutions [Rhyne and Christen
(1999)]. In nutshell, conceptualisation of poverty as lack of access of
the poor to financial capital, the extension of financial capital for
self-employment and productive utilisation of credit, marginalisation of
poor by the traditional banks due to lack of physical collateral,
exploitation of the poor by informal credit sources and focus of
business models on alleviation of poverty are some of the factors which
initiated microfinance activities across the globe.
The operations of microfinance institutions can be broadly observed
into two contexts. First, Microfinance institutions can be observed as
financial intermediaries such as they collect deposits from the clients
and non-clients, they provide saving facilities to the clients and then
mobilise the funds among the clients who need them [Christen and Drake
(2002); Qayyum and Ahmad (2006)]. In this context, microfinance
institutions are more or less similar to conventional banks in terms of
their operations. Second, microfinance institutions can be treated as
production units [Gonzalez (2007); Haq (2008)]. MFI institutions use
certain inputs such as credit officers, capital and produce outputs such
disbursement of loans, generating revenue and targeting the poor clients
[Armendariz and Morduch (2010); Qayyum and Ahmad (2006)]. Production
approach seems more appropriate than intermediary approach because all
MFIs do not provide the facility of saving and deposits, except
microfinance banks (which is only one kind of MFIs) thus, this approach
does not fit to maximum MFIs [Gutierrez, Serrano-Cinca, and Molinero
(2007)]. Efficiency theories, to test financial or social efficiency and
overall performance of microfinance thus do not seem good in production
approach. The neo classical theory of production and production
efficiency seems more suitable when the MFIs are assumed as productive
units (such as firms), while they are producing almost same products,
working in the same regulatory and environment, using more or less same
inputs. Based on the assumptions of neoclassical economists, producers
always operate efficiently in terms of both technical aspects and
economic aspects as well [Kokkinou (2010)]. For example, technical
efficiency means optimisation by not wasting productive resources while
economic efficiency means producers optimise by solving allocation
problem involving prices. The difference in production may result from
the differences in;
(i) Technology of production.
(ii) Differences in the efficiency of the production process.
(iii) Differences in the environment where production is taking
place.
There is a fair chance of difference in production even when
technology and production environment are almost the same, firms or
industries may exhibit different productivity levels due to differences
in their production efficiency [Kokkinou (2010)]. Thus, this study
attempt to investigate social and financial efficiency of the MFIs under
the assumption of constant return to scale (input oriented CCR-model)
and variable return to scale (input oriented BCC-model). The following
Figure 1 shows theoretical framework of the study.
[FIGURE 1 OMITTED]
Inputs: Factors which are used to produce something or deliver a
service. These can affect the production process, Industry
characteristics can be affected from external factors.
Production Process: This is a link between factors input and
output. This may compromise the quality and quantity of inputs,
exogenous factors, and industry characteristics while it can affect
output and industry characteristics in turn. This may encompasses
production technology, internal environment, scale of production.
Output: This may be in the form of physical production or the
provision of service. Output is affected by inputs through the
production process and affects organisational performance.
External Factors: Factors which are exogenous such as intervention
of the government through regulation polices, donors, rating agencies.
These factors may affect the whole process-starting from input selection
to operational performance.
Industry Characteristics: Industry characteristics such as the
number of FMIs in the industry, capital or labour intensity of the
industry, what product is being produced or what service is being
offered. Industry characteristics are affected by and also affect
inputs, production process, output and organisational performance.
3. MATERIAL AND METHODS
The sample size of the study consists of all Pakistani MFIs,
available with latest complete information on Microfinance Information
Exchange (MIX). The study therefore; uses cross sectional data for the
year 2012. The selection of input and output variables is based on the
literature [Gutierrez-Nieto, et al. (2007); Hassan and Sanchez (2009);
Mamiza Haq, Michael, and Shams (2010)]. After going through the
literature three inputs (Assets, Operating Costs (OC) and Loan Officers
(LO)) and four outputs (two financial variables such as Gross Loan
Portfolio (GLP), Financial Revenue (FR) and two social variables such as
Women Borrowers (WB) and indicator of poorer clients' index (P))
selected to investigate how efficiently MFls in Pakistan transform the
selected inputs to achieve their twin objectives; optimal social and
financial efficiency. Based on Gutierrez, et al. (2007) calculating
poverty index requires to weight each MF1 as; w = [{l-(Ki-Min (k)) /
Range of K] where i represents the number of a particular MFI. Min (k)
is the minimum of Average Loan per Borrower (ALPB) while max (k) is the
maximum of ALPB. The range represents the difference between maximum and
minimum (Max (K) - Min (k)). Based on the weight (w) assigned to each
MFI, the indicator of poverty has been thus obtained. Pi is an index of
support of the poor, based on ALPB. This index favours those MFIs which
have smaller ALPB. Pi for a specific MFI can be obtained when its weight
(w) is multiplied by number of borrowers (B); wB = [{(l-(Ki-Min (k) /
Range of K} * B], It is a combination of two outreach indicators; width
of outreach (number of borrowers) and depth of outreach (ALPB). Women
borrowers and poverty index, both, are used as social indicators of
MFIs.
MFIs in Pakistan consist of seven specialised microfinance banks,
three Non-Banking Financial Institutions (NBFI) and nineteen NGOs.
Keeping into consideration the limitation of same input and output
variables for DEA models, this study adopted a production approach and
avoided deposits with MFIs as input because the majority MFIs
(particularly, NGO are mostly not regulated and thus are not able to
mobilise savings and collect deposits from their clients) do not provide
the facilities of saving or deposit collections. The following Table 1
represents input and output variables, their definitions and measurement
units.
Data envelopment analysis efficiency score, with the help of the
selected input and output variables, is estimated under BCC [Banker,
Chames, and Cooper (1984)] and CCR [Chames, Cooper, and Rhodes (1978)]
input based models through 19 different specifications. Each
specification of input/s and output/s represents a unique combination to
reveal the sources of efficiency or inefficiency for each MFI. For
example, the input variables represent three dimensions; asset (capital
structure), operating cost (cost structure) and loan officers (the
quality of human resources) and the output variables represent financial
indicators (gross loan portfolio and financial revenue) and social
indicators (indicator of poverty and targeting the women clients). First
12 models (Al, A2, A3, A4, B1, B2, B3, B4, C1, C2, C3, and C4) represent
a corresponding one to one relationship among the input and output
variables. This will help to identify the channels of efficiency for
each MFI. The next four models (ABC1, ABC2, ABC3 and ABC4) represent the
combinations of all inputs with respect to financial and social
indicators. Subsequent two models (ABC 12 and ABC34) represent financial
and social efficiency models. These are more comprehensive models of
financial and social efficiency than the previous models because they
take into consideration all input and output variables, which make an
MFI efficient or otherwise. The final model (ABC 1234) represents
overall efficiency based on all the selected input and output variables.
These models were estimated through DEA, a non-parametric
technique, used for calculation of social and financial efficiency
without prior information about the shape of the distribution of a data
set. This technique allows the researchers to calculate social or
financial efficiency with multiple inputs and outputs [Gutierrez, et al.
(2007); Gutierrez and Lezama (2011); Flaq, Skully, and Pathan (2010);
Kabir and Benito (2009)]. This technique is equally beneficial for
commercial and non-commercial DMUs. Both input-oriented (IO) and
output-oriented (OO) versions of the DEA methodology have been applied
to the data for the sake of efficiency score comparison. In order to
specify the mathematical formulation of the IOM, if there are K MFIs (in
the language of DEA it is called DMUs) using N inputs to produce M
outputs then inputs are denoted by [X.sub.jk](j=l .... n) and the
outputs are represented by [y.sub.ik] (i=l .... m) for each MFI k (k=1
.... K). The efficiency of the DMU can be measured as shown by [Coelli,
Rao, and Battese (1998); Qayyum and Ahmad (2006); Shiu (2002);
Worthington (1999)].
Technical Efficiency = (Sum of weighted output/Sum of weighted
input)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)
Where [y.sub.ik] is the quantity of the 7th output produced by the
kth MFI, XjS is the quantity of jth input used by the kth MFI, and
[u.sub.i] and Vjare the output and input weights respectively. The DMU
maximises the efficiency ratio, TEk, subject to;
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)
The above Equation (2) indicates that efficiency measures of an MFI
cannot exceed 1, and the input and output weights are positive. The
weights are selected in such a way that the MFI maximises its own
efficiency. To select optimal weights the following mathematical
programming (output-oriented) is specified [Coelli, et al. (1998);
Qayyum and Ahmad (2006); Shiu (2002); Worthington (1999)].
4. CCR AND BCC INPUT ORIENTED MODELS
Input-orientated DEA model looks at the amount by which inputs can
be proportionally reduced, where the amount of output is supposed to be
fixed. On the contrary, the output-orientated model looks at the amount
by which outputs can be proportionally expanded, where the amount of
input is supposed to be fixed. The DEA can be conducted under the
assumption of constant returns to scale (CRS) or variable returns to
scale (VRS)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
Subject to
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
Where [theta]o is the proportion of DMUo's inputs needed to
produce a quantity of output equivalent to its benchmarked DMU output
identified and weighted by the [lambda]i. [S.sup.*.sub.i] sr +is the
slack variables of input and output respectively. [[lambda].sub.j] is a
(n x 1) column vector of constants and indicate benchmarked DMUs.
The CCR model developed by Charnes, et al. (1978) estimate the
efficiency of DMU with the assumption of Constant Return to Scale (CRS).
This assumption may fail in imperfect markets. The CRS assumption is
only appropriate when all firms are operating at an optimal scale. The
use of the CRS specification when all firms are not operating at the
optimal scale results in measures of Technical Efficiency (TE) which are
confounded by scale efficiencies (SE). The use of the VRS specification
permits the calculation of TE devoid of these SE effects.SE can be
calculated by estimating both the CRS and VRS models and looking at the
difference in scores. VRS model is essentially the CRS with an
additional constraint added to the LP problem.
The BCC model developed by Banker, et al. (1984) is a modified
version of CCR. This model helps to investigate scale efficiency. If
striction Y?k=i = 1, is connected, then CCR model becomes BCC [Banker,
Charnes, Cooper (1984)] model.
The modified form of CCR can be written as:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
Subject to
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
5. ESTIMATION TECHNIQUES AND METHODOLOGICAL CONCERNS
Following the model specification suggested by Cinca and Molinero
(2004) the study estimated DEA efficiency for each feasible
specification. Thus 19 different specifications of input and output
variables have been estimated. Finally, super efficiently for all
specified models have been estimated to rank the efficient MFIs [Lovell
and Rouse (2003)]. In terms of input oriented models, super efficiency
of a DMU represents the maximum possible proportional increase in an
input vector retaining the DMU efficiency [Khodabakhshi (2007)].The DEA
efficiency and super efficiency of all selected models have been
estimated in the Efficiency Measurement System (EMS 1.3 version). The
following Tables 2 and 3 show the results of various DEA efficiency
models under the BCC and CCR models. The BCC models are used to estimate
pure technical efficiency while CCR models are used to estimate overall
technical efficiency. The ratio of CCR and BCC are then used to estimate
scale efficiency and returns to scale [Banker and Thrall (1992);
Ruggiero (2011)]. The maximum value of a technical or pure technical
efficient MFI is 100. It means that the MF1 is 100 percent efficient to
transform inputs into outputs. Any MFI for which the efficiency score is
less than 100 is considered inefficient in managerial and technical
aspects [Charnes, et al. (1978)].
6. RESULTS AND DISCUSSION
Tables 2 and 3 show efficiency scores, which resulted from input
oriented CCR and BCC models for 29 MFIs with 19 specifications, to
comprehend that what makes an MFI efficient or otherwise. The last three
columns (ABC 12, ABC34, ABC 1234) of the Tables 2 and 3 present
financial, social and overall efficiency respectively. None of the MFIs
is 100 percent efficient under all specifications. A total of 10 out of
29 MFIs while only 2 FMIs were found 100 percent efficient on social,
financial and overall efficiency dimensions under BCC and CCR models
respectively. An MFI, which is efficient on social or financial
dimensions is also 'overall efficient'. Under both model
structures (BCC & CCR) the number of efficient MFIs increases when
it has used more input and output variables. This is evident from the
last three columns of the Tables 2 and 3. As these models involve more
input and output variables, therefore the numbers of efficient MFIs are
also higher than the rest of models' results.
The efficiency result of MFIs also varies across the return to
scales. Under the CCR models, assuming a constant return to scale, only
two out of twenty nine MFIs are overall efficient (means efficient on
social, financial and overall dimensions-including all input and output
variables) while under BCC models, assuming variable returns to scale,
ten MFIs are efficient on social, financial and overall dimensions. This
finding of the study cautions about misleading results, resulting from a
single specification of DEA efficiency estimated for a DMU.
Notwithstanding, such a single specification may not reveal the sources
of efficiency or inefficiencies. The difference between the results of
the CCR and BCC models of efficiency reveals the difference between
managerial, technical and scale efficiencies. The MFIs, which are
socially, financially and overall efficient under CCR models such as
ASA- Pakistan and Orangi are at least efficient by either managerial or
scale dimensions. Relaxing the assumption of constant return to scale
enhanced the number of efficient MFIs. This reflects that majority of
MFIs are efficient based on the managerial and technical skills but not
on the scale dimensions. Thus the difference between BCC and CCR
efficiency models reveal the sources of inefficiency , which resulted
from the scale of the DMUs. The findings reveal that 2 out of 10
efficient MFIs, based on three comprehensive specifications (ABC 12, ABC
34, ABC 1234) under CCR are efficient based on managerial and scale
dimensions (Please see Table 3 last three columns). Estimating
efficiency of DMUs with a single specification and from full dataset
will not reveal that how a particular DMU has achieved efficiency?
Similarly, if a DMU is inefficient we shall not be able to detect the
reasons of inefficiency.
Super efficiency for all 19 specifications of models has been
estimated to know the rank of the efficient MFIs. As super efficiency of
inefficient MFIs remains the same therefore, this technique only helps
to rank the efficient MFIs [Scheel (2000)]. Based on the CCR input
efficiency model, the super efficiency of Oranagi (an NGO based MFI) is
216.60 percent followed by ASA- Pakistan (an NFBI) with a 120.90 percent
score. It can be interpreted as keeping the same output level; an
increase in the inputs usage by Orangi and ASA- Pakistan by 116
percentage points and 20 percentage points respectively will not affect
the efficiency level of these MFIs.
7. CONCLUSION AND POLICY IMPLICATIONS
The assessment of MFIs' efficiency is imperative for all
stakeholders for optimal policy measures. Data envelopment analysis is a
popular non-parametric, non-stochastic, liner programing based
efficiency technique. This paper concentrates on the technical aspects
of DEA efficiency score that how it varies across the selection of
inputs and outputs, the number of inputs or outputs and the selection of
DEA estimation technique. The sample size of this study consists of all
MFIs in Pakistan. We have modelled all feasible and meaningful
specifications. After 19 different specifications with the help of three
input and four output variables, representing various dimensions of MFIs
such as cost structure, financial structure and organisational
characteristics, we have used input oriented BCC and CCR data
envelopment analysis oriented models. We have also estimated super
efficiency for all MFIs to rank them according to their potential. This
study attempted to investigate financial and social level of efficiency
of MFIs and to gauge tracks to efficiency.
The study attempted to achieve the required objectives using
appropriate methodology. The study used Data Envelopment Analysis
technique to investigate social and finical efficiency. The findings of
the study revealed that NGOs and NBFI were more efficient, based on the
achievements of social and financial objectives than microfinance banks.
Financial and social efficiency of MFIs were estimated by two ways to
reveal information about 'managerial and technical' aspects of
MFIs. The study revealed that none of the microfinance institutions was
found 100 percent efficient under all financial and social efficiency
models. There were 13 MFIs, which were pure technically efficient in
financial aspects out of the 29 MFIs. Bukhsh foundation scored highest
(77.7 percent) and remained financially efficient under 15 of 19
different pure technical efficiency models. Subsequently, non-banking
financial institutions and microfinance banks stood second in financial
efficiency ranking (55.5 percent) based on pure technical score.
Like financial performance of MFIs, there was also a difference in
social performance of MFIs, which resulted from variation in
institutional characteristics. Twelve MFIs were found socially efficient
based on input oriented pure technical efficiency models. Out of total
socially efficient MFIs, nine were NGOs, one microfinance bank (Khushali
bank) and two non-banking financial institutions (ASA- Pakistan, Orix
leasing). The study reveals and recommends the following; The study
reveals that efficiency score resulted from DEA, is sensitive towards
the choice of inputs, outputs, functional form and number of inputs and
outputs. Based on the sensitivity of this technique, the study warns
against single specification of DEA and recommends multiple
specifications of DEA efficiency models to conclude whether a particular
DMU is efficient or otherwise. It was noticed that two MFIs could yield
the same efficiency score, however; their way to achieve efficiency was
quite different from each other. The MFIs had used different channels,
which were considered their strengths, such as controlling operational
cost or optimal utilisation of loan officers and Assets. It was also
noticed that MFIs were more efficient in their managerial and technical
skills rather than the scale of operation of MIFs. It is recommended to
estimate pure technical and scale efficiencies separately, to comprehend
the sources of efficiency or inefficiency about various DMUs to identify
peers for corresponding MFIs accordingly. The overall super efficiency
result of an MFI, based on collective social and financial output
variables (variable 1,2,3, and 4), is at least as efficient as financial
or social super efficiency models for that MFI. Increasing the number of
input and output variables changes the efficiency score of DMUs. This is
evident from Tables 2 and 3. The higher the number of input and output
variables, the higher the efficiency chance for an MFI and vice versa.
In this case the estimation of super efficiency is important along with
technical and scale efficiencies. This allows the researchers to rank
the MFIs, based on super efficiency score. Technical and scale
efficiency in isolation cannot rank MFIs according to their
corresponding efficiency levels.
Zahoor Khan <zahoor.khan@imsciences.edu.pk> is Assistant
Professor, Institute of Management Sciences, Hayatabad, Peshawar.
Jamalludin Sulaiman is Professor, School of Social Sciences, Universiti
Sains Malaysia, Malaysia.
REFERENCES
Armendariz, Beatriz, and Jonathan Morduch (2010) The Economics of
Microfinance. MIT press.
Banker, Rajiv D. and Robert M. Thrall (1992) Estimation of Returns
to Scale Using Data Envelopment Analysis. European Journal of
Operational Research 62:1, 74-84.
Banker, Rajiv D., Abraham Chames, and William Wager Cooper (1984).
Some models for estimating technical and scale inefficiencies in data
envelopment analysis. Management Science 30:9, 1078-1092.
Berger, Allen N. and David B.Humphrey (1997). Efficiency of
Financial Institutions: International Survey and Directions for Future
Research. European Journal of Operational Research 98:2, 175-212.
Charnes, Abraham, William W. Cooper, and Edwardo Rhodes (1978)
Measuring the Efficiency of Decision Making Units. European Journal of
Operational Research 2:6, 429-444.
Christen, Robert Peck and Deborah Drake (2002) Commercialisation:
The New Reality of Microfinance. The Commercialisation of Microfinance:
Balancing Business and Development 2-20.
Cinca, C. Serrano and C. Mar.Molinero (2004) Selecting DEA
Specifications and Ranking Units via PCA. Journal of the Operational
Research Society 55:5, 521-528.
Coelli, Tim, D. Rao and G. Battese (1998) An Introduction to
Efficiency and Productivity Analysis. Kulwar Academic Publishers,
Massachusetts, USA.
Engberg-Pedersen, Lars, and Helle Munk Ravnborg (2010)
Conceptualisations of Poverty. DIIS Reports/Danish Institute for
International Studies.
Gonzalez, Adrian (2007) Efficiency Drivers of Microfinance
Institutions (MFIs): The Case of Operating Costs. Microbanking Bulletin
15.
Gutierrez, B., C. Serrano-Cinca, and C. Mar. Molinero (2007) Social
Efficiency in Microfinance Institutions. Journal of the Operational
Research Society 60.T, 104-119.
Gutierrez, Jorge and Beatriz Goitisolo Lezama (2011) Profitability
and Social Performance of Microfinance Institutions: Empirical Evidence
of Relations between Different Types of Variables. revista de economla
mundial 27, 189-214.
Gutierrez-Nieto, Begofia, Carlos Serrano-Cinca, and C. Mar.
Molinero (2007) Social Efficiency in Microfinance Institutions. Journal
of the Operational Research Society 60:1, 104-119.
Gutierrez-Nieto, Begona, Carlos Serrano-Cinca, and Cecilio Mar
Molinero (2007) Microfinance Institutions and Efficiency. Omega 35:2,
131-142.
Haq (2008) Microfinance Industry Assessment: A Report on Pakistan.
Islamabad Pakistan Microfinance Network (PMN).
Haq, Skully, Michael, and Shams Pathan (2010) Efficiency of Micro
finance Institutions: A Data Envelopment Analysis. Asia-Pacific
Financial Markets 17.1, 63-97.
Hassan, M. Kabir, and Benito Sanchez (2009) Efficiency Analysis of
Microfinance Institutions in Developing Countries. Networks Financial
Institute at Indiana State University Working Paper, 12.
Hulme, and Andrew Shepherd (2003) Conceptualising Chronic Poverty.
World Development, 31:3, 403-423.
Kabir, Hassan and Sanche Benito (2009) Efficiency Analysis of
Microfinance Institutions in Developing Countries. Networks Financial
Institute Working Paper.
Khodabakhshi, M. (2007) A super-efficiency Model based on Improved
Outputs in Data Envelopment Analysis. Applied Mathematics and
Computation 184/2, 695-703.
Kokkinou, Aikaterini (2010) A Note on Theory of Productive
Efficiency and Stochastic Frontier Models. European Research Studies
Journal 13:4, 109-118.
Ledgerwood, Joanna (1999) Microfinance Handbook: An Institutional
and Financial Perspective: World Bank Publications.
Lovell, CAK and APB Rouse (2003) Equivalent Standard DEA Models to
Provide Super-efficiency Scores. Journal of the Operational Research
Society 54.T, 101-108.
Mamiza Haq, Michael, Skully and Shams Pathan (2010) Efficiency of
Microfinance Institutions: A Data Envelopment Analysis. Asia-Pacific
Financial Markets 17:1, 63-97.
Mixmarket (2012) www.mixmarket.org.
Qayyum, Abdul, and Munir Ahmad (2006) Efficiency and Sustainability
of Micro Finance Institutions in South Asia. South Asian Network of
Economic Research Institutes (SANEI).
Rhyne, Elisabeth, and Robert Christen (1999) Microfinance Enters
the Marketplace: USAID.
Ruggiero, John (2011) Introduction Frontiers in Major League
Baseball (pp. 1-5): Springer.
State Bank of Pakistan (2011) Microfinance Department Strategic
Framework for Sustainable Microfinance in Pakistan. Karachi: State Bank
of Pakistan.
Scheel, Holger (2000) EMS: Efficiency Measurement System
User's Manual. Operations Research and Witshaftinsformetik.
University of Dortmund, Germany.
Shiu, Alice (2002) Efficiency of Chinese Enterprises. Journal of
Productivity Analysis 18:3, 255-267.
Worthington, Andrew (1999) Performance Indicators and Efficiency
Measurement in Public Libraries. Australian Economic Review 32:1, 31-42.
(1) Please see Table 1.3 and 1.4.
Table 1
Inputs and Outputs and Their Definitions and Measurement
Symbol Variable Name Variable Definition Unit
Input (A) Total Assets Total of all net asset $
accounts
Input (B) Operating Cost Expenses related to
operations, such as all $
personnel Expenses, rent
and utilities,
transportation, office
supplies, and depreciation
Input (C) Number of loan The number of individuals who Number
officers are actively employed by
the MFI to disburse loan
and collect repayments.
Output (1) Gross loan Gross loan portfolio
portfolio outstanding principal $
balance of all of the MFI's
outstanding loans including
current, delinquent and
restructured loans, but not
loans that have been
written off.
Output (2) Financial Financial revenue generated
revenue from the gross loan $
portfolio and from
investments plus other
operating revenue
Output (3) Indicator of Poverty Index, it is a
Benefit to combination of two outreach
the poorest indicators; width of
outreach (number of
borrowers) and depth of
outreach (ALPB).
Output (4) Number of women Number of active borrowers Number
borrowers who are female
Source: Adopted from [Gutierrez, et al. (2007); Gutierrez and
Goitisolo Lezama (2011)].
Table 2
DEA Efficiency of MFIs Based on BCC Input Oriented Models
DMUs A1 A2 A3 A4 B1 B2 B3
Akhuwat 54 33 41 15 45 37 43
Apna MF Bank 83 74 9 10 42 59 15
ASA Pakistan 100 100 100 100 79 90 100
Asasah 8 53 79 79 16 16 18
BRAC-PAK 73 95 58 60 20 34 20
Buksh Foundation 100 100 100 100 29 29 29
CSC 62 85 31 38 22 47 15
CWCD 82 80 28 14 28 41 15
DAMEN 77 89 30 4 59 99 27
FFO 79% 75 63 66 32 42 30
FMFB--Pakistan 48 68 10 5 45 64 12
GBTI 33 40 10 12 37 61 25
JWS 69 74 36 41 29 51 20
Kashf Bank 57 81 1 1 22 36 1
Kashf Foundation 100 98 100 100 63 65 85
Khushali Bank 100 76 67 11 83 55 39
NRSP 99 84 100 50 100 100 100
NRSP Bank 63 85 13 3 59 80 16
Orangi 6 32 50 5 100 86 100
Orix Leasing 62 53 45 45 46 63 42
POMFB 18 42 6 4 12 40 7
PRSP 28 52 20 11 29 72 26
RCDS 59 76 34 36 27 58 21
SAFWCO 64 61 56 28 35 57 40
SRSO 86 61 48 51 75 68 47
SRSP 85 48 97 94 69 69 69
Sungi 98 79 100 100 100 100 100
TMFB 100 100 1 4 100 100 1
TRDP 83 61 82% 51 69 81 96
DMUs B4 C1 B2 C3 C4 ABC1 ABC2
Akhuwat 18 21 10 30 16 61 38
Apna MF Bank 15 34 26 13 14 90 79
ASA Pakistan 100 22 17 59 82 100 100
Asasah 23 43 30 52 78 88 56
BRAC-PAK 24 14 14 23 38 73 95
Buksh Foundation 29 100 100 100 100 100 100
CSC 22 39 40 31 52 72 93
CWCD 15 25 21 18 13 82 80
DAMEN 50 69 60 41 100 100 100
FFO 35 32 28 35 47 81 79
FMFB--Pakistan 7 45 45 26 19 79 83
GBTI 25 58 56 39 46 61 81
JWS 30 37 31 29 52 79 80
Kashf Bank 1 52 53 5 5 82 98
Kashf Foundation 100 27 19 75 100 100 98
Khushali Bank 10 61 42 100 39 100 71
NRSP 95 17 12 100 39 100 100
NRSP Bank 4 43 41 27 7 96 100
Orangi 18 99 46 100 28 100 88
Orix Leasing 54 65 50 65 100 82 76
POMFB 7 28 41 20 16 29 54
PRSP 16 15 20 21 16 35 78
RCDS 29 36 35 30 54 69 83
SAFWCO 28 29 22 33 31 71 66
SRSO 62 47 26 49 89 100 69
SRSP 69 90 80 100 100 100 100
Sungi 100 43 38 52 62 100 100
TMFB 5 100 100 5 27 100 100
TRDP 72 31% 19 54 50 99 81
ABC ABC ABC
DMUs ABC3 ABC4 12 34 1234
Akhuwat 52 22 61 52 61
Apna MF Bank 20 20 91 20 91
ASA Pakistan 100 100 100 100 100
Asasah 100 100 88 100 100
BRAC-PAK 58 60 95 60 95
Buksh Foundation 100 100 100 100 100
CSC 48 62 93 62 93
CWCD 30 21 82 30 82
DAMEN 54 100 100 100 100
FFO 69 72 81 72 81
FMFB--Pakistan 26 19 83 31 85
GBTI 41 46 81 46 81
JWS 52 64 82 64 82
Kashf Bank 5 5 98 5 98
Kashf Foundation 100 100 100 100 100
Khushali Bank 100 28 100 100 100
NRSP 100 95 100 100 100
NRSP Bank 30 7 100 30 100
Orangi 100 30 100 100 100
Orix Leasing 80 100 82 100 100
POMFB 20 16 54 20 54
PRSP 30 21 78 30 78
RCDS 53 62 83 62 83
SAFWCO 74 42 73 74 82
SRSO 72 95 100 95 100
SRSP 100 100 100 100 100
Sungi 100 100 100 100 100
TMFB 4 18 100 18 100
TRDP 100 77 99 100 100
Source: Authors' own calculations.
Table 3
DEA Efficiency of MFIs Based on CCR, Input Oriented Models
DMUs A1 A2 A3 A4 B1 B2 B3
Akhuwat 54 30 34 13 36 36 35
Apna MF Bank 81 66 5 7 37 55 3
ASA Pakistan 100 89 82 88 54 90 69
Asasah 77 48 70 73 12 14 17
BRAC-PAK 73 85 48 53 16 34 16
Buksh Foundation 60 100 24 19 3 10 2
CSC 61 77 28 35 19 45 14
CWCD 78 72 27 7 22 37 12
DAMEN 77 80 26 36 51 98 26
FFO 74 68 59 63 22 37 27
FMFB--Pakistan 39 47 8 5 29 64 9
GBTI 31 36 7 11 25 55 9
JWS 68 66 31 37 27 49 19
Kashf Bank 57 71 0 0 16 36 0
Kashf Foundation 81 71 63 68 40 65 48
Khushhali Bank 61 51 27 9 32 49 22
NRSP 65 57 38 35 62 100 55
NRSP Bank 51 59 11 3 37 80 12
Orangi 65 29 42 5 100 82 100
Orix Leasing 60 48 39 41 39 59 40
POMFB 17 38 5 2 9 38 5
PRSP 27 47 17 10 23 72 22
RCDS 58 68 30 32 26 57 20
SAFWCO 63 55 47 26 34 55 39
SRSO 86 55 40 45 57 67 40
SRSP 66 40 80 75 18 20 33
Sungi 85 75 100 100 46 75 84
TMFB 53 64 1 4 32 72 1
TRDP 82 55 68 45 64 78 81
DMUs B4 C1 B2 C3 C4 ABC1 ABC2
Akhuwat 16 19 8 22 15 61 37
Apna MF Bank 6 24 16 2 7 87 77
ASA Pakistan 87 21 16 32 62 100 100
Asasah 21 24 13 41 77 84 56
BRAC-PAK 21 13 13 16 32 73 93
Buksh Foundation 2 10 13 7 10 60 100
CSC 20 27 28 23 52 70 92
CWCD 4 14 11 9 4 78 80
DAMEN 44 61 52 38 96 100 100
FFO 34 15 12 22 43 74 76
FMFB--Pakistan 6 44 44 17 17 58 73
GBTI 16 28 27 12 33 42 56
JWS 27 29 23 24 52 77 80
Kashf Bank 0 49 49 0 1 76 90
Kashf Foundation 62 26 19 38 74 88 83
Khushhali Bank 9 61 41 50 31 83 65
NRSP 60 16 12 17 29 85 100
NRSP Bank 4 42 40 17 7 68 82
Orangi 14 84 31 100 21 100 84
Orix Leasing 49 44 29 53 100 77 60
POMFB 2 16 29 9 6 23 50
PRSP 15 13 18 15 15 34 73
RCDS 26 29 28 27 53 67 83
SAFWCO 25 23 16 32 31 70 65
SRSO 55 43 23 37 76 100 69
SRSP 37 17 9 39 66 69 46
Sungi 100 13 9 28 50 85 84
TMFB 4 100 100 4 23 100 100
TRDP 64 27 15 40 49 98 80
DMUs ABC3 ABC4 ABC12 ABC34 ABC1234
Akhuwat 52 22 61 52 61
Apna MF Bank 6 10 88 10 88
ASA Pakistan 99 100 100 100 100
Asasah 97 100 84 100 100
BRAC-PAK 52 57 93 57 93
Buksh Foundation 25 20 100 25 100
CSC 47 62 92 62 92
CWCD 30 8 80 30 80
DAMEN 54 96 100 96 100
FFO 67 71 76 71 76
FMFB--Pakistan 18 17 73 24 73
GBTI 1 33 56 33 56
JWS 50 64 80 64 80
Kashf Bank 0 1 90 1 90
Kashf Foundation 91 100 90 100 100
Khushhali Bank 57 26 86 63 87
NRSP 6 60 100 65 100
NRSP Bank 23 7 82 23 82
Orangi 100 24 100 100 100
Orix Leasing 79 100 77 100 100
POMFB 12 6 50 12 50
PRSP 30 21 73 30 73
RCDS 51 62 83 62 83
SAFWCO 73 41 71 73 81
SRSO 70 93 100 93 100
SRSP 100 96 69 100 100
Sungi 100 100 85 100 100
TMFB 3 16 100 16 100
TRDP 100 77 98 100 100
Source: Authors own calculation.
Table 4
DEA Super Efficiency of MFIs Based on CCR Input Oriented Models
DMUs A1 A2 A3 A4 B1 B2 B3 B4
Akhuwat 54 30 34 13 36 36 35 16
Apna MF BK 81 66 5 7 37 55 3 6
ASA Pakistan 117 89 82 88 54 90 69 87
Asasah 77 48 70 73 12 14 17 21
BRAC-PAK. 73 85 48 53 16 34 16 21
Buksh Found 60 112 24 19 3 10 2 2
CSC 61 77 28 35 19 45 14 20
CWCD 78 72 27 7 22 37 12 4
DAMEN 77 80 26 36 51 98 26 44
FFO 74 68 59 63 22 37 27 34
FMFB--Pak 39 47 8 5 29 64 9 6
GBTI 31 36 7 11 25 55 9 16
JWS 68 66 31 37 27 49 19 27
Kashf Bank 57 71 0 0 16 36 0 0
Kashf Found 81 71 63 68 40 65 48 62
Khushali Bk 61 51 27 9 32 49 22 9
NRSP 65 57 38 35 62 102 55 60
NRSP Bank 51 59 11 3 37 80 12 4
Orangi 65 29 42 5 157 82 119 14
Orix Leasing 60 48 39 41 39 59 40 49
POMFB 17 38 5 2 9 38 5 2
PRSP 27 47 17 10 23 72 22 15
RCDS 58 68 30 32 26 57 20 26
SAFWCO 63 55 47 26 34 55 39 25
SRSO 86 55 40 45 57 67 40 55
SRSP 66 40 80 75 18 20 33 37
Sungi 85 75 121 114 46 75 84 115
TMFB 53 64 1 4 32 72 1 4
TRDP 82 55 68 45 64 78 81 64
DMUs C1 C2 C3 C4 ABC1 ABC2 ABC3 ABC4
Akhuwat 19 8 22 15 61 37 52 22
Apna MF BK 24 16 2 7 87 77 6 10
ASA Pakistan 21 16 32 62 117 111 99 108
Asasah 24 13 41 77 84 56 97 106
BRAC-PAK. 13 13 16 32 73 93 52 57
Buksh Found 10 13 7 10 60 112 25 20
CSC 27 28 23 52 70 92 47 62
CWCD 14 11 9 4 78 80 30 8
DAMEN 61 52 38 96 105 126 54 96
FFO 15 12 22 43 74 76 67 71
FMFB--Pak 44 44 17 17 58 73 18 17
GBTI 28 27 12 33 42 56 16 33
JWS 29 23 24 52 77 80 50 64
Kashf Bank 49 49 0 1 76 90 0 1
Kashf Found 26 19 38 74 88 83 91 104
Khushali Bk 61 41 50 31 83 65 57 26
NRSP 16 12 17 29 85 102 61 60
NRSP Bank 42 40 17 7 68 82 23 7
Orangi 84 31 189 21 181 84 209 24
Orix Leasing 44 29 53 104 77 60 79 108
POMFB 16 29 9 6 23 50 12 6
PRSP 13 18 15 15 34 73 30 21
RCDS 29 28 27 53 67 83 51 62
SAFWCO 23 16 32 31 70 65 73 41
SRSO 43 23 37 76 106 69 70 93
SRSP 17 9 39 66 69 46 107 96
Sungi 13 9 28 50 85 84 122 115
TMFB 120 192 4 23 100 136 3 16
TRDP 27 15 40 49 98 80 103 77
DMUs ABC12 ABC34 ABC1234
Akhuwat 61 52 61
Apna MF BK 88 10 88
ASA Pakistan 121 108 121
Asasah 84 106 106
BRAC-PAK. 93 57 93
Buksh Found 112 25 112
CSC 92 62 92
CWCD 80 30 80
DAMEN 128 96 140
FFO 76 71 76
FMFB--Pak 73 24 73
GBTI 56 33 56
JWS 80 64 80
Kashf Bank 90 1 90
Kashf Found 90 104 104
Khushali Bk 86 63 87
NRSP 105 65 108
NRSP Bank 82 23 82
Orangi 181 209 217
Orix Leasing 77 120 120
POMFB 50 12 50
PRSP 73 30 73
RCDS 83 62 83
SAFWCO 71 73 81
SRSO 106 93 106
SRSP 69 107 107
Sungi 85 122 122
TMFB 136 16 136
TRDP 98 103 103
Source: Authors' own calculation.
