The Effect of Increasing Competition on the Efficiency
and Sustainability of Microfinance Institutions
Anjli Vatvani
1467301
February 2010
Master Thesis
Dr Aljar Meesters
Abstract
The thesis analyzes the relationship between increasing competition and the efficiency and sustainability of microfinance institutions, using a dataset of 1318 observations for 435 institutions. Cost efficiency is used as a measure for financial sustainability. I apply stochastic frontier analysis and find significant results for a negative relationship between increasing competition and the efficiency of microfinance institutions. To be exact, a higher profit margin goes along with increased efficiency. A higher profit margin implies more market power and a decreasing level of competition among institutions.
Key words: Microfinance institutions, competition, efficiency, sustainability
1. Introduction
The topic of the thesis is the relationship between increasing competition and the efficiency and sustainability of microfinance institutions, MFIs henceforth. Microfinance is considered one of the most effective and flexible strategies in the fight against global poverty (Grameen Foundation). Microfinance consists mainly of giving small loans to individuals to establish or expand a small and self-sustaining business. The loans are usually less than $200 and the individuals are mostly women. These borrowers do not have access to credit by the regular banking system due to credit rationing. Since the late 1970s, these borrowers have increasingly gained access to small loans with the help of microfinance programmes, which are supported by MFIs (Hermes and Lensink, 2007a). Microfinance programmes have been initiated in many developing countries all over the world.
selection and moral hazard problems concerning individual-based lending is the requirement of adequate collateral.
Since MFIs have a social purpose, a large number of these institutions get financial support from NGOs, western donors or commercial banks. Providing microfinance is costly due to the high transaction and information costs that are involved in the business (Hermes and Lensink, 2007a). First of all, the costs of screening and monitoring the activities of the poor and the enforcement of their contracts are high. Moreover, they are not able to put up acceptable collateral. These grounds make lending to the poor relatively unprofitable. The higher costs cannot be passed on to the borrowers, because that would make lending from the MFIs unfeasible. This would in turn stand in the way of the main purpose of MFIs, namely providing credit to the poor. Many MFIs are not financially sustainable and this has become an issue of increasing concern within microfinance (Cull et al., 2007). This thesis studies the supply side of microfinance, thus from the perspective of the MFIs. More explicitly, I examine the effect of increasing competition on the efficiency and sustainability of MFIs.
Recently the microfinance industry has been confronted with a number of developments. One of these developments is increasing competition among MFIs (Hermes et al., 2008). The quantity of MFIs with reporting data in 1997 amounted to 57. In 2007 this number increased to 1178 MFIs (www.mixmarket.org). A point to be noted is that not every MFI has reporting data from its commencement. Hence, the numbers merely give an indication and do not give a complete picture of the increase in competition among MFIs. The commercialization of microfinance has attracted commercial banks and investors to invest in the industry. The involvement of traditional commercial banks in microfinance is growing rapidly around the world. In several developing countries, large state banks and private banks have started to provide services related to microfinance (Hermes and Lensink, 2007b).
rate charged to borrowers. This is why MFIs must strive to maintain their costs as low as possible. Here it is assumed that efficiency eventually leads to sustainability, which implies that an MFI can stand on its own feet. In contrast, increasing competition may result in higher levels of indebtedness of the borrowers as they may take up multiple loans from different sources at the same time (McIntosh et al., 2005 and Vogelgesang et al., 2003). Multiple loan taking is possible due to a reduction in the amount of credit rationing that takes place. This phenomenon may lead to lower repayment rates, which can endanger the sustainability of the programmes in the long run. In the next section, the terms efficiency and sustainability, with respect to MFIs, will be explained more elaborately.
From a policy perspective it is extremely important to apprehend whether and to what extent increasing competition, which focuses on strengthening market forces, contributes positively or negatively to the efficiency of MFIs. The aim of this study is to bridge this gap in the literature and to examine the relationship between competition and the efficiency and sustainability of MFIs in a correct and systematic manner, using a large dataset containing more than 1300 observations. Stochastic frontier analysis (SFA), a method which has not been used widely in microfinance, is applied to measure the efficiency of each MFI. To determine competition two methods, namely the Lerner Index and Hirschman-Herfindahl Index, are employed. The data on MFIs is collected from the Microfinance Exchange website (www.mixmarket.org) and the observed period is January 1997 to December 2007. To my knowledge, a study of this kind has not been conducted as yet.
2. Literature Review
Microfinance has experienced a strong success in terms of the number of borrowers as well as lenders. The quantity of MFIs rose from 618 to 3133 between December 1997 and December 2005. The amount of borrowers increased from 13.5 million to 113.3 million throughout the same period (Daley-Harris, 2006). At present, in the face of increasing competition, the industry appears to have reached a stage where MFIs need to implement new strategies to maintain a good repayment performance by their borrowers (McIntosh et al., 2005). For instance, MFIs can offer long-term services to its borrowers to maintain high repayment incentives. Microfinance took off in 1976 when Mohammad Yunus established the Grameen Bank of Bangladesh (Hermes and Lensink, 2007b). In the first decade of growth in microfinance, MFIs were mostly operating as local monopolists and often divided up countries into areas where each operated exclusively. As more and more MFIs started entering the business, monopolistic positions fell and today in many countries MFIs are found competing for the same borrowers (McIntosh et al., 2005). By the end of 1990s, MFIs were increasingly overlapping in their geographical coverage. Moreover, more profit-oriented MFIs have entered the field, changing the nature of competition between MFIs and enlarging the choices for borrowers.
To start with, I shall explain how competition can have a negative impact on the sustainability of MFIs. The increasing number of MFIs has lead to an immense inflow of capital in the microfinance market during the last 15 years (Vogelgesang et al., 2003). This implies an increase in the supply of loans by MFIs. Initially, an MFI can select borrowers that can generate high returns with low risks. Eventually as these borrowers are served, borrowers with higher risks are considered as well. This change in the borrower structure towards riskier borrowers should lead to increasing default rates. This in turn can endanger the sustainability of the MFIs. To be precise, the kind of sustainability referred to here is financial sustainability. An MFI that is financially sustainable is self-sufficient; it can reach its target clientele and cover administrative and other costs (Srinivas, accessed 2009). The previously mentioned finding by Vogelgesang et al. (2003) is also supported by Marquez (2002), who analyzes competition in the banking industry. Marquez (2002) draws attention to the fact that competition reduces the ability of screening of the incumbent bank, which in turn increases the share of low quality borrowers among clients. The intuition is that a larger number of competing banks goes in conjunction with dispersion of borrower-specific information, because every bank is notified about a smaller pool of borrowers. This in turn reduces banks’ ability of screening as more low-quality borrowers obtain financing.
loan to smooth the timing of repayment of loans and to maintain cash flow. Moreover, combining a number of smaller loans leads to a lower overall cost of credit, due to declining-balance interest calculations that are common in the industry.
In contrast to the negative effect, increasing competition can have a positive impact on the repayment behavior of borrowers as well. The analysis of payments, done by Vogelgesang et al. (2003), shows that a borrower with given characteristics is more likely to have a better repayment behavior at a time and in a branch with higher competition and a higher supply of microloans. Vogelgesang et al. (2003) present two possible explanations for this finding. To start with, in an environment with a high availability of microloans, borrowers with one other loan, for example, are more likely to pay on time. The borrowers could possibly be more aware of the significance of timely repayment in an environment where microloans are provided on a routine basis. Moreover, the awareness of the probable negative incentive effects of increasing competition and supply by MFIs could have led to the development of higher incentives for repayment or more efficient screening to compensate.
Hence, Vogelgesang et al. (2003) show that there are two opposing effects of increasing competition on the repayment behavior of borrowers. The evidence that competition has caused an increase in late payments and defaults in the study by Vogelgesang et al. (2003) is mixed. On one hand, higher competition and supply lead to increasing levels of indebtedness which result in difficulties in repayment. On the contrary, borrowers with a given level of obligations may pay on time in such an atmosphere. Similar to Vogelgesang et al. (2003), McIntosh et al. (2005) find that borrowers take loans from multiple sources at the same time. McIntosh et al. (2005) study the increasingly intense interactions between MFIs to analyze how borrowers respond to competition between different types of MFIs. The dataset is from Uganda’s largest incumbent MFI, specifically FINCA that gives loans in the form of village banking, and consists of 780 groups during the years 1998 to 2002.
perspective of one country. McIntosh et al. (2005) study the effect that competition has on borrower behavior and find that increasing competition does not lead to a change in the rate of client dropout. But competition from individual and solidarity group MFIs in particular, induces a decline in repayment performance of FINCA borrowers which, as mentioned above lends in the form of village banking. Thus, borrowers are more prone to taking multiple loans from a different kind of MFI at the same time. However, this result doesn't hold for other village banking MFIs. FINCA is less affected by this phenomenon from other MFIs of the same type. A possible reason behind this is that frequent meetings by village banking MFIs make multiple loans difficult to obtain at the same time. Thus, there seems to be an advantageous impact on repayment behavior by having many of village banking MFIs, in this case. An advantage of the study’s results by McIntosh et al. (2005) is that they are not subject to specification. Moreover, the study applies robust methodologies, al though the data used are relatively noisy.
prices. For example, the MFIs in Bolivia have become more efficient and this is a country which has faced an increase in the level of competition.
In addition, Hermes et al. (2008) mention certain developments which have contributed to the improvement of the efficiency and sustainability of microfinance. First, the latest banking technology, for instance the use of ATM machines, mobile phones and the internet have facilitated a reduction in costs and an improvement in the delivery of services. Furthermore, the liberalization of financial markets in numerous developing countries, as well as the implementation of regulations aimed at improving the stability in the microfinance industry. Moreover, the commercialization of microfinance requires MFIs to become financially sustainable as well as improve their efficiency. The summarized developments are relevant for the efficiency of MFIs. The developments all are means through which MFIs can become more cost efficient and eventually attain financial sustainability.
Al though there seems to be no literature available on the effects of competition among MFIs on their efficiency, this does seem to be the case for the banking industry. In a recent study, Casu and Girardone (2004) study the banking industry in the European Union and use 11.000 observations between 1997 and 2003. The evidence that Casu and Girardone (2004) find is a twofold one. They find that increasing competition results in more efficiency. Banks have attained higher efficiency through rationalization processes and cost cutting. However, Casu and Girardone (2004) also find that increased efficiency does not lead to a more competitive banking system. This has to do with the fact that the most cost efficient banks have tried to improve profitability and expand by acquiring less efficient banks. This in turn reduces the amount of competitors, implying a higher concentration ratio in the industry. Casu and Girardone (2004) conclude that the relationship between competition and efficiency is not a clear-cut one.
Overall, smaller banks have become less efficient and larger, foreign-owned banks have turned out to be more efficient in the face of increasing competition. The larger banks are possibly better able to exploit market power and benefit from economies of scale. Casu and Girardone (2004) and Hauner and Peiris (2005) both apply data envelop analysis to compute efficiency.
It can be concluded that none of the above studies explicitly study the effect of increasing competition on the efficiency and sustainability of MFIs. Al though a number of studies, such as McIntosh et al. (2005) and Vogelgesang et al. (2003), question the effect that competition has on the behavior of borrowers, which in turn has implications for the sustainability. Moreover, the majority of studies apply data based on one area, which have been acquired from one country or one region. A comprehensive study is hence of highest importance. The aim of this study is to bridge this gap in the literature and to examine the relationship between competition and the efficiency and sustainability of MFIs, using a large dataset of over 1300 observations during the years 1997-2007. This leads to the following research question:
Does an increase in competition among MFIs have an effect on the efficiency and sustainability of these institutions?
The subsequent null-hypothesis and alternative hypothesis can be derived.
H0: An increase in competition among MFIs does not have an effect on the efficiency and sustainability of these institutions.
H1: An increase in competition among MFIs does have an effect on the efficiency and sustainability of these institutions, which could be either a positive or negative one.
3. Methodology
some advantages compared to DEA. First of all, DEA assumes all deviations from the frontier are due to inefficiency. If any noise is present, it is likely to influence the placement of the DEA frontier to a larger extent than would be the case with the stochastic frontier approach. However, SFA is a parametric technique that controls for measurement errors and other random effects. Though, a disadvantage of a parametric model is that the technology must be specified. Another benefit of the SFA is that tests of hypotheses regarding the existence of inefficiency as well as the structure of the production technology can be performed in a SFA (Coelli, 2005). Recently it has been proven that this can be done with DEA as well, but in SFA it is done in a more intuitive manner. Considering the advantages as well as the shortcomings, SFA is selected for measuring efficiency of production.
In SFA models, cost efficiency is measured in terms of how close the actual costs of the lending activities of an MFI are to what the costs would have been of an MFI that is efficient, given it produces identical output under the same conditions. The efficiency of an organization consists of two components: technical efficiency and allocative efficiency (Coelli, 2005). Technical efficiency reflects the ability to obtain maximal output from a given set of inputs. Moreover, allocative efficiency casts the ability of an MFI to use the inputs in optimal proportions, given their respective prices. The economic efficiency which is composed of both the technical and allocative efficiency, hence the complete picture, is of importance here.
model is comparatively most suitable here and is frequently used in research (Abdul-Majid et al., 2009 and Sedik et al., 1999); therefore I select the model.
The Battese and Coelli (1995) model specification for a stochastic cost frontier is as:
t i t i t i t i t i t i
C
y
w
q
u
v
C
,(
,,
,,
,;
)
, ,ln
=
β
+
+
(1)where the dependent variable Ci,t is the total cost of MFI i at time t. C(yi,t,wi,t,qi,t;β)
represents the cost frontier, in which yi,t is the logarithm of output of MFI i at time t,
t i
w, is a vector of the logarithm of input prices of MFI i at time t and qi,tare MFI specific control variables. Furthermore, β is a vector of all estimated parameters. The term µi,t measures cost inefficiency and is independent and identically distributed with
a truncated normal distribution. νi,t captures measurement errors and random effects
and is distributed as a standard normal variable. The following formulas belong to the latter two variables.
)
,
(
~
, 2 ,t it u iN
m
u
+σ
(2)
)
,
0
(
~
2 ,t v iN
v
σ
(3) To model the inefficiency of an MFI, the following model is applied:
t i n n n t i
z
m
,=
δ
0+
∑
δ
,,(4) where mi,t measures the inefficiency of MFI i at time t, the deltas are the coefficients
and z stands for variables that determine the inefficiency. To solve equations (1) and (2) the method of maximum likelihood is employed. The equations are subsequently solved in one step.
and the interest expenses of holding money (R) are the input variables, while the gross loan portfolio (GLP) of an MFI represents total output. The price of a unit of labor is measured as the total operating expenses per employee of an MFI. The interest expense per unit of deposits held is the MFI’s total financial expenses per dollar of deposits. The nature of the cost function is that of a translog one. In this cost function, the individual input and output variables, the square of these variables and combinations of these variables are considered. The variables in the cost function are taken in logs and the cost function can be specified as:
t i t i t i t i t i j j t i t i t i t i t i t i t i t i t i t i t i t i t i
v
u
LLR
EQUITY
MFITYPE
GLP
R
GLP
SALARY
R
SALARY
GLP
R
SALARY
GLP
R
SALARY
TC
, , , 15 , 14 , 13 10 , . 9 , , 8 . , 7 2 , 6 2 , 5 2 , 4 , 3 . 2 , 1 0 ,)
ln(
)
ln(
)
ln(
)
ln(
)
ln(
)
(
)
ln(
)
ln(
)
ln(
)
ln(
)
ln(
)
ln(
)
ln(
+
+
+
+
+
+
+
+
+
+
+
+
+
+
=
∑
=β
β
β
β
β
β
β
β
β
β
β
β
β
(5)As stated previously, the aim of the thesis is to examine the relationship between competition and the efficiency and sustainability of MFIs. An empirical model is specified to study the relationship between competition and efficiency and sustainability. The measure of inefficiency of an MFI, mi,t which has already been identified, forms
the dependent variable in the model. Several measures of competition and control variables that might influence the inefficiency should be included as well.
Regarding measures of competition, there are various methods to measure competition. For instance, the three measures of competition used by McIntosh et al. (2005) are the presence of any competitor known to the group, the number records of competitors and the proximity. The final measure estimates the distance of the closest competitor to the FINCA group in kilometers. Vogelgesang et al. (2003) calculate the fraction of clients with concurrent loans from other MFIs by branch or by quarter. These measures are not suitable for a study at a large scale. These measures are rather difficult to obtain, especially in a study which is a cross-country analysis and to be exact does not examine the effect of competition from the perspective of simply one MFI or country. Moreover, there are alternatives which are equally good. Hence, I shall now discuss several alternative techniques to measure competition that could be employed to financial institutions. Berger et al. (2008) use alternative measures of bank competition, namely the Lerner Index and Herfindahl-Hirschman Index using loans as well as deposits. I begin with the Lerner Index. The Lerner Index is based on the deviation between price and marginal costs. The Lerner Index measures the market power of an organization and is calculated as:
i i i
i
P
MC
P
L
=
(
−
)
/
(6)
where Li is the Lerner Index of institution ι, P is the market price set by the relevant
of perfect competition, the price is set equal to marginal cost, resulting in a value of 0 of the Lerner Index. In this case the institutions are price takers in the market. On the contrary, when the Lerner Index has the value 1, it indicates a monopolist who experiences no competition and could set the highest price to maximize profits.
Another measure of competition is the Hirschman-Herfindahl Index. The Index gives an indication of the market concentration and is measured at the national or industry level (Berger et al., 2008). The thought behind the measure is that when the level of competition in a given market is high, there is a low concentration ratio. The index is measured as:
HHI
=
∑
= n i i MS 1 2(7)
where
HHI
is the Hirschman-Herfindahl Index, MS is the market share of institution i in the market and n is the amount of institutions. Analogous to the Lerner Index, the Hirschman-Herfindahl Index can obtain a value ranging from 0 to 1. A high ratio implies that a number of firms have large market shares and dominate the market. Due to the sum of squares in the model, greater emphasis is placed on larger firms. In the extreme case of monopoly, the value 1 point towards an institution which holds 100% of market share. On the other hand, a market consisting of many institutions and each with a small market share, results in a Hirschman-Herfindahl Index close to 0.An additional way to measure competition is called Profit Elasticity (Boone, 2004). The Profit Elasticity calculates the fall in the percentage of an institution’s profit due to a percentage increase in the marginal costs. The theory suggests that in a competitive environment, an increase in cost leads to a reduction in the profits either by an increase in the selling price, which leads to a decrease in demand, or by a lower price-cost margin. Therefore, in terms of profit, inefficient firms are affected more severely. The Profit Elasticity is calculated as:
i
i
ln
MC
where π is the profit of institution i, MC equals the marginal cost and β estimated as the profit elasticity; the percentage drop in profit caused by a percentage increase in marginal costs. High levels of β correspond with a high level of competition, as profits are affected to a further extent due to an increase in marginal costs.
In the previous paragraphs, the different measures of competition, which can be applied to financial institutions, are discussed. Each measure has its own advantages as well as limitations. In the last methodβ must be estimated for each MFI. This can be done by using the method of Ordinary Least Squares. However, there are not enough data points to obtain reliable estimates. For this reason, I leave out this measure of competition for my analysis.
Regarding the Lerner Index, the price in the formula refers to the interest rate which is charged to borrowers. The computation of the marginal cost requires, among others, the evaluation of the risk premium charged to properly measure the marginal cost for different types of bank loans, Oliver et al. (2006). In practice, however, this calculation is difficult, because the risk premia of different loans are not directly observable. Whereas the Lerner Index is the price-cost margin, the profit margin gives an indication of how much a firm earns with the sale of each unit sold. The profit margin is calculated by dividing the Net Operating Income by the Financial Revenue (www.mixmarket.org). Here average costs are used instead of marginal costs, which are used for calculating the Lerner Index. The calculation of the profit margin closely coincides with that of the Lerner Index and hence appears to be a good approximation. Therefore, the profit margin is employed as a proxy for the Lerner Index due to availability of data directly on the MixMarket.org website. Oliver et al. (2006) show a disadvantage of using the profit margin instead of the Lerner Index. In their study, the estimated welfare loss from market power resulted to be twice as high with the use of the profit margin in the calculation as compared to the Lerner Index. Nevertheless, the profit margin is the most suitable and available alternative.
as well as the gross loan portfolio of individual MFIs. These variables are selected because they give a fair impression of the scale and size of an MFI. The Hirschman-Herfindahl Index is estimated per country-year. This approach is chosen to account for changes in market power over time. A disadvantage is that the case of a single available observation in a given year in a given country, may lead to a wrong interpretation of the results. But I attempt to solve this problem by running the model with and without these single observations. I shall return to this in the results section.
So, the Lerner Index and the Hirschman-Herfindahl Index are selected as measures for competition. An advantage of alternative measures for competition is that it serves as a check for the robustness of the results. The general specification of the inefficiency equation is calculated as:
t i t i i t i t i t i t i
AGE
LOANTYPE
HHI
HHI
PM
m
, , 7 .. 4 , 3 , 2 , 1 0 ,+
+
+
+
+
=
=δ
δ
δ
δ
δ
(9)where mi,t measures the extent to which an MFI is considered to be inefficient. mi,t
estimates the first moment of the inefficiency distribution for MFI i at time t. The first two variables are measures of competition, as discussed previously. As the dependent variable measures the degree of inefficiency, I expect either a positive or a negative relation between both measures of competition and the inefficiency. As indicated earlier, the profit margin is used as an approximation of the Lerner Index. A higher profit margin as well as a higher Hirschman-Herfindahl Index, indicate more market power and a declining level of competition and vice versa.
of years from the time when the MFI was established. Regarding AGE, two scenarios are possible. On one hand, it is possible that the older MFIs work more efficiently due to experience. On the contrary, it is also likely that new MFIs can profit from the knowledge which has been gained and accumulated with time by the older MFIs, making the new MFIs more advanced in terms of efficiency. In this sense, old MFIs do not lose their knowledge, however old MFIs may already have invested in less efficient technology, whereas new MFIs invest directly in the latest, more efficient technology.
4. Data
The data are collected from the MixMarket website, the source for data on the global microfinance sector (www.mixmarket.org). The period of study is January 1997 to December 2007 and the entire sample is composed of 1318 observations for 435 MFIs. For most of the MFIs, it is the case that data is available for just a part of the time span studied for the thesis. This is why every year is considered to be an independent value. The correlation results of all variables employed can be found in the table in the appendix.
Mean
Standard
deviation Minimum Maximum Observations
SALARY 12154.240 12707.770 0 431489.300 3280
R 1.897 15.086 0.000 380 1360
GLP 1.71E+07 9.72E+07 464 3.04E+09 4472
EQUITY 0.381 0.423 -18.352 1.034 4489 LLR 0.041 0.064 -0.123 0.993 4391 PROFIT MARGIN -0.384 7.089 -413.987 1 4317 HHI BORROWERS 0.396 0.265 0.056 1 4763 HHI GLP 0.413 0.252 0.049 1 4340 AGE 10.546 9.573 0 112 4753
Table 1: Descriptive statistics for the variables in the cost function, the numerical control variables and the measures of competition.
Table 2 provides descriptive statistics for the profit margin with respect to the different types of loans. The Hirschman-Herfindahl Indexes have been left out here as they have been estimated on the basis of country-years, while loan types are concerning individual MFIs. The individual loan type seems to have the highest profit margin, implying the highest amount of market power and a lower level of competition, compared to the rest of the loan types. Furthermore, the average profit margin for village lending type is the lowest, indicating a lower level of market power and a higher level of competition.
INDIVIDUAL GROUP VILLAGE ALLTYPE Total
PROFIT Mean 0.013 -0.005 -0.012 0.001 -0.384
MARGIN Standard Deviation 0.074 0.223 0.325 0.296 7.089
Observations 421 200 183 583 1387
Table 2: Descriptive statistics for the profit margin with respect to the different types of loans.
Type of MFI Non-bank Rural bank Bank Non-governmental organization Cooperative Other
Graph 1: Types of MFIs
5. Results
This section presents the results of the relation between increasing competition and the efficiency and sustainability of MFIs. The results have been estimated by applying the SFA model by Battese and Coelli (1995). In this model the cost frontier and the inefficiency equation are calculated at once. The focus here lies on the inefficiency equation and more specifically on the relation between competition and efficiency. Different models are run, each with a distinct specification of the inefficiency equation, to clearly examine the sensitivity of the results. I first start with a simple model. If econometric problems are found, the model is made more complicated with the intention to solve the occurred problem. This is also called a specific-to-general approach, Hoover and Perez (1999).
run with and without adjustments made to the Hirschman-Herfindahl Index based on the number of active borrowers as well as the gross loan portfolio of individual MFIs. The diverse models can be found in table 3.
Inefficiency equation 1 2 3 4 5 6 PROFIT MARGIN -0.442*** -0.448*** -0.452*** -0.480*** (0.030) (0.033) (0.033) (0.034) HHIBORROWERS 1.893 0.088 0.060 -0.038 (2.831) (0.076) (0.076) (0.082) HHI GLP 20.643 -0.029 0.005 0.115 (63.047) (0.083) (0.082) (0.088) INDIVIDUAL 0.012 0.017 (0.037) (0.038) GROUP -0.144** -0.137** (0.062) (0.062) VILLAGE 0.229** 0.240** (0.112) (0.111) ALLTYPE 0.058 0.078** (0.038) (0.039) AGE 0.005*** 0.005*** (0.001) (0.001) CONSTANT 1.459*** -9.000 -294.993 1.452*** 1.374*** 1.382*** (0.046) (13.032) (427.945) (0.051) (0.053) (0.055) SIGMA -1.805*** 0.879 4.241*** -1.813*** -1.843*** -1.860*** (0.041) (1.251) (1.437) (0.043) (0.043) (0.045) LAMBDA 26.142 2.905** 6.269*** 26.088 25.838 24.822 (237.544) (1.300) (1.442) (303.511) (293.447) (208.367) OBSERVATIONS 1316 1265 1156 1147 1147 1080
Table 3: Estimation results for the cost frontier and inefficiency equation using a dataset of 1318 observations for 435 MFIs. The standard errors can be found in parentheses. * , ** and ** stand forrespectively, significant at 10%, 5% and 1%.
The first column in the table presents the results derived from incorporating solely the profit margin in the model. In the following two models the Hirschman-Herfindahl Index, based on the number of active borrowers and the gross loan portfolio respectively, are introduced as the independent variable. In the fourth model all measures of competition are included. The fifth model consists of the measures of competition accompanied by the control variables. The final model takes account of changes in the Hirschman-Herfindahl Index, as explained above and in the methodology section.
The coefficients for SALARY and GLP are positive and significant at all times. This is as expected. However, the coefficient for R is negative at times as well as always insignificant. The interaction of R with SALARY is always positive and significant at times. The interaction with GLP is mostly negative and the quadratic term appears to have an insignificant effect. R is perhaps negative due to subsidies received by NGOs, western donors or commercial banks. Except for R, it can be said that the specification of the cost frontier matches theory quite well. The dummies for the different types of MFIs are significant, except for in the last model. The EQUITY and LLR, which stand for the equity to total assets ratio and loan loss reserves divided by gross loans outstanding respectively, are significant at all times. This shows that the risk taking strategies as well as the type of MFI certainly influence the cost frontier.
The second part of the table turns to the inefficiency equation. In all cases, the coefficients for the Hirschman-Herfindahl Index, based on the number of active borrowers as well as the gross loan portfolio, prove to have an insignificant effect. On the contrary, the coefficients for the profit margin are significantly negative. The results for the profit margin strongly support the negative relation between increasing competition and the efficiency and sustainability of MFIs. As the profit margin increases, the inefficiency of the MFIs reduces. A higher profit margin implies a decreasing level of competition. In other words, increasing competition leads to increasing inefficiency on the part of MFIs. This finding is not in line with the null-hypothesis that increasing competition among MFIs does not have an effect on the efficiency and sustainability of the MFIs.
encourages screening, monitoring and enforcement of contracts among borrowers, hereby reducing the agency costs of the MFI (Hermes and Lensink, 2007a and Raj, 1998). Therefore, screening and monitoring of group members is efficient and economic, due to existing social ties and the short distance between the houses of the members. Hence this finding is in line with the existing literature. Moreover, the coefficient for AGE is positive and significant. This indicates that new MFIs are more efficient. More recently established MFIs may have benefitted from knowledge acquired by older MFIs throughout the years. This is how the newer MFIs may advance when it comes to the efficiency of their activities.
So, I find evidence for a negative relation between increasing competition and the efficiency and sustainability of MFIs. The results remain significant also after control variables are taken account of. These findings are in line with the results by McIntosh et al. (2005) and Vogelgesang et al. (2003), although these studies analyze repayment performance. These two studies seem to be most relevant to this work. They find that increasing competition goes together with an increase in the supply of loans by MFIs. Moreover, increasing competition could lead to an increase in the share of low quality borrowers. Furthermore, they find evidence for ‘douple-dipping’ among borrowers.
This study is an important contribution to the existing literature. To my knowledge, it is the only one on the effect of increasing competition on the efficiency and sustainability of MFIs. First of all, it's a cross-country analysis, instead of one that focuses on one particular country or MFI. Moreover, I use different measures for competition. Furthermore, I apply cost efficiency of as a measure for the sustainability of MFIs.
6. Conclusions
between increasing competition and the efficiency and sustainability of MFIs. This finding is not in line with the hypothesis and leads to the rejection of the null-hypothesis.
Moreover, the results seem to support the findings of McIntosh et al. (2005) and Vogelgesang et al. (2003). These authors find that increasing competition is associated with an increase in the supply of loans by MFIs. Moreover, rising competition could lead to an increase in the share of low quality borrowers. Furthermore, they find evidence for ‘douple-dipping’ among borrowers. The outcome of the thesis suggests that increasing competition has an adverse effect on the efficiency and in turn on the sustainability of MFIs. Overall, the disadvantages of increasing competition seem to outweigh the benefits. Specifically, growing competition could lead to more cost efficiency and eventually bring about the sustainability of MFIs. However it could also be otherwise, increasing competition may result in higher levels of indebtedness of the borrowers as they may participate in an activity called ‘double-dipping’. The finding could pose a significant problem in the view of the large growth in the number of MFIs and the commercialization of the industry. This could also imply that in the future MFIs will become more dependent of the subsidies that they obtain. Improved government regulation and coordination among MFIs are probably needed in the industry. According to McIntosh et al. (2005) increasing competition among MFIs requires institutional innovations for information sharing on the levels of client indebtedness to avoid declining repayment rates. A more formal mechanism for information sharing is needed than now exists in countries, such as a credit bureau. Moreover, the reinforcement of timely repayment is important for the sustainability of MFIs. A number of branches in Bolivia strongly reinforced timely repayment in 1999 and 2000 and the enforcement had the desired effect of clients paying more punctually in those branches (Vogelgesang et al., 2003).
period of study. Further research could try to incorporate the year as a control variable to gain additional insights. Furthermore, I obtain significant results for one measure of competition. Perhaps future research could apply various measures as a check for the robustness of the results. Moreover, a similar analysis of the effect of increasing competition on the efficiency and sustainability of MFIs could be carried out on a country level to gain country-specific insights.
References
Abdul-Majid, M., Saal, D.S. and Battisti, G. (2009). ‘Efficiency in Islamic and conventional banking: An international comparison’, Journal of Productivity Analysis, vol. 10, pp. 1-19.
Aigner, D., Lovell, C.A.K. and Schmidt, P. (1977). ‘Formulation and estimation of stochastic frontier production function models’, Journal of Econometrics, vol. 6, pp. 21-37.
Battese, G.E. and Coelli, T.J. (1995). ‘A model for technical inefficiency effects in a stochastic frontier production function for panel data’, Empirical Economics, vol. 20, pp. 325-332.
Berger, A.N., Klapper, L.F. and Turk-Ariss, R. (2009). ‘Bank Competition and Financial Stability’, Journal of Financial Services Research, vol. 35, pp. 99-118.
Boone, J. (2004). ‘A New Way to Measure Competition’, Discussion Paper Center,
Tilburg University, no. 31, pp. 1-34.
Boone, J., Ours, van J. and Wiel, van der H. (2007). ‘How (not) to measure competition’, Discussion Paper Center, Tilburg University, no. 32, pp. 1-48.
Casu, B. and Girardone, C. (2004). ‘Bank Competition, Concentration and Efficiency in the Single European Market’, Working Paper, University of Wales and
University of Essex, pp. 1-28.
Coelli, T.J. ‘A Guide to FRONTIER Version 4.1: A Computer Program for Stochastic Frontier Production and Cost Function Estimation’, Centre for Efficiency and
Productivity Analysis (CEPA) Working Papers, no. 7, pp. 1-33.
Creusen, H., Minne, B. and Wiel, van der H. (2006). ‘Measuring Competition in the Netherlands: A comparison of indicators over the period 1993-2001’, CPB Netherlands
Bureau for Economic Policy Analysis, no. 163, pp. 1-68.
Cull, R., Demirgüç-Kunt, A. and Morduch, J. (2007). ‘Financial Performance and Outreach: A Global Analysis of Leading Microbanks’, The Economic Journal, vol. 117, pp. F107-F133.
Daley-Harris, S. (2006). ‘State of the Microcredit Summit Campaign Report 2006’, pp. 1-66.
Dietsch, M. and Lozano-Vivas, A. (2000). ‘How the environment determines banking efficiency: A comparison between French and Spanish banks’, Journal of
Banking and Finance, vol. 24, pp. 985-1005.
Fries, S. and Taci, A. (2005). ‘Cost efficiency of banks in transition: Evidence from 289 banks in 15 post-communist countries’, Journal of Banking and Finance, vol. 29, pp. 55-81.
Grameen Foundation,
http://www.grameenfoundation.org/what_we_do/microfinance_in_action/?gclid=CKryy c_30pkCFV0A4wodUS_Wug
Grigorian, D.A. and Manole, V. (2006). ‘Determinants of commercial bank performance in transition: An application of Data Envelop Analysis’, Comparative
Economic Studies, vol. 48, pp. 497-522.
Hauner, D. and Peiris, S.J. (2005). ‘Bank Efficiency and Competition in Low-Income Countries: The Case of Uganda’, IMF Working Paper, pp. 1-29.
Hermes, N. and Lensink, R. (2007a). ‘The Empirics of Microfinance: What do we know?’, The Economic Journal, vol. 117, pp. F1-F10.
Hermes, N. and Lensink, R. (2007b). ‘Impact of Microfinance: A Critical Survey’,
Economic and Political Weekly, February 10, pp. 462-465.
Hermes, N., Lensink, R. and Meesters, A. (2008). ‘Outreach and Efficiency of Microfinance Institutions’, Centre for International Banking, Insurance and Finance
(CIBIF), University of Groningen and University of Nottingham, pp. 1-34.
Lensink, R., Meesters, A. and Naaborg, I. (2008). ‘Bank efficiency and foreign ownership: Do good institutions matter?‘, Journal of Banking and Finance, vol. 32, pp. 834-844.
Lozano-Vivas, A., Pastor, J.T. and Hasan, I. (2001). ‘European bank performance beyond country borders: What really matters?’, European Finance Review, vol. 5, pp. 141-165.
Marquez, R. (2002). ‘Competition, Adverse Selection, and Information Dispersion in the Banking Industry’, The Review of Financial Studies, vol. 15, pp. 901-926.
McIntosh, C., de Janvry, A. and Sadoulet, E. (2005). ‘How Rising Competition among Microfinance Institutions affects Incumbent Lenders’, The Economic Journal, vol. 115, pp. 987-1004.
Meeusen, W. and Broeck, van den J. (1977). ‘Efficiency estimation from Cobb-Douglas production functions with composed error’, International Economic Review, vol. 18, pp. 435-444.
MixMarket, www.mixmarket.org
Oliver, A.M., Fumás, V.S. and Saurina, J. (2006). ‘Risk premium and market power in credit markets’, Economics Letters, vol. 93, pp. 450-456.
Pagura, M., Graham, D. and Meyer, R. (2001). ‘Determinants of Borrower Dropout in Microfinance: An Empirical Investigation in Mali’, The Rural Finance program.
Chicago, Illinois, pp. 1-18.
Ray, D. (1998). ‘Development Economics’, Chapter 14: Credit, pp. 578-584.
Sealey, C.W. and Lindley, J.T. (1977). ‘Inputs, Outputs, and a Theory of Production and Cost at Depository Financial Institutions’, Journal of Finance, vol. 32, pp. 1251-1266.
Sedik, D., Trueblood, M. and Arnade, C. (1999). ‘Corporate Farm Performance in Russia, 1991–1995: An Efficiency Analysis’, Journal of Comparative Economics, vol. 17, pp. 514-533.
Vogelgesang, U. (2003). ‘Microfinance in Times of Crisis: The Effects of Competition, Rising Indebtedness and Economic Crisis on Repayment Behaviour’,
Appendix
TC SALARY R GLP BANK COOPE- NON- NGO RURAL- LLR EQUITY YEAR INDIVI- GROUP VILLAGE ALL- AGE PROFIT HHI HHI RATIVE BANK BANK DUAL TYPE MARGIN BRR GLP TC 1.000 SALARY 0.126 1.000 R -0.018 -0.056 1.000 GLP 0.936 0.158 -0.019 1.000 BANK 0.315 0.403 -0.048 0.323 1.000 COOPERATIVE -0.096 -0.005 -0.025 -0.099 -0.307 1.000 NONBANK -0.072 -0.026 0.040 -0.070 -0.257 -0.386 1.000 NGO -0.065 -0.233 0.034 -0.061 -0.192 -0.288 -0.241 1.000 RURALBANK -0.067 -0.167 0.003 -0.074 -0.161 -0.242 -0.202 -0.151 1.000 LLR 0.003 0.042 -0.023 -0.012 0.043 -0.064 0.058 -0.016 -0.013 1.000 EQUITY -0.118 -0.161 0.017 -0.122 -0.137 -0.071 0.250 0.047 -0.127 -0.083 1.000 YEAR -0.012 0.107 0.059 0.037 -0.058 0.050 -0.063 -0.001 0.105 -0.046 -0.075 1.000 INDIVIDUAL 0.210 0.128 -0.047 0.180 0.123 -0.068 -0.022 -0.127 0.135 0.000 -0.143 -0.098 1.000 GROUP -0.044 -0.177 -0.001 -0.045 -0.105 -0.158 0.355 -0.033 -0.083 -0.005 0.258 -0.008 -0.096 1.000 VILLAGE -0.020 -0.071 0.020 -0.024 -0.050 -0.076 0.103 0.063 -0.040 -0.057 -0.053 -0.044 -0.046 -0.026 1.000 ALLTYPE -0.030 -0.035 0.002 -0.015 0.095 -0.143 0.080 0.099 -0.116 -0.040 0.201 -0.150 -0.171 -0.097 -0.046 1.000 AGE 0.608 0.022 -0.015 0.573 0.036 -0.072 -0.121 -0.037 0.262 0.024 -0.163 0.085 0.307 -0.130 -0.047 -0.051 1.000 PROF MARGIN 0.059 0.024 0.009 0.075 0.064 -0.024 -0.012 -0.134 0.126 -0.240 0.009 0.048 0.118 -0.026 -0.046 0.052 0.144 1.000 HHI BRR 0.086 -0.028 -0.055 0.093 0.008 0.064 -0.064 -0.003 -0.039 -0.008 -0.047 -0.046 -0.028 -0.109 -0.025 -0.006 0.036 -0.002 1.000 HHI GLP 0.042 -0.002 -0.045 0.078 -0.001 0.064 -0.069 -0.051 0.037 -0.009 -0.038 -0.008 -0.044 -0.097 -0.024 -0.035 0.010 0.025 0.796 1.000
