Determinants of Margin in Microfinance

Institutions

Beatriz Cullar Fernndez, Yolanda Fuertes-
Calln, Carlos Serrano-Cinca and Begoa
Gutirrez-Nieto

Microfinance institutions (MFIs) lend to the poor, fostering these individuals
financial inclusion. However, microfinance clients suffer from high interest rates,
a type of poverty penalty. Reducing margins and lowering interest rates should
be a target for MFIs with a strong social commitment. This paper analyzes the
determinants of margin in MFIs. A banking model has been adapted to the case
of MFIs. This model has been empirically tested using 9-year panel data. Some
factors explaining bank margin also explain MFI margin, with operating expenses
being the most important factor. Specific microfinance factors are donations and
legal status, as regulated MFIs can collect deposits. It has also been found that
MFIs operating in countries with a high level of financial inclusion have low
margins.


JEL Classifications: G21, C23, R51
Keywords: Microfinance institutions, banking, net interest income, outreach,
financial inclusion.







CEB Working Paper N 12/030
October 2012







Universit Libre de Bruxelles - Solvay Brussels School of Economics and Management
Centre Emile Bernheim
ULB CP114/03 50, avenue F.D. Roosevelt 1050 Brussels BELGIUM
e-mail: ceb@admin.ulb.ac.be Tel. : +32 (0)2/650.48.64 Fax : +32 (0)2/650.41.88
Determinants of Margin in Microfinance Institutions
Beatriz Cullar Fernndez
Department of Accounting and Finance
University of Zaragoza, Spain

Yolanda Fuertes-Calln
Department of Accounting and Finance
University of Zaragoza, Spain

Carlos Serrano-Cinca
Department of Accounting and Finance
University of Zaragoza, Spain

Begoa Gutirrez-Nieto*
Department of Accounting and Finance
University of Zaragoza, Spain

This version: October 2012
ABSTRACT
Microfinance institutions (MFIs) lend to the poor, fostering these individuals
financial inclusion. However, microfinance clients suffer from high interest rates, a type of
poverty penalty. Reducing margins and lowering interest rates should be a target for MFIs
with a strong social commitment. This paper analyzes the determinants of margin in MFIs.
A banking model has been adapted to the case of MFIs. This model has been empirically
tested using 9-year panel data. Some factors explaining bank margin also explain MFI
margin, with operating expenses being the most important factor. Specific microfinance
factors are donations and legal status, as regulated MFIs can collect deposits. It has also
been found that MFIs operating in countries with a high level of financial inclusion have
low margins.
JEL classification: G21, C23, R51
Keywords: Microfinance institutions, banking, net interest income, outreach, financial
inclusion.

*
Corresponding author. Gran Va 2, 50005 Zaragoza, Spain. Telephone: +34 876 554643. Fax: +34 976 761
769. E-mail: bgn@unizar.es
Determinants of Margin in Microfinance Institutions
1. Introduction
Microfinance institutions (MFIs) lend to the poor, who are traditionally excluded
from financial services. Their presence has been a step forward in poverty alleviation as
well as the empowerment of women. However, many MFIs charge very high interest rates,
which is controversial. This practice is an example of poverty penalty: the poor pay high
interest rates to enter the credit market (Prahalad and Hammond, 2002). This paper,
adapting a banking model, studies the determinant factors of MFIs margin. The main
motivation of this paper is to identify the factors driving this margin. If these factors are
known, a way to reduce the margin will be available, and entities with a social mission will
be able to reduce their interest rates.
Since the seminal work by Ho and Saunders (1981), several studies have analyzed the
factors determining bank margin. Allen (1988) and Angbazo (1997) include factors such as
interest and credit risk; Maudos and Fernndez de Guevara (2004) include bank efficiency;
Saunders and Schumacher (2000) study solvency regulations and Carb and Rodrguez
(2007) or Lepetit et al. (2008) incorporate product diversification. Maudos and Sols (2009)
test in a comprehensive model the main additions to the Ho and Saunders (1981) model.
The research question in this paper is as follows: what are the margin determinants in
microfinance? To the best of our knowledge, this question has not been previously studied.
Financial margin is a MFIs performance measure and is included as an independent
variable in several works, such as Mersland and Strm (2008), Mersland and Strm (2009),
Mersland (2009) and Ahlin et al. (2011).
Commercial banks maximize profits. Profits can be maximized by reducing costs or
by increasing revenues. There is a consensus to reduce costs in MFIs. However, there is a
debate over whether to increase revenues in MFIs (Gulli, 1998), which is still unresolved
(Hermes and Lensink, 2011). The financial systems approach emphasizes the idea of MFI
sustainability: if MFIs are simply considered to be banks with poor clients, conventional
banking models could be directly applied to the MFIs case. The poverty lending approach
favors subsidized interest rates. It supports sustainability, but individuals sustainability, not
institutions sustainability.
MFIs have specific characteristics in their funding structure that may explain their
margin. Representing an example are donations, whose role has been widely discussed in
the microfinance literature since Morduch (1999). The relationship between donations and
margin must be studied.
Another specific aspect of microfinance is regulation (Hartarska and Nadolnyak,
2007). Many MFIs only lend; they do not collect deposits because they are not supervised
by the monetary authorities. They only perform half of the banking business. Regulation
allows them to access a cheap funding source: deposits. However, the collection of small
amounts implies high costs. It is important to study how deposits influence the margin.
Another potential thread of research is the study of the relationship between the type
of institution and margin. There are NGOs in the microfinance market, but there are also
conventional banks downscaling into the sector, credit cooperatives and Non-Banking
Financial Institutions (NBFI). Mersland and Strm (2009) and Servin et al. (2012) have
studied the relationship between the type of institution and efficiency.
The contribution of this paper is twofold. First, it formulates a model to explain
margin determinants in microfinance. Second, it empirically tests this model by using
standard panel data techniques with a 9-year sample of MFIs. It includes variables that
influence the margin in commercial banks. It also includes microfinance-specific variables:
funding-related variables, such as donations or deposits, social performance variables and
the type of institution.
The most recent debate in microfinance concerns mission drift. MFIs focused on
financial objectives run the risk of losing their social objectives, Mersland and Strm
(2010). Abnormally high margins can be a type of mission drift. Processing many small
loans can be a justification because the administrative costs of doing so are high. To shed
light on this question, the empirical study in this paper analyzes a subsample of MFIs that
only lend to the poorest; those with an average loan size under $300. These MFIs have not
drifted from their mission in terms of their target but may have drifted from their mission in
terms of their margins, which may go beyond reasonable limits. The model has been tested
with this subsample of pure MFIs as well as with the subsample of MFIs with an average
loan size over $300.
The next section presents the literature revision and describes the theoretical model.
Section 3 justifies the empirical model. Section 4 presents the empirical results. In the final
section, the conclusions are discussed.

2. Theoretical model
The interest margin model used in this paper is based on the original proposal by Ho
and Saunders (1981) and further extensions by Angbazo (1997) and Maudos and Fernndez
de Guevara (2004). The financial institution is considered a risk-averse agent that operates
as a financial intermediary in the loans and deposits market, with the aim of maximizing a
mean-variance objective function in end-of-period wealth. The wealth of the institution (W)
is obtained as the difference between its assets (loans, L and investments in the monetary
market, M) and its liabilities (deposits, D):
W
0
= L
0
-
0
+ H
0
(1)
A period later, this wealth can be expressed as follows:
W
1
= (1+r
I
+Z
I
)( L
0
-
0
) +H
0
(1+r
M
+Z
M
)C=W
0
(1 + r
w
) + (I
0
-
0
)Z
I
+ H
0
Z
M
C
(2)
Where r
I
=

L
L
0
-
D

0
L
0
-
0
represents the profitability of the net loan balance and r
w
=
r
I
L
0
-
0
W
0
+ r
M
M
0
W
0
represents the average profitability of the institution. Z
M
is the institutions
risk from interest rate volatility in the money market, distributed as a random variable
Z
M
_N(0,
M
2
). Z
I
= Z
L
L
0
L
0
-
0
+ Z

0
L
0
-
0
is credit risk with Z

0
L
0
-
0
= u. This risk also
follows a random distribution Z
L
_N(0,
L
2
). C are operating costs associated with loans and
deposits.
The bank sets its interest rates by incorporating a spread over its funds cost on the
money market (r
D
=r
M
-a and r
L
=r
M
+b). The sum of both margins (a+b) is the pure interest
margin. Deposits and loans arrive randomly according to Poisson processes influenced by
the spreads fixed by the entities in their operations.

L
= o - [b

= o + [o (3)

l,d
represents the probability of giving a loan or collecting a deposit. measures the
loans and deposits demand sensibility to margin variations.
First-order conditions for a and b in the utility function allow obtaining the optimal
interest margin (see Maudos and Fernandez de Guevara, 2004, for a full development of the
model):
s = o +b = [
u
L
[
L
+
u
D
[
D
+
1
2
R((I + 2I
0
)o
L
2
+(I + )o
M
2
+2(H
0
- I)o
LM
2
) +
1
2
[
C(L)
L
+
C()

(4)
The margin is obtained by three terms:
i) The first term, /, is a measure of the pure margin in a risk-neutral institution
(risk-neutral spread). Ho and Saunders (1981) interpret it as a measure of market power;
markets with the most inelastic demand may be monopolist and obtain higher margins than
a competitive market.
ii) The second term is the risk premium. Higher degrees of risk aversion (R=-1
2 u(w )u(w ) ), interbank interest rate volatility

credit risk (
L
) and a stronger
relationship between interest rate risk and credit risk (
ML
) imply higher margin. Moreover,
higher values of the institutions transaction amount (L+D), total loan volume (L+2L
0
) and
excess of funds kept in money markets over the new disbursed loans (M
0
-L) imply a higher
impact of risk on the margin, where M
0
represents the assets in monetary markets.
iii) Representing the last term are average operating costs, which increase the
institutions margin and are not affected by the low competitiveness or risk neutrality of the
markets.



3. Empirical model
The empirical model first includes the determinants of the financial margin
commonly used in commercial bank studies. Its applicability to microfinance is assessed.
The empirical model also adds exogenous variables to the theoretical model to consider
microfinance-specific characteristics.
NARu
t
= o

+ [
]
IH
t
]
]=1
+ y
k
HFIS
t
K
k=1
+e
t
(5)
t=1,T, T number of periods and i=1,.I, I number of MFIs. MARG is the
financial margin, TM are the margin determinants according to the theoretical model and
MFIS are the specific microfinance margin determinants.
Table 1 shows each variable definition. The dependent variable is Net Interest
Margin, calculated as the difference between Financial Income and Financial Expenses in
relation to Total Assets. The independent variables are defined as measured as follows:
***Table 1***
Operating costs. Its inclusion is justified from the theoretical and empirical points of
view. In fact, according to Maudos and Sols (2009), operating costs are the most relevant
determinant of the intermediation margin. The operating costs ratio is defined as Operating
Expense/Total Assets. Operating costs are expected to positively influence MFIs margin.
Risk aversion. Solvency has been related to the degree of risk aversion (McShane and
Sharpe, 1985). The relationship between solvency and margin is positive in banks. The
same relationship is expected in MFIs. Following McShane and Sharpe (1985), Maudos
and Fernndez de Guevara, (2004) and Maudos and Solis (2009), the Capital Assets Ratio
(CAR), defined as Equity/Total Assets, is used to measure the risk aversion.
Credit risk. The model applied in banks predicts a positive relationship between
margin and credit risk and the same could be expected in MFIs. The default probability of
MFI borrowers must be correctly managed to minimize its impact on the margin. Giving
loans to poor people without a credit history was the main innovation of the Grameen Bank
by Yunus (1999), who implemented the proverb: the poor always pay back. MFIs
monitor risk by using alternative systems to secure repayment, such as solidarity groups or
peer monitoring (Krauss and Walter, 2009). This makes the relationship between risk and
margin unclear. The variable Loan Loss Rate is used to measure credit risk, defined as
(Write-offs - Value of Loans Recovered)/ Loan Portfolio.
Size and age. Recently created MFIs are improving their systems and reducing costs.
When MFIs mature and their loan portfolio grows, they gain efficiency, among other
issues, by reducing information asymmetries, Behr (2011). Large and mature MFIs are
expected to have low margins. The Gross Loan Portfolio is chosen to measure size. The
number of years elapsed since the MFIs creation date is chosen to measure age.
Risk neutral spread. Competition is a key margin determinant in the financial sector.
Banks that face higher competition within a given country have lower margins (Demirg-
Kunt and Huizinga, 1999). This paper proposes the degree of financial inclusion as a
measure of competition. In developed countries, most adults report that they have an
account at a formal financial institution. In developing countries, where MFIs operate, the
level of financial inclusion is low, indicating a low level of financial culture and a low level
of banking competence. The World Banks Development Research Group has developed
the Global Financial Inclusion Database (Demirg-Kunt and Klapper, 2012), which
measures the populations use of financial services in a given country. The paper uses the
variable Account at a Formal Financial Institution to measure financial inclusion.
The specific margin determinants for MFIs are as follows:
Average loan size. The average loan balance per borrower is an outreach indicator in
microfinance (Navajas et al., 2000). Small loans retain high fixed administrative costs.
MFIs giving small loans are expected to have high margin due to fixed costs. Average loan
size is measured as Loan Portfolio Gross / Number of Active Borrowers and is corrected by
per capita GNI. This variable is commonly used as a social performance indicator to
measure MFI mission drift.
% Women borrowers. Many MFIs include in their mission the empowerment of
women. Women are associated with poverty in the countries where MFIs operate;
consequently, the percentage of women borrowers is considered a social performance
indicator. Alesina et al. (2008) find that women pay a higher interest rate, although women
are not riskier than men. Bellucci et al. (2010) and Beck et al. (2011) also demonstrate that
female entrepreneurs face tighter credit availability. A high percentage of women
borrowers is expected to be associated with a high financial margin.
Donations. Donations are a key difference in the capital structure of MFIs. Not all
MFIs receive donations. Start-up MFIs are usually funded with grants and when maturing,
they are more commercially oriented. Many of them are regulated and reduce donations in
favor of deposits, long-term loans or private equity. Bogan (2011) affirms that MFIs
accessing donor funds have pressure to obtain financial sustainability, which implies low
operational efficiency. Many MFIs use donations to serve their poorest clients in rural
areas, which also reduces the margin (Armendariz and Morduch, 2005). A negative
relationship is expected between margin and donations. The variable is Donations/Income.
% Deposits. MFIs must be regulated to capture deposits. They also suffer from high
external pressure to operate under financial sustainability criteria, Bogan (2011). According
to de Sousa-Shields and Frankiewicz (2004), deposit-taking MFIs finance their growth by
efficient management and not by charging high interest rates on the loans to the poor. A
negative relationship between margin and deposits is expected. The variable is
Deposits/Loans.
Type of entity. MFIs can operate as Non-Governmental Organizations (NGOs), Credit
Unions, Non-Bank Financial Institutions (NBFI) or Commercial Banks. It can be expected
that NGOs, more socially driven than banks, charge higher interest rates and operate with
lower margins. However, they target the poorest and their small loans have high costs,
which would justify high margins. Credit cooperatives, whose members are co-
proprietaries, generally do not attempt to maximize profits unconditionally (Smith et al.
1981). By contrast, it is expected that NBFI, which do not capture deposits, have high
margins.
4. Sample and results
Data come from the MixMarket database, which provides the annual statements of
the microfinance sector. The panel includes data from 2002 (322 MFIs) to 2010 (1,035
MFIs). Only the 2011 financial inclusion data were available because the World Banks
Global Financial Inclusion Database was recently built. The MFIs sample was divided into
two groups, according to the average loan size: under or over $300. The group under $300
(ALS < $300) contained pure MFIs, those that had not drifted from their mission due to
only serving poor clients. MFIs in the other group (ALS > $300) are more similar to
commercial banks, at least with regard to the type of clients served. These firms are
expected to be more market oriented.
Table 2 presents the exploratory analysis, which shows the mean of each variable, for
every year, for the two groups of MFIs. A T-test for means was performed to identify the
differences between the two groups.
*** Table 2***
The margin was higher in the ALS < $300 group than in the ALS > $300 group and
the differences are statistically significant for every year. The operating costs are higher in
the ALS < $300 group than in the ALS > $300 group and the differences are statistically
significant for every year. Margins and operating costs decrease each year, especially in the
ALS < $300 group.
There are significant differences in size, and MFIs with ALS < $300 are smaller than
MFIs with ALS > $300. Differences in risk are significant for only two years. Differences
in age are significant for five years. There are significant differences in the percentage of
women borrowers; MFIs with ALS < $300 serve a higher percentage of women and are
more mission centered than MFIs with ALS > $300. Donations only have data available
from 2006. MFIs with ALS < $300 receive more donations than MFIs with ALS > $300,
but MFIs receive fewer donations each year. There are significant differences in deposits;
MFIs with ALS < $300 capture fewer deposits than MFIs with ALS > $300. The last row
shows the profitability results, measured by the return on assets (ROA). This variable has
not been included in the model because it is not a determinant but a result of margin. It is
remarkable that MFIs with ALS < $300 have negative ROA, except for the last year.
Table 3 shows the pooled Pearsons correlation coefficients. The correlation
coefficient for each group of MFIs is shown. Both groups present positive correlations
between margin and operating costs, between margin and solvency and between margin
and ROA. There are negative correlations between margin and size, margin and age,
margin and donations, and margin and deposits. The correlation between risk and margin is
zero; MFIs with ALS < $300 have the same delinquency level as MFIs with ALS > $300.
All the results are coherent with the theoretical model. Some differences can be
appreciated between both groups of MFIs, explained by their social commitment. For
example, the group with ALS > $300 presents a negative correlation between margin and
loan size; that is, a smaller loan is associated with a higher margin (-0.21). However, in the
group with ALS < $300, a smaller loan is associated with a lower margin, although the
correlation coefficient is low (0.1). A similar pattern is observed in the other social
indicator. In the group with ALS > $300, a higher percentage of women borrowers implies
a higher margin (0.20); however, the other group, with ALS < $300, does not present a
significant correlation (0.03). An explanation can be based on the social character of pure
MFIs, which give small loans, mainly directed at women. They try to keep their margins
low with the aim of not charging high interest rates to their clients. They can do this
because they receive donations. Significant differences can be found here: in MFIs with
ALS < $300, higher donations lead to lower margin. The group of MFIs with ALS > $300
does not present significant correlations.
In MFIs with ALS > $300, higher deposits imply lower margin. The other group
hardly shows any correlation. This finding may be interpreted as a difficulty of gaining
profitability from small deposits.
*** Table 3***
McShane and Sharpe (1985) and Angbanzos (1997) approaches are followed to
estimate the explanatory model. The variables in the theoretical model and the exogenous
variables are estimated in a single step. The equation is estimated including the random
effects with robust errors, according to the panel data methodology. The random model has
been chosen because of the dichotomous nature of some variables; a fixed effects model
cannot include them. Moreover, and according to Bhargava and Sargan (1983), in panels
with a large number of individual units (MFIs) who are observed only for a few time
periods, random effects model estimation may be the most appropriate, and fixed effects
estimation may be inconsistent.
Table 4 shows the 8 analyzed models. Each model displays the results obtained in
both groups of MFIs. A Chow test is performed to assess whether the explained variance of
the variables is different in each group and a Smith-Satterthwaite test is performed to
measure the differences between the estimated coefficients in each regression. Both tests
will show whether the variables behavior is identical between the group of MFIs with ALS
> $300 (market oriented MFIs) and the group of MFIs with ALS < $300 (pure MFIs).
*** Table 4***
Model 1 contains the variables of the banking theoretical model. The results show
that the most determining factor is operating costs, in line with the theoretical model and
previous evidence. High operating costs imply high margin, but the effect is lower in MFIs
with ALS < $300. The beta coefficient is 0.63 in the group of MFIs with ALS > $300 and
0.25 in the other group. The Smith-Satterthwaite test shows statistically significant
differences on this variable in both groups.
A higher solvency implies a higher margin, in line with the theoretical model and the
results of other empirical studies. No relationship is found between credit risk and margin, a
finding consistent with the exploratory analysis.
Age has a negative relationship with margin but is only significant in entities with
high loan size. A higher loan size and a higher age of the MFI imply a lower margin. This
finding means that in this sector, maturity contributes to reduce the margin. Age and size
are correlated; therefore, to avoid multicollinearity problems, size has not been included in
the regression model.
Finally, financial inclusion presents a negative and significant relationship with
margin in both groups. This finding means that a higher degree of financial inclusion is
associated with a lower margin, as the model establishes.
The R2 of Model 1 is 0.57 for MFIs with ALS > $300, those closest to banks.
However, it is only 0.33 for the other group, with pure MFIs. The Chow test confirms the
statistically significant differences in the models explanatory capacity for both groups. It
seems clear that for pure MFIs, which are more different from commercial banks than the
other group of MFIs, the classical model has less explanatory power. This finding justifies
the inclusion of specific variables.
Model 2 incorporates donations. Higher donations are associated with a lower margin
in both groups. Donations effectively lower the margin. The inclusion of this variable is
especially relevant in the group of MFIs with ALS < $300, as the models R2 increases
from 0.33 to 0.44. In the other group, the increase is less pronounced, from 0.57 to 0.60.
Donations matter in the case of pure MFIs.
Model 3 includes deposits. A higher value of deposits to loans implies a lower
margin. Not every MFI can collect deposits; only those that are regulated can. The model
confirms that the decision of being regulated to capture deposits would lead to a reduction
in margin, which may represent a means of reducing the poverty penalty.
Model 4 includes an outreach measure, the percentage of women borrowers. A
positive and significant relationship has been found in the group of MFIs with ALS > $300,
meaning that a higher percentage of women borrowers is associated with a higher margin.
This does not happen in the other group, the one with pure MFIs. Notice that this variable is
significant according to the Smith-Satterthwaite test of differences.
Finally, the rest of the models, from 5 to 8, include the MFIs legal status. Neither
NGOs nor banks present a significant relationship with the margin. Credit cooperatives
have the lowest margin, which is coherent with their type of ownership, which is composed
of members. By contrast, the margin is high in NBFI.
Some final reflections may be made in an attempt to answer the research question
posed in the introduction. The poverty penalty exists: the margin is higher in those MFIs
with ALS < $300 than in the other group. However, this finding is not necessarily
associated with mission drift but rather to high operating costs, which are the main margin
determinants. The microfinance sector does not obtain high returns; in fact, most MFIs
have a negative ROA. MFIs willing to reduce their margin and thus the poverty penalty can
follow different strategies. The first one is to follow a turnover strategy, maximizing their
outreach by offering many small loans while improving their efficiency. This strategy is
consistent with the social mission of many MFIs and represents a means of achieving self-
sustainability because profitability can be broken down into two factors: margin and
turnover. Technologies to reduce operating costs in small loans should play a key role. The
sector is already doing so because the margin is decreasing each year. The growth of MFIs
leads to scale and learning economies. The second strategy is related to donations. Its role
is positive because donations contribute to reduce the margin. Although many claim self-
sustainability, MFIs serving the poorest with very small loans are those receiving
donations. In our opinion, donations are totally justified for MFIs fighting on the front lines
against financial exclusion. The third possibility is to reduce the margin by capturing
deposits through regulation, which represent a cheap funding source. Again, efficient
technology to reduce the high administrative costs of small deposits should be
implemented.
4. Conclusions
MFIs provide loans to the financially excluded. Because they are financial
institutions, their margin an important issue to keep in mind. However, unlike commercial
banks, which try to maximize their margin, many MFIs are socially oriented, with the aim
of maximizing outreach. This characteristic makes it necessary to adapt the banking models
explaining the margin to this special case, which may present different margin
determinants. This paper attempts to develop an explicative model for the microfinance
margin.
An empirical study has been performed with panel data from over one thousand MFIs
over 9 years. The sample has been divided into two groups: those with an average loan size
over $300, and those with an average loan size under $300. This figure represents a
threshold separating pure poverty orientation from market orientation.
The presence of a poverty penalty is confirmed: MFIs with low loans have high
margins. This fact does not imply mission drift because this high margin is caused by high
operating costs. These operating costs are the key variable determining margin, which is in
line with studies on commercial banks. If MFIs wish to lower their margin to alleviate the
poverty penalty suffered by their clients, the best way is by reducing these operating costs.
MFIs can choose a turnover strategy, giving many small loans. In addition, in every
turnover-based business, keeping costs under control and being efficient and productive are
key factors. It has also been found that solvency ratios follow the same pattern in
microfinance as in commercial banks.
The study has discovered differences in size and age with respect to commercial
banks previous studies. A larger and more mature bank has a higher margin. However,
microfinance is a young sector with an average life of less than 10 years in the dataset
analyzed. MFIs are still in a learning process, lowering their costs, and it has been observed
that the oldest and largest ones have the lowest margin, which is good news for these
socially oriented institutions.
The role of donations remains important, and in general, entities receiving donations
use them correctly because their margin is low. The same happens with deposits, which are
a cheap funding source, and those regulated entities capturing deposits have low margin.
However, the effect of deposits on the margin is less important than that of operating costs,
according to the regression coefficient.
The legal status matters because it is often associated with the social character of
MFIs. Credit cooperatives, owned by their members, operate with the lowest margin. Non-
Bank Financial Institutions operate with the highest margin.
Finally, where the level of accessibility of banking services is low, the margin is high.
As the populations financial literacy grows and the spread of financial services leads to
competition, the margin will decrease, as will the poverty penalty. This phenomenon
represents an invitation to make microcredit more accessible and more affordable.

Acknowledgements: The work reported in this paper was supported by grant ECO2010-
20228 of the Spanish Ministry of Education and Science and the European Regional
Development Fund and by grant Ref. S-14 (3) of the Government of Aragon.

References
Ahlin, C., Lin, J., Maio, M., 2011. Where does microfinance flourish? Microfinance
institution performance in macroeconomic context, Journal of Development
Economics 95 (2), 105-120.
Alesina, A., Lotti, F., Mistrulli, P., 2008. Do Women Pay More for Credit? Evidence from
Italy. NBER Working Paper 14202, Cambridge, MA: National Bureau of Economic
Research.
Allen, L., 1988. The determinants of bank interest margins: A note. Journal of Financial
and Quantitative Analysis 23, 231235.
Angbazo L., 1997. Commercial bank net interest margins, default risk, interest-rate risk,
and off-balance sheet banking. Journal of Banking & Finance 21, 55-87.
Armendariz, B. and Morduch, J., 2005. The economics of microfinance. MIT Press,
Cambridge, MA.
Beck, T.H.L., Behr, P., Madestam, A., 2011. Sex and Credit: Is There a Gender Bias in
Microfinance? Discussion Paper 2011-101, Tilburg University, Center for Economic
Research.
Behr, P., Entzian, A., Gttler, A., 2011. How do lending relationships affect access to credit
and loan conditions in microlending? Journal of Banking & Finance 35(8), 2169-2178.
Bellucci, A., Borisov, A., Zazzaro, A., 2010. Does gender matter in bankfirm
relationships? Evidence from small business lending, Journal of Banking & Finance
34(12), 2968-2984.
Bhargava A., Sargan, J.D., 1983. Estimating dynamic random effects models from panel
data covering short time periods. Econometrica 51, 1635-1660.
Bogan, V.L., (2011). Capital Structure and Sustainability: An Empirical Study of
Microfinance Institutions. Review of Economics and Statistics, forthcoming, DOI
10.1162/REST_a_00223.
Carb, S., Rodriguez, F., 2007. The determinants of bank margins in European banking.
Journal of Banking & Finance 31, 20432063
De Sousa-Shields, M., and Frankiewicz, C., 2004. Financing Microfinance Institutions: The
Context for Transitions to Private Capital. Micro Report 32, Accelerated
Microenterprise Advancement Project, USAID.
Demirg-Kunt A., Huizinga, H., 1999. Determinants of Commercial Bank Interest
Margins and Profitability: Some International Evidence. World Bank Economic
Review 13(2), 379-408.
Demirg-Kunt, A., Klapper, L.F., 2012. Measuring Financial Inclusion: The Global
Findex Database. World Bank Policy Research Working Paper 6025.
Gulli, H., 1998. Microfinance and Poverty: Questioning the Conventional Wisdom. Inter-
American Development Bank, Washington D.C.
Hartarska, V., Nadolnyak, D., 2007. Do regulated microfinance institutions achieve better
sustainability and outreach? Cross-country evidence. Applied Economics 39, 1207
1222.
Hermes, N., Lensink, R., 2011. Microfinance: Its Impact, Outreach, and Sustainability.
World Development 39(6), 875-881.
Ho, T., Saunders, A., 1981. The determinants of banks interest margins: theory and
empirical evidence. Journal of Financial and Quantitative Analysis XVI (4), 581600.
Krauss, N., Walter I., 2009. Can Microfinance Reduce Portfolio Volatility? Economic
Development and Cultural Change 58 (1), 85-110.
Lepetit L., Nys, E., Rous, P., Tarazi, A., 2008. The expansion of services in European
banking: Implications for loan pricing and interest margins. Journal of Banking &
Finance 32, 23252335.
Maudos, J., Fernndez de Guevara, J., 2004. Factors explaining the interest margin in the
banking sectors of the European Union. Journal of Banking & Finance 28, 2259-2281.
Maudos, J., Sols, L., 2009. The determinants of net interest income in the Mexican
banking system: An integrated model. Journal of Banking & Finance 33, 1920-1931.
McShane, R.W., Sharpe, I.G., 1985. A time series/cross section analysis of the
determinants of Australian trading bank loan/deposit interest margins: 19621981.
Journal of Banking & Finance 9, 115136.
Mersland, R., 2009. The Cost of Ownership in Microfinance Organizations. World
Development 37(2), 469-478.
Mersland, R., Strm, R.., 2008. Performance and trade-offs in Microfinance
Organisations-does ownership matter? Journal of International Development 20(5),
598-612.
Mersland, R., Strm, R.., 2009. Performance and governance in microfinance institutions.
Journal of Banking & Finance 33, 662-669.
Mersland, R., Strm, R.., 2010. Microfinance Mission Drift? World Development 38(1), 28-
36.
Morduch, J., 1999. The role of subsidies in microfinance: Evidence from the Grameen
Bank. Journal of Development Economics 60, 229-248.
Navajas, S., Schreiner, M., Meyer, R L., Gonzlez-Vega, C., 2000. Microcredit and the
poorest of the poor: Theory and evidence from Bolivia. World Development 28(2),
333-346.
Prahalad, C.K., Hammond, A., 2002. Serving the world's poor, profitably. Harvard
Business Review 80(9), 4857.
Saunders, A., Schumacher, L., 2000. The determinants of bank interest rate margins: an
international study. Journal of International Money and Finance 19(6), 813-832.
Servin, R., Lensink, R., van den Berg, M., 2012. Ownership and technical efficiency of
microfinance institutions: Empirical evidence from Latin America. Journal of Banking
& Finance 36(7), 2136-2144.
Smith, D.J., Cargill, T.F., Meyer, R.A., 1981. Credit Unions: An Economic Theory of a
Credit Union. Journal of Finance 36(2), 519-528.
Yunus, M. (1999): Banker to the poor: Microlending and the battle against world poverty.
Public Affairs, New York

Variable Definition
Margin MARG Net Interest Margin =(Financial income Financial Expenses) / Total Assets
Operating costs OE Operating Expense / Total Assets
Solvency CAR Capital Assets Ratio =Equity / Total Assets
Risk LLR Loan Loss Rate =(Write-offs - Value of Loans Recovered)/ Loan Portfolio
Size lnGLP Gross Loan Portfolio Logarithm
Age YEARS The number of years elapsed since the MFIs creation date
Financial inclusion FIN Account at a formal financial institution (% age 15+)
Outreach
ALS
Average Loan Balance per Borrower =Loan Portfolio Gross / Number of Active
Borrowers corrected by the per capita GNI.
WB % Women Borrowers
Donations DON Donations / Income
Deposits DEP Deposits / Loans
Type of entity TYPE Dummy variables on the legal status
NGO: Non-Governmental Organization
BANK: Commercial Bank
COOP: Cooperative / Credit Union
NFBI: Non-Bank Financial Institution
Profitability ROA Return on Assets =(Net Operating Income - Taxes) / Period Average Assets

Table 1. Variables and their definitions.

Table 2. Descriptive statistics. Each column contains the mean for both groups (ALS > $300 for Average Loan Size over $300 and
ALS < $300 for Average Loan Size under $300). It also shows the results from a T-test for means and its significance. ** significant at
5% level; *** significant at 1% level.
2002 2003 2004 2005 2006 2007 2008 2009 2010
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
ALS
>$300
ALS
<$300
Margin
N 178 144 292 240 445 323 546 415 624 440 680 388 772 391 713 387 697 338
Mean 0.22 0.26 0.21 0.23 0.21 0.25 0.21 0.24 0.21 0.23 0.20 0.22 0.21 0.24 0.21 0.23 0.21 0.23
T-test 3.48 *** 2.76*** 3.14*** 3.16*** 2.41** 2.44** 3.04*** 2.81*** 1.97**
Operating
costs
N 178 144 292 240 445 323 545 416 626 440 681 389 776 395 713 387 698 340
Mean 0.19 0.26 0.18 0.24 0.18 0.25 0.17 0.22 0.17 0.23 0.16 0.22 0.18 0.24 0.17 0.23 0.17 0.21
T-test 4.93*** 5.18*** 6.34*** 5.91*** 7.63*** 7.39*** 6.00*** 5.48*** 3.51***
Solvency
N 302 150 459 279 607 375 687 453 750 527 775 521 875 441 797 477 723 447
Mean 0.41 0.47 0.38 0.45 0.37 0.44 0.33 0.39 0.34 0.35 0.33 0.33 0.32 0.31 0.33 0.32 0.34 0.31
T-test 3.48*** 2.76*** 3.14*** 3.16*** 2.41** 2.44** 3.04*** 2.81*** 1.97**
Size
N 304 280 468 382 612 457 695 533 759 533 791 445 888 483 819 464 774 394
Mean 13E6 4E6 10E6 5E6 12E6 6E6 20E6 7E6 27E6 9E6 36E6 10E6 42E6 13E6 67E6 19E6 73E6 24E6
T-test 8.61*** 9.31*** 11.33*** 10.40*** 9.70*** 8.53*** 7.41*** 6.75*** 5.94***
Risk
N 159 128 285 235 429 303 525 392 613 427 674 378 760 377 709 387 710 353
Mean 0.015 0.029 0.014 0.019 0.061 0.018 0.033 0.029 0.012 0.013 0.012 0.019 0.013 0.018 0.020 0.021 0.022 0.016
T-test 1.56 0.59 1.22 0.30 0.38 2.51** 1.93 0.02 2.00**
Age
N 276 250 421 345 552 412 629 478 678 477 691 395 749 409 680 384 625 323
Mean 9.87 8.86 10.16 9.78 11.10 10.39 11.37 10.24 11.72 10.10 11.87 10.29 12.98 11.04 13.69 12.14 14.30 13.06
T-test 0.40 0.04 1.00 1.88 3.04*** 3.22*** 3.35*** 2.53** 2.00**
Average Loan
per borrower
N 304 280 468 382 612 457 695 533 759 533 791 445 888 483 819 464 774 394
Mean 953 116 1213 123 1234 119 1256 127 1586 138 1922 148 1968 151 1960 155 2053 162
T-test 9.93*** 11.76*** 12.72*** 12.26*** 13.85*** 14.06*** 15.62*** 17.72*** 17.05***
Women
borrowers
N 254 252 387 343 504 412 609 489 657 493 653 411 718 386 651 365 677 368
Mean 0.64 0.77 0.61 0.78 0.62 0.81 0.61 0.80 0.59 0.80 0.56 0.81 0.53 0.82 0.53 0.84 0.54 0.85
T-test 10.50*** 13.58*** 17.45*** 18.70*** 18.89*** 18.67*** 21.45*** 22.28*** 22.31***
Donations
N - - - - - - - - 680 477 694 396 762 418 689 397 638 334
Mean - - - - - - - - 0.14 0.40 0.13 0.36 0.14 0.23 0.07 0.22 0.08 0.09
T-test - - - - - - - - 4.97*** 4.49*** 1.24 1.77 0.31
Deposits
N - - 171 124 251 153 352 234 525 324 662 357 846 459 809 453 757 388
Mean - - 0.39 0.28 0.33 0.20 0.45 0.33 0.44 0.31 0.40 0.29 0.44 0.30 0.43 0.29 0.40 0.26
T-test - 2.03** 3.22*** 2.02** 2.97*** 3.29*** 4.20*** 4.14*** 4.10***
ROA
N 178 144 292 240 445 323 547 416 627 441 681 388 778 395 713 387 698 340
Mean 0.01 -0.02 0.00 -0.03 0.01 -0.03 0.02 -0.01 0.02 -0.02 0.02 -0.03 0.01 -0.02 0.01 -0.03 0.02 0.00
T-test 2.05** 3.10*** 4.17*** 3.62*** 5.87*** 6.19*** 4.03*** 3.05*** 1.51

M
a
r
g
i
n

O
p
e
r
a
t
i
n
g

c
o
s
t
s

S
o
l
v
e
n
c
y

S
i
z
e

R
i
s
k

A
g
e

A
v
e
r
a
g
e

L
o
a
n

p
e
r

b
o
r
r
o
w
e
r

W
o
m
e
n

b
o
r
r
o
w
e
r
s

D
o
n
a
t
i
o
n
s

D
e
p
o
s
i
t
s

R
O
A

Margin 1


Operating
costs
0.63
1

[0.54]

Solvency
0.27 0.17
1

[0.30] [0.20]

Size
-0.19 -0.31 -0.28
1

[-0.09] [-0.28] [-0.19]

Risk
0.00 0.02 0.03 -0.05
1

[0.03] [0.04] [0.00] [-0.06]

Age
-0.19 -0.21 -0.18 0.31 -0.03
1

[-0.09] [-0.18] [-0.13] [0.39] [-0.03]

Average Loan
per borrower
-0.21 -0.20 -0.01 0.19 0.00 -0.01
1

[0.10] [-0.01] [0.00] [0.21] [-0.03] [0.01]

Women
Borrowers
0.20 0.23 0.02 -0.20 -0.02 -0.02 -0.31
1

[0.03] [-0.01] [-0.11] [0.09] [0.00] [0.12] [-0.25]

Donations
-0.06 0.21 0.13 -0.14 0.01 -0.08 -0.01 0.01
1

[-0.12] [0.30] [0.04] [-0.14] [0.05] [-0.09] [-0.07] [0.01]

Deposits
-0.27 -0.18 -0.39 0.10 0.00 0.22 0.02 -0.14 -0.04
1

[-0.06] [-0.04] [-0.20] [-0.07] [0.00] [0.05] [0.02] [-0.15] [-0.04]

ROA
0.23 -0.57 0.09 0.15 -0.03 0.07 0.01 -0.06 -0.33 -0.05
1
[0.22] [-0.67] [0.05] [0.23] [-0.02] [0.14] [0.09] [0.04] [-0.45] [0.00]

Table 3. Pooled Pearsons correlation coefficients for both groups. The top figures
correspond to ALS > $300 group, with Average Loan Size over $300, and the figures in
brackets correspond to the ALS < $300 group, with Average Loan Size under $300.


Table 4. Dependent variable: net interest margin. Panel data, random model. ** significant
at 5% level; *** significant at 1% level

Variable
(1) (2) (3) (4)
ALS>$300 ALS<$300 ALS>$300 ALS<$300 ALS>$300 ALS<$300 ALS>$300 ALS<$300
Operating
costs
0.6315*** 0.2466*** 0.6534*** 0.3305*** 0.6577*** 0.3999*** 0.6357*** 0.3311***
(14.12) (3.31) (15.09) (3.67) (14.89) (3.52) (13.93) (3.44)
Solvency 0.1006*** 0.0483*** 0.1077*** 0.0578*** 0.0957*** 0.0579*** 0.0926*** 0.0539***
(6.82) (2.71) (6.49) (3.1) (5.47) (3.18) (5.16) (2.84)
Risk -0.0241 -0.0602 -0.0409 -0.0324 -0.0339 -0.0191 -0.0446 -0.1017
(-0.62) (-0.69) (-1.12) (-0.33) (-0.9) (-0.14) (-1.18) (-0.99)
Age -0.0013*** 0.0003 -0.0012*** -0.0007 -0.0011*** -0.0005 -0.0012*** -0.0005
(-4.81) (0.59) (-4.94) (-1.18) (-4.78) (-0.88) (-5.03) (-0.87)
Financial
inclusion
-0.0005*** -0.0008*** -0.0004*** -0.008*** -0.0004*** -0.0008*** -0.0004** -0.0008**
(-3.13) (-3.22) (-2.71) (-2.99) (-2.74) (-2.95) (-2.86) (-2.82)
Donations -0.0432*** -0.0281*** -0.0430*** -0.0281*** -0.0385*** -0.0282***
(-5.65) (-5.20) (-4.60) (-5.22) (-5.23) (-5.08)
Deposits -0.0099*** -0.0174*** -0.0106** -0.0180***


(-2.73) (-3.32) (-2.55) (-3.38)
Women
borrowers


0.0336*** -0.0130
(3.24) (-0.91)
Constant 0.1045*** 0.1889*** 0.0995*** 0.1799*** 0.1062*** 0.1815*** 0.0835*** 0.1737***
(9.36) (7.39) (9.05) (5.95) (8.61) (5.64) (7.49) (5.77)
N obs
R
2

3028 1810 2523 1445 2500 1431 2267 1341
0.5671 0.3257 0.6031 0.4362 0.6123 0.4365 0.6086 0.4294
Wald Chi
2
400.68 68.05 372.61 105.77 499.20 120.45 545.73 65.88
p-value
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Chow test 134.34 46.40 39.90 37.21
p-value 0.000 0.000 0.000 0.000


Variable
(5) (6) (7) (8)
ALS>$300 ALS<$300 ALS>$300 ALS<$300 ALS>$300 ALS<$300 ALS>$300 ALS<$300 S-S Test
Operating
costs
0.6343*** 0.3265*** 0.6314*** 0.3255*** 0.6211*** 0.3238*** 0.6266*** 0.3265***
2.87***
(13.83) (3.39) (13.77) (3.39) (13.42) (3.37) (13.63) (3.41)
Solvency
0.0942*** 0.0538** 0.0918*** 0.0527** 0.0895*** 0.0522*** 0.0928*** 0.0529** 1.50
(5.17) (2.78) (5.12) (2.79) (4.98) (2.76) (5.22) (2.80)
Risk
-0.0438 -0.1026 -00463 -0.1004 -0.0521 -0.1033 -0.0484 -0.1029 0.52
(-1.17) (-1.00) (-1.23) (-0.97) (-1.39) (-1.01) (-1.31) (-1.00)
Age
-0.0011*** -0.0005 -0.0012*** -0.0006 -0.0013*** -0.0007 -0.0009*** -0.0002 1.10
(-4.87) (-0.90) (-5.06) (-1.07) (-4.89) (-1.14) (-4.14) (-0.36)
Financial
inclusion
-0.0004** -0.0008** -0.0004** -0.0008** -0.0004** -0.0009** -0.0003** -0.0009*** 1.15
(-2.86) (-2.9) (-3.00) (-2.92) (-2.46) (-3.07) (-2.42) (-3.08)
Donations
-0.0377*** -0.0282*** -0.0379*** -0.0283*** -0.0378*** -0.0283*** -0.0377*** -0.0281***
1.04

(-5.15) (-5.12) (-5.08) (-5.13) (-5.16) (-5.14) (-5.27) (-5.11)
Deposits
-0.1402*** -0.0189*** -0.0111** -0.0181*** -0.0071* -0.1629*** -0.0087** -0.0165** 1.08
(-3.23) (-3.54) (-2.64) (-3.38) (-1.72) (-3.31) (-2.07) (-2.93)
Women
borrowers
0.0386*** -0.0131 0.0351*** -0.0152 0.0339*** -0.0157 0.0362*** -0.0116 2.72**
(3.58) (-0.87) (3.38) (-1.05) (3.28) (-1.08) (3.49) (-0.79)
NGO
-0.0135* -0.0047
0.67
(-1.9) (-0.42)
BANK
-0.0067 -0.0277
0.61
(-0.50) (-0.87)
COOP
-0.0271*** -0.0425**
0.82
(-4.61) (-2.41)
NBFI
0.0271*** 0.0305**
0.22
(3.83) (2.13)
Constant
0.0974*** 0.2033*** 0.0979*** 0.2041*** 0.1033*** 0.2088*** 0.0819*** 0.1911***
2.65**
(7.79) (5.51) (7.70) (5.50) 7.95 (5.51) (6.52) (5.30)
N obs 2265 1335 2265 1335 2265 1335 2265 1335
R
2
0.6140 0.4305 0.6107 0.4337 0.6148 0.4329 0.6187 0.4310
Wald Chi
2
587.46 117.7 541.71 111.75 628.01 117.68 554.81 109.87
p-value 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Chow test 30.13 30.33 30.37 30.45

p-value 0.000 0.000 0.000 0.000
Table 4 (continued). Dependent variable: net interest margin. Panel data, random model.
S-S test: Smith-Satterthwaite test. * significant at 10% level ** significant at 5% level; ***
significant at 1% level