Content uploaded by Sunil Sangwan
Author content
All content in this area was uploaded by Sunil Sangwan on Jul 20, 2020
Content may be subject to copyright.
Factors influencing the borrower
loan size in microfinance group
lending: a survey from Indian
microfinance institutions
Sunil Sangwan and Narayan Chandra Nayak
Department of Humanities and Social Sciences,
Indian Institute of Technology, Kharagpur, India
Abstract
Purpose –The purpose of this paper is to analyze the impact of the cost of microfinance intermediation on
borrowers’loan size. The identified transaction cost and credit risk factors tell about what a lender takes into
accounts while screening and allocating loan amounts to the borrowers, where the lender has limited
information about the client’s ability to repay.
Design/methodology/approach –The analysis is based on the primary data collected from a sample of
498 microfinance institutions (MFI) linked group clients covering two microfinance leading states of India.
Findings –Empirical findings suggest that the cost of microfinance intermediation has an impact on
borrowers’loan size. To reduce the cost, the MFIs lend big loans to clients having a high income, assets, land
size, lower informal borrowings and having longer loan experiences. In MFI lending, the younger and less
educated people are the ones who demand bigger loan amounts. The geographical distance of borrowers’
location from MFI offices, group size and interest rate are the other factors that influence the loan size.
Originality/value –The past empirical works seem to have not focused on how the cost of microfinance
intermediation creates loan size variation among the borrowers in joint liability group lending. The
endogeneity problem has not been resolved. The present article thus identifies the factors that influence the
individual member loan size by using two-stage least squared regression to tackle the issue of endogeneity.
Keywords Microfinance institutions, Payment methods, Group lending, India, Loan size
Paper type Research paper
1. Introduction
Of late, the microfinance institutions (MFIs) have emerged as the supplementary
intervention to formal financial institutions to achieve the target of financial inclusion. MFIs
are especially characterized by their provisions of financial services primarily to low-income
groups, who are otherwise excluded from the mainstream financial services (Ranjani and
Kumar, 2018;Singh and Padhi, 2017). In India, to a large extent, these MFI loans are
provided to low-income households’women members. The lending happens by way of
group formation. Access of credit to women is stated to have fostered their economic and
social empowerment, as well as an increase in the overall household well-being (Armend
ariz
and Morduch, 2010;Mohapatra and Sahoo, 2016).
Along the positive aspect, the cost of microfinance intermediation is a concern for the
MFIs while assuring a special focus on lending to the financially excluded low-income
sections of society (Swamy, 2019). In microfinance lending, the loan size is a vital attribute
that is closely related to the cost of financial intermediation. It is argued that when the loan
size is bigger, the transaction cost incurred by the MFI to serve the targeted financially
Indian
microfinance
institutions
Received 3 January2020
Revised 23 April 2020
Accepted 14 June2020
Journal of Financial Economic
Policy
© Emerald Publishing Limited
1757-6385
DOI 10.1108/JFEP-01-2020-0002
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/1757-6385.htm
excluded population tends to be lower (Shankar, 2007;Swamy, 2019). For MFIs, the lower
transaction cost can improve efficiency and financial performance and can make it easy to
increase the client outreach (Sangwan and Nayak, 2019;Yimga, 2018). Putting thrust on
financial sustainability, self-sufficiency and income generation, it becomes imperative for
MFIs to recover the cost of lending through loan portfolios. Efficient loan size policy within
the institutional framework may help the MFIs to create profitable loan portfolios to achieve
a high level of financial performance (Srinivasan, 2015). The efficient loan size led financial
performance can make it feasible for MFIs to make their large scale client outreach and
continuation of operation in the long run (Ranjani and Kumar, 2018;Yimga, 2018). In India,
as per Sa-dhan (2018) report, the MFI’s average loan size seems to have grown exponentially
over the years from INR 7,481 in the year 2011 to INR 14,700 in 2018, reflecting their focus
on financial sustainability.
Nonetheless, determining an efficient loan size becomes more relevant when MFIs’
alternative channel to recover the cost of microfinance intermediation, the interest rate is
regulated and has a ceiling. In India, Reserve Bank of India has regulated a margin cap of
10% in case of large NBFC-MFIs (loan portfolio of INR 100 core or more) and 12% in case of
smaller MFIs. Still, in general, the interest rates charged by the MFIs are much higher than
that fixed by the conventional formal financial institutions (Armend
ariz and Morduch, 2010;
Dehem and Hudon, 2013). The proponents of social welfare are concerned about the high-
interest rates as microfinance intermediation is meant for the low-income population
(Mendoza, 2011;Sangwan et al., 2020;Shankar, 2007). There is thus a need for interest rate
rationalization. It is argued that the impact of the higher interest rate on the poor is two-fold.
On the one side, if they receive loans at a high cost of borrowing, they default more often
than not, leading to their over-indebtedness. On the other side, if they are deprived, they
become the victims of discrimination (Sangwan et al., 2020).
However, Helms and Reille (2004) and Fernando (2006) suggest that the interest rate
ceilings will hamper the financial sustainability of the MFIs and will retard the long term
availability of financial services to the targeted population. In case when the MFIs are not
able to recover their lending costs, they are compelled to exit the market. The phenomenon
would assert the prospective borrowers’dependence on informal sources of finance
(Shankar, 2007). Under such constrained conditions, it becomes imperative for the MFIs to
focus on their loan size and design it in such a way that it reduces the burden of the cost of
microfinance intermediation and helps them lower down the interest rates in a sustainable
manner.
Though the existing literature has substantially discussed the impact of regulations
(Haldar and Stiglitz, 2016), competition (Hossain et al.,2020), governance structure
(Mersland and Strøm, 2009) and the interest and non-interest expenses of financial
institutions (Swamy, 2019) on the cost of microfinance intermediation, the studies estimating
how the cost of microfinance intermediation affects loan size are in dearth. Moreover, most
of the existing empirical works are based on the secondary data self-reported by the MFIs.
The present study, however, uses primary cross-section data collected from a field survey of
two leading microfinance states in India.
The research is motivated by the concern that the MFI lending program can have a
limited impact if it fails to fulfil the genuine loan demands and keep the borrowers’credit
constraints concentrating on the concern of the cost of microfinance intermediation. We aim
to analyze the causal relationship between the different components of the cost of
microfinance intermediation and the loan size. A reduction in the cost of microfinance
intermediation through the loan size may eventually result in the lowering of interest rates
to the borrowers and raising the demand for credit that would lead to an increase in MFI
JFEP
lending. Following the concept of the economies of scale, it may be argued that the increased
lending even can encourage these MFIs to further enlarge their outreach to financially
excluded people located at far off areas. Undertaking the economic comparative advantage
rationale, it will provide competition to reduce the interest rates of moneylenders, which
seems not affordable by low-income households.
The remaining part of the article is organized into five sections. Section 2 provides an
overview of the literature and presents an empirical framework to examine the association
between the cost of microfinance intermediation and the loan size. Section 3 outlines the data
and methodology. Section 4 presents the empirical results and discussion. Section 5
concludes the study.
2. Overview of literature
Modern finance theory has helped in better comprehension of the lender credit allocation
decisions and provides a framework to analyze how the loan size differs among borrowers.
Borrowers’differential loan size can be translated as a type of bias against particular classes
of clients and that may be attributed to the cost of microfinance intermediation. The latter
predominantly comprises the transaction cost and credit risk. In the literature, the
transactional costs for the MFIs comprises components such as lending methodology, group
size, loan size, repayment period, borrowers’geographical location and monitoring and
supervision of loan repayments (Ahlin, 2015;Shankar, 2007;Srinivasan, 2015). Of all the
costs, the doorstep delivery is the very compelling component of the transaction cost in
microfinance intermediation (Srinivasan, 2015). MFIs employ loan officers that provide
doorstep credit facilities to the borrowers at their residences or nearby community places. It
incurs high operational costs for staffing and traveling besides the opportunity cost of time
spent (Shankar, 2007). Unlike banks that provide loan disbursements through checks or
bank transfers, MFI lending involves cash disbursements and cash repayments through
weekly/fortnightly or monthly meetings at locations where borrowers are situated. The
transaction costs of the MFIs in such kind of financial services are higher on account of the
periodic meetings (Srinivasan, 2015). Further, the frequency of meetings or the repayment
period (weekly/fortnightly/monthly) for these cash transactions determines the scale of the
transaction cost. It is observed that the incurred costs in these frequent repayment meetings
are higher compared to the monthly collection periods (Shankar, 2007;Srinivasan, 2015).
However, frequent visits ensure high repayments in the MFI lending (Shu-Teng et al., 2015).
Another component in MFI lending is interest rates charged upon the borrowers. The
interest rates significantly affect the transaction costs and can influence the loan size
(Dehem and Hudon, 2013). Though an increase in interest rate lowers the transaction cost to
the lenders (Swamy, 2019), simultaneously the rise in the cost of borrowing discourages the
borrowers’total loan borrowings (Jumpah et al.,2019). Contrarily, Tiwari et al. (2008)
ascertain that borrowers are more concerned about their loan demands but not with the
interest rates.
The empirical literature (Ahlin, 2015;Shankar, 2007;Singh and Padhi, 2017) identifies
some group attributes affecting the cost of microfinance intermediation and, in turn, the
borrowers’loan size. Among the different group characteristics, group size causes a positive
impact on the reduction of transaction costs. For the loan officers, the groups’unit cost per
member decreases significantly as the number of members in the group increases (Shankar,
2007). The reduction in the transaction cost can be passed on to the members in terms of
their bigger loan size. From the credit risk perspective, Ahlin (2015) finds that the
probability of the loan repayments increases when the loans are distributed in larger groups
provided the group size should be large enough to fully overcome the informational friction.
Indian
microfinance
institutions
Group cohesiveness is another group characteristic that supports better repayments among
group members and reduces the cost of credit risk (Sangwan et al.,2020;Zeller, 1998). High
group cohesiveness enhances the effectiveness of peer monitoring and social sanctions
among the group members, the two key moral hazard mitigation measures in group lending
(Zeller, 1998). Following the above two arguments, one can hypothesize that a member
belonging to a bigger group size having higher group cohesiveness is expected to receive a
bigger loan size.
The empirical literature (Mason, 2014;Singh and Padhi, 2017) also ascertains the impact
of group members’loan experiences on their loan size. A borrower’s loan experience
symbolizes the long-term credit relationship with the lender (White and Alam, 2013). Mason
(2014) finds that a borrower having a long loan relationship with a lender tends to get a
bigger loan size.
The borrowers’differential loan size can also be explained based on the theory of
asymmetric information. The asymmetric information is primarily affiliated to the credit
risk in MFI lending. It creates problems for MFIs making them incapable to sort loan
applicants between creditworthy and non-credit worthy clients (Stiglitz and Weiss, 1981).
However, the empirical literature suggests that the problem of asymmetric information and
associated credit risk can be resolved through local information. The local information is
easily available to the group members but not to the MFIs, which the former use while self-
selecting the peer members (Stiglitz and Weiss, 1981;Zeller, 1998). The information that a
group or MFI typically uses to determine the borrowers’loan size is related to the
creditworthiness or credit risk of the clients. The information usually comprises different
household characteristics that can play a significant role in evaluating the loan size.
Among the different household characteristics, the household income is the key feature
that reflects clients’repayment capacity and attracts the MFIs for loan disbursements. It is
argued that a household experiencing more incidences of poverty tends to pose higher credit
risk (Dorfleitner et al., 2017). Consequently, the MFI is anticipated to lend a smaller loan size
to such a household. In addition to income, an MFI requires the clients to report their asset
holdings in loan applications, which usually act as potential securities for the credit lending
agencies (Sangwan et al., 2020;Simtowe et al.,2008). As the asset value appreciates, the
borrowers’tendency to loan repayments upsurge fearing the confiscation of pledged
securities. Though microfinance intermediation does not ask for collateral, still borrowers’
assets are the perceived collateral for the lenders to reduce the cost of credit risk. Simtowe
et al. (2008) suggest that landholding with secure property rights acts as collateral and can
be a significant determinant to avail of credit access in the financial market. Besides, the
affordability of credit is an issue in deciding the borrower’sloansize(Mason, 2014). A
borrower’s affordability of credit or financial stress can be assessed from its level of
indebtedness (Singh and Padhi, 2017). Sangwan et al. (2020) find that indebtedness is a
serious threat to the repayments in microfinance services. Hence, the MFIs are anticipated to
reduce their loan disbursements to households that are already having other informal
borrowings.
Among the demographic characteristics, the borrowers’age turns out to be an important
factor to determine the loan size. Bandyopadhyay and Saha (2011) argue that younger
clients tend to show low credit risk as these people have an advantage over the old ones in
terms of better earning opportunities and greater longevity in income-generating activities.
So, it can be hypothesized that MFIs disburse bigger loans to young households.
Education is another household variable that seems to affect the credit risk and loan size.
Shu-Teng et al. (2015) find that education has a positive impact on loan repayments.
Education makes clients familiar with the financial products, enhances their comprehension
JFEP
of the available information and helps them make the right business decisions (Bhatt and
Tang, 2002;Singh and Padhi, 2017). Okurut et al. (2005) find education as a significant
positive indicator of loan size. The type of economic activity a borrower is engaged in can
also be hypothesized to determine borrowers’loan size. It is argued that the people engaged
in the agriculture are usually having their untimely demands, and the earnings are irregular
and seasonal that lead them to make unsystematic loan repayments (Basu and Srivastava,
2005). Contrarily, people indulged in the non-farm sector tend to earn a comparatively
regular income that enables them to make smooth periodic repayments (Sangwan et al.,
2020). Hence, the clients engaged in non-agricultural activities are expected to receive bigger
loan amounts. The regional characteristics also influence the MFIs’loan disbursements. The
regions tend to differ according to climate, socioeconomic structure, infrastructure and
governance features, which may impact the MFIs’cost of microfinance intermediation and
lead them to make differential loan disbursements (Sangwan and Nayak, 2019).
2.1 Empirical framework
As envisaged earlier, the borrowers’loan size can be conceptualized as a function of the cost
of microfinance intermediation. The cost of microfinance intermediation primarily
comprises two components, namely, transaction cost and credit risk. Under transaction cost,
the doorstep delivery is a very compelling component of microfinance intermediation. The
geographical distance of borrowers’locations from MFIs’offices includes travel cost, staff
cost and monitoring and supervision cost (Srinivasan, 2015). The transaction cost also
varies according to the lending mechanisms such as group lending, size of the group and
repayment period (frequency of repayment) (Shankar, 2007). These factors entail peculiar
costs such as group formation costs, costs on training the group members’regarding the
procedures to be followed, cost of supervision and frequency of installment collections
(weekly/fortnightly/monthly). The interest rate is also closely related to the transaction cost.
To measure the cost of credit risk, the credit risk associated with MFI group lending may
be conceptualized as a function of borrowers’repayment capability, the enabling
environment and the moral hazards to which the clients are subjected to. Household
characteristics comprising economic (income, assets, land size and primary occupation)
factors can be considered as capability indicators, while demographic (age and education)
factors represent the enabling conditions.
Yet another closely related issue to credit risk is the problem of moral hazard. Under the
condition of information asymmetry, moral hazard may involve recourse to private
information (Arrow, 1963). Theoretical literature suggests group lending as the mechanism
to mitigate the problem of moral hazard (Ahlin, 2015). In a situation where supervision is
poor, moral hazard may have repercussions on the borrowers’ability and willingness to
repay. The indicators considered under this are the client’s informal borrowings, group
members’loan experiences with MFI borrowings, and group cohesiveness. Moreover, the
credit risk could have regional ramifications. In India, the socio-political, economic and
cultural features vary across states. Consequently, the borrowers may tend to exhibit
different credit risk behavior according to the states they belong to. We thus capture this
factor as a regional dummy.
An analytical framework showing the plausible association between selected variables of
cost of microfinance intermediation and the borrowers’loan size is provided in Figure 1. The
operationalization of all the variables along with the hypothetical relationship is presented
in Table 1.
Indian
microfinance
institutions
The understanding of measurement of group cohesiveness merits attention here. To
measure the group cohesiveness, an index is created following Zeller (1998) that involves the
following five components:
(1) the members are similar in their economic background;
(2) they operate the income-generating activities in proximity;
(3) their residents are in the vicinity;
(4) the communication happens among the group members regularly; and
(5) they belong to the same caste.
The selected parameters are measured as binary (Yes = 1, No = 0). Then, applying principal
component analysis, we created a group cohesiveness index for each client.
3. Data and methodology
3.1 Data
The following method was used to select the study area and collect the cross-sectional
primary household-level data. The primary survey was carried out in two microfinance
leading states of India namely West Bengal and Odisha. The field survey was conducted
during March-April, 2016 and February-March, 2017 in West Bengal and Odisha
respectively. Though both the states are located in the eastern part of India, they differ quite
visibly on socio-economic characteristics. To be specific, Odisha is reported to be the poorest
Figure 1.
An empirical
framework exhibiting
the linkage between
the cost of
microfinance
intermediation and
the loan size
Informal Borrowing
Land Size
Asset
Income
Interest Rat e
Repayment Period
Group Size
Geographical Distance
Credit Risk FactorsTransaction Cost Factors
Cost of Microfinance Intermediation
Primary Occupation
(Agriculture/Non
Agriculture
Education Level
Age
Loan Experience
Loan Size
Group Cohesiveness
JFEP
state of the country with 45.9% of its population living below poverty line [Government of
India (GoI), 2014). As MFIs are intervened to serve the poor, the inclusion of Odisha may
provide useful information regarding the role of MFIs in a least-developed state. Contrarily,
West Bengal is relatively better in position with a much lower incidence of poverty (29.7%)
(GoI, 2014). Interestingly, in both the states, there has been a dramatic growth in
microfinance outreach in recent years. It seems relevant to understand how the microfinance
intermediation at the aggregate level gets reflected at the individual household level as well.
To conduct primary survey, from each state, three different districts were selected. The
selection of districts was based on two different criteria, namely, the extent of MFI outreach
and district per capita income. In the first step, districts were ranked according to the MFI
outreach and the districts that reported outreach above the corresponding state average
were identified. In the next step, randomly three districts from the identified ones were
selected having respectively high, medium and low per capita incomes (Table 2). A total
sample of 498 microfinance households having more or less balanced representation of both
rural and urban locations were chosen for personal interviews. The data was collected
through questionnaire-cum-personal interviews. The sample households were disbursed
loans under group lending mechanisms from different MFIs covering NBFC-MFIs,
cooperative societies and trusts. The profile of the sample borrowers/households is
presented in Table 3.
Table 1.
Operationalization
and hypothetical
relationship of the
variables with
loan size
Characteristics Parameters Operationalization
Expected
impact
Notation in
the model
Loan size Natural log of loan amount disbursed
to a borrower
Size
Cost of microfinance intermediation
Transaction cost
factors
Geographical
distance
Borrower’s residence physical distance
from the MFI office
þdis
Group size Number of borrower members in the
group
þgsize
Repayment period/
frequency
Weekly = 1, Others = 0 /þrep
Interest rate Loan interest rate int
Credit risk
factors
Income Natural log of household monthly per
capita income
þinc
Asset Index based upon total household asset
measured assigning relative weights to
household items (weights are allotted
on basis of asset probable market price)
þasset
Land Ownership of land (Yes = 1, No = 0) þland
Informal borrowings Other loans by household (Yes = 1, No
=0)
infb
Primary occupation Non-Agriculture = 1, Agriculture = 0 þocc
Age Average age of family adult members age
Education level Mode year of schooling attained in a
family
þedu
Members’loan
experience
Number of years’experience of MFI
borrowings
þexp
Group cohesiveness Index based upon group characteristics þgcoh
State West Bengal = 0, Odisha = 1 /þstate
Source: Author estimates
Indian
microfinance
institutions
The size of the groups varies from a minimum of four members to a maximum of 20
members. All the members of the groups were women only. In the sample, the loan size
varied widely among group members ranging from INR 5,000 to 170,000, while interest rates
varied with a range from 20.50% to 27%.
3.2 Model specification
To analyze the impact of different components of the cost of microfinance intermediation on
the borrowers’loan size, the following ordinary least squared (OLS) regression model is
applied [equation (1)].
Yi¼X
j
b
jxj;iþ
m
i(1)
Table 2.
Survey details
State
Districts
(per capita income) Sample size Urban Rural
West Bengal 24 North Parganas (High) 59 100 115
Murshidabad (Medium ) 108
Birbhum (Low) 48
Sub Total 215 100 115
Odisha Khurda (High) 97 142 141
Nayagarh (Medium) 83
Balangir (Low) 103
Sub total 283 142 141
Total 498 242 256
Table 3.
Profile of the sample
households
Characteristics Mean SD Minimum Maximum
Loan size (INR) 34,993.52 20,119.85 5,000.0 1,700,00.00
Geographical distance (Km) 9.25 1.98 0.50 30
Group size (number) 8.16 1.56 4 20
Repayment period (days) 7.00 30.00
Interest rate (%) 23.80 1.83 20.50 27.00
Monthly instalment (INR) 4,943.35 1735.66 445.00 13,580.00
Monthly per capita income (INR) 7,389.97 1,588.42 1,657.14 12,350.00
Monthly per capita expenditure (INR) 2,937.55 487.62 654.32 5,628.50
Household asset index (number) 31.20 8.11 10.12 60.00
Average age (age >14)(years) 33.37 5.56 21 64.50
Mode years of schooling (number) 11.13 3.21 0.00 17.00
Household loan experience (years) 4.55 2.79 0.00 15.00
Households having land ownership (%) 221 (44.38%)
Households having informal borrowings (%) 187 (37.55%)
Households having major occupation as non-agriculture (%) 408 (81.92%)
Households having weekly repayments (%) 244 (49.01%)
Source: Author estimates
JFEP
where i=1,2... 498 refers to cross-section units, Y
i
refers borrowers’loan size,
b
j
refers the slope coefficients and x
ij
is cross-sectional unit for j-th variables.
m
i
is the
random error term.
We segregate the cost of intermediation into transaction cost factors and credit risk
factors and run different regressions to see their separate impacts on loan size. To check the
impact of transaction cost factors on borrowers’loan size, the operational OLS regression
model (Model 1) is as follows:
LoanSizei¼
b
0þ
b
1gdisiþ
b
2gsizeiþ
b
3repiþ
b
4intiþ
b
5stateiþ
m
i(2)
LoanSize
i
represents the loan size of the i-th borrower. All the notations are defined in
Table 1.
However, to analyze the impact of credit risk factors on borrowers’loan size, similar
OLS regression model cannot be employed. There seems to be the presence of
endogeneity problem in the model and that may result in biased and inaccurate
estimates. The problem of endogeneity arises when there happens to be the presence of
an unobserved factor in the error term and that correlates with an independent variable
and the interaction influences the dependent variable (i.e. loan size). In MFI lending,
there is one substantial unobserved factor that could create an endogeneity problem.
There is a possibility of client selection bias on the part of the loan officers, which
remains uncontrolled. Loan officers may be reluctant to disbursing big loan size to low-
income households accounting for their low creditworthiness and high credit risk
(Maitrot, 2018). The unobservable factor seems to influence the household loan size and
may result in biased estimators.
However, the problem of endogeneity can be checked and corrected by using a valid
instrument variable. In the present study, we consider the client’s income as the endogenous
variable as it is a fundamental parameter that loan officers tend to count and become the
basis of selection bias to determine the loan size. Income acts as a direct measure of a client’s
creditworthiness. For estimation, we employ the two-stage least squared regression (2SLS)
that contains an instrument variable and that is correlated with the endogenous factor
(household income) but uncorrelated with the error term. The 2SLS regression model (Model
2) is expressed as follows:
LoanSizei¼
b
0þ
b
1inciþ
b
2assetiþ
b
3landiþ
b
4infb þ
b
5occiþ
b
6ageiþ
b
7edui
þ
b
8expiþ
b
9gcohiþ
b
10stateiþ
m
i
(3)
inci¼
a
þ
a
1ammiþvi(4)
The instrument taken for endogenous household income is the number of male adult
family members, as it reflects the household’s earning capacity and diversified
sources of livelihood. Adult male members (amm) are considered purposefully as we
observe the patriarchal nature of the Indian families and relatively greater economic
dependence on them. The family members having age above 14 years are selected
decisively as in India the working age is above 14years. The loan officer is expected
to easily quantify the household income by observing the number of adult male
members.
Indian
microfinance
institutions
To see the impact of both transaction cost and credit risk factors together, we run a
similar 2SLS regression due to the presence of endogenous household income variable
in the model. The operational model (Model 3) can be expressed as follows:
LoanSizei5
b
0þ
b
1gdisiþ
b
2gsizeiþ
b
3repiþ
b
4intiþ
b
5inciþ
b
6asseti
þ
b
7landiþ
b
8infbiþ
b
9occiþ
b
10ageiþ
b
11eduiþ
b
12expi
þ
b
13gcohiþ
b
14stateiþ
m
i(5)
inci¼
a
þ
a
1ammiþvi(6)
It can be argued that the geographical distance of borrowers’locations from the MFI
offices may increase both the transaction cost and credit risk and that may discourage
the MFIs to distribute loans at such locations. However, if a client’s household income
is high, then MFIs may consider it a perceived security to reduce the credit risk and
may disburse larger loan size. For MFI, the client’s household income signifies the
regular cash inflows that the borrowers can use to regularize their repayments. So to
examine how the client’s household income overcomes the constraints of geographical
distance and influences the MFI loan disbursements, we introduce an interaction term
[gdis
i
*inc
i
] of these two variables. Accordingly, the 2SLS regression model (Model 4)
is proposed as follows:
LoanSizei¼
b
0þ
b
1gdisiþ
b
2gdisiinci
½
þ
b
3gsizeiþ
b
4repiþ
b
5intiþ
b
6inci
þ
b
7assetiþ
b
8landiþ
b
9infbiþ
b
10occiþ
b
11ageiþ
b
12eduiþ
b
13expi
þ
b
14gcohiþ
b
15stateiþ
m
i(7)
inci¼
a
þ
a
1ammiþvi(8)
To obtain robust estimates, the study uses the Durbin–Wu–Hausman test to check the
presence of endogeneity. The significant test statistics indicate that the household income
must be treated as endogenous (Table 4). For the selected variable to become a valid
instrument, it must be sufficiently correlated with the endogenous household income and
uncorrelated with the error term. The first stage regression provides the statistics to decide
the explanatory power of the identified instrument. The significant F-statistics justify that
the identified instrument is sufficiently correlated with the endogenous household income.
For all the three endogeneity models (Models 2, 3 and 4), the estimated F-statistics are above
10, which makes the 2SLS highly reliable (Stock et al.,2002). The first stage regression also
produces the Cragg and Donald minimum eigenvalue statistics to test for the weak
instrument. The test statistic of above 10 proves that the instrument is not weak (Staiger
and Stock, 1997).
As estimations are carried out on cross-section data, the problems of heteroscedasticity
may not be ruled out. Consequently, the 2SLS regression with robust standard error is
applied. There are no problems of multicollinearity as in all the models the values of the VIF
are within tolerable limits (<5).
JFEP
Determinants Model 1 Model 2 Model 3 Model 4
Distance 0.028(1.68)
**
0.048 (2.03)
**
0.039 (1.85)
**
Distance
*
Income 0.020 (2.05)
**
Group size 0.012 (3.12)
***
0.011 (2.33)
***
0.011 (2.26)
**
Repayment period 0.010 (1.51)
*
0.005 (1.10) 0.005 (1.11)
Interest rate 0.015 (1.58)
*
0.006 (1.38)
*
0.004 (1.33)
*
Income 0.457 (2.04)
**
0.538 (2.27)
***
0.579 (2.58)
***
Asset 0.015 (1.67)
**
0.013 (1.55)
*
0.013 (1.54)
*
Land 0.060 (1.98)
**
0.076 (2.24)
**
0.080 (2.22)
**
Informal borrowing 0.005 (1.88)
**
0.006 (1.65)
**
0.005 (1.53)
*
Primary occupation (agriculture/non-agriculture) 0.023 (0.20) 0.035 (0.31) 0.034 (0.31)
Age 0.012 (2.44)
***
0.013 (2.71)
***
0.013 (2.65)
***
Education 0.015 (1.95)
**
0.013 (1.67)
**
0.012 (1.64)
**
Loan experience 0.027 (2.61)
***
0.015 (1.34)
*
0.014 (1.32)
*
Group cohesiveness 0.020 (1.56)
*
0.059 (0.43) 0.052 (0.37)
State 0.309 (6.71)
***
0.302 (4.01)
***
0.343 (4.53)
***
0.357 (4.11)
***
F-stat F
(5,492)
= 33.91
***
Wald
x
2
Wald
x
2
(10) = 233.94 Wald
x
2
(14) = 364.11 Wald
x
2
(15) = 361.00
Prob. >
x
2
0.001 0.001 0.001
R
2
0.24 0.26 0.42 0. 40
Observations 498 498 498 498
Robustness checks
Endogeneity
Durbin test
x
2
=18.28, p= 0.001
x
2
= 17.76, p= 0.001
x
2
= 18.40, p= 0.001
Wu-Hausman test F
(1,484)
= 18.52, p= 0.001 F
(1,482)
= 17.83, p= 0.0011 F
(1,481)
= 18.45, p= 0.001
Weak instrument identification
Cragg-Donald F-statistic 34.23 33.51 33.34
Multicollinearity
Variance inflation factor 1.15 1.11 1.15 2.11
Notes: 2SLS = two-stage least squares,*10%, **5%, ***1% level of significance respectively. Figures in parentheses are t-values
Source: Author estimates
Table 4.
OLS and 2 SLS
Regression estimates
to determine
borrowers’loan size
(dependent variable)
Indian
microfinance
institutions
4. Results and discussion
The regression results that estimate the impact of different components of transaction cost
and credit risk on borrowers’loan size in group lending mechanisms are presented in
Table 4. As far as the transaction cost factors are concerned, the estimates show that as the
geographical distance of the borrowers’locations from MFIs’offices increases the loan size
of the borrowers becomes smaller. The result may seem to be contrary to the common belief
that the loan size of distantly located clients should be larger to cover the increased
transaction cost of large distances. However, the results may not be surprising as it may be
argued that for MFIs large distance tends to upsurge the cost of monitoring, supervision and
credit risk and if bigger loans are distributed to the distantly situated clients, the MFIs may
seem to be at loss on any or all such fronts.
For MFIs, the fear of credit risk seems to override the transaction cost. This argument is
reinforced by the results of the interaction term between the geographical distance the
household income (a credit risk factor) (Model 4). Interestingly, the coefficient of the
interaction term is positive, which signifies that the loan size increases with distance
provided the client’s household income is high. The income by itself also exerts positive
influence on the loan size. For MFIs and for that matter, for any financial institutions,
income is considered as a primary source of creditworthiness of the clients. Income acts as a
collateral and indicates regular inflows of cash and regularization of the periodic
repayments. The result corroborates the findings of Dorfleitner et al. (2017), who find that
MFIs exhibit a higher tendency of disbursing bigger loan sizes to households with higher
incomes. The bigger loan size can be attributed to the fact that wealthier households are
likely to be considered more creditworthy both from the group members and the MFIs.
Contrarily, the credit risk is perceived higher with the low-income households (Dutta and
Banerjee, 2018). From the demand perspectives, one may argue that the loan demand of
higher-income clients are bigger and they are anticipated to use the credit more
productively, while the low-income clients may be utilizing the credit more in contingency
spending such as medical emergencies, marriage, school fee, disburdening of other loans,
which may further weaken their loan repayment capacity (Dutta and Banerjee, 2018;
Sangwan et al.,2020).
The coefficient of group size shows that borrowers’belonging to bigger groups tend to
receive smaller loan amounts. The result corroborates the findings of Singh and Padhi (2017)
that group size has a negative impact on borrowers’loan size. The larger group sizes seem to
invite internal management issues, the problems of free-rider, moral hazard and increased
credit risk (Ahlin, 2015;Singh and Padhi, 2017). The frequency of repayments of loans does
influence the loan size but does not seem to be that important as it turns out to be significant
only in the first estimation model (Model 1).
Turning to interest rate, the size of the household borrowings tends to decrease with the
increase in interest rates. In MFI lending, higher interest rates are transferred to clients
either in the form of their big loan installments or an increase in loan tenure. Though
microfinance borrowers may be little aware of the interest rates, they are well aware of their
weekly installments and loan tenures (Tiwari et al., 2008). In both the cases, clients perceive
a high burden of periodic repayments. Perceiving the high repayment burden, the clients
reduce their borrowing amount and the loan size becomes smaller. The other plausible
argument for this inverse relationship can be the cross-subsidization. When the MFIs
manage to reduce their transaction cost by disbursing the bigger loan amount, they seem to
subside the interest rate.
Among the other factors of credit risk, loan size seems to be influenced by clients’assets,
land size, informal borrowings, age, education and loan experience. For financial
JFEP
institutions, clients’asset holdings act as potential securities. In rural areas, MFIs consider
borrowers’land ownership as prime collateral and for the MFIs, an increase in the value/
volume of the clients’asset and land size mitigates the cost of credit risk (Simtowe et al.,
2008).
The MFIs seem to disburse bigger loan sizes to the households endowed with relatively
younger people. It can be attributed to the MFIs’perceived lower credit risk for the younger
people than the older ones. The likely cause could be young clients’better job opportunities
and greater longevity for the services, that enhance their level of income generation
(Bandyopadhyay and Saha, 2011;Sangwan et al.,2020). Following the life-cycle hypothesis,
it may be further argued that the young clients, in quest of high investment, tend to exhibit
high loan demands, while the chances of older ones to venture into investment activities are
minimal, and if any, they are more likely to invest their past savings (Banerjee and Duflo,
2011;Mpuga, 2010).
The findings further show that an household’s education level has negative impact on
the loan size, hence corroborating the findings of Tiwari et al. (2008),whofind that less
educated people are the major clients of MFI lending. They are more concerned with the loan
amounts than the cost of borrowing. Higher educated households, on the contrary, prefer
borrowings from formal financial institutions. The interactions of the study team with the
relatively more educated clients during the field survey also revealed that these people
possess better reasoning abilities and consequently, they tend to make a benefit-cost
analysis of borrowings. They are well-informed of the alternative sources of credit and the
relative cost differences.
The group members with earlier MFI borrowing experiences tend to receive bigger loan
amounts. It may be attributed to the clients’loan experiences reflecting their long-term
credit relationships with the lenders. Clients having good credit repayment history are
believed to be trusted by the lenders and the latter may be inclined to levy concessional
interest rates or provide larger loan sizes. Besides, the positive relationship of borrower’s
loan experience with loan size suggests the persistent use of dynamic incentives where the
client’s credit limit increases as the time spent increases between the borrower and lender
(Singh and Padhi, 2017).
The regional characteristics are also found to have an impact on the loan size. Between
the two states, MFIs in Odisha seem to be disbursing a relatively larger credit amount
compared to their counterparts in West Bengal. From our observations during the field
survey, it can be ascertained that the MFI operations are almost saturated in West Bengal,
because of their vast outreach. Consequently, the MFIs are gradually expanding their
operations to neighboring state of Odisha, as the latter is still having a much lesser outreach.
Given the lesser outreach and large untargeted population in the state, the MFIs are possibly
encouraged to disburse the larger loan size.
5. Conclusions
This study analyzed how the cost of microfinance intermediation creates an impact on the
borrowers’loan size. It was based on cross-sectional primary survey data collected from a
sample of MFI clients covering two eastern states of India. It identified transaction cost and
credit risk factors that the lenders tend to take into account while screening and allocating
loan amounts to the prospective borrowers. The identified factors may help the lenders to
minimize the adverse impact of asymmetric information over the clients’ability to repay.
One may ascertain that under asymmetric information, the loan officers’decisions to screen
and sanction the loan amount for the borrowers may be based on the measures of household
income, asset, land size, age, education and loan experience. The geographical distance and
Indian
microfinance
institutions
group size also seem to affect the decision to determine borrowers’loan size. The rise in
income plays an overriding role over the geographical distance, reflecting thereby the
critical importance of credit risk in loan size. It is equally important that MFIs focus on the
interest rate rationalization to increase their loan disbursements.
One may, thus, infer that the reduction in the cost of microfinance intermediation seems
to be the underlying motive behind the increase in MFI loan size for the borrowers. A
reduction in transaction costs and credit risks may help MFIs lower down the interest rates
that may eventually lead to a higher demand for credit in the financial market. Besides, the
reduction in the cost of microfinance intermediation will encourage the MFIs to increase the
outreach to new locations and to provide financial services to more number of financially
excluded low-income households.
References
Ahlin, C. (2015), “The role of group size in group lending”,Journal of Development Economics, Vol. 115,
pp. 140-155.
Armend
ariz, B. and Morduch, J. (2010), The Economics of Microfinance, 2nd edition, MIT press.
Arrow, K.J. (1963), “Uncertainty and the welfare economics of medical care”,The American Economic
Review, Vol. 53 No. 5, pp. 941-973.
Bandyopadhyay, A. and Saha, A. (2011), “Distinctive demand and risk characteristics of residential
housing loan market in India”,Journal of Economic Studies, Vol. 38 No. 6, pp. 703-724.
Banerjee, A.V. and Duflo, E. (2011), Poor Economics: A Radical Rethinking of the Way to Fight Global
Poverty, Public Affairs.
Basu, P. and Srivastava, P. (2005), Scaling-up Microfinance for India’s Rural Poor, The World Bank.
Bhatt, N. and Tang, S.Y. (2002), “Determinants of repayment in microcredit: evidence from programs in the
United States”,International Journal of Urban and Regional Research, Vol. 26 No. 2, pp. 360-376.
Dehem, T. and Hudon, M. (2013), “Microfinance from the clients’perspective: an empirical enquiry into
transaction costs in Urban and Rural India”,Oxford Development Studies, Vol. 41 No. sup1,
pp. S117-S132.
Dorfleitner, G., Just-Marx, S. and Priberny, C. (2017), “What drives the repayment of agricultural micro loans?
Evidence from Nicaragua”,The Quarterly Review of Economics and Finance,Vol.63,pp.89-100.
Dutta, A. and Banerjee, S. (2018), “Does microfinance impede sustainable entrepreneurial initiatives among
women borrowers? Evidence from rural Bangladesh”,Journal of Rural Studies, Vol. 60, pp. 70-81.
Fernando, N. (2006), Understanding and Dealing with High Interest Rates on Microcredit: A Note to
Policy Makers in the Asia and Pacific Region, Asian Development Bank.
Government of India (GoI) (2014), Report of the Expert Group to Review the Methodology for
Measurement of Poverty, Planning Commission, New Delhi.
Haldar, A. and Stiglitz, J.E. (2016), “Group lending, joint liability, and social capital: insights from the
Indian microfinance crisis”,Politics and Society, Vol. 44 No. 4, pp. 459-497.
Helms, B. and Reille, X. (2004), “Interest rate ceilings and microfinance the story so far, consultative
group to assist the poor”, available at: www.cgap.org (accessed 28 October 2019).
Hossain, S., Galbreath, J., Hasan, M.M. and Randøy, T. (2020), “Does competition enhance the double-bottom-
line performance of microfinance institutions?”,Journal of Banking and Finance, Vol. 113, pp. 1-17.
Jumpah, E.T., Osei-Asare, Y. and Tetteh, E.K. (2019), “Do farmer and credit specific characteristics
matter in microfinance programmes’participation? Evidence from smallholder farmers in Ada
west and east districts”,Agricultural Finance Review, Vol. 79 No. 3, pp. 353-370.
Maitrot, M. (2018), “Understanding social performance: a ‘practice drift’at the frontline of microfinance
institutions in Bangladesh”,Development and Change, pp. 1-32.
JFEP
Mason, D.R. (2014), “Who gets what? Determinants of loan size and credit rationing among microcredit
borrowers: evidence from Nicaragua”,Journal of International Development, Vol. 26 No. 1,
pp. 77-90.
Mendoza, R.U. (2011), “Why do the poor pay more? Exploring the poverty penalty concept”,Journal of
International Development, Vol. 23 No. 1, pp. 1-28.
Mersland, R. and Strøm, R.Ø. (2009), “Performance and governance in microfinance institutions”,
Journal of Banking and Finance, Vol. 33 No. 4, pp. 662-669.
Mohapatra, S. and Sahoo, B. (2016), “Determinants of participation in self-help-groups (SHG) and its impact
on women empowerment”,Indian Growth and Development Review,Vol.9No.1,pp.53-78.
Mpuga, P. (2010), “Constraints in access to and demand for rural credit: evidence from Uganda”,
African Development Review, Vol. 22 No. 1, pp. 115-148.
Okurut, F.N., Schoombee, A. and Berg, A. (2005), “Credit demand and credit rationing in the informal
financial sector in Uganda”,The South African Journal of Economics, Vol. 73 No. 3, pp. 482-497.
Ranjani, K.S. and Kumar, S. (2018), “An investigation of mission drift in Indian MFI”,International
Journal of Social Economics, Vol. 45 No. 9, pp. 1305-1317.
Sa-dhan (2018), “The Bharat microfinance report 2018”, Sa-Dhan, New Delhi.
Sangwan, S. and Nayak, N.C. (2019), “Do outreach approaches differ between self-help group-bank
linkage and microfinance institution-based microfinance? Evidences from Indian states”,Journal
of Social and Economic Development, Vol. 21 No. 1, pp. 93-115.
Sangwan, S., Nayak, N.C. and Samanta, D. (2020), “Loan repayment behavior among the clients of
Indian microfinance institutions: a household-level investigation”,Journal of Human Behavior in
the Social Environment, Vol. 30 No. 4, pp. 474-497.
Shankar, S. (2007), “Transaction costs in group microcredit in India”,Management Decision, Vol. 45
No. 8, pp. 1331-1342.
Shu-Teng, L., Zariyawati, M.A., Suraya-Hanim, M. and Annuar, M.N. (2015), “Determinants of
microfinance repayment performance: evidence from small medium enterprises in Malaysia”,
International Journal of Economics and Finance, Vol. 7 No. 11,pp. 110-120.
Simtowe, F., Diagne, A. and Zeller, M. (2008), “Who is credit constrained? Evidence from rural Malawi”,
Agricultural Finance Review, Vol. 68 No. 2, pp. 255-272.
Singh, V. and Padhi, P. (2017), “Dynamic incentives and microfinance borrowers: a comparative study
of self-help groups and joint liability groups: a case study of Mirzapur district in India”,Journal
of Land and Rural Studies, Vol. 5 No. 1, pp. 67-92.
Srinivasan, N. (2015), “Transaction costs of lending to vulnerable people”,Microfinance Review, Vol. 8
No. 2, pp. 212-219.
Staiger, D.O. and Stock, J.H. (1997), “Instrumental variables regression with weak instruments”,
Econometrica, Vol. 65 No. 3, pp. 557-586.
Stiglitz, J.E. and Weiss, A. (1981), “Credit rationing in markets with imperfect information”,American
Economic Review, Vol. 71, pp. 393-410.
Stock, J.H., Wright, J.H. and Yogo, M. (2002), “Asurveyofweakinstrumentsandweakidentification in
generalized method of moments”,Journal of Business and Economic Statistics, Vol. 20 No. 4,
pp. 518-529.
Swamy, V. (2019), “Macroeconomic significance of transaction costs in microfinance intermediation”,
Management Decision, Vol. 57 No. 9, pp. 2307-2324.
Tiwari, A., Khandelwal, A. and Ramji, M. (2008), “How do microfinance clients understand their
loans?”, Institute for Financial Management and Research, Centre for Micro Finance Working
Paper Series, Vol. 25.
White, R. and Alam, S.A. (2013), “Asset ownership and microloan repayment: examining data from
Bangladesh”,Development Policy Review, Vol. 31 No. 3, pp. 343-368.
Indian
microfinance
institutions
Yimga, J. (2018), “Microfinance expansion and its effects on cost efficiency”,The Quarterly Review of
Economics and Finance, Vol. 69, pp. 205-216.
Zeller, M. (1998), “Determinants of repayment performance in credit groups: the role of program design,
intragroup risk pooling, and social cohesion”,Economic Development and Cultural Change,
Vol. 46 No. 3, pp. 599-620.
Corresponding author
Sunil Sangwan can be contacted at: challengersangwan@gmail.com
For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com
JFEP
