Comments on "Effects of Social Capital on Access to Microcredit in Ethiopia: A Study on

Small Scale Farmers in Bahir Dar District"

Overall, I enjoyed reading the paper.

1. I have made substantial changes using track changes. Please have a look at them and
change or reject the suggested changes
2. There are some works that have been cited in the study that miss references (e.g.
Klapper and Singer, 2013). Please ensure that all citations are referencerd
3. There are a number of repetitions; please avoid them.
4. You used only the variables you constructed as social capital—what about other indicators?
a. Wouldn’t this imply that you have omitted variable problem? In fact, muticollinearity
may not be a problem as increasing the sample size or dropping some (irrelevant
explanatory) variables could be a solution in cross section data. This is a problem in
time series data!
b. Homoskedasticity is a more serious problem in cross section data but you do not
seem to consider this.
5. I have inserted specific comments for the other issues
 Effects of Social Capital on Access to Microcredit in Ethiopia: A Study on
Small Scale Farmers in Bahir Dar District

Aderaw Gashayie (PhD)*

ABSTRACT
The purpose of this study was is to identify effects of social capital on access to microcredit
access in Ethiopia: . It uses data obtained from a study on small scale farmers living in five
villages; namely, Wogelsa, Yibab, Yinsa Sostu, Wondata, and Sebatamit, that are located
in Bahir Dar district. In this study, social capital variables (density of membership,
heterogeneity index, labor contribution, cash contribution, meeting attendance index and
decision making index) .were the hypothesized factors to predict small scale farmer’s
access to microcredit in Ethiopia. To this end, the study conducted in Bahir Dar district in
five selected villages: Wogelsa, Yibab, Yinsa Sostu, Wondata, and Sebatamit. The data
were collected using By a two stage random sampling method, . The sample consists of the
data were collected from 226 rural small scale farmers. A sStructured interview
questionnaire were was the instruments used for primary data collection of the sampled
farmers. Moreover, semi-structured questionnaires were used to obtain information from
by interview, focus group discussions, and key informants. discussion. Secondary data
were also collected from annual reports, minutes and company websites. The data
collected were analysed by using logistic regression. Variables used to indicate social
capital are: density of membership, heterogeneity index, labor contribution, cash
contribution, meeting attendance index and decision making index Findings from the data
analysis indicates ed that decision making index and heterogeneity indicesex significantly
affects access to microcredit access of small scale farmers. The study found that social
capital increased the probability of households having access to microcredit. These
findings suggest that there is a need for encouraging stakeholders (both private individuals
and enterprises) to use informal networks to increase microcredit access.

Keywords: Ethiopia, microcredit, accessibility, logistic regression

CHAPTER ONE
[1.] IntroductionBackground and justification of the study
Inadequate resources and the difficulty in mobilising whatever is availablethem, that lead to
coupled with inadequate access to financial resources, notably credit, are is a common
challenge in most sSub-Saharan African (SSA) countries. In this part of the world, mMost
people in the region are excluded from the mainstream financial system (Klapper and Singer,
2013). The most affected of them are the poor, (those who mainly live in rural areas) and
women. According to the United Nations Development Programme (UNDP, 2011), in 2011,
one billion poor people in the world did not have access to basic financial services. In
developing countries, two-thirds of the working population do not always have access to
financial services, and 80% of households and enterprises are excluded from the formal
financial system.

*
In SSA countries, in 2013, only 23% of adults had a bank account, compared to 41% in
developed countries. This rate has remained low despite a rise in 2014 to 34% in SSA,
against 62% in the world as a whole (Demirguc-Kunt et al., 2014).
The awarding of the Nobel Peace Prize to Muhammad Yunus in 2006 was a form of
recognition of the contribution of microfinance to the development of financial markets in
developing countries and to poverty reduction. Undoubtedly, microfinance institutions
(MFIs) have facilitated access to financial services and are considered today as one of the
tools that contribute to development. However, enormous effortsmuch more needs to still
have to be done made by the MFIs, as two-thirds of the unbanked people in developing
countries still lack access to formal financial services (Duflo and Parenté, 2009).
In Ethiopia, after more than two decades, following the enactment of the law on freedom of
association and that on cooperative societies and joint initiative groups, access to credit by
households was is still limited;, this is especially in rural areas. From the supply side,
Aaccording to NBE annual report (2020), the Commercial Bank of Ethiopia (CBE),
Development Bank of Ethiopia (DBE), construction and business bank, and other 16 private
banks have a total of 6511 branches and the ratio of population to bank branch is 15701,
Commercial bank branches per 100,000 adults is 2.95 and Commercial bank branches per
1,000 km square is 1.53. This shows that not only the bank branches do not cover a number
of districts but also in localities where bank branches exist, the majority of the population has
no access to financial services, due to high collateral requirements (NBE, 2020).
In order to reach a large number of the population, particularly those residing in rural areas,
To address this issue, the government of Ethiopia established microfinance institutions
(MFIs) to reach a large number of rural poor. WithIin the past 22 years, 31 microfinance
institutions (MFIs) have been established that are and have been operational. Their main
focus is ng towards resolving the ensuring access to credit access problem ofor the poor
particularly those who engage in petty business (Befekadu, 2007). The MFIs, however, have
no clear rules and criteria for targeting the poorest of the poor indicating that MFIs are
drifting away from their original mission of reaching the poor (Ejigu, 2009). There is
evidence which shows that a large number of rural farmers are marginalized, and thereby do
not have access to micro credit (due to high transaction cost associated with small size of the
loan and different requirements imposed by the lenders). Besides, due to high default and
lack of effective enforcing mechanisms lenders restrict supplying credit to limited borrowers
(Stieglitz and Weiss, 1981). Because of this, in rural Ethiopia only 34% of the credit demand
of the poor is reached by microfinance institutions (Ejigu, 2009). It is estimated that over 2.9
mMillion economically active poor people are outside the reach of the conventional financial
service in Amhara regional state. ACSI and other smaller MFIs and Ssaving and Credit
Cooperatives (SACCOs) in the region only manage to reach between 10% and 12% of the
demand. There are many economically active poor people that are still un-reached (Getaneh,
2005).
From this state of affairs arises the questions: of wWhat the factors that can do lenders use
explain thein granting of credit by lenders in Ethiopia? are, and, on the other hand, w What
the deters rents against households from applying for credit by households are?. Is it

iii
socioeconomic, geographical, demographic, and financial factors, or is it the membership of a
network and access to information, i.e., social capital, that influence access to credit(social
capital)? (Ajani and Tijani, 2009; Guerin et al., 2011; Osotimehin et al., 2011). It is likely that
aAll or some elements of social capital these factors may cause lenders to exclude individuals
with weak social capitalvaluable stakeholders from the financial market. Stiglitz and Weiss
(1981) These authors argue that the formation of associations or other social networks can
reduce information asymmetry and thus allow the lenders to finance the best projects,
because information asymmetry leads to adverse selection and to moral hazard (Stiglitz and
Weiss, 1981).
Muchany of the rResearch has already been conducted in Africa and has shows n that social
capital; (as measured by the fact of belonging to an institution, to associations, and, recently,
by the homogeneity index, the social-network index, the contract-compliance index, the
collective membership index, and the decision-making index) and household wages (as
measured by the income of the head of household) are the main determinants of access to
credit (Bastelaer, 2000; Ajani and Tijani, 2009; Lawal et al., 2009; Heikkilä et al., 2009;
Balogun and Yusuf, 2011; Guerin et al., 2011; Karlan and Zinman, 2011; Osetimehin et al.,
2011; Kangogo et al., 2013; Sadick et al., 2013; Banerjee et al., 2015). However, practically
no similar research has yet been done concerning Ethiopia households. Most of the research
conducted in Ethiopia has generally focused on the characteristics of the head of household,
particularly on his/her age, income, and socio-professional category. Studies on social capital
as an explanatory factor for access to microcredit are rare.
This study was conducted to identify the effect of social capital on microcredit access in
Bahir Dar districtthe study area. Specifically the study addresses the ‘’" what is the effect of
social capital on access to microcredit access"’’?
The objective of the study is to identify effect of social capital on microcredit access in
Ethiopia on small scale farmers in Bahir Dar district.
To address the basic research questions and research objectives, the following hypothesis
were are put forward suggested and tested:
Hs1: high frequency of meeting participation positively influences farmer’s access to
microcredit.
Hs2: Participation of households in decision making in associations positively influences
farmer’s access to microcredit.
Hs3: farmers who are members of associations have more access to credit from cooperative
source.
Hs4: homogeneity increases the amount of credit secured from different microcredit sources.
Hs5: increased cash contribution to any association increases the amount of credit secured
from different microcredit sources.
Hs6: Increased labour contribution to any association increases the amount of credit
secured from different microcredit sources

iv
While the aim of thisThe study aims a to identifying identify the effects of social capital on
smallholder farmer’s access to microcredit in Ethiopia., its The scope of this research is
restricted to Bahir Dar district/zuriya, West Gojjam zone in Ethiopia. Data used in this
research are is collected from a small-scale household survey covering different areas within
the province. Therefore, the results of this research may not be applicable to the whole
country. In addition, this research focuses on the microcredit programme implemented by in
the district only. Therefore, the results cannot represent the characteristics (such as client
target) and economic/social influences of the overall Ethiopian microfinance sector which
consists of different types of institutions with different business goals and objectives. This is
mainly because of limited availability of resources and time to undertake the study on a wider
scale.
The major significance of the study includestudy's importance includes, among others:s:
 It is Important in Pprovidesing information that enables to takeformulation of
effective decisions making processes for financial regulators, financial donors, and
microfinance service providers to improve access to credit.
 It is useful to Iidentify innovative options and institutional arrangements that would
serve as an input for policy makers in formulating rural credit policy.
 It aAdds some empirical knowledge of Ethiopia to the existing rural financial
knowledge in the world in the country and the world at large.
After this background and justification of the study chapter, tThe remainder of the paper is
structured as follows: the next section Chapter 2 reviews theoretical and empirical literatures
related to effect of social capital on microcredit access; Chapter 3which will be followed by a
presentation of presents research methods. Chapter 4It then presents deals with presentation
and of data and its analysisanalysis of data. Chapter 5Finally, presents it presents a
summaryies of the major research findings, conclusions and& recommendations of the study.
[2.] CHAPTER TWO
[3.] Literature review
1.1. Concepts of social capital
In this chapter a review of available literature, both theoretical and empirical, has been done
for gaining a perspective on the problems and issues related to the effect of social capital on
microcredit access in Ethiopia.

2.1 concepts of social capital

Social capital as defined by Portes (1998) stands for ability of people to secure benefits by
virtue of their membership in social networks or other social structures. However, according
to Narayan and Pritchette (1999) provide a more general description and identify it as, social
capital refers to the internal and cultural coherence of society, the norms and values that
govern interaction among people and institutions, in which they are embedded. For Clercq
and Dakhli (2004) social capital Social capital as posited by Clercq and Dakhli (2004) is
generated from informal institutions, and is considered as an additional factor of production.

v
Social associations have a direct bearing on the contacts that help in accessing resources and
helping the economic processes. An example of such associations is the Grameen micro-
credit program in Bangladesh where village women benefit from their affiliation with a
particular group by obtaining loans without collateral (Woolcock and Narayan, 2000). Social
capital as posited by Clercq and Dakhli (2004) is generated from informal institutions, and is
considered as an additional factor of production. AlsoMoreover, individuals with high level
of social capital possessors can gain direct access to economic resources like subsidized
credit, protected market etc.; consequently, and can increasing e their cultural capital through
contacts with experts and affiliate with institutions that can give institutional support (Portes,
1998). A growing body of literature is adding weight to the concept that social capital plays
an important role in financial services to the peasant small-scale farmers. This would lead to a
relatively healthier process and consequently theof rural development process.
1.2. Measurement dimensions of social capital
2.2 SOCIAL CAPITAL MEASUREMENT DIMENSIONS
Different literature on social capital discusses various approaches to measure social capital.
Measuring social capital is a notoriously difficult exercise partly caused by the co-existence
of multiple definitions of what comprises social capital and partly because it is based on
elusive and intangible proxies rather than facts. Different scholars have used measur
indicators of ed social capital based on specific aspect of the problem they sought to
addressthat they approached;, this leads to diverse indicators for its measurement based on
both in terms of structural and cognitive terms.
The effectiveness with which social capital, in the form of local associations, can fulfill its
role in disseminating information, reducing opportunistic behaviour and facilitating collective
decision making depends on many aspects of the association, reflecting its structure, its
membership and its functioning. This study used the following six indices in measuring social
capital: density of membership, heterogeneity index, labor contribution, cash
contribution, meeting attendance index and decision making index which are adopted
from Grootaert (1999) and the measurements of each is described below:
1. Density of membership: A complete inventory of all associations at local level
institutions was presented and each household was asked which associations they
belong to. This was measured by the number of active household members in existing
associations, and the number of active membership in each household was then
normalized by household size.
2. Group Heterogeneity index. The questionnaire identifies the two most important
associations for each household. For those associations, a number of supplementary
questions were asked including internal homogeneity of the group. This was rated
according to ten criteria: neighborhood, kin group, same income group, same religion,
same sex, age, level of education, belief and cultural practices, and trust, hence, for
each of the factors a ‘yes’ response was coded ‘2’ while a ‘no’ was coded ‘1’. A
maximum score of 20 for each association represents the highest level of
homogeneity. The score of the two associations was averaged for each household by
dividing by the maximum score 40 to obtain the index. The resulting index was then

vi
 multiplied by 100 (whereby a zero value represents complete heterogeneity and 100
correspond to highest homogeneity).
3. Decision making index. It has been argued that associations which follow a
democratic pattern of decision making are more effective than others. This is
calculated by summation of the subjective responses of households on their rating in
the participation in decision making of the two most important institutions to them.
This response was scaled from 4 to 0 respectively and averaged across the three most
important groups in each household and multiplied by 100 for each household.
4. Cash contribution index. This was achieved by taking records of payment of
membership dues and other contributions. The summation of the total cash
contributed to the various associations which the household belong was calculated.
The actual contributions for each household were rescaled by dividing the amount by
the maximum fee in the data and the resultant fraction was multiplied by 100.
5. Labor contribution index. This is the number of days that an individual member
belonging to an association claimed to have worked for the association. This
represents total numbers of man-days worked by household members. This was also
scaled to 100 using the same method of cash contribution.
[6.] Meeting attendance index. This index was measured by finding the number of times
members of an association actually met as a group over a period of time. This was
obtained by summing up attendance of household members at meeting and relating it
to the number of scheduled meetings of the associations. The value was then
multiplied by 100.i.e the ratio of number of times meetings were attended to the
number of meetings were called.
[3.1.] 2.3 the The effect of social capital on microcredit access
Studies on the effect of social capital on access to microcredit access have been were
conducted in different countries including Indonesia, Nigeria, Bangladesh, Kenya, Vietnam
and Uganda. Generally, most all of the researcherssuch studies considered social capital as
networks that belonged to thewhereby individuals are members, actors and werewere utilized
for obtaining the credit through the transaction of the network. However, tTheir
dissimilarities among them were the type of approach for studies.
Okten and Osili (2004) started at the context of strong social network in Indonesia first and
then tested its effect on credit access of local people. Evidently, Indonesia has good
conditions for studying the role of family and community network as well asto its effect on
credit market. This was also examined by Grootaert(1999) when he found that each
household’s member participated in five associations on the averagely and number of time
that each member participated in association meeting was 6.0 in three-month period.
Moreover, other indicators for measuring social capital in Indonesia such as index of
contribution and index of making decision were also high Grootaert (1999). For those
reasons, it is believed that the people in Indonesia hasve strong networks among local people.
In addition, the credit system has been diversified in terms of formal and informal institution
especially the success of Bank Rakyat Indonesia through using the role of community in loan
procedure (Okten & Osili, 2004).

vii
Balogun & and Yusuf (2011) examined critically the key factors influencing demand for
micro credit among rural households in Ekiti and Osun states, Nigeria. A multistage sampling
was employed for the study. Ekiti and Osun states were randomly selected from the six states
in South-western Nigeria. This was followed by random selection of two Local Government
Areas (LGAs) from each senatorial district of the states. Thirty microcredit groups (MGs)
were randomly selected from each of the selected LGAs based on probability proportionate to
the size of the MGs. Lastly, 399 respondents were randomly selected from the MGs. Data on
household demographic characteristics, social capital and microcredit variables were
collected with structured questionnaire. The data were analyzed using descriptive and
multinomial logit model. Their result of multinomial model shows that household social
capital variables (membership density index, meeting attendance index, cash contribution
index and& heterogeneity index), dependency ratio and credit variables (credit distance &
and interest rate) significantly explained households demand for credit. Social capital
variables significantly influenced the amount of credit available from different sources.
Ho Le Phi Khanh (2011) examined the relation between social capital factor and capacity
for obtaining the credit sources of local people in coastal area of Thua Thien Hue province in
Vietnam. Moreover, it It analyzed the role of social capital to access different credit sources
as well as explores the effect of network types horizontally and vertically on obtaining the
credit from Bank, “hui” groups and moneylenders. The study applied a qualitative approach
based on focus group discussion and in-depth interviews. The findings indicate that social
capital affects the access to credit access from the bank through group lending which is
assessed by participating in popular organizations. Neighbor network, which reflectis
exposed by generalized trust, reputation, balanced reciprocity, and mutual aid activity,
facilitates to establish “hui” group in order to obtain rotating credit. On the other hand,
vertical relation, which is reflectsexposedmirrors by trust, generalized reciprocity, and
reputation, supports to receive the loans from moneylenders. The study concludes that, the
social capital, which is addressed by horizontal and vertical network help for obtaining credit
from both formal and informal credit sources.
Daniel et.al (2013) studied the social capital dimensions and other determinants influencing
household participation and level of participation in micro-credit groups in Uasin Gishu
Kenya specifically Moiben division. A structured questionnaire was used to gather
information from 174 households from the division, using the multistage sampling technique.
Heckman selection model was applied to identify factors that influenced households to join
and the level of participation in the micro-credit group. The results indicate that age, gender,
education farm size, household size, farm income and distance to the nearest financial
institution influenced household decision to join the micro-credit groups. On the other hand
age, farm size, total income, heterogeneity index, density of membership, years of experience
in group borrowing and decision making index significantly influenced the level of
participation in [microcredit].
Ivanda et al (2014) examined the effect of social capital on access to credit among cassava
farming households (CFHs) in Ogun State, Nigeria. One hundred and twenty CFHs were
surveyed using a multi-stage sampling technique. Analyses included descriptive statistics and
regression technique. Social capital dimensions considered are density of membership index,
viii
cash contribution index, labor index, decision making index, meeting attendance index, and
heterogeneity index. Logistic regression analysis of access to credit revealed that increasing
values of decision making index, age, and payback period correspond to increasing having
access to credit. Conversely, increasing values of heterogeneity index and household size
correspond to decreasing of having access to credit.
Sarker and& Islam (2014) examined the effects of social capital on access to microcredit in
rural Bangladesh; . A cross-sectional study was conducted in Tuker Bazar Union Parishad in
Sylhet district, Bangladesh. and data was collected through face to face interview by social
survey method. Descriptive statistics was applied to present the data and probit-regression
was applied to measure the effects of social capital on access to microcredit. Their study
results shows that all dimensions of social capital were low in this area. The results show that
age of the household head was positively associated but size of the family was negatively
associated with access to microcredit. The study revealed that aggregated social capital was
positively associated with access to microcredit. Among the social capital dimensions:
density of membership, cash contribution and labor contribution were positively
associated with access to microcredit in this area. So, this study revealed that the social
capital enhance the probability to access to microcredit in rural areas.
To sum up, all of researches have expressed the relation between social capital and credit
access through economic model that were formed by indicators for measurement. However,
there is no study conducted how social capital affect access to microcredit in Ethiopia.
[3.2.] 2.4 study fFramework for the study
From theoretical and empirical literature reviews made above and the author'sand information
obtained from the FGDs and KIIs knowledge of theabout credit schemes of available in the
study area the following study framework developed and it depicted the most important
variables expected to influence smallholder farmers’ access to microcredit in the study area.

Social capital
variables
Frequency of attending
meeting
Level of participation in
decision making small scale farmers
Membership in access to micro
associations credit
Level of heterogeneity in
associations
Amount of Cash
contribution
Amount of Labor
contribution

ix
Dependant variable
The logit model dependent variable is dichotomous nature indicating rural households’ access
to microcredit. since direct measurement of microcredit accessibility is impossible , the
‘accessibility’ is measured by using observations on household borrowings such as ‘obtained
micro loans’ and ‘did not obtain micro loans’. This is in accordance with previous studies
which adopted observable formal or informal borrowings as indicators of credit accessibility
(see for example, Yehuala, 2008; Ravi, 2003; Mohamed, 2003; Vaessen, 2001).
Specifically, the dependent variable (Yn) takes a value of ‘1’ for households who have
secured micro loans from rural microcredit providers and ‘0’ for households who have never
secured micro loans from rural microcredit providers.
Explanatory variables of the study
Review of literatures on factors influencing small scale farmers’ access to microcredit, past
research findings and the author's knowledge of the credit schemes of the study area are used
to establish working hypotheses of this study. In other words, among a number of factors,
which have been related to small scale farmers’ access to formal credit in this study, the
following, six main variables (Frequency of attending meeting, Level of participation in
decision making, Membership in associations, Level of heterogeneity in associations,
Amount of Cash contribution and Amount of Labor contribution) are hypothesized to explain
the dependent variable.

[4.] CHAPTER THREE

[5.] Research methodology
In this Chapter, the research methodology used in the study is described and it contains nine
sections. The first and second sections describe the research design and study area
respectively. In sections three, four and five study population, sample size and sampling
methods are presented respectively. Section six and seven presents data collection methods
and analysis respectively. Finally model specification in section eight and data accuracy and
reliability in section nine are presented.
3.1 research design
The study adopted a cross-sectional research design (Ranjit, 1996) which included both
survey and ethnographic methods. Using this research design, data was collected from more
than one respondent in both five districts at a single point in time. The study examined
patterns of similarities and differences across a moderate number of survey household
respondents. The design enabled to examine the effect of social capital on microcredit access
in credit and non-credit users using: survey respondents, focus group discussion participants
and key informants (KI). The choice of this design is due to the fact that it is cheap in terms
of time and human resources as data was collected simultaneously from both districts at a
single point in time. The design further enables ed me to to triangulate between the
quantitative survey questionnaires and the qualitative FGDs and key informants
interviewsKII.
x
3.2 THE STUDY AREA
The study was conducted in Bahir Dar Zuriya district, west Gojam zone, Amhara regional
state, Ethiopia, in North West of Ethiopia (3.1b), about 565 km from Addis Ababa. Bahir
Dar Zuria includes the forested Zege Peninsula, known for its numerous medieval churches,
of which the best known is Ura Kidane Mehret, and associated monasteries. Other points of
interest include the Tis Issat falls, and Dilde, better known as the Portuguese Bridge, over the
Abay at Alata, about half a mile below the falls. Bahir Dar Zuria district has 32 Kebele
Administrations (KAs) and has an area of 1,283.56 km2 (CSA, 2007).
3.3 STUDY POPULATION
The study populations in the locality consists ed of individuals both men and women of
diverse socio-economic background from different households. Based on the 2007 national
census conducted by the Central Statistical Agency of Ethiopia (CSA), this woreda has a total
population of 182,730 with a density of 142 persons per km 2, of whom 93,642 are men and
89,088 women; no urban inhabitants were reported. A total of 40,893 households were
counted in this woreda, resulting in an average of 4.47 persons to per a household, and 40,097
housing units.
3.4 SAMPLE SIZE
Although there are more complex formulae, the general rule of thumb is no less than 50
participants for a correlation or regression with the number increasing with larger numbers of
independent variables (IVs). Greene (2003) provides a comprehensive overview of the
procedures used to determine regression sample sizes. He suggests N > 50 + 8 m (where m is
the number of IVs) for testing the multiple correlation and N > 104 + m for testing individual
predictors (assuming a medium-sized relationship). In this study, there are 6 IV as the result
of this rule of thumb the required sample size is 110 (104+6). But to make more reliable I
used survey general formula which takes only three variables (margin of error, confidence
interval and proportion for the two groups in this case assumed .5 to .5) to estimate sample
size. The desired sample size is calculated based on the formula given by Selvanathan et al
(2007):
2
Z √ pq
n=⌊ α/2 ⌋
B
Where, n is the desired sample size; p is sample proportion for one group of respondents, and
q is the sample proportion for the second group of respondents which is 1-p; B is acceptable
error margin and z is the degree of confidence. As suggested by Selvanathan et al. (2007)
taking Z=.05 and B=0.05 and by assuming equal proportion between borrower and non-
borrower households p=q=.5, the desired sample size derived from equation is 385(estimated
with a confidence level of 95% (i.e., zα/2 = 1.96) and an acceptable error margin of 0.05).
Since majority of respondents cannot read and write I used structured interview questionnaire
this makes the number too large to manage. As the result of this I used the sample size that
was used by previous researcher which is 250. The sample size must be larger than the
calculated sample response to take into account sample attrition. Literature indicates that
sample response rates based on survey questionnaires are normally between 60% and 90%

xi
(see for example, Atieno, 2001; Husain, 1998). The survey interview in our study was
conducted with the help of a group of trained survey assistants. Using a conservative
estimated response rate of 90%, the calculated working sample size for was 277
(277=250/0.9).
3.5 SAMPLING METHOD
This section describes how the research site, key informants and households were selected.
Research site selection: The district of Bahir Dar Zuriya district was selected by purposive
sampling among the 16 districts found in West Gojam zone purposively. The district as it
was selected for the study because ofis the only district with no city in west Gojjam
administrationthe Zone, and its proximity of the district to Bahir dar city and familiarity of
the researcher to the district. An exploratory ion survey was made carried out in different
Kebeles Administrtion (KA) units of the district to obtain an good insight into microcredit
access and to choose the specific study sitesKebeles.
The selection criterion for selectingof KAs among others was largely as based on the
familiarity of the researcher and the proximity respective distance of the KAs from to Bahir
dDar city. Accordingly, the following ; five KAs were selected:, named by Wogelsa, Yibab,
Yinesa-Sositu, wondata and Sebatamit were identified and selected among the 32 KAs in
the district. They are all located in a radius of around 8 km from Bahir Dar: Yibab and
Yinesa-Sositu KA are about 8 km from Bahir city following the main Asphalt road towards
to Addis Ababa; , Wogelsa KA is towards found about 8km from Bahir dar city and found
towards Bahir dar-Zegie gravel road;, and Wwondata and Sebatamit are found about 8 km
from Bahir Dar towards Bahir dar-Adet gravel road.
Key informant selection: Key informants (KI) in this study are defined as persons who are
knowledgeable about microcredit usage and who have continuously lived for a long period of
time in the villages. The selection of the key informants was done by the snowball method.
At least five farmers were asked to identify and give names of seven key informants as
defined above. Then the identified key informants were ranked and the most frequently
appeared cited top five persons were assigned taken as key informants in each selected
village. A total of 25 key informants were selected and used in the study. The purpose of
selecting KI’s was to provide information on access to microcredit. This information was
used for triangulating facts which was obtained by questionnaire.
Household selection: In this study, a household (HH) is defined as a basic unit of production
and consumption, made up from the persons who manage common landholdings and live
under one central decision-maker, the household. A household was randomly selected when
an individual come and visit any social gatherings in the village. Key informants were used to
identify whether ho was the individual i.e. is he/she a household or not? In such a way a total
of 277 samples HH’s were selected randomly.
3.6 DATA COLLECTION METHODS
Different methods and techniques were used in the data collection process. The study used
both household survey questionnaire and ethnographic methods. The ethnographic study
methods used include key informant interviews, FGDs and observations.

xii
3.6.1 Household survey questionnaires: A structured questionnaire was prepared and
Household survey questionnaires were administered to on the 277 randomly selected
household respondents. who had got and never got a loan facility. The survey questionnaires
consisted of both open and close-ended questions that focused on themes and sub-themes
such as identifying the effect of social capital factors on microcredit access in Ethiopia. This
saved time and other resources, and it also ensured that I collected even secretive information
which could not be got through FGDs, as Aanonymity and confidentiality was assured to the
respondents.
3.6.2 Key Informant Interviews: KII interviews were conducted with Key informant
interviews were conducted with the help of a key informants guide. The guide was organized
in form of research questions on various topics or topics likeincluding the effect of the effect
of elements of social capital factors on microcredit access in Ethiopia. The guide was used
when interviewing the KI. The guide included topics on the research subject and was
categorized according to the themes and sub themes as identified in the research questions.
This gave created me room to probe and prompt respondents, there by yielding to detailed
and in-depth data.
3.6.3 Focus Group Discussions: This involved the use of the FGD guide to facilitate and
ensure that the discussions are not off track. The FGD guide was constructed based on the
themes and sub themes (such as the effect of social capital factors on microcredit access) to
have comprehensive discussions. I facilitated or moderated the discussions with the help of a
note taker. The FGDs helped me to get detailed and comprehensive data for comparative
purposes, since this instrument allowed serious brain storming with the participants.
3.7 DATA ANALYSIS
Both qualitative and quantitative techniques used to analyze the data.
3.7.1 Qualitative Data Analysis: Ethnographic data/ Qualitative data/ was analyzed using
content analysis. Qualitative data which were obtained by observation, focus group
discussion, and key informants were organized in the field. This involved developing themes
and sub-themes in line with the objectives of the study; that is,; the effect of social capital
factors on microcredit access. Data from the field was then organized into themes and sub
themes, and the field data were arranged to ensure that no information is left outmissed.
Similar responses were put together under one theme or sub theme in order to avoid generic
and uncoordinated information.
3.7.2 Quantitative Data Analysis: After thoroughly cross checking ed the household survey
questionnaires to establish completeness, accuracy, and consistency and uniformity of the
answers given, the survey information was then coded with the help of a coding frame. For
proper data analysis, the coded data was entered in the computer using a proper SPSS
window 20.00 computer package and this enabled me to tabulate the data. Quantitative data
analyzed using descriptive statistics such as mean, percentage, standard deviation, tabulation,
ratio and frequency distribution. A binary logit, model which best fits the analysis for
determinant factors that affects small scale farmers access to microcredit was employed.

xiii
Lastly, both ethnographic data and household survey data were merged togethersynthesized
for a complete research report and other related literature were integrated in the dissertation.
3.8 mModel specification
This study is intended to analyze which and how much the hypothesized regpressors’ are
related to the small scale farmers’ access to microcredit. As already noted, tThe dependent
variable is a dummy, which takes a value of zero or one depending on whether or not small
scale farmers had access to micro credit. HoweverOn the other hand, the independent
variables are both continuous and discrete.
There are several methods to analyze the data involving binary outcomes. However, for this
particular study, logit model selected over discriminativee and linear probability models. If
the independent variables are normally distributed the discriminate-analysis estimator which
follows ordinary least square procedures (OLS) is more efficient than the logit model which
requires maximum-likelihood method. However, if the independent variables are not normal,
the discriminate-analysis estimator is not consistent, whereas the logit MLE is consistent and
therefore more robust. The linear probability model (LPM) which is expressed as a linear
function of the explanatory variables is computationally simple. However, despite its
computational simplicity, as indorsed by Gujarati (2003), it has a serious defect in that the
estimated probability values can lie outside the normal 0-1 range.
Hence logit model is advantageous over LPM in that the probabilities are bound between 0
and 1. Moreover, logit best fits the non-linear relationship between the probabilities and the
explanatory variables.
In the analysis of studies involving qualitative choices, usually a choice has to be made
between logit and probit models. According to Amemiya (1981), the statistical similarities
between logit and probit modelsthem make the choice between them difficult. The
justification for using logit is its simplicity of calculation and that its probability lies between
0 and 1. Moreover, its probability approaches zero at a slower rate as the value of explanatory
variable gets smaller and smaller, and the probability approaches 1 at a slower and slower
rate as the value of the explanatory variable gets larger and larger (Gujarati, 2003).
Hosmer and Lemeshew (1989) pointed out that the logistic distribution (logit) has got
advantage over the others in the analysis of dichotomous outcome variable in that it is
extremely flexible and easily used model from mathematical point of view and results in a
meaningful interpretation. Hence, the logistic model is selected for this study.
Logistic regression (logit) analysis is a uni/multivariate technique which allows for
estimating the probability that an event occurs or not, by predicting a binary dependent
outcome from a set of independent variables. In this regard, access to credit as defined in this
study could be either unconstrained or constrained. The dependent variable is access to credit
and since probability ranges between 0 and 1, small scale farmers with unconstrained access
to credit were assigned 1 and the ones that were constrained have 0 assigned to them. The
linear probability model is depicted as:

xiv
 ( )
❑❑=E Y ❑ =()❑❑ ❑❑❑❑Where P is a vector of probabilities that Y takes the value of 1,
❑❑
Xi is the a matrix explanatory variables and Y = 1 if means that the small scale farmers has
unconstrained access to credit,α is the intercept, and β is a vector of parameters to be
estimated. .pt

The logit model, on the other hand is given by:

Considering the following representation of access to credit:
1
(
❑❑ ❑ ) (
❑❑=E Y ❑ = ❑❑❑ ❑ ❑❑ ❑❑❑
❑ ❑
❑ ❑ ❑ ) ❑ ❑ ❑❑
=
1+ e
(−z )
i
1

Where, with similar notations given in the linear probability model

Pi is the probability that an individual used formal credit or does not use given Xi;
e denotes the base of natural logarithms, which is approximately equal to 2.718;
Xi represents the ith explanatory variables; and
α and βi are parameters to be estimated
Equation (1) is known as the (cumulative) logistic distribution function. In the logit model,
Here Zi Z ranges from -∞ to +∞; Pi ranges between 0 and 1; however, Pi is non-linearly
related to Zi (i.e. Xi ) thus satisfying the two conditions required for a probability model. As
a non-linear model, OLS cannot be used to estimate the parameters.
In satisfying these requirements, an estimation problem has been created because Pi is
nonlinear not only in X but also in the β’s. This means that one cannot use OLS procedure to
estimate the parameters.
Hence, the maximum likelihood method is used to estimate the parameters in the equation.
ere Pi is the probability of having unconstrained access to credit and is given by
1
(z ¿¿i¿);¿
1+ e ¿
Then (1- Pi), the probability of having constrained access to credit, is
1
(−Z ¿¿i)
1+ e ¿
Therefore, one can write
(z ¿¿ i)
Pi e
=1+ (−z¿¿ i)
¿ 2
(1−Pi ) 1+e
z
=e ¿ i

Pi/(1-Pi) is the odds ratio in favor of having access to credit i.e.; the ratio of the probability
that a small scale farmers will have unconstrained access to credit to the probability that it
will not have. By taking the natural log of (2), we obtain

xv
 Li=¿
[ ]
Pi
1−Pi
= Z i=❑1 +¿2 X i ¿ 3

That is, the log of the odds ratio is not only linear in X, but also linear in the parameter. L is
called the Logit, and it is thus specified below,
Y= log P/1-P = b0+b1X1+b2X2+b3X3...b17X17+µ 4
Where: Y= Access to credit by small scale farming households. µ= error term, b 0= Constant
term
An odds ratio equal to 1 suggests that the explanatory variable leaves the dependent variable
unchanged. If the odds ratio is greater (less) than 1, it implies that the effect of explanatory
variable is to increase (reduce) the dependent variable (Balogun et al., 2011).
The explanatory variables included in the Access to Credit model include the following
variables: Hs1=Meeting attendance of households to associations (%),Hs2=Decision making
index (%),Hs3=Density of membership index (%),Hs4=Heterogeneity index of associations
(%),Hs5=Cash contribution index of households (%) and Hs6= Labor contribution index of
households to associations (%)
3.9 Validity and Reliability of instruments
Initially, tThe data gathering instrumentquestionnaire was subjected to a small “piloted on a
small group. test” (n=5)before circulating the questionnaire widely, This was done a small
number of test responses were sought, in order to gauge see the validity and reliability of the
questions, the overall “sense” of the questions, and the time required to complete the
questionnaire. Feedback from this was used to refine and update the questionnaire before
printing. Important to any instrument of this nature was validity and reliability.
To obtain acceptable content-validity, the researcher used a panel of five judges. ;
Iindividuals serving as judges were individuals with knowledgeable of microcredit practices,
procedures and concerns but were not part of the research population. Each judge was asked
to rate the items for relevance and appropriateness on a five-point scale. This process
provided ratings for the whole items separately and the instrument as a whole. Judges were
also asked to comment on the time taken to complete the questionnaire, identify
“bad/irrelevant” items, and to add items they ‘strongly’ feel should be included on the
instrument. This procedure was accepted by the researcher and used many times over the
years (Litwin, 1995).
Reliability using statistical methods was : This was carried out done in aon the survey pilot
test (N=5). Employing tThe split-half reliability correlation method was employed and, it was
found that the reliability of the instrument is 90% as shown in apendix1. Therefore, this was
consistent with Nunnaly (1978) who reported there is not a generally agreed cut-off but
usually 0.7 and above is acceptable. Once a coefficient was calculated for half of the
instrument, the Spearman-Brown Correction Factor was used to determine reliability for the
total instrument. Ferguson (1998), opined notes that this to beis an effective method to
establish reliability of tests or surveys.

xvi
 Reliability Statistics

1. Cronbach' N of Items
s Alpha

.904 23

CHAPTER FOUR
2.[6.] Presentation and analysis of data
Under tThis section presents his chapter,t results ofrom data analysis based on the
questionnaires, key informants and focus group discussion were presented. The first
sectionwe start with a presentation s of summary statistics of respondents who participated in
the study. This is followed by testing the Section two involved six hypotheses of the
studymentioned in background and justification of the study by using logistic regression
analysis.
2.1. Descriptive statistics of respondents

4.1 Summary Statistics of the Respondents

The study was conducted in Ethiopia Amhara state Western Gojjam zone, Bahir Dar District
in five randomly selected villages: Wogelsa, Yibab, Yinsa Sostu, Wondata, and Sebatamit.
The targeted respondents were small scale households who live in Bahirdar district.
The distributed and collected questionnaires distributed and collected was are presented in
tTable 4.1.
Table 4.1: questionnaire distributed and collected
No Name of Village distributed Collected Response rate %
1 Wogelsa 56 46 82.1
2 Yibab 56 41 73.2
3 Yinesa Sositu 55 45 80.3
4 Wondata 55 46 82.1
5 Sebatamit 55 48 85.7
Total 277 226 80.7
Out of 277 total questionnaires distributed to the Kebeles/respondents Total returned
questionnaires 228 were 228 out of 277(82.3%) were returned. questionnaires distributed to
respondents. All The returned filled survey questionnaires were checked for errors, omissions
and inconsistencies prior to entry into a computerized statistical program for analysis. and
tTwo questionnaires were not found worth analysis and rejected. As the result, of this only
226 observations were used in the analyseis, which represents a net response rate of 81.6%.
This which wasis greater more than 70%, a response rate as recommended by Earl Babbie
(1973). According to Babbie a 70% return rate is considered very well for survey studies,

xvii
60% return is good but 50% return is worth analysis. The goal for this study was 70% or
greater and actually 80.7% returned. The SPSS version-20 was used for this study.
Table 4.2 shows the profile of the respondents. The overall distributions of males in the
sample were was 71 percent and for females were 29. In this analysis, the age of the
respondents was grouped into six categories; the respondents’s age ranged from 18 to 65 and
above. The age of the respondents were grouped into six categories. In reference to
respondents microcredit usage experience indicates that the majority of respondents have no
experience in microcredit usage. On the other hand, education level of the respondents
revealed that higher percentage level of respondent’s (61%) has no education. 14% of
respondents has primary school level education. The rest 12% has above primary school
education.
Table 4.2: Profile of the Respondents
Gender % Age % Education level % Microcredit %
usage
experience
Male(161) 71 18-25(N=11) 5 Not educated 61 yes(54) 24
(N=138)
Female(65) 29 26-35(N=25) 11 Primary school 27 No(172) 76
(N=62)
36-45(N=70) 31 Above primary 12
school(N=26)
46-55(N=64) 28
56-65(N=40) 18
>66(16) 7

4.2 Econometric/logit model results

A problem arises when at least one of the independent variables is a linear combination of the
others. The existence of multi-collinearity might cause the estimated regression coefficients
to have the wrong signs and smaller t-ratios that might lead to wrong conclusions. There are
two measures that are often suggested to test the presence of multi-collinearity. These are:
Variance Inflation Factor (VIF) and Tolerance value for dummy variables Gujarati (2003).
The larger value of VIF and smaller value of Tolerance indicates multi-collinearity problem.
Menard (1995) suggests that a tolerance value less than 0.1 almost certainly indicates a series
collinearity problem. Myers (1990) also indicates that a VIF value greater than 10 also
indicates a series collinearity problem.
Table 4.3: Variance Inflation Factor and Tolerance Value for Explanatory Variables
Variables Tolerance VIF
Membership .954 1.049
Density of Membership in % .913 1.095
Meeting attendance index % .643 1.554
cash contribution index % .662 1.511
labor contribution index % .980 1.021
homogeneity index .799 1.252

xviii
 Decision making index in % .843 1.187

Based on the VIF and Tolerance results shown in Table 4.3, the data were found to have no
serious problem of multi-co linearity and therefore all explanatory variables were retained in
the model.
A logistic regression analysis was carried out to predict microcredit access of small scale
farmers for the date using the social capital variables as predictors. The method used is
maximum likelihood estimation technique. Six social capital variables were hypothesized to
predict smallholder farmer’s access to microcredit. The social capital variables and
hypothesis tested were Hs1 to Hs6, as defined earlier.
A test of the full model against a constant only model was statistically significant, indicating
that the predictors as a set reliably distinguished between small-scale farmers who access
microcredit and did not access. The Omnibus Test of Model Coefficients “goodness of fit”
(chi-square = 34.059, df = 6, p = .000) indicates the improvement of the full model over the
base model. The Hosmer and Lemeshow Test (chi-square = 7.330, df = 8, p = .501) indicated
that the full model is a good fit.
The final logit model as depicted in Table 4.4, based on equation 4 in methods section is:

Logit (access to microcredit) = 0.002Densitymemb+0.014cashcontindex-0.06

Homoindex +0.033Dmindex +3.067

Table 4.4: Effects of Social Capital Variables on the Probability of Microcredit access
Variable B S.E. Wald Df Sig. Exp(B) 95% C.I.for
A EXP(B)
Lowe Upper
r
Density 0.002 0.011 0.044 1 0.834 1.002 0.981 1.024
Meeting -0 0.012 0.017 1 0.895 0.998 0.974 1.023
Cash contribution 0.014 0.018 0.632 1 0.427 1.014 0.98 1.05
Labor contrribution 0 0.011 0.001 1 0.975 1 0.98 1.021
Homogeneity -0.06 0.021 7.871 1 0.005 0.943 0.905 0.982
Decision making 0.033 0.007 23.943 1 0.000 1.033 1.02 1.047
Constant 3.067 1.615 3.604 1 0.058 21.474
Table 4.4 shows that based on the Wald criterion only decision making index (p=.000) and
homogeneity index (p=.005) are statistically significant in the determination of access to
microcredit. The odds ratio of homogeneity index ((Exp (B)=.943)) indicates that when
homogeneity of small scale farmers score increases by one unit the odds ratio decreases
by .943 times, therefore householders are 0.943 times less likely to access microcredit.
Similarly, the Odd ratio of decision making index((Exp (B)=1.033)) indicates that when
decision making index of small scale farmers score increased by one unit the odds ratio is
1.033 times more and therefore householders are 0.943 times more likely to access

xix
microcredit. Other four variables (density of membership index, meeting attendance index,
cash contribution index and labor contribution index) were not a significant predictors of
access to microcredit. The results are similar to the findings of Daniel (2013) and Ivanda et al
(2014).
5. Summary conclusions and recommendations

CHAPTER FIVE
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
This section chapter consists deals with of three sectionsissues: . Section one presents a
summary of the major findings of the research. ; The conclusions and policy implications of
the research findings are discussed in Section two. Section three presents as well as
recommendations for further research.
5.1 Summary of major findings
According to the Wald criterion only decision making index (p=.000) and homogeneity
index (p=.005) made a significantly contribute ion to access microcredit by small scale
farmers in the study area. Odd ratio of homogeneity index ((Exp (B) =.943)) indicates that
that there is an inverse relation with odds ratio of access to credit.when homogeneity of small
scale farmers score increased by one unit the odds ratio is .943 times decrease. SimilarlyOn
the other hand the relationship , between the oOdds ratio of decision making and odds ratio of
access to credit index((Exp (B)=.1.033)) indicates that when decision making index of small
scale farmers score increased by one unit the odds ratio is 1.033 times increase. Other The
remaining four variables (density of membership index, meeting attendance index, cash
contribution index and labor contribution index) were not a significant predictors of access to
microcredit.
5.2 Conclusion and Policy Implications
Based on the results The findings of this research have important implications for academics,
rural households, ACSIs, and policy makers. Thus, the following conclusion and policy
implications are forwardedmade:
[1.] Group lending becomes the most is probably the most important method of providing
rural credit to the poor who could do not have bring material collateral. However, for
poor farmers, especially the very poor farmers onesfind, group lending seems to be
inconvenient to access credit from MFI since they are not easily accepted in the process
of rejected from the group formationby others. Therefore, there should be a policy
environment whereby individuals may have access to MFI credit, without forming
groups, by means of using land use right certificates and also
[2.] The majority of the rural households’ especially female headed households and the very
poor farmers did not have access use to credit from microcredit suppliers. Therefore, high
emphasis should be given in screening potential borrowers and to address the very poor
and female headed households in the microcredit market. Participatory wealth ranking

xx
 can be carried out to select and reach those who should be first prioritized forbeneficiaries
of the service.
[3.] The heterogeneity of the socio-economic status of rural people and diverse nature and
scale of their economic activities imply that the demand for financial services by rural
enterprises and households cannot be met by a single financial institution or through a
uniform approach. Thus, this calls for institutional mix, product innovations, and
appropriate methodology. Owing to high costs and associated risks with the early stages
of the development of RFMs, market forces alone cannot bring about the required
financial sector development. Hence some kind of government and donor facilitation is
called for. It is, however, critical to note that such facilitation should be based on a
coherent RFMs development sStrategy that involves various actors in the market so as to
make them own the development process. Such an envisaged RFMs should be an
integrated one, so that the various players in the financial system operate as one guided by
the same regulation and supervision framework. Financial viability of the “infant
financial intermediaries” is critical. The government has to ensure that such institutions
are nurtured through appropriate incentive mechanisms, appropriate financial
infrastructure – legal, regulatory and information system that minimizes transactions costs
and market risks. In addition a well-designed financial innovation reduces transaction
costs and brings about widening, deepening and integration of RFMs. In turn this process
stimulates economic growth through enhanced savings, investment and production.
5.3 Recommendations for Further Research
There are specific issues that call for further studies. These include:
1) development of realistic rural development strategy that covers, among others, the
development of the financial markets,
2) rural property ownership rights in order to determine how these can be used
productively, through say mortgage, collateral, and/or sale for cash income, and
3) Strategies for loan enforcement mechanisms in rural areas – especially on how the
informal finance experiences can be utilized.

REFERENCES
Ajam, O. & Tijani, G. (2009). The Role of Social Capital in Access to Micro Credit in Ekiti State,
Nigeria. Pakistan Journal of Social Sciences 6(3),125-132.
Banerjee, A., E. Duflo, R. Glennerster and C. Kinnan. 2015. “The miracle of micro-finance? Evidence
from a randomized evaluation”. American Economic Journal: Applied Economics, 7(1): 22–
53.
Balogun, O and Yusuf, S. (2011). Determinants of demand for microcredit among the rural
households in south-western states,Nigeria. Journal of Agricultural and Social Sciences, 7(2),
41-48.
Bastelaer, T.V. (2000). Imperfect information, social capital and the poor's access to credit. Center
for Institutional Reform and the Informal Sector (IRIS) at the University of Maryland. ISBN

xxi
Befekadu B. Kereta, 2007. Outreach and financial performance analysis of microfinance institutions
in Ethiopia: African economic conference United Nations conference center (UNCC), Addis
Ababa, Ethiopia.
Clercq D.D & Dakhli M. (2004). Human capital, social capital and Innovation. A multi country study.
Enterpreneurship and Regional Development, 16(2).
Daniel, K. K., Job, L. K, Ithinji, G. K. (2013). Social Capital Dimensions and other Determinants
Influencing Household Participation in Micro-credit Groups in Uasin Gishu County, Kenya.
Developing country studies. Volume 3(3).
Ejigu, l. (2009). Performance analysis of a sample microfinance institution of Ethiopia. Self, 4(5),
287-298.
Getaneh Gobezie, 2005. Livelihoods through Micro-enterprise Services; Assessing Supply and
Demand Constraints for Microfinance in Ethiopia (With Particular Reference to the Amhara
Region)
Guérin, I., M. Roux, M. Roesch, J-Y. Moisseron and P. Ould-Ahmed. 2011. “Analyse des
Déterminants de la Demande de Services Financiers dans le Maroc Rural”. Agence Française
de Développement. Série d’Analyses d’Impact, No. 6.
Social capital as posited by Clercq and Dakhli (2004) is generated from informal institutions,
and is considered as an additional factor of production. Social capital as posited by
Clercq and Dakhli (2004) is generated from informal institutions, and is considered as
an additional factor of production. Greene, W. H. (2003). Econometric analysis (5th ed.).
Stern School of Business, New York University, New York: Prentice Hall
Grootaert, C. (1999). Social capital, household welfare and poverty in Indonesia. Local Level
Institutions Study Working Paper, vol. 6 World Bank, Washington, DC
Gujarati, Damodar N. (2003) Basic Econometrics. 2nd edition. McGraw-hill Book Company. New
York.
Ho Le Phi Khanh (2011).the role of social capital to access rural credit: A case study of two districts.
Swedish university of agricultural science.
Hosmer, D.W. And lemeshew, S. (1989). Applied logistic regression. A Wiley-inter science
publication, New York.
Ivanda, J.O. , Afolami, C.A., Obayelu, A.E., Ladebo , O.J.(2014). Social Capital and Access to Credit
among Cassava Farming Households in Ogun State, Nigeria. Journal of Agriculture and
Environmental Sciences. 3(2). American Research Institute for Policy Development.
Karlan, D. and J. Zinman. 2011. “Microcredit in theory and practice: Using randomized credit scoring
for impact evaluation. Science, 6035(332): 1278-84.
Kangogo, K.D., K.J. Lagat and K.I. Gicuru. 2013. “Social capital dimensions and other determinants
influencing household participation in micro-credit groups in Uasin Gishu County, Kenya”.
Developing Country Studies, 3(3): 213–22.
Maddala, G. S. (1983). Introduction to econometrics. Chichester; New York: john Wiley.
National Bank of Ethiopia (NBE), (2011). A study on the need for establishing rural and urban micro-
financial institutions in Ethiopia, January, Addis Ababa.

xxii
Narayan, D. & Pritchett, L. (1999). Cents and sociability: household income and social capital in rural
Tanzania. Economic Development and Cultural Change 47(4), 871 -897.
Okten, C. & Osili, U. (2004). Social networks and credit access in Indonesia. World Development
32(7), 1225-1246.
Osotimehin, K.O., C.A. Jegede and B.H. Akinlabi. 2011. “Determinants of microfinance outreach in
South-Western Nigeria: An empirical analysis”. Interdisciplinary Journal of Contemporary
Research in Business, 3(8): 780–97.
Litwin, M. (1995). How to measure survey reliability and validity. Thousand Oaks, CA: Sage
Publications.
Mohamed, K. (2003). Access to formal and quasi-formal credit by smallholder farmers and artisanal
fishermen: a case of Zanzibar (research report no. 03.6). Dares salaam, Tanzania: research on
poverty alleviation
Portes, A. (1998). Social Capital: Its origins and applications in modern sociology. Annual Review of
Sociology, 24, 1-25.
Sadik, M., S.F. Irene and D.P.K. Amegashie. 2013. “Social capital and access to credit by farmer
based organization in the Karraga District of Northern Ghana”. Journal of Economics and
Sustainable Development, 4(16): 146–55.
Stieglitz, J and Weiss, a. (1981). Credit rationing in markets with imperfect information. American
economic review, 71(3), 393-410.
Sarker M.M & Islam M.S.(2014). Social Capital and Access to Microcredit: Evidence from Rural
Bangladesh. Developing Country Studies. Volume 4(7).
Selvanathan, E.A, S. Selvanathan, G. Keller and B. Warrack(2007). Australian Business Statistics,
Abridged 3rd Edition, Nelson Thompson Learning Australia.
UNDP, 2011. The state of the world population: United Nation Population Program,
Vaessen, J. (2001). Accessibility of rural credit in northern Nicaragua: the importance of networks of
information and recommendation. Savings and development, 25(1), 5-32.
Yehuala, S. (2008). Determinants of smallholder farmers access to formal credit: the case of Metema
Woreda, north Gondar, Ethiopia. (Master’s thesis, Haramaya University, 2008

xxiii