You are on page 1of 29

Giordano Dell-Amore Foundation

FACTORS INFLUENCING REPAYMENT OF AGRICULTURAL INPUT LOANS IN ETHIOPIA: THE


CASE OF TWO REGIONS
Author(s): Bekele Hundie, Kassa Belay and Mulat Demeke
Source: African Review of Money Finance and Banking, (2004), pp. 117-144
Published by: Giordano Dell-Amore Foundation
Stable URL: http://www.jstor.org/stable/23026296 .
Accessed: 17/06/2014 01:20

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Giordano Dell-Amore Foundation is collaborating with JSTOR to digitize, preserve and extend access to
African Review of Money Finance and Banking.

http://www.jstor.org

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
FACTORS INFLUENCING REPAYMENT OF AGRICULTURAL
INPUT LOANS IN ETHIOPIA: THE CASE OF TWO REGIONS

Bekele Hundie
Ethiopian Civil Service College
Kassa Belay
Alemaya University
Mulat Demeke
Addis Ababa University

1. Introduction

Ethiopia is endowed with favourable natural environment for the production of


various crops and livestock. The agricultural sector accounts for nearly 50% of the
GDP and provides employment for more than 80% of the population. It also accounts
for 85% of the export revenue and satisfies 70% of raw material requirements of the

country's industries (Assefa and Heidhues, 1996).


In spite of the huge agriculturalpotentials, growth in agriculturalproduction has not
been able to keep pace with that of the demand. In fact, a great proportion of cultiva
ted land is owned by subsistence farmers who produce about 97% of the national

agricultural output (Welday, 1999). The small-scale farmers, however, produce a little

'surplus' which cannot adequately feed the population and as a result, the country has
been importingfood grains since the 1950s.
Credit is considered as a lubricant that helps to provide a push to the development

process (Singh, 2000). Steady agricultural development depends, among other


things, upon the continuous increase in farm investment. Most of the time, especially

during the take-off stage of agricultural development, heavy investment cannot be


made by the farmers out of their own savings because of their low level of incomes.
The gap between the owned and the required capital calls foroutside funding (Singh
et al., 1985). Nevertheless, supplying large amount of money to farmers on credit
basis is not an end by itself.The loan funds need to be repaid so that the money can
be used in successive production periods.

In agricultural loan administration, the increasing default rate has been one of the

major problems for all the financial institutions. Increasing defaults in the repayment of
loans may have various implications. Primarily, defaults discourage financial institu
tions to refinance the defaulting members. Consequently, the overall agricultural pro
duction may diminish. Therefore, a thorough investigation of the various aspects of
loan defaults is of utmost importance both for policy makers and the lending institu
tions.
The objectives of this paper are: to identifythe determinants of loan repayment
performance of smallholder farmers in two regions of Ethiopia; and to estimate the
relative importance of these factors in influencing loan repayment performance of

sample households.

117

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

The rest of this paper is organised in three sections. Section 2 reviews rural credit

policy and other issues related to loan provision in Ethiopia. Section 3 presents the
study areas, data sources, methods of data analysis, explanatory variables and wor

king hypotheses. Section 4 discusses the findings of the study.The final section sum
marises the findings and discusses their policy implications.

2. Review of Rural Credit Policy and Agricultural Loan Provision in Ethiopia

2.1. General financial policy environment

During the era of the command economic system (1974-1991), banks/financial


institutionswere entirely governed by credit regulations issued by the National Bank
of Ethiopia (NBE).The NBE was responsible forthe unified and systematic allocation
of all credit (Mauri and Caselli, 1986). Accordingly,the Development Bank of Ethiopia
(DBE) was given exclusive mandate to address rural credit needs before 1986. In
June 1986, the Commercial Bank of Ethiopia (CBE) which previously concentrated its
credit operations on urban commercial and related activities, was given additional
responsibilities to meet rural development credit demand. Moreover, the Ethiopian
Insurance Corporation (EIC) was given responsibility to put in place a limited scheme
of insurance forselected crops and livestock and to support the credit policy in close
consultation and collaboration with the DBE and the Ministryof Agriculture (NBE,
1986).
However, the geographical coverage of these financial institutions and their servi
ces in relation to the loan demand in the rural sector were not adequate. This could
be attributed not only to the limited capacity of the financial institutionsbut also to the
limited credit absorption capacity of the rural sector as a whole. Hence, this had
necessitated the provision of effective extension services by the financial institutions
in addition to those provided by other development agencies. To this effect,the NBE
issued the "revised rural credit policy" in August 1988 (NBE, 1988).
Since the market reformof 1991, the NBE has lost its power and responsibilities
forproviding guidelines on which formal financial institutionsmust depend to allocate
loans across various sectors of the economy. Financial institutions have been, rather,

guided by the market signals to extend loan funds to their clients. However, for the pur

118

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

pose of fostering monetary stability and credit and exchange conditions for the balan
ced growth of the country,the NBE has been given responsibilities of issuing directi
ves on credit and other financial transactions of formal financial institutions.

2.2. Interest rate determination

During the reign of the militarygovernment (1974-1991), financial institutionsused


their credit facilityas an instrumentto strengthen and expand the socialist economy.
More precisely, banks used to give preferential treatment to state-owned institutions.
Even after the market reform of 1991 (till August 1994), different interest rates were
used to extend loans to different sectors of the economy. In September 1994, the NBE
fixed a non-differentiatedlending rate for all sectors of the economy (14 - 15%). The
interest rate for the central government loans was allowed to remain between 12 and
13 per cent (both figures inclusive).
In January 1995, the NBE abandoned its practice of fixingspecific interest rates
and decided to fix the minimum deposit rates and the maximum lending rates for
financial institutions in order to make the credit system more flexible. Accordingly, the
financial institutionswere entrusted with the responsibility of fixingtheir own deposit
as well as lending rates as long as they are withinthe regulation-frameof the NBE. In
1998, the NBE made the credit system more liberal, i.e. it liftedthe maximum lending
rate, except for the Federal Government loan. This enabled financial institutions to fix

lending rates which can better reflect the existing market conditions and opportunity
cost of capital.

2.3. Situation of formal loan for smallholders

2.3.1.Formal credit sources and volume

As already noted, the DBE was the sole supplier of input loans for smallholders
before 1986. However, after the issuance of the rural credit policy in June 1986, CBE,
which untilthen had concentrated its credit operations on urban commercial and rela
ted activities, joined DBE in meeting ruraldevelopment credit demand. Until1997 both
the CBE and DBE had been involved in the provision of input loans to smallholders.
However, in 1998 the DBE ceased the provision of credit to smallholders. Since this

119

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

time the CBE has become the sole source of formal input loans for smallholder far
mers.

As far as the volume of credit provided to smallholders is concerned, the CBE has
taken the lion's share. For instance, the volume of credit provided by CBE to small
holders of the Amhara National Regional State and Oromiya National Regional State
in 1995 was more than three-fold of that provided by the DBE. The total volume of cre
dit extended to smallholders increased at an average rate of 17.5% per annum bet
ween 1995 and 1999. The growth rate was the highest (35.4%) between 1995 and
1996 but itsharply decreased thereafter perhaps because the DBE gave up providing
credit for smallholders. Thus, the average growth rate after 1996 was only 8.4% per
annum.

. Loan provision and collection mechanism

Agricultural input loans are provided in kind (mostly in the form of fertilizersand
improved seeds). The CBE has been the major financier of input credit to smallholder
farmers since 1994/1995. From 1994 to 1996, credits were given to farmers through
their cooperatives without the Regional Government providing guarantees for full loan

refunding. As of 1996, the CBE began to sign loan contracts with the Regional
Governments directly. Due to the lack of collateral held by rural households and to

guarantee loan repayments, the CBE requested the Regional Governments to use
their budgets as collateral. In doing so, the Regional Governments vouched for the
beneficiaries of the loans, and thus became responsible to collect loan repayments
from farmers. More precisely, input loans are taken up by the Regional Governments
and channeled into the District Administration Offices. Farmers then receive credit, in
kind, via the District Agricultural and Finance Offices. In the input loan provision
system, input supply and credit are dealt with in one transaction. The procedures
involved in input loan disbursement are as follows. The Regional Government borrows

directlyfromthe CBE and relies on its administrative machinery and peasant organi
zations to disburse and collect the loan. Farmers have to apply via their service co

operatives, which submit applications forcredit to the DistrictAgriculturalOffice. The


District Finance Office is also involved. An agreement is signed between the Finance
Office and the Co-operatives. The service co-operatives collect a 25% down payment
of the input prices, withthe balance due after harvest. The down payment and signing
result in a delivery order by the Finance Office, which the co-operatives use to collect

120

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

their stock from the designated input supplier. Borrower farmers pay a 10.5% interest
rate on the input loan. This credit delivery mechanism, which was introduced in 1996,
allows local government officials (administrators and agricultural office staff) to parti
cipate in screening borrowers' applications.
Loan repayments are effected in the following harvest, immediately after harvest.
Farmers would be notifiedto settle theirdebts by extension agents before the specified
due date. If farmers fail to repay loans within the normal repayment schedule, admini
strative enforcement starts to operate. To strengthen the enforcement, loan collection
committees were established at differentadministrative levels. Members of the com
mittees include administrators and other officials, the police force, and elder farmers.

3. Methodology

3. 1. An overview of the study regions

This paper is based on data collected from a total of 315 household heads ran
domly selected from two regional states of the country1 . More specifically, data were
collected fromseven sites located in the Amhara National Regional State (ANRS) and
the Oromiya National Regional State (ONRS).

3.1.1. The Oromiya National Regional State (ONRS)

The ONRS is the largest of the nine regional states of Ethiopia. It covers the cen
tral part of the country and extends from the west across the central highlands to the
south eastern part of the country. Its total area is estimated to be 353,010 km2, which
is about 28.2% of the total area of Ethiopia. It has 12 administrative zones and 180
districts. Four of the seven study sites are located in the ONRS. More precisely, the
four study sites are located in the central part of this region.

1Withthe
Change in government in 1991, on the basis of ethnic, linguisticand cultural identity,the countrywas
divided into9 semi-autonomous administrativeregions, one federal capital (Addis Ababa) and one special administra
tive division (Dire Dawa). According to the Ethiopian federal democratic republic administrativehierarchy,the regional
states are divided into zones, districtsand kebeles (local administrationunits) in that order.

121

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING - 2004

According to CSA (2002), the total population of the region in July 2002 was 23.7
million.The majority of the population (88%) were reported to live in rural areas and
depended directlyor indirectlyon agriculture fortheir livelihood. Agriculture accounts
for about 73% of the region's gross domestic product and about 79 % of the farmers
were smallholders engaged in mixed farming practices (OBOPED, 1996).
Ninety-nineper cent of the total farm households practised rain-fedagriculture and
the rest practised both rain-fed and irrigationagriculture. Out of the total land area
under private peasant holdings, about 73% was used for production of temporary

crops in the 1997/98 cropping season (CSA, 1998a). Moreover, 271,840 hectares
(6%) of the cultivated land were planted with perennial crops such as coffee, fruits,
chat and enset2.

The ONRS alone accounts for about 55% of inorganic fertiliser consumption in

Ethiopia (FAO, 1997). The average fertiliserconsumption of this region between 1991
and 1997 was estimated to be 106,149 tons per annum (OADB, unpublished docu

ment). The wholesale and retail sales of farm inputs are carried out mainly by a public
organisation known by the name of AgriculturalInput Supply Enterprise (AISE) and a
private firm called Ethiopian Amalgamated Private Limited (EAL). Moreover, the
Dinsho Trading Private Limited Company was participating in retail sales of farm
inputs mainly as an agent of AISE and in some areas as an agent of EAL. Formal

input credit for smallholders is mainly supplied by the CBE. The ONRS is by far the
largest client of CBE in terms of borrowing short-term loans for the procurement of

agricultural inputs.

3.1.2. The Amhara National Regional State (ANRS)

The ANRS is the thirdlargest region of the nine regional states of Ethiopia, with a
total area of 159,170 km2, next to the ONRS and the Somali National Regional State.
Its boundary extends from the central part to the north western part of the country.
The region is divided into 11 administrative zones and 105 districts.Three of the seven

2 Chat is a natural stimulant


plant,which reaches heights from3 to 7 meters. Fresh chat leaves, which are typically
chewed like tobacco, produce a mild cocaine- or amphetamine-like euphoria that is less potent than either substance.
Chat is widely used in eastern and southern parts of the country.Enset (Ensete ventricosum) is a banana-like peren
nial plant grown throughoutthe Southern highlands of Ethiopia as the major staple food crop by many culturalgroups.

122

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

study sites are located in the ANRS. More precisely, the three study sites are located
in the western, southern and central parts of this region.

According to CSA (2002), the total population of the region in July 2002 was 17.2
million, of which about 89 % were reported to reside in rural areas. Ninety-three per
cent of the total population were engaged in agriculture while the rest depended on
other economic activities such as wholesale and retail trade, hotels and restaurants

(CSA, 1998b).
About 99% of the farmers practised rain-fedagriculture and the rest practised both
rain-fed and irrigation agriculture. Nearly 90% of the total cultivated land
(2,501,450ha) was allotted to the cultivation of temporary crops in the 1997/98 crop
ping season and the rest (2,870 ha) was planted withpermanent crops (CSA, 1998a).
There are three supplying agencies forchemical farm inputs (fertiliserand pesti
cides) in the region. These are AISE, Ambassel Pvt. Ltd. and EAL. Moreover, Fertiline
Company supplies fertiliser in some areas of North Shewa zone. However, it is esti
mated that Ambassel Pvt. Ltd. accounts for more than 80% of fertiliser distribution in
the region. Formal input credit for smallholders is mainly supplied by the CBE. The
ANRS is the second importantclient of CBE next to the ONRS.

3.2. Data sources

The data used in this paper were obtained from the fifth Ethiopian Rural
Household Survey conducted by the Economics Department of Addis Ababa

University in collaboration with USAID. Four successive


surveys had been conducted

prior to this one and all were aimed at generating time-series data on the Ethiopian
rural households. The fifth survey was carried out in 1999 and the collected data refer
to the 1998/1999 cropping season. Secondary data were also extracted from relevant
sources to supplement the primary information. Some of the secondary data sources
were CBE, DBE, NBE, Central Statistical Authority(CSA) and other relevant publica
tions.
As this study is aimed at assessing loan repayment performance of smallholder
farmers engaged in the production of annual crops (mainly cereals), of the eighteen
sites which the survey covered, only the data pertaining to the cereal-based sites
were used for analysis. These sites are seven in number and four of them, namely

123

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

Shashemene, Debre Zeit, Eteya, and Bako-Tibe are located in the ONRS; whereas
three and Debre Berhan - are found in the
the remaining -Yetmen, Shumsha-Lalibela,
ANRS.
The survey covered a total of 721 sample households in the seven sites. Out of

these, 406 could not be used for this study because of the unavailability of data on
loan repayment status. Therefore, the remaining 315 sample households (all of whom
were borrowers of input loans) were retained for final analysis.

3.3. Selection of the empirical model

Farmers' loan repayment performance is influenced by a complex set of socio

economic, demographic, technical, and institutional factors. Conceptually, the model


used to examine relationship between loan repayment status and determinants of
loan repayment performance involves a mixed set of qualitative and quantitative data.
The response (dependent) variable is dichotomous taking on two values, 1 ifthe far
mer settles his/herinput loan on time and 0 ifhe/she does not. Estimation of this type
of relationship entails the use of Qualitative Response Models. In this regard, the
linear probability, logit and probit models are possible alternatives. However, several
estimation problems arise particularly when ordinary least squares (OLS) regression
and linear probability models are employed (Aldrich and Nelson, 1990; Feder et al.
1985 and Maddala, 1992). The OLS regression technique, when the dependent varia
ble is binary, produces parameter estimates that are inefficient and a hetroscedastic
error structure. Consequently, hypothesis testing and construction of confidence inter
val become inaccurate and misleading. Likewise, a linear probabilitymodel may gene
rate predicted values outside the admissible 0-1 bound, which violate the basic tenets
of probability.To alleviate these problems and produce relevant empirical outcomes,
the most widely used qualitative response models are the logit and probit models
(Amemiya, 1981).
Both the probit and logit models yield similar parameter estimates and it is difficult
to distinguish them statistically (Aldrich and Nelson, 1990). However, Maddala (1983)
and Kmenta (1986) reported that the logistic and cumulative normal functions are very
close in the mid-range, but the logistic function has slightly heavier tails than the
cumulative normal function. Gujarati (1988) and Pindyck and Rubinfeld (1981) poin
ted out that the logistic and probit formulations are quite comparable, the chief diffe

124

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

rence being that the former has slightly fatter tails; that is, the normal curve approa
ches the axes more quickly than the logistic curve.

However, because of the fact that the logistic distributionfunction (logit model) is
relatively simple from a mathematical point of view and lends itself to a meaningful
interpretation, it is used in the present study.

3.3.1 Specification of the Logit Model

As
already noted, the main purpose of this study is to analyze which, how and
how much the hypothesized regressors are related to the loan repayment performan
ce of farmers. The dependent variable in this case is a dummy variable, which takes
a value of zero or one depending on whether or not a farmer defaults. However, the
explanatory (independent) variables are of both types, i.e. continuous or binary.

Following Pindyck and Rubinfeld (1981) the cumulative logistic probability func
tion is specified as

Pi = F (Zi) = F a +£ fiiXi (1)


-
1+ e (a +113iXi)

Where e represents the base of natural logarithms (2.718)

X( represents the ith explanatory variable

is the probabilitythat ithindividual will make a certain choice (in this


p(j)
study defaults or does not default) given m explanatory variables (X-)
a & Pj are parameters to be estimated

Interpretationof the coefficients will be understandable ifthe logistic model is written


in terms of the odds and log of odds (Hosmer and Lemeshow, 1989). The odds ratio
implies the ratio of the probabilitythat an individual would choose an alternative (Pi)
to the probabilitythat he/she would not choose it (1-P/).

125

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

But,

-
(1 Pi) = (2)
1 + e Zi

Therefore,

1 + e Zi = eZi (3)
[^] 1 + e -Zi

Or,
■ 1

i1 + CDNj
—-4. = e (a+^ P'Xi) (4)
a: 1 + e ~z'

Taking the natural logarithms of equation (4) will result in what is known as the logit
model as indicated below:

+ P1X1+ P2X2 + ... +jimXm

If the disturbance term Ui is taken into account, the logit model becomes

m
Zi = a + [IiXi + Ui (6)
^
i=1

Hence, the above econometric model was used in this study and was treated

against potential variables assumed to affect loan repayment performance. The model
was estimated using the iterative maximum likelihood estimation procedure. The lat
ter yields unbiased and asymptotically efficientand consistent parameter estimate.

126

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

In reality, the significant explanatory variables do not all have the same level of

impact on the loan repayment decision of farmers. The relative importance of quanti
tative explanatory variables in loan repayment decision can be measured by exami
ning loan repayment elasticities, defined as the percentage change in probabilities
that would resultfrom a percentage change in the value of these variables. To com

pute the elasticity, one needs to select a variable of interest, compute the associated

Pj, vary the of interest by some small amount and re-compute the P,, and then
Xp
measure the rate of change as

dPj/
/dXpi

where

and dPj stand for percentage changes in the explanatory and in the
dXp variable(Xpi)
associated probability levels (Pj), respectively. When dXp is very small, this rate of
change is simply the derivative of P, with respect to Xp and it is expressed as follows
(Aldrich & Nelson, 1990):

dP, _ 1 *
exp (Z)
'
dX, 1+ exp (Z) 1+ exp(Z) (7)

(P,)( 1-R)-^ (8)

The impact of each significant explanatory variable on the probability of non


defaulting is calculated by keeping the continuous variables at their mean values and
the dummy variables at their most frequent values (zero or one).

3.3.2. Definitionof Variables and Working Hypotheses

Once the analytical procedure and its requirements are known, it is necessary to

identifythe potential explanatory variables and describe their measurements and


represent them in symbols.

127

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

The Dependent Variable of the Model (DEFAULT), Yi: The dependent variable for
the logit analysis has a dichotomous nature representing the observed loan repay
ment status of the household. It takes a value 1 for non-defaulters and 0 for defaul
ters.

The Independent Variables of the Study: Several factors have been hypothesized
to affect the loan repayment performance of farmers. Based on literature review and
the discussion held with subject matter specialists in the area of agricultural credit, in
this study a total of 15 variables were hypothesized to explain loan repayment perfor
mance in the study areas. In what follows a brief explanation of the explanatory varia
bles selected for this study and their influence on loan repayment performance will be

presented.

Age of household head (AGE): This refers to the number of full years since the
time of birth that the farm household head had completed at the time of the survey.

Through time farmers acquire experience in formal credit use. Moreover, older farmers

may accumulate more wealth than younger ones. Therefore, this variable is hypothe
sized to have positive impacts on the loan repayment performances of farmers.

Familysize (FMSIZE): Refers to the number of people living in the same resi
dence. Large households are better credit risks because they may have more diversi
fied sources of income (Schreiner & Nagarajan, 1997). On the other hand, large family
size may imply self-insufficiency because large households consume more than do
small households. This is usually true if the dependency ratio of the household is

large. Therefore, familysize, as a variable, is included in the model in a sense that it


would increase or decrease the rate of loan default.

Formal loan (FORLOAN): This refers to the amount of formal loan in Birr that the
household received in the 1998/99 production year3 . Local leaders play a great role
in assessing and determining the amount of input loans to be disbursed to individual
farmers. To do so, they consider various personal and socioeconomic characteristics
of a borrower farmer. Larger loans are permitted for those who have enough capacity
to repay the loans. Therefore, this variable is expected to have a positive relation with
the dependent variable.

3 Birris the
Ethiopian national currency.The exchange rate is currentlydetermined by inter-bankexchange of cur
rencies and it is around 1 US = 8.7 Birr.

128

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

Informalloan (INFLOAN):): This refers to the amount of informalloan in Birrthat


the household received in the 1998/99 production year. Farmers take loans from
various informal sources such as relatives, friends and village moneylenders. Funds
obtained from these sources are usually used for nonproductive purposes. Hence,
they may reduce farmers' capacity to repay formal loans especially if their repayment
time coincides with that of formal loans. On the contrary,the availability of informal
loans may reduce rate of loan diversion, which is commonly reported as a problem to
institutionalloan repayment. In so doing, it may improve the loan repayment perfor
mance of farmers. Therefore, the coefficient of this variable is expected to appear with
either positive or negative sign.
Total livestock owned in monetary terms (VTOTLIVE): Livestock constitute accu
mulated wealth, as they area source of cash, and security against crop failure.
Farmers owning more livestock can settle their debts even if they encounter crop fai
lure by selling out theiranimals. This variable is defined in monetary units because the
alternative approach (defining in terms of the Tropical Livestock Unit) tends to equali
ze livestock having differentmarket values as it is based on the feed consumption
level of a specified herbivore. For instance, horses and mules are given the same
standard based on the latter approach while mules have usually higher market values
than horses. In this study, loan repayment performance of farmers is expected to
increase with the value of total livestock owned.

Grain crop production (GRNCROP): Defined as the amount of grain crops


(cereals, pulses, and oil crops) produced expressed in kilo calories. Physiological
needs, particularlyfood, usually get priorityto social, political or any other obligations.
In fact, the primary objective of smallholders is to produce sufficient food crops that
can sustain their family members year-round. When other obligations (such as loan

repayment) counteract this primary objective, farmers may not honor their prior com
mitments. Therefore, the expectation is that households producing more grain crops
will be better in food self-sufficiency and hence will have more marketable surplus to

repay loans.

Yield loss (YIELDLOS): Refers to the percentage reductions in revenue fromcrops


due to natural shocks and bad weather. Agriculturalproduction is a biological process.
Its success is influenced by several factors among which natural factors such as unex

pected floods, pests, disease, and poor rainfalldistributionare the prominent ones.
Needless to say that bad weather conditions, insect infestation and disease reduce

129

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

crop production. In the Ethiopian context, the loss is sometimes so large that farmers
run out of grains to feed their families. Therefore, this variable is included in the model
expecting that itwill have negative relationship with the dependant variable.
Celebration of social ceremony (CERMONYD): This is a dummy variable, which
takes a value 1 if the household celebrates social ceremonies and 0 otherwise. The

expenses on traditional social ceremonies such as weddings, funerals, engagements


and circumcisions are sometimes too large relative to farmers' economic status. As
this variable proxies the use of farm income for nonproductive purposes, it is expec
ted to have negative impact on loan repayment performance of smallholders.

Extension contact (EXTCONTD): This is a dummy variable, which takes a value


1 if the household head has extension contacts and 0 otherwise. Agricultural exten
sion services provided by agricultural development bureaus play an important role in

improving farmers' productivity. The development agents in rural areas play a great
role in agitating farmers to repay institutional loans in time. Therefore, ceteris paribus,
this variable is expected to influence loan repayment positively.

Expenditure on health care (MEDEXD): This is a dummy variable, which takes a


value 1 if the household spends some amount of money on medical treatments of its
members and 0 otherwise. Expenses on health care are expected to reduce farmers'

capacity to repay institutionalloans.


Income from off-farmactivities (OFFINCMD): This is a dummy variable, which
takes a value 1 if any member of the household involves in off-farm activities for pay
ment and 0 otherwise. Off-farmactivities are additional sources of income for small
holders. The income generated from these activities may help farmers to cover expen
ses, which would otherwise consume receipts from crops or livestock. Hence, house
holds involved in off-farmactivities tend to be more capable to repay input loans in
time. On the contrary, non-farm income may increase willful default as income is gene
rated fromvarious sources (Zeller and Sharma, 1996). Therefore, itwas expected that
this variable would have either positive or negative impact on delinquency.

Timely supply of farm inputs (TIMEINP): This is a dummy variable, which takes a
value 1 ifthe household receives farm inputs at the righttime and 0 otherwise. Farm
operations such as planting need to be performed in the righttime to boost produc
tion. For this, the necessary inputs and materials should be made available in time.
Farmers who get inputs behind schedule may sell the inputs and use the cash for non

130

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

productive purposes. Therefore, timely loan disbursement may reduce the rate of loan
diversion and, consequently, reduces the rate of loan default.

Education level of the household head (EDUCATD): This is a dummy variable,


which takes a value 1 if the household head is literate and 0 otherwise. Education
increases farmers' ability to get, process and use information. For example, literate
farmers may seek information on prices more than the illiterate ones and conse

quently sell their produce at reasonable prices. Moreover, education may enable far
mers to be more aware of the importance of formal loan and hence may reduce will
ful default. Therefore, ceteris paribus, education is expected to reduce the rate of
default.

Sex of the household head (GENDER): This is a dummy variable, which takes a
value 1 if the household head is male and 0 otherwise. Gender differentials in the farm
households play a significant role in the economic performances of a given house
hold. More specifically, gender differentialscan be related to access to credit. In this
case, one may expect that female-headed households are less experienced in formal
credit and hence will be defaulters for they know little about the consequences of loan
default. The
opposite expectation may be that female borrowers tend to be more loyal
to the lenders than male borrowers. This may arise from the fact that females are more

responsible for childcare and home management and hence they may fear more than
males the punishments arising from loan default, which may be detention. Therefore,
this variable is expected to take both positive and negative signs.

Regional Differentials(REGION): This is a dummy variable, which takes a value


1 if the household is in the ANRS and 0 otherwise. Regional differentials may influen
ce the rate of loan recovery at least in two ways. First, regions may be different in natu
ral climate, which is a prominent exogenous factor determining annual receipts in agri
culture. Second, regions may be dissimilar in efficiency during loan collection, which

may result from differences in administrative structure, qualified staff, rules and regu
lations. Therefore, the rate of delinquency is expected to vary with regions.

131

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

4. Results and Discussion

4.1. Socio-economic and institutionalcharacteristics

As already noted, this study is based on the information collected from a total of
315 respondents. One hundred ninety three respondents were from ONRS and the
rest (122) were from ANRS. The sample was composed of both male- and female
headed households. More precisely, 89.5% of the sample respondents were males
and 10.5% were females. About 52% of the household heads were literate. With

respect to loan repayment performance, 262 (164 from ONRS and 98 from ANRS)
respondents reported that they repaid all institutional loans (were non-defaulters).
Whereas the remaining 53 respondents (29 from ONRS and 24 from ANRS) indica
ted that they failed to repay their loan.
Expenses on traditional social ceremonies and health care were hypothesised to
affect the rate of loan default. Forty per cent of the households reported to have cele
brated at least one traditional social ceremony (wedding, funeral and circumcision)
during the year under consideration. The proportion of sample households, which
celebrated these ceremonies, was higher for the defaulters' group than for the non
defaulters' group. Similarly, 37.5% of the respondents reported to have spent some

money for medical care of their familymembers.


About 37% of the sample households reported that at least one of their family
members had been participating in off-farm activities. Larger proportion of defaulter
households (47.2%) sent their members to off-farm activities as compared to the non
defaulter households (34.7%).
The sample farmers used commercial inputs such as fertiliser, herbicides, and

pesticides forcrop production. The majorityof the respondents (73.9%) reported that
they procured chemical fertilizers through peasant associations (PAs) or service
cooperatives (SCs) while a few (15.5%) purchased from input dealers (such as AISE
and Dinsho) or private traders. Similarly,more than 80% of the improved seed growers
procured the seeds either through peasant association/service cooperatives or exten
sion agents. The inputs supplied through PAs/SCs or extension agents were purcha
sed mainly on credit. Chemicals were also supplied to the farmers mostly using the
same channels used for fertilisers. About 22% of the sample farmers reported that

they received agricultural inputs late. A larger proportion of the defaulters (34%) recei

132

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

ved inputs late in the schedule than the non-defaulters (21%).

Only about 20% of the sample households reported that they received extension
services either individuallyor collectively during the 1998 /1999 cropping season. With
regard to the frequency of contact with extension agents, among those who received
extension services, more than 70% had less than five contacts per year with the

agents.
Table 1 shows that defaulter and non-defaulter borrowers of agricultural input
loans differ in various aspects. On the average, the defaulters were younger (47.3
years) than the non-defaulters (48.9 years). The non- defaulters had slightly larger
family size than the defaulters. The non-defaulters and defaulters borrowed on the

average Birr 260 and Birr 169 from formal and informal credit sources, respectively.
Credit from formal sources was meant to be used only for production purposes and it
took the form of short-term input financing. In fact, all of the sample farmers reported
that they took credit in the formof fertiliserand in addition to the fertilisercredit about
39% of the sample farmers reported that they received credit in the form of other

inputs, namely chemicals and improved seeds.

Informal lenders
(moneylenders, merchants, friends, relatives and local social

organisations) were
reported to be other important credit sources for the sample

respondents. About 40 % of the sample households reported that they borrowed from
informal lenders. The majority of those farmers who reported to have borrowed from
informallenders (80.5%) belonged to the non-defaulter group. The respondents sta
ted that informal credit was received in cash form. The average amount of money bor
rowed from informal sources was Birr 121 for the non-defaulters and Birr 71.6 for the
defaulters. The cash obtained from informal lenders was used for different purposes,
namely to purchase food grains, farm inputs, farm implements, and to pay for health

expenses.

133

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

Table 1: Descriptive Statistics of the Sample Farmers


Variable Unit % with a Mean
Non Defaulters Total
Code or type value 1
defaulters
CERMONYD dummy 40.6
EDUCATD dummy 52.1
EXTCONTD dummy 20.2
GENDER dummy 89.5
MEDEXD dummy 37.5
OFFINCMD dummy 36.8
REGION dummy 38.7
TIMEINP dummy 76.8
AGE years 48.9 47.3 48.6
FMSIZE number 7.9 7.2 7.8
FORLOAN birr 260 169 244.7
INFLOAN birr 121 71.6 112.7
FARMSZ hectares 1.63 1.34 1.6
GRNCROP Unit* 25.1 14.6 23.3
VTOTLIVE birr 3671 2450 3465.7
YIELDLOS percentage 10.9 19.4 12.3
* One unitis
equal to345,678Kcalwhichis equivalenttoone quintalofgrain.The conversion
was done based on thestandardsofEthiopian
Healthand Nutrition ResearchInstitute
(EHNRI).
Source:Surveyresults

Land is the basic asset of farmers. The average size of cultivated land owned by
the sample respondents was 1.6 ha. Non-defaulters owned, on average, larger area
of cultivated land (1.63 ha) than
defaulters (1.34 ha). Forty of the
households
percent
cultivated other farmers' land
through either sharecropping or renting. On the other

hand, nearly 11% of the sample households stated that they leased out their cultiva
ted land to others through either renting or sharecropping arrangements. Five per cent
of the sample households reported that they cultivated farm plots obtained as giftsor
borrowed fromothers (relatives, neighbours).
It may be of interest to note that in this study farm size is not considered as one of
the potential factors affectingloan repayment performance because of the fact that in
Ethiopia land constitutionally belongs to the state. Farmers have only the right to use
their land in their lifetime;as a result the size of land holding can not be considered as
a measure of wealth as is the case in many countries. In reality,because of the fact
that land constitutionally belongs to the state, farmers are rather skeptical to invest in

134

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

long-term land improvement practices. In this regard, recent studies in different parts
of the country found that tenure insecurity generated by fear of further redistribution
was the principal factor explaining farmers' unwillingness to invest effort in measures
to improve soil conservation and enhance fertility
(Girma, 2001; Mulugeta, et al., 2001).
The survey results show that the average total grain production of the sample hou
seholds was about 23 units. The non-defaulters reported to have produced more grain
(25.1 units) than the non-defaulters (14.6 units). Nearly 50% of the sample house
holds were not self-sufficientin terms of feeding their family members. Surprisingly,
none of the defaulting households could consume even the minimum energy require
ments recommended by Ethiopian Health and Nutrition Research Institute and only
37.7% of them produced above the minimum standard4. That is, each family member
in the defaulter households could utmost get about 1927 kcal/day (173 kcal of energy
below the minimum daily requirement) from his/herfamily's grain stock. On the con
trary, each member of the non-defaulter households could get about 913 kcal of

energy above the fixed minimum standard for one day from own grain stock. This may
imply that non-defaulters are better than their counterparts as far as food self-suffi

ciency is concerned.

Livestock are important assets for rural households in Ethiopia. They are used as
sources of food, draft power, income, and energy. Moreover, livestock are indices of
wealth and prestige in rural areas. More than 95% of the sample households kept live
stock, which constituted cattle, small ruminants, and pack animals. Estimation of the
monetary value of livestock owned by the sample respondents reveals that the ave
rage value of livestock owned by the non-defaulters (3671 Birr) was higher than that
of the defaulters (2450 Birr).
The survey results reveal that poor rainfall distribution, insect infestation, diseases,
frost, hail, weeds, wild animals and birds are major factors, which cause crop losses
in the study sites. Based on the estimates of the farmers, crop losses due to natural
factors could reach as high as 100% of the total volume of production. The non-defaul
ters and defaulters reported that they lost, on the average, 10.9% and 19.4%, respec
tivelyof their crops due to these factors.

4
According to EHNRI, the minimumenergy requirementforan adult person per day is 2100 kcal.

135

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

4.2. Analytical findings

To determine the best set of variables influencing loan repayment performance of


the sample respondents, the logistic regression model was estimated6. The model
results confirm the a priori expectation that loan repayment performance is influenced

by the interaction of several factors. The Chi-square value shows that the parameters
included in the model taken together are significantlydifferentfromzero at 1% signi
ficance level. Another measure of goodness of fit is based on a scheme that classifies
the predicted value of DEFAULT as 1 if > 0.5 and 0 otherwise. The model correctly
P(i)
predicts 271 of 315, or 86.1% of the observations. The signs of all the coefficients,
except those of education and age, turned out to be consistent with the a priori expec
tations.

The maximum likelihood estimates of the logistic regression model are presented
in Table 2. Of the 15 variables hypothesised to affect loan repayment performance,
three were found to be significantat 1% probabilitylevel. These include amount of for
mal loan per hectare (FORLOAN), value of total livestock holding (VTOTLIVE), and
timeliness of input supply (TIMEINP). Table 2 shows also that three variables were

significantat 5% probabilitylevel. These variables include participation in off-farmacti


vities (OFFINCMD), food grain production level (GRNCROP), and proportion of yield
loss due to natural factors (YIELDLOS). The amount of informalloan (INFLOAN) was
found to be significant at 10% probabilitylevel. Seven of the fifteenexplanatory varia
bles that were hypothesised to affect loan repayment performance did not have stati

stically significant effects.

5A
technique called variance inflationfactor(VIF) was used to measure the degree of linear relationships among
the quantitative explanatory variables. Moreover, contingency coefficientwere computed for each pair of qualitative
variables to check forthe degree of association among the qualitative variables. As the results show very small degree
of collinearityamong the explanatory variables, all of the qualitative and quantitativevariables were included in the esti
mation of the model.

136

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

Table 2: Parameter Estimates of the Logistic Regression Model

Explanatory Variables Estimated Odds Ratio Wald statistic


Coefficients
AGE -0.0065 0.9935 0.2193
CERMONYD -0.5604 0.5710 2.1856
EDUCATD -0.3601 0.6976 0.8425
EXTCONTD 1.4626 4.3173 1.2862
FMSIZE 0.0534 1.0549 0.9283
GENDER -0.5030 0.6047 0.7211
FORLOAN 0.0034 1.0034 7.9424***
INFLOAN 0.0021 1.0021 3.2406*
MEDEXD -0.6529 0.5205 2.4152
OFFINCMD -0.7938 0.4521 3.8676**
GRNCROP 0.0393 1.0401 6.0753**
REGION -0.0219 0.9784 0.0025
VTOTLIVE 0.0002 1.0002 6.6462***
TIMEINP 1.2538 3.5036 7.8254***
YIELDLOS -2.2681 0.1035 5.3116"
CONSTANT -0.1440 0.0182
Percent correctly predicted3 86.08
Chi-square value 57.566***
*" at 10 %, 5% and 1% probability
levels,respectively.
*,**, Significant
classification
aBased on a 50-50 probability scheme.
Source: Modeloutput

As expected, the amount of formal loan that the sample household received (FOR
LOAN) was found to have a significant positive effect on loan repayment. The possi
ble explanation is that local leaders were effective in screening loan applications. In
actual practice, local screening teams take several factors such as past debt history,
ability to manage farm, and personal behaviour (like attitude towards work) to appro
ve individual loan applications. Accordingly, larger loans are disbursed to those far
mers who are trustworthy perhaps because of their economic position, hardworking
nature, and social position. This result is in a complete agreement with that of Vigano

(1993), who in a study about Burkina Faso, stated that large loan amount receivers
would be better payers ifthey were established bank customers. Similarly,Belay and
Belay (1998) reported negative relationship between loan default and loan amount.
This result suggests that sample respondents who took small amount of input loans
were less likelyto settle their loan on time.
Food grain production level (GRNCROP) has a positive and significant relations

137

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

hip with the dependent variable showing that households producing more food grain
crops have better loan repayment performance. In other words, the probability of sett

ling input loans on time, if all other factors are kept constant, increases with food grain

production level. This variable proxies food self-sufficiency and may indicate how
much poverty influences institutional loan repayment. As discussed earlier, defaulters

produced less than the minimum recommended calorie requirement while non-defaul
ters produced more thantheir consumption requirement. This clearly shows that hou
seholds that are self-sufficient in food production tend to settle input loans on time.

The value of total livestock holding (VTOTLIVE) which indicates the level of wealth
is associated positively and significantlywith loan repayment performance. The impli
cation is that farmers with higher value of total livestock holding tend to be more likely
to settle input loans on time. In other words, the probabilityof paying back input loans,
if all other factors are kept constant, increases with the value of livestock holding. The

possible explanation for this is that farmers who own more livestock are better dispo
sed to repay their loans even when their crops fail due to natural calamities. This result
is in complete agreement with the findings of Belay and Belay (1998).
As expected, timeliness in the supply of commercial farm inputs (TIMEINP) has a
positive and significant effect on loan repayment performance. This result suggests
that input loan delivered at the righttime (when farmers need them) will facilitate effi
cient utilisation of the loans and probably prevent farmers from diverting it for non-pro
ductive purposes. The result of a study by Jama and Kulundu (1992) shows that the

proportion of loan diverted to unintended purposes was higher for farmers who recei
ved their loan funds late. This result is also in complete agreement with the findings of

Vigano (1993), Hunte (1996) and Zeller (1998). The odds in favour of settling input
loans on time, if all other factors are kept constant, increases by a factor of 3.5 as far
mers procure farm inputs at the right time.

The variable INFLOAN, which stands for the amount of informal loan received, is

positively related with loan repayment performance. The coefficient forthis variable is
significant at 10% probability level. The implication of this result is that farmers who
borrow large amount from informal sources tend to settle institutional (formal) loans
on time. The possible explanation forthis relationship can be given from two angles.
First, some of the sample farmers might have used informal loans for productive pur
poses. More precisely, about one-third of the respondents reported that they used the
funds obtained from informalsources to purchase farm inputs. In this respect, it can

138

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

be concluded that informal loans contribute to the production process. Second, infor
mal loans might have helped to reduce the rate of diversion of input (productive)
loans. Again, this can be evidenced by the fact that nearly 60% of the borrowers from
informal sources received the loans between June and August out of which one-third
used the funds to purchase food grains. Informal loans were hypothesised by some
researchers (e.g. Belay and Belay, 1998) to reduce farmers' capacity to repay institu
tional loans mainly because they were expected to be used for consumption smoo

thening. But just because a loan is used for consumption purposes does not imply that

repayment falters. In this respect, it is worth noting that consumption loans in


Cameroon and Madagascar were found to have the same or even higher repayment
rates than production loans (Schrieder and Heidhues, 1995; and Zeller, 1995).

Participation in off-farmactivities (OFFINCMD) is associated negatively and signi


ficantly with loan repayment performance. This result suggests that household that
have members who are involved in off-farmactivities are less likely to repay input
loans on time. The possible explanation is that households having members involved
in off-farm work give less attention to farm activities because they have alternative
income sources. In other words, households who generate income from off-farm sour
ces tend to be wilfuldefaulters because the punishment, which could be inhibitionof
access to credit in the following season, might be less painful to them as they were
less dependent on farmactivities. This result is consistent withthe results of Zeller and
Sharma (1996), who showed that non-farm income might have negative relation with
loan repayment performance because wilful default would happen when income is

generated from different sources. The other possible explanation is that households
who took part in off-farm activities might divert input loans to supplement the off-farm
business.

As expected, yield loss (YIELDLOS) is negatively and significantlyrelated withthe


dependent variable, implying that as yield loss due to natural factors increases far
mers tend to default. More precisely, farmers who encounter relatively high yield los
ses will be less likely to repay their input loans on time. Adeyemo (1984) and Ike
(1986) reported that natural calamities were importantfactors, which reduced the rate
of loan recovery in Nigeria.

139

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

4.3. Sensitivity analysis

As already discussed, all dummy and continuous variables do not have the same
level of impact on farmers' loan repayment performance. To rank these variables a

"typical farmer"is defined by the most frequent values of the dummy variables inclu
ded in the model. Accordingly, a typical farmer is a male farmer (89.5%) who is litera
te (52.1%) and procures farm inputs at the righttime (76.8%). The probabilitythat this
will settle his input loan on schedule (measured at the sample mean of the significant
quantitative variables) is estimated to be 0.9397 (Table 3). By contrast, farmers who
are typical in all respects except that they do not procure farm inputs in time have a
much lower probabilityof settling input loans on time (0.8164 or a reduction of 13.1%
as compared to that of the typical farmers). Similarly,the probability of farmers who
are typical in all respects except that they have family members involved in off-farm
work to settle their loan on time is 0.8757 (a reduction of 6.8% as compared to that of
the typical farmer).

Likewise, the relative


importance of significant continuous explanatory variables
on loan repayment performance was measured by computing variable elasticities.

Accordingly, a 10% increase in the amount of formal loan (from the average of 244.7
to 269.17) will increase the probability of settling input loan on time by 0.48% over that
of the typical farmer. Similarly, the probability of non-defaulting is predicted to increa
se by 0.4% if the value of total livestock holding increases by 10% over that of the typi
cal farmer. The impacts of other significant continuous variables are shown in Table 3.

Table 3: The effects of significant explanatory variables on probability of loan repay


ment decision
Description Probability Change in Percentage
probability change in
probability
Typical farmer 0.9397
Typical farmer but does not procure farm inputs in time 0.8164 0.1233 13.1
Typical farmer but involved in off-farmactivities 0.8757 0.0640 6.8
Typical farmer but a 10% increase in amount of formal loan 0.9442 4.5x10'3 0.48
Typical farmer but a 10% increase in amount of informalloan 0.9410 1.3x10"3 0.14
Typical farmer but a 10% increase in percentage of yield loss 0.9380 -1.7X10"3 0.18
Typical farmer but a 10% increase in value of total livestock holding 0.9435 3.8x10"3 0.40
Typical farmer but a 10% increase in level of grain crop production 0.9447 5x10"3 0.53
Source: Computedfrommodeloutput.

140

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

5. Conclusion and Policy Implications

The hypothesis that input loan repayment performance is determined by several


farm and non-farm characteristics was examined using the logistic regression model.
The results of the study reveal that loan repayment performance was influenced by
the amount of formal loan received, value of total livestock holding, timeliness of input
supply, participation in off-farmactivities, food grain production level, proportion of
yield loss due to natural factors and the amount of informal loan received.

The model output shows that individuals who took larger loans had better repay
ment performances than those who took smaller loans. This indicates that individual
loan applications were carefully evaluated and approved by the local screening com
mittees members who prioritize beneficiaries and decide on the amount of loan to be

provided. Available evidence shows that whenever screening committees discharge


their duties properly the loan repayment rate is very high. Therefore, local screening
committees should be strengthened and encouraged to set a minimum uniform stan
dard for screening loan applications. Group lending is also another way of acquiring
information on creditworthiness of potential borrowers.

Timeliness of input supply is the other factor influencing the loan repayment per
formance of sample farmers. Late disbursement of inputs purchased by the loan funds
was reported to be an important factor to agricultural production by 22% of the sam

ple farmers. The problem of late disbursement of inputs is in most cases related to the
late arrival of the inputs from abroad (fertilisers and agro-chemicals are imported). The

delay has also to do with long-drawn-out procedures of making bids, lengthy decision
making process at different administrative levels and poor infrastructure. Therefore,
conducive administrative environment should be created for timely supply of farm

inputs on credit basis. This includes efforts to reduce/avoid late imports of inputs as
well as early and efficientbidding procedures. Similarly, great attention should be
given to improve the existing rural infrastructure, especially roads, so that inputs reach
the desired locations in time.

The findings of this study reveal that livestock are important farm assets that
improve the farmers' repayment performance. As livestock are sources of income and
serve as security against crop failure,the result of this study is consistent with the a
priori expectation. It is, therefore, importantthat more attention be given to the live

141

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

stock sector at least in the following areas: feed resource improvement and manage
ment; genetic resource improvement; control and/or prevention of animal diseases
and parasites; and development of marketing facilitiesforanimal and animal products.
The model results show also that food grain production is one of the determinants
of institutionalloan repayment. As already discussed, about 50% of the sample hou
seholds were not self-sufficientin food production. This can be attributable to the fact
that the sample households, particularly the defaulter ones, use backward farming
technologies. Therefore, it is importantthat concerted effortsbe made to disseminate
improved farming technologies that are suitable to the areas where they are to be
introduced so that the average output per unit area of land increases.

142

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
B. HUNDIE, K. BELAY,M. DEMEKE - REPAYMENTOF AGRICULTURALLOANS

References

Adeyemo, R. (1984) 'Loan Delinquency in Multi-Purpose Cooperative Union in Kwara State, Nigeria', Savings and
Development, V, III, No. 3.

Aldrich,J.H. and Nelson, F.D. (1990). Linear probability,logit and probitmodels. Sage Publications, London.

Amemiya, T. (1981) 'Qualitative Response Model: A Survey', Journal of Economic Literature19:1483-1536.

Assefa, A. and Heidhues, F. (1996) 'Estimation of Technical Efficiencyof Smallholder Farmers in the Central
Highlands of Ethiopia', Ethiopian Journal of AgriculturalEconomics, 1(1):18-37.

Belay, Kebede and Belay, Kassa (1998) 'Factors AffectingLoan Repayment Performance of Small-holders in the
Central Highlands of Ethiopia: The Case of Alemgena District',Ethiopian Journal of AgriculturalEconomics. 2(2):61-89.

CSA (Central Statistics Authority)(2002) Statistical Abstract 2001, Addis Ababa.

(1998a) 'AgriculturalSample Survey: Report on Land Utilisation (Private Peasant Holdings)', Statistical
Bulletin,Vol. IV, No. 193.

(1998b) The 1994 Population and Housing Census of Ethiopia: Results for Amhara Region. Analytical
Report, Vol. II.
FAO (1997) Ethiopia: Study on Deregulation Fertilizer Prices and Withdrawal of Subsidy. Technical Report;
ETH/94/100, Rome.

Feder, G., Just R.E. and Zilberman, D. (1985) 'Adoption of agriculturalinnovations in developing countries: A sur
vey', Economic Development and Cultural Change, 33(2):255-298.

Girma, T. (2001) "Land Degradation: A Challenge to Ethiopia", EnvironmentalManagement, 27 (6): 815-824.

Gujarati, D. N. (1988) Basic Econometrics. 2nd edition. New York: McGraw-HillBook Company.

Hosmer, D.W. and Lemeshow, S. (1989) Applied logistic regression. A Wiley-Interscience Publication, New York.

Hunte, C. K. (1996) 'ControllingLoan Default and Improvingthe Lending Technology in Credit Institutions',Savings
and Development. Quarterly Review, (1):45-59.

Ike, D.N. (1986) The Problem of Loan Default in Nigerian Agriculture:An Economic and Financial Analysis', Indian
Journal of Economics. 66(262):409-422.

Jama, M.A. and Kulundu, D.M. (1992) 'Smallholder Farmers Credit Repayment Performance in Ugari Division,
Kakamaga District,Kenya', East AfricanEconomic Review. 8(2):85-91.
Kmenta, J. (1986) Elements of Econometrics. 2nd ed. New York: Macmillan Pub. Co. USA.

Maddala, G.S. (1992) Introductionto econometrics. 2nd ed. Macmillan, New York.

(1983) LimitedDependent and Qualitative Variables in Econometrics. New York:Cambridge UniversityPress.

Mauri, A. and Caselli. C. (1986) Money and Banking in Ethiopia, Finafrica,Milan.

Mulugeta, E., Belay, K. and Legesse, D. (2001) 'Determinants of Adoption of Physical Soil Conservation Measures
in Central Highlands of Ethiopia: The Case of Three Districtsof NorthShewa Zone', Agrekon,40 (3):313-335.

NBE (National Bank of Ethiopia) (1986) Credit Regulation, NBE/CR/1/86,Addis Ababa, Ethiopia.

(1988) Revised Rural Credit Policy,NBE/CR/7/88,Addis Ababa, Ethiopia.

143

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions
AFRICAN REVIEW OF MONEY, FINANCE AND BANKING- 2004

OADB (Oromiya AgriculturalDevelopment Bureau). Unpublished Document.

OBOPED (Oromiya Bureau of Planning and Economic Development) (1996) Socio-economic Profileof Oromiya;
November, Finfinne.

Pindyck, R. S. and Rubinfeld, D. L. (1981) Econometric Models and Economic Forecasts. 2nd Edition. New York:
McGraw-Hill Book Co.

Schrieder, G. and Heidhues, F. (1995) 'Reaching the Poor Through Financial Innovations', Quarterly Journal of
InternationalAgriculture,34(2) :132-148.

Schreiner, M. and Nagarajan, G. (1997) 'Predicting Creditworthiness with Publicly Observable Characteristics:
Evidence fromASCRAs and RoSCAs in the Gambia'. Savings and Development, 22(4):399-413.

Singh, B. (2000) 'Nature and Extent of InstitutionalCredit in Rural Punjab', In: R.S. Bawa and P.S. Raikly (eds.)
Punjab Economy- Emerging Issues. Amristar:Guru Nanak Dev University.

Singh, G., Sidhu, J.S. and Singh, B. (1985) 'A Study on Repayment Performance of Borrowers in Punjab',
Financing Agriculture,April-June,India.

Vigano, L. (1993) 'A Credit Scoring Model for Development Banks: An African Case Study', Savings and
Development, 17(4) :441 -482.

Welday, A. (1999) 'Improved Seed Marketingand Adoption in Ethiopia', Ethiopian Journal AgriculturalEconomics,
3(1) :41 -87.

Zeller, M. (1995) The Demand forFinancial Services forRural Households: Conceptual Framework and Empirical
Findings', QuarterlyJournal of InternationalAgriculture,34(2): 149-170.
Zeller, M. (1998) Determinants of Repayment Performance in Credit Groups: The Hole of Programme Design,
Intra-groupRisk Pooling, and Social Cohesion. InternationalFood Policy Research Institute.Reprint No. 384.
Zeller, M and Sharma, M. (1996) Repayment Performance in Group-Based Credit Programmes in Bangladesh: An
Empirical Analysis. Food Consumption and NutritionDivision, InternationalFood Policy Research Institute,Washington
D.C., USA.

Abstract

This paper examines the determinants of agricultural input loan repayment in two

regions of Ethiopia. The paper is based on analysis of data collected from315 hou
sehold heads selected from seven sites in two regional states, namely, the Amhara
National Regional State (three sites) and the Oromiya National Regional State (four
sites). The results of the study show that agricultural loan repayment by borrower far
mers is influenced by a number of factors. More precisely, the results of the econo
metric analysis show that timeliness of input supply, amount of formalloan, food grain
production level, participation in off-farmactivities, value of total livestock holding,
amount of informalloan, and yield loss due to natural calamities were important fac
tors influencing loan repayment performance of sample respondents.
KEY WORDS: default, informalloan, Input loan, performance, repayment.

144

This content downloaded from 195.78.109.162 on Tue, 17 Jun 2014 01:20:21 AM


All use subject to JSTOR Terms and Conditions

You might also like