1

Credit access, irrigation technology adoption and repayment performance: reflections on the role of risk perception on creditdemand andrepayment

Fitsum Hagos*a, Gebrehaweria Gebregziabhera, Nicole Leforeb and Amare Haileslassiea aInternational Water Management Institute,East Africa and Nile Basin sub-regional office, PO Box 5689,

Addis Ababa b International Water Management Institute, Southern Africa sub-regional Office, South Africa, Private

Bag X813, Silverton 0127, Pretoria, South Africa Abstract

The paper addresses three questions, first whether access to credit services has significant impact on adoption small scale irrigation technologies, secondidentifying determinants of participation and extent of credit use and third identify factors that influence repayment performance. +++ Conclusions and policy implications are drawn.

Key words: credit service, irrigation technology adoption, loan demand, repayment, propensity score matching, Double Hurdle model, Ethiopia, Africa

1. Background

Many studies indicate that Irrigation development is closely linked withpoverty reduction (Namara et al., 2010; Hagos et al, 2012; Jin et al., 2012).Although irrigation development needs huge infrastructure investment in the form storage and conveyance structures, there are evidences that farmers could harvest irrigation water with application of different water lifting (WL) technologies like manual pumps, motorized and solar pumps. These technologies only require households having access to shallow wells or perennial streams (rivers) which may not require huge and costly infrastructure investmentsper se.

However, small holder farmers are constrained to adopt these WLtechnologies because of lack immediate cash.The working capital requirements and services to operate these lifting technologies is widely reported as one of the reasons for limiting the performance of irrigated agriculture (Gebregziabher et al. 2014) or poverty of the small holder in general (Kereta, 2007).Stiglitz (1990)and Godquin (2004) highlighted the difficulties in obtaining capital, and the high cost of capital when it can be obtained, may act as important impediments to adoption.

As a response to ease these liquidity constraints, the main intervention in many developing countries is institutionalizing microfinance institutions. Ethiopia, for instance, has promoted microfinance institutions (MFIs) (FDRE, 2009) and member-owned financial cooperatives to alleviate credit constraints of the smallholder farmers. MFIs have developed products to respond to smallholders loan demand. But it is not clear whether the microfinance services and products are convenient to farmers’ demands and how much are farmers making use of these services.

Adeoti (2009) and Gebregziabher et al. (2014) investigated the factors that determine adoption of irrigation technology, treadle pump and motor pump, respectively. Getacher, et al. (2013) also investigated the factors that determine adoption of irrigation technology, both treadle pump and motor pumps and reported that credit access among other variables was important in determining adoption. Hailu et al., (2014) investigated adoption and impact of agricultural technologies, fertilizers and improved varieties, and reported that credit access, among other variables, was important for adoption. Evidence byLiverpool and Winter-Nelson (2010) indicate that microfinance has positive effects on both consumption and asset growth as well as on the use of improved technology.Tadesse, et al. (2016)

* Corresponding author: f.hagos@cgiar.org

2

reported that access to institutional finance has a significant positive impact on both the adoption and extent of agricultural technology. Most adoption studies that indicated above investigated the general determinants without focusing on the influence of one particular factor.

The novelty of this study is estimating average treatment of access to credit on adoption of irrigation technologies (measures by the cost of irrigation technology) by creating comparable groups of participating households and non-participating households using matching with propensity scores. Moreover, this study ses unique dataset collected 2016 on 193 treated and 207 non-treatment households.

But small scale farmers in Ethiopia are not very keen to take loans from MFIs or other sources, even if there are evidences that show the promoted technology are financially feasible(Gebregziabher et al., 2017). Explaining factors for participation, including the role of risks - crop failure risk (due to bad weather), price risk, health risk, etc – in farmers’ demand for loan is important.

Evans et al., (1999) while indicating that rates of participation in microcredit are higher among poorer households in Bangladesh, lack of female education, small household size and landlessness are factors for nonparticipation. Amin et al., (2003) indicated that microcredit is successful at reaching the poor but it is less successful at reaching the vulnerable suggesting evaluation of targeting.

While credit access is presumed as one of the most important factors for adoption of new technologies (like inputs, irrigation technologies, machinery, etc), loan repayment performance is vital for the functioning of microfinance. Successful repayment of loan takenis paramount importance to have sustainable agricultural development and financial institutions (Brau and Woller, 2004; Tucker, 2001).High repayment rates is important for the MFIs to cut the interest rate it charges toborrowers, thus reducing the financial cost of credit and allowing more borrowers to have access to it. Improving repayment rates might also help reduce the dependence on subsidies of the MFIs and burrowers (Godquin, 2004).

Most MFIs in Ethiopia, as it the case many developing economies, provide credit services to burrowers through group lending program. With joint liability lending the group is responsible for the repayment of the loan of its members. Nonrepayment of loan by any group member means that all group members will be denied future access to loans by the MFI. In this manner, group loan creates an incentives to individual members to screen and monitors other members (within the group) and to enforce repayment to minimize the risk of default and contributing to the repayment of other members’ loans and future credit access. Most of the theoretical works, in the literature, focused on how group liability lending stimulates screening, monitoring and enforcement of repayment within the group.For example Stiglitz (1990) showed how group lending leads to more effective in overcoming information asymmetry and moral hazard. Peer selection with substantial cross-guarantees may enhance the effectiveness of rural credit markets (Stiglitz, 1990). Provisions for cosigning may be important not only for the increased effective collateral but also for the induced peer monitoring (Stiglitz, 1990).Ghatak (1999) model focuses on the effect of group lending on adverse selection and screening. Besley and Coate (1995) explored the role of social ties within the group leading to better repayment performance.

Many empirical evidences alsoreported on how group lending helped to reduce asymmetric information, enhanced monitoring and reinforcement of repayment. Wenner (1995) indicates that the repayment

3

performance of the group in Cota Rica improves when groups have written (formal) rules stating how members should behave. Another finding of this research is that if the groups are located in remote areas,this reduces their possibilities to access to alternative sources of credit. Zeller (1998) based on 146 credit groups in Madagascar reports that groups with strong social ties show higher repayment rates. Moreover, groups with internal rules and regulations demonstrate better repayment rates as already indicated by Wenner (1995). Moreover, heterogeneity in asset holdings among members, and related intragroup diversification in on farm and off-farm enterprises, enables members to pool risks so as to better secure repayment of the loan (Zeller, 1998). Hermes and Lensink, (2007) showed empirical results on the nature and determinants of the trade-off between financial performance and outreach of microfinance institutions. Nawai and Shariff (2012) explored the factors affecting repayment performance in microfinance programs providing private lending, in Malaysia, indicate that gender, formal religious education, distance to the lender office, business formality, total sales per month, total loan received, loan monitoring and loan disbursement lag havesignificantly affected the burrow’s repayment performance. Moreover, no pressure from microfinance institutions on the burrowers to pay their loan, delayed the clients’payment or just pay at a minimum amount and not fully utilizing the loan given for business makes them finally default. Further developments were focused on overcoming problems, in the works just cited, such as model specification, endogeneity of social ties and errors in measurement some of the variables, and gather more empirical evidence on determinants of performance in joint liability lending (Ahlin and Townsend, 2007;Karlan, 2007; Cassar et al. 2007). Ahlin and Townsend (2007) resultindicate that social ties between group members are not necessarily positive in promoting group repayment, which contrasts with generally accepted view in the literature. Karlan (2007) by creating groups exogenously (circumventing the endogeneity problem) using experiment, finds out group members who have a stronger social connections are more likely to repay their loans and to save more, indicating the importance of group performance and social connections in monitoring and enforcement within the group. He was able to distinguish between strategic defaultsfrom default due to external shocks, which is relevant for the current study. Cassar et al., (2007) results,also based on experiment, provide clear evidence that specific trust between the group members is more important for the group performancethan trust in the society as a whole. Moreover, social and cultural homogeneity of group members improves performance.They also find that past positive experience with other members helping an individual to repay her loan, provides incentives to this individual to help others repaying their loans in the future. Finally, they indicate the importance of disentangling different aspects of social capital when explaining group repayment performance. Berhane (2009) using panel data examined how joint liability risk and the threat of losing future access to credit affect households’ burrowing decisions. His results indicate that contractual risk in joint liability negatively affects participation in burrowing indicating that households are willing to pay a positive risk premium to avoid joint liability contract in favor of individual liability contract (Berhane, 2009). Moreover, his result indicates that successful repayment rates in joint liability can also rise under risk heterogeneity. Some authors report that the role of uncontrollable factors such as natural disasters and personal crises affecting the ability to repay a loan (Sterns, 1995) and adverse shocks affecting the borrower, or to the low performance of institutions, like justice (Godquin, 2004). The vital research question is, hence, although almost all the theoretical and empirical evidences (cited) indicate group liability, hailedas the most important innovation of microfinance, improves repayment performance, semi-structured interviews, part of this study, from Ethiopia attest that repayment rates

4

are quite low. Does weak group performance explain the low repayment or do different types of risk explain farmers’ low performance rate? The main objective of this study are, thus,

1. Whether access to credit enhances adoption of irrigation technologies, 2. Identify determinants of farmers‘participation in credit markets and demand for loan (including

the role of different types of risks),and 3. Identify different factors, including risks that explain the repayment performance of burrowers.

The subsequent sections are organized as follows. Section two presents a conceptual framework followed by description of data and study site and description of empirical methodologies used in this paper.. Section 3 presents results and discussions and the final part concludes and draws policy implications.

2. Research Methodology 2.1 Conceptual/theoretical framework

The conceptual framework is described below in Figure 1. Household’s loan demand is normally influenced by the household’s capital requirements for capital investments (farm inputs, new production and irrigation technologies, non-farm investment, etc) and consumption needs. However, households may or may choose to apply for loan from MFIs. Households may prefer private loan for fear of the risk of joint liability. Besides, this option is not available in Ethiopia unless households access informal source (from neighbor, friend or wholesaler)†. Households can access loans from MFIs but they may choose not to apply for fear of default (as a group or individual) because of weather, market, health, etc shocks and can’t secure future access of loans.Households’ loan demand is expected to be influenced by asset holdings (farm size, livestock, labor holdings and value of other farm assets), access to irrigation and other household and village characteristics like distance to office of MFI and extension services. Households’ loan repayment is also influenced positively by group liability because of screening, monitoring and enforcement mechanisms of the group. Private liability, depending on the type of collateral used, could improve repayment performance as well. However, the use of collateral in rural setting in Ethiopia is difficult, especially where land title is not used as collateral.As was the case with loan demand, repayment performance is influenced by asset holdings (farm size, livestock, labor holdings and value other farm assets), access to irrigation, expected income and saving, and other household and village characteristics like distance to office of MFI and extension services. However, the presence of external shock such as weather risk, market risk, health risk, etc could make group responsibility nonfunctional because the group as a whole may default due to those risks. The effect of these risks on private loan repayment could bethe same, resulting in default. Access to irrigation could be vital in minimizing crop failure risks but seasonal volatility of horticultural crops is rampant in Ethiopia. The effect of irrigation development on minimizing health risks (because of better nutrition) is not significant as some evidences indicate (Hagos et al., 2017).

† There are cases where MFIs grant individual loans for SMEs (Hagos et al., 2016).

5

Figure 1: Conceptual framework

To reiterate the above statements in the form of hypotheses: H1: Capital requirements increases the loan demand (whatever the lending system is) , H2: Generally households prefer private lending contrary to the classical cases of failure in private lending the call for interventions in MF with group liability, H3: Households with larger asset holdings (farm size, livestock, labor holdings and value of other farm assets), access to irrigationmay demand more loan for capital investment and are more able to repay the loan. H4: Group liability improves repayment performance with improved screening, monitoring and enforcement effects expected from the group. H5: External shocks may increase loan demand for consumption and nonfarm investments purposes and decrease the repayment performance of the group or the individual household.

Theempirical questions outlined above (in the form of hypotheses) and their answer could provide evidences that could contribute to literature by examining particularly the roles of external shocks on the functionality of group lending. The study could also draw policy implications on what is required to support credit services to make it more functional.

The expected sign of the variables listed is summarized in Table 1.

Loan demand

New agricultural technologies

Build assets (capital goods)

Consumption needs

Shocks (weather, price, personal, etc)

Group liability

Individual liability

Loan repayment

6

7

Table 1: Variables and their expected sign

Loan demand

Variables Description Expected sign

Access to irrigation Accessing irrigation through WL or otherwise +

Total farm size Total land holdings, both own-operated and rented-in/sharecropped-in (in timad)

+

Livestock holding Number of livestock (in TLU) +

Value of assets Value of farm assets (in ETB) +

Weather shocks/crop failure

Higher incidence of crop failure mainly due to rainfall failure

-

Price shock Higher incidence of price failure -

Health risk Higher incidence of the household member being sick

+/-

Group lending Group liability in lending -

Individual lending Private loan service +

Loan size Amount of loan taken (in ETB) -

Interest rate Annualinterest rate -

Loan duration Length of loan duration (in yrs) -

Loan repayment

Variables Description Expected sign

Total farm size Total land holdings, both own-operated and rented-in/sharecropped-in (in timad)

+

Size of irrigated land Total irrigated land holdings, both own-operated and rented-in/sharecropped-in (in timad)

+

Livestock holding Number of livestock (in TLU) +

Value of assets Value of farm assets (in ETB) +

Weather shocks/crop failure

Higher incidence of crop failure mainly due to rainfall failure

-

Price shocks Higher incidence of price failure -

Health risks Higher incidence of the household member being sick

+/-

Group lending Group liability in lending -

Individual lending Private loan service +

Interest rate Annual interest rate -

Loan duration Length of loan duration (in yrs) -

2.2 Study site and data description

The study covered four sites of the Feed the Future (FtF) - Innovation Laboratory for Small Scale Irrigation (ILSSI) in Ethiopia including Robit-Bata and Dangila in Amhara, Lemo in SNNP and AdamiTulu in Oromia regional states, respectively (see Figure 1). ILSSI promoted irrigation technologies such as Pulley, Rope& Washer pump, motor pump and solar pump technologies for small scale farmers in 4 target

8

kebeles‡through provision of revolving funds. ILSSI selected 193 target households administratively from these sites and the survey selected 207 control households randomly (in total 400 households) to assess whether access to credit,from ILSSI or other sources, promotes use of irrigation technologies. The purpose it to create comparable households before exploring the impact of access to credit on adoption of irrigation technologies.

Figure 1: Location of study sites

Parallel to the household survey semi-structured interview was conducted in all ILSSI sites the result of

which is reported below.

2.3 Description of data used and its organization

Different dependent variables were used in this study. In assessing impact of credit access on adoption of small scale technologies, application for loan (yes/no) was used as dependent variable and cost of the technologies was used as an outcome variable. In Explaining determinants of participation and level of participation, participation was measured by binary variable (yes or no) and level is measured by max amount of loan taken last year. Moreover, repayment performance was measured interms of the degree of arrearsmeasured as a ratio of paid loan to total loan received. Various variable like household characteristics (like head’s sex, age, educational status of the head), labor endowment, c-worker ratio, plot size, both rain fed and irrigated, livestock holding (expressed in terms of Tropical livestock unit (TLU)) asset holding, village level variables related to access to services and infrastructure, credit service related variables, risk variables were used, according to their specification, in the different models. 2.4 Empirical models

‡ Kebele or peasant association is the lowest administrative unit in Ethiopia. It covers an area which inhabits about

1000 households.

9

Three empirical model are used in this study, a description of their main features are given below. When assessing the impact of certain intervention on technology adoption, propensity score matching provides a promising tool to find comparable groups of treated and control groups, that is, applicants of and non-applicants of loans. Various works have been devoted to explain the structure of the model(Wooldridge, 2010; Khandker et al., 2010), why it is preferred (;Ravallion, 2003; De Janvry et al., 2010), its application (....; Hagos et al., 2012), including user-written commands (Becker andIchino, 2002). For matching to be validcertain assumptions must hold.The primary assumption underlyingmatching estimators is the Conditional IndependenceAssumption (CIA). CIA stated that the decision to adopt israndom conditional on observed covariates x .This assumption imply that the counterfactual outcome inthe treated group is the same as the observed outcomes fornontreated group.This assumption rules out selection into the program on thebasis of unobservables gains from access. The intuition is that two households with the same probabilityof adoption will show up in the treated and untreated samples.Once the propensity score (pscore) is estimated, the data is split intoequally spaced intervals (also called common support).Within each of these intervals the mean pscore and ofeach covariate do not differ between treated and control plots.This is called the balancing property. The second empirical methodology used is Cragg’s double -hurdle approach. Given the objective of assessing the factors that determine the participation and level of participation in credit market, double-hurdle model proposed by Cragg (1971) is the most appropriate econometric model. Double-hurdle model accounts for the existence of a significant number of desired demand for credit, but farmers are too constrained to adopt them. In other words, the double-hurdle considers the possibility of zero outcomes in the second hurdle even if the first hurdle is passed. Mal et al., (2012) indicated the assumptions in the Heckman selection model (Heckman, 1987)are too restrictive. A farmer passes two hurdles: taking a loan or not, and once the decision is made to participate then the farmer decides the loan size from the given options. The model integrates and simultaneously estimate the probability of participation in the credit market and the truncated normal model for loan size. Thus, double-hurdle model is specified as follows. Specification- first probit model

xxwP 1 (2)

Where P is the probability, w is a binary variable of participation in credit market, is the commutative

normal distribution, x is a vector of household characteristics, asset holdings including labor, livestock and land holding, credit service characteristics, risk variables - that includes weather risk, crop damage

risk, personal risk ( household member being sick or death), price risk, etc and are the coefficients to be estimated.

The level of loan demanded,*Y , assumed to have a truncated normal distribution as indicate below:

ixY *

(3)

Where *Y represents the amount of loan demanded last year, x is a vector of household and service

and infrastructure characteristics, credit services (like interest rate, duration, possibility of rescheduling,

etc), risk variables, regional dummies and are estimated parameters and i is the error term. The double-hurdle model is estimated in STATA using the user-written command craggit (Burke, 2009). Binary and truncated (tobit) regression models could be estimatedthe probability of repayment and the repayment performance respectively, we refrain to explain the structure of these model as they are widely described in the literature (See Wooldridge, 2010).

10

3. Results and discussions 3.1 Summary statistics

The summary statistics is reported in Table 2. The average age of the household head is 46 years there is no significant difference in age between those applied for loan and not. Male household-head consist almost 82.5 precent of the households interviewed. Households who applied for loan last year had about 4 adult while those who didn’t apply for loan had about 3.5 adults, which could be a measure of labor holding, are significantly associated with loan application. The c-worker ration is about 0.8, for both households who applied for loan and those not, is not significantly associated with loan application.

Table 2: summary statistics (n= 2903)

Variable

name

Description of the variable Applied for

loan

Not

applied

t-test/X2

test

age Age of household head/spouse 45.9 46.6 1.0702

sex Sex of the household-head

(reference female)

82.52 -

Adults Male and female adults between

the age of 14-65 years

3.83 3.58 -3.0685***

c-worker ratio Consumer-worker ratio 0.823 0.825 0.0775

Land holding Cultivated land both rainfed and

irrigated , in timad§)

0.84 0.93 2.4250**

Irrigland13/14 Irrigated land in 2013/14 (in

timad)

0.20 0.70

Irrigland12/13 Irrigated land in 2012/13 (in

timad)

0.17 0.47

TLU Household livestock holding (in

TLU)

3.51 2.55 -10.303***

Oxen holding Number of oxen the household

owns

1.06 0.80 -9.5777***

Large ruminant Number of large ruminants in

TLU the household owns

2.78 2.002 -

12.3960***

Small

ruminants

Number of small ruminants in

TLU the household owns

0.119 0.13 1.3381

Equines Number of equines in TLU the

household owns

0.608 0.42 -4.7717***

TLU less oxen Household livestock holding

less oxen (in TLU)

2.83 1.88 -9.5976***

Asset value Value of asset holding 100031.4 58965.41 -2.3504**

Applied loan Applied for loan for different

purposes (%)

72.35 - -

Able to obtain Able to obtain loan (%) 84.06 - -

Max amount Max amount of finance that

could be obtained (in ETB)

22446.97 8792.39 -1.6656*

Interest rate Annual interest rate (%) 16.93 - 5.92

§ A timad is quarter of hectare.

11

irrigexp Households with irrigation

experience (%)

0.34 0.32 0.2268

Year irrigation

experience

Number of years of irrigation

experience

2.48 2.018 -2.0275**

irrigtech Use of irrigation technology (%

of irrigators)

0.57 0.44 .0185

Irrigcost Cost of irrigation technology 1779.537 2574.69 5.1025***

distancemm Distance to major market (in

km)

24.55 54.90 14.2455***

distanceawr Distance to all weather roads (in

km)

19.6 10.86 -1.4937

distancemfi Distance to microfinance

institution (in km)

7.143 16.60 5.6351***

Amhara Households from Amhara region

(%)

60.25 - -

Oromia Households from Oromia (%) 12.95 - -

SNNP Households from SNNP region

(%)

26.79 - -

Source: survey 2016

The average land holding is 0.8 and 0.9 timad for applicants and non-applicants, which was significantly different, implying that households with larger average land holding were non-applicants. Irrigated land holding was about 0.2 of timad in 2013/14. Livestock holding, in tropical livestock unit (TLU) was 3.5 units for loan applicants and 2.5 units for loan non-applicants. When disaggregated, oxen holding, large ruminants and equines are significantly higher for loan applicants. Likewise, value of asset holding was significantly higher for applicants compared to loan non-applicants. As far is loan demand is concerned, close 72 percent of the households have applied last year. The mean loan was about 16000 ETB. About 34 percent of the households have irrigation experience, an average of about 6 years, ranging from 1 to 30 years. Households with significantly higher irrigation experience (close to 3 years) applied for loan compared to households with 2 years of irrigation experience which didn’t apply. The average irrigation technology cost is ETB 2000. Applicants covered significantly higher irrigation costs than non-applicants. ILSSI target households invested about ETB 3867.60 compared to to non-target households ETB 694.50 which was significantly different at 1 percent level of significance. About 46 percent of the households apply various irrigation technologies (not indicated in the table), 12.96 % motor pump, 13.07 % rope & washer pump, 19.22 % pulley, 1.19 % solar pump, 0.17 % electric Pump, 0.91 % treadle pump and 6.43 using other technologies like bucket, hose and watering can. The dominant water application is using Bucket/hose/ watering (73.5 %) can followed by surface flooding (13.87 %), furrow irrigation (7.05 %) and drip technology (3.30 %). Major credit sources in the study sites are are microfinance institutions (76.32 %), followed by friends/relatives (4.22 %) and farmers groups (3.73%). Formal banks and private lenders do play a very minimal role. Disaggregate by purposes, loan requested are for oxen purchase (32.30%), purchase of farm inputs (24.72 %), such as improved seeds and agro-chemicals, to cover health expenses (19.45%), small scale irrigation (17.96 %) and for other/nonagricultural businesses (3.42 % ). There is limited credit service for covering family events and consumption expenses. The important reasons for not applying for loan are fear failure of repayment (50.75%); the household did not require extra cash (31.0 %), households didn’t apply for another loan because outstanding loan and granting another loan (in the face of outstanding loan) requires collateral (6.40 %), the loan request of the household last year was not granted (5.44%), the household didn’t know there were credit service possibilities around (2.45%) and other reasons (3.94 %).

12

The average distance to major markets (in km) was 25 and 55 kms for applicants and non-applicants respectively while average distance to microfinance institution is about 7 and 16 kms for applicants and non-applicants respectively. Both variables were significantly different implying that access to market and microfinance institution is important for participating in loan service. In general, cross tabulation results indicate that labor, land, asset holding, livestock holdings, and closeness to major markets and microfinance institution were important for loan application. Whether this holds true will be explored in multivariate setting as we control for more covariates in the subsequent sections.

3.2 MFI situation in Ethiopia

The National Bank Ethiopia (NBE) developed the legal framework to further enhances the development and soundness of the micro-financing business in Ethiopia (FDRE, 2009). The main purpose of a micro-financinginstitution, according to the Micro-Financing Business Proclamation No. 626/2009, is to collect deposits andextend credit to rural and urban farmers and people engaged in other similar activities as well as micro and small scale rural and urbanentrepreneurs, the maximum amount of whichmay be determined by the National Bank (FDRE, 2009). According to the same proclamation, loans may be made without collateral, secured by collateral or secured by group or individual guarantees as appropriate and at thediscretion of the institution.The NBE issued directives on licensing and supervision of MFIs following the proclamation (NBE, 2013). Several micro finance institutions (MFIs) were established and have been operatingtowards resolving the credit access problem of the poor (Kereta, 2007). As at the end of June 2007, twenty-seven Microfinance institutions operate in the country, obtaining license from National Bank of Ethiopia (Kereta, 2007).Wiedmaier-Pfister et al., (2008) show the regional distribution of MFIs in Ethiopia. Currently the Association of Ethiopian Microfinance Institutions (AEMFI), an umbrella organization of MFIs, has 30 members. Moreover, savings and credit cooperatives, which are not member of AEMFI, are growing in number and size in the country (Wiedmaier-Pfister et al., 2008). NBE removed all interest rate ceilings in the financial sector in 1998, but this has not led to free market-based determination of interest rates in all cases. Lending rates of Ethiopian MFIs range between a 9% a year to a 24% flat rate**, and are thereby relatively lower than other sub-Saharan countries Wiedmaier-Pfister et al., (2008). MFIs affiliated with the government or international NGOs are dominant, which is a source of success but also concern. Finally, almost all MFIs offer micro lending and saving to small holder farmers, although the lending and saving products are limited and some MFIs, mainly the larger ones such as ACSI, DECSI and OMO, offer a limited range ofmicro insurance, leasing, pension fund and transfers products. The results of the semi-structured interviews, covering only 4 sites, reflect general picture which already depicted the above paragraphs. The credit system is cash based in Adami Tulu where the maximum loan size was ETB 6000 for irrigated system, compared to ETB 3000 for rainfed, have increasing its loan size and reaching more farmers from year to year. In Lemo the maximum loan size is ETB 10000, for water lifting technology, and ETB 2000 for rainfed, while the interest rate is 12% in both sites. Oromia and Omo microfinance organizations practice group lending. Credit is provided on rainfed and irrigated systems, mainly focusing for the purchase of farm inputs. The repayment duration is usually a year for farm inputs and quarter of year for fattening and poultry, it shows no significant differences in

**A group loan from the largest MFI (ACSI) costs 18% a year and DECSI’s group loan costs 15%. In both MFIs, an individual MSE loan costs 9% to 12.5%. MFIs affiliated to international NGOs or the privately owned Aggar charge a flat rate of 15% to 24% respectively. MFIs, which are donor

13

repayment rates, low repayment rates in both sites and for both 12 % and lower in Lemo (10%), between rainfed and irrigated system is reported. The main challenges in both Adami tulu and Lemo are crop failure due to drought, flooding or hailstorm, market price volatility, high interest rate and perception of farmers of loans as gift. Perception of farmers of loans as gift emanated the microfinance history, which many of them transformed from humanitarian-oriented organisations or affiliated to regional governments were not initially focused on financial sustainability and outreach(Kereta, 2007; Kassa, 2010). Proximity to district capital, access to extension agents (also called development agent) and water source (like rivers and lakes) is appraised as important factors that increase the repayment rates. Workers of MFIs in Adami tulu are trained twice annually on credit appraisal and they appraised to have the knowledge to design appropriate financing products and farmers are likewise informed about existing loan products by front line workers of MFIs and Das, besides information provided through brochures, radio and television programs. In Lemo workers of MFIs are trained once every year on customer service. The extension department is appraised as one of supporting structure, in selection of loan beneficiaries and granting extension advice on the use of farm inputs. In Lemo the cooperative office is appraised as supporting in selection for loan taking and provide training on how to manage and use the resources. There are some organizations such as Rift Valley Project and Productive Safety Net Program, which helped the Oromia microfinance institution in loan provision to farmers. In Lemo the Household Asset Program, ILSSI project, extension and cooperative Bureaus do provide support to Omo microfinance institution. Workers in Adami Tulu see seed money management, provided by projects, as something outside of their major task although more farmers could benefit from revolving the fund in accessing water lifting technologies. But repayment rate of seed money was not reported. Albina credit and saving cooperative, operating in Dangishta kebele and serving members only, offering loans in cash only stated that the average loan size is currently ETB 5256, have increased its loan size and reaching more farmers from year to year. The loan is given to members in the form of private loans for fattening and purchase of livestock. There are no credit products for Irrigation or other services.Finance providers have the capacity and knowledge to develop suitable loan/financing products, although not for irrigation. Credit worthiness of lender is appraised of being a member, saving experience of six months and six times saving, collateral – fixed assets and personal guarantee are used. The interest rate is 14.5%, loan duration ranges between 0.5year for agricultural inputs (rainfed) and 5 years for non-agricultural and the repayment rate was 100% in the last two years. Having ground rule (bye-laws) of the association, membership grant and private lending, strict loan rule and strict follow up and strong stakeholder involvement, is the reason for maintaining 100% repayment rate. Various offices are involved in the credit program offered, the local administration help to enforce repayment of the loan; the cooperative agency is active in follow up, training, and office provision. The cooperative agency provides support to the cooperative in terms training and credit appraisal. Albina had no experience in seed money management; no members have benefited from seed money. The main challenge faced giving out loans to farmers was shortage of capital and staff problem. The lack of staff makes effective monitoring of loans difficult. Even though there is no loan default, crop failure due to bad weather, flooding, hailstorm, etc market prices volatility, willful default and wrong timing of credit delivery are important.

3.3 Credit access on cost of irrigation technology used

As was explained earlier, we created comparable groups of control and treatment using propensity score matching (psmatch2). The scoring variable used was access to loan last year. Of the total 1839

14

observations 604 untreated and 1220 treated(99 percent of the total) were on the common support.The balancing property is, thus, satisfied in the model. Table 3: Propensity score variables

Dependent variable: apply to loan (yes=1 no=0) Coef. SE.

Project (ILSSI) target farmer 0.048 0.154

Educational status of household 0.009 0.0037***

Male-headed household (reference Female headed) 0.481 0.167***

Male adult labor -0.018 0.008**

Female adult labor 0.028 0 .0115***

# of large to ruminant 0.234 0.036***

# of equines 0.018 0.060

# of small ruminants -0.082 0.033***

Asset value (in ETB) -0.00002 0.000004***

Land holding (in timad) 0.0006 0.0013

Irrigation technology use (reference no=0) 0.439 0.151***

Distance to FTC (in km) -0.008 0.002***

Distance to major markets (in km) -0.023 0.002***

Cons- 0.304 0.198

Number of obs = 1839 LR chi2(13) = 462.49 Prob > chi2 = 0.0000

Log likelihood = -932.96692 Pseudo R2 = 0.1986

*, **,*** significant at 10, 5, and 1 percent level of significance Source: survey 2016

Several propensity score variables were identified (Table 3). Households that apply irrigation technology are more likely to apply to loan services than households that did not use irrigation technologies, even if they have access to irrigation. Households’ heads with higher educational attainment were found more likely to apply for loan. Male-headed households are found to be more likely to apply for loan, compared to female-headed households, which is true the socio-cultural environment in Ethiopia.However, households with female adult labor are more likely to participate, and households with more male adult labor were not. Households with more large ruminants are more likely while households with small ruminants less likely to participate in credit market, partly because larger ruminants are used as crdit collateral. Unexpectedly, households with more asset value are less likely to participate, perhaps indicating lower capital requirement of such households. Finally, households located far from farmers’ training center (FTC) and major market are less likely to apply for loan indicating the importance of access to extension and market.

15

Table 4: Estimate of ATT on technology cost

Variable Observations Treated Control t-test

Technology cost Unmatched 2012 2267 -1.45

ATT 2018 1595 1.66

Source: Survey 2016

Based on the calculated average effect of treatment for the treated ATT vales (see Table 4), the mean difference in cost of technologies was significantly different between the treated and control households, indicating the importance of access to credit for adoption of irrigation technologies.

3.4 Determinants of participation in credit market and level demanded

The factors that explain the participation and level of loan demanded is reported in Table 5. Educational attainment of the head have a positive influence on participation. Female adult labor, however, negatively influence participation.Asset holding, mainly owning small ruminants affect participation positively. Issues related to credit provision services also influence participation. Length loan repayment duration, higher annual interest rate, unexpectedly, possibility of payment rescheduling and private loans do influence participation positively. Risk factors like whether last year was a good year and higher likelihood of drought in the next five years do influence participation positively. Village level factors like access to extension and market services do have an influence on participation. Distance to microfinance institutions and to farmers training center (with in the village) have positive effect while distance to major district market has negative effect for participation.

Table 5: Determinants of participation and level credit demanded

Dependent variable: apply to loan (yes=1 no=0) Dependent variable: apply to loan (yes=1 no=0)

Variables Coef. SE. Coef. SE.

Age of the household head -0.018 0.033 -109.657 28.577***

Educational status of household

0.081 0.045* 22.007 20.48

Male-headed household (reference Female headed)

0.208 0.856 880.393 880.825

C-worker ratio - - 266.064 455.985

Livestock holding (in TLU) - - 528.281 124.755***

# of large to ruminant (in TLU) -0.439 0.538 - -

# of small ruminants 16.387 8.626** - -

Asset value (in ETB) 0.00001 0.0001 -0.007 0.003**

Land holding (in timad) 0.0757 0.0687 - - Irrigated land ratio - - 696.586 771.36 Irrigation technology use (reference no=0)

0.988 0.988* -1117.68 651.294*

16

Distance to MFI 0.172 0.093* 166.311 94.38*

Distance to FTC (in km) 1.281 0.606** -763.55 243.22*** Distance to major markets (in km)

-0.152 0.0594*** 1.396 12.565

# good seasons in the past 5 yrs 2.418 0.764*** -1468.108 367.654*** Annual interest rate (%) 1.254 0.386*** 312.215 64.196*** Duration -0.403 0.154*** -127.558 56.588***

Private loan (reference group loan)

- - -2045.167 801.124***

Repayment rescheduling 1.683 0.959* -1488.47 802.530* Good growing season (reference bad)

25.114 8.926*** -7283.08 1096.86***

Likelihood of drought in the nest 5 yrs

-0.269 0.693 1062.804 443.901**

Likelihood of crop damage (due to hail)in the nest 5 yrs

1.037 0.653 -981.314 443.901**

2 yrs ago Price fall of a major crop

11.637 174.155 856.654 1103.653

Kilel1 - - 273.137 861.055

Kilel33 - - 1130.08 906.265 Cons- -27.927 9.700*** 18473.21 3171.744***

*, **,*** significant at 10, 5, and 1 percent level of significance Source: survey 2016

The level of loan demanded is positively determined by the age of the household, implying that households headed by older heads demand more loans perhaps indicating the year of experience in agriculture and having more labor endowments than younger households. Households with larger livestock holding (in TLU) demand more loans. On the other hand, households with larger valued assets demand lower loans. Similarly, households using irrigation technologies demand lower loans. Village level factors like access to extension and market services do have an influence on the loan demanded. In contrast to the decision to participate, distance to microfinance institutions and to farmers training center (with in the village) have negative effect while distance to major district market has a positive negative effect for loan demanded. As far loan provision services are concerned, length of payment, annual interest rate and repayment rescheduling and private loan have negative effect of the amount of loan demanded. The first two are important variables to decide loan size although the influence latter two on loan size is not intuitively clear. Contrary to participation, risk factors like whether last year was a good year, unexpectedly, and higher likelihood of drought in the next five years do influence participation negatively.

3.5 Determinants of the repayment performance

As explained earlier, the dependent variable was generated as ratio of the paid amount to the amount loan received. According to our calculation, 27 of the observation were left-censored (with ratio less than zero) and the rest 354 observations were uncensored. The factors that explain repayment performance is reported in Table 6. Several household factors like age, sex and educational status of the head were significant in explaining repayment performance.

17

Households with older heads have a higher repayment performance. Male-headed households have a better repayment performance than female-headed households. Households with heads with higher educational status, do have a higher repayment performance. The story that is emerging is households with relatively with better educational status are more likely to participate (not significant in loan size demand) and have higher repayment performance. Households with more female adult labor have higher repayment performance while households with larger male adult labor have lower repayment performance. Table 6: Determinants of repayment (Tobit regression)

Dependent variable: repayment ratio Coef. SE.

Project (ILSSI) target farmer 0.518 0.107***

Age of the head 0.009 0.002***

Educational status of household -0.003 0.025

Male-headed household (reference Female headed) 0.490 0.122***

Male adult labor -0.077 0.045*

Female adult labor 0.186 0.061***

Livestock holding (in TLU) -0.031 0.0188*

Asset value (in ETB) -0.000003 0.000007

Private loan (reference group loan) -1.184 0.252***

Loan diversion) (reference used for the same purpose) -0.010 0.125

Annual interest rate -0.008 0.009

Loan duration 0.128 0.008***

Payed once a year (reference payed twice a year) 1.000 0.227***

Payed at once in the repayment duration (reference payed twice a year)

1.562 0.202***

Rescheduling ( reference no possibility) -0.407 0.154***

Likelihood of drought in the next 5 yrs 0.129 0065

Crop damage in the next 5 yrs 0.0298 0.066

Price fluctuation of major crops 2 yrs ago -0.115 0.159

Kilel1 0.155 0.124

Kilel2 0.217 0.141

Cons- -3.440 0.498***

Number of obs = 381 LR chi2(20) = 376.47 Prob > chi2 = 0.0000

Log likelihood = -400.83253 Pseudo R2 = 0.3195

*, **,*** significant at 10, 5, and 1 percent level of significance Source: survey 2016

Contrary to role of livestock in loan demand, households with larger livestock holding have lower repayment performance. Loan provision services, such as private loanservice, loan diversion, higher annual interest rate and possibility of repayment rescheduling have negative effect on repayment.Compared togroup loans, households taking private loans have lower repayment performance. This study is in line with the mainstream economics that argue that group liability is necessary in ensuring enforcement of repayment (Stiglitz 1990;Ghatak, 1999; Besley and Coate, 1995). The evidences in the literature that indicate that loan diversion and higher interest rate make repayment more difficult (...). Rescheduling is usually is

18

understood as one of the mechanisms to ensure repayment. Our result is counter intuitive. On the other hand, duration of payment, payment schedule - payment once a year and payment once in the repayment duration(contrary to payment twice a year) - have a positive effect on repayment performance. While the positive effect of the length of duration of repayment on repayment performance is understandable, the effect of payment schedule on loan repayment is not clear. Risk factors and regional dummies were not significant in explaining repayment performance.

4. Conclusions and recommendations Accessing water lifting technologies enhances household’s access to irrigation water without the requirement of making huge investments on irrigation storage and conveyance structures. There are studies that indicate the economics of some of the water lifting technologies (Gebregziabher et al., 2016; Hagos et al, 2016). However, purchasing irrigation technologies may require availability of credit service.

The study explores whether having access to credit market is important for the adoption and level of small scale irrigation technologies used. The study also examines the determinants ofparticipation and level of loan demanded. Moreover, the study assessed determinants of repayment performance.

The outreach of the existing microcredit services, considering only the magnitude of the household served, is encouraging, reaching about 72 percent of the households mainly for the purchase of farm inputs, livestock purchase and fattening, irrigation technologies and non-agricultural investments. Considering MFI and saving and credit cooperatives, the former have higher average loan size reaching more households, the number increasing every year, but the latter although small are also promising.

Products served by MFIs and saving and credit cooperatives is limited, however, and the credit instrument is predominantly cash. Semi-structured interview indicate that the repayment rate in Adami tulu and Lemo is low, about 12% although it is 100% in Dangishta, served by a saving and credit cooperative. The presence a conducive regulatory framework and considerable government and donor support for MFIs is also appreciated. However, another element which is reported by the literature, not fully supported by evidence of this study, is there is a lack of innovative, demand-driven financial services and sustainable institutions that can cater for the huge unmet demand of poor households (Wiedmaier-Pfister et al.,2008).

The econometric results indicate that loan access is one of the important variable for adoption of small scale technologies. This study shows that the mean difference in cost of technologies was significantly different between the treated and control households, indicating the importance of access to credit for adoption of irrigation technologies.

Various factors influence credit participation such as educational attainment of the head, female adult labor and asset holding, repayment duration, interest rate, possibility of payment rescheduling and private loans. Risk factors like whether last year was a good year and higher likelihood of drought in the next five years do influence participation positively. Moreover, access to extension and market services do have an influence on participation. Distance to microfinance institutions and to farmers training center (with in the village) do have positive effect, which is counter intuitive, while distance to major district market has negative effect for participation. The level of loan demanded is positively determined by the age of the household, larger livestock holding (in TLU) enhanced loan demand. Village level factors like access to extension and market services do have an influence on the loan demanded. In contrast to the decision to participate, distance to microfinance institutions and to farmers training center (with in the village) have negative effect while distance to major district market has a positive negative effect for loan demanded.

19

Loan provision services such as length of payment, annual interest rate and repayment rescheduling and private loan have negative effect of the amount of loan demanded. The first two are important variables to decide loan size although the influence latter two on loan size isintuitively lessclear. Risk factors like whether last year was a good year, unexpectedly, and higher likelihood of drought in the next five years do influence participation negatively.

Households with older heads have a higher repayment performance. Male-headed households have a better repayment performance than female-headed households. Households with heads with higher educational status, do have a higher repayment performance. Households with more female adult labor have higher repayment performance while households with larger male adult labor have lower repayment performance. Loan provision services, such as private loan service, loan diversion, higher annual interest rate and possibility of repayment rescheduling have negative effect on repayment. Compared togroup loans, households taking private loans have lower repayment performance. The evidences in the literature that indicate that loan diversion and higher interest rate make repayment more difficult (...).On the other hand, duration of payment, payment schedule dummies have a positive effect on repayment performance. Risk factors and regional dummies were not significant in explaining repayment performance.

Policy decisions should undertake measures to further increase the reach and effectiveness, although the later not fully addressed in this study, of microfinance to enhance irrigation development in Ethiopia. Moreover, promoting irrigation technology through credit service may also require improvement of the input and output markets and strengthening extension through farmers’ training services, as it is usually the case. These measures will ensure the development of Irrigation sustainably, thereby, increase the demand for credit.

There is need of linking credit with insurance because households usually refrain from taking loan due to crop fail and prices risks. this will provide another future area of research in countryes like Ethiopia.

Acknowledgement

This study was supported Feed the Future(FtF)Innovation Lab on Small Scale Irrigation (ILSSI). We are grateful to the respondents for being willing and taking their time to respond to our long questionnaire. The errors in analysis, if any, are of the authors only. The results of this study do not reflect views of the project or funding agency.

20

References

Adeoti, I. A. 2009. Factors Influencing Irrigation Technology Adoption and its Impact on Household Poverty in Ghana. Journal of Agriculture and Rural Development in the Tropics and Subtropics, 109(1): 51–63. Ahlin, C. and Townsend, R. 2007. Using repayment data to test across models of joint liability lending, Economic Journal, 117: F11-F51. Amin, S. Rai, A. S. Topa, G. 2003. Does microcredit reach the poor and vulnerable? Evidence from northern Bangladesh, Journal of Development Economics 70: 59– 82. Becker, O.S., and Ichino, A. 2002. Estimation of average treatment effects based on propensity scores. The Stata Journal 2(4): 358-377. Berhane, G. T. 2009. Econometric analysis of microfinance credit group formation, contractual risk, and welfare impacts in Northern Ethiopia, PhD dissertation, Wageningen University. pp. 160 Besley, T. and Coate, S. 1995. Group lending, repayment incentives and social Collateral, Journal of Development Economics Vol. 46: 1-18.al of Development Ec Brau, J. C.; Woller, G. M. 2004. Microfinance: A comprehensive review of the existing literature, Journal of Entrepreneurial Finance, JEF, ISSN 1551-9570, Vol. 9, Iss. 1, pp. 1-27. Burke, W. J. 2009. Fitting and interpreting Cragg’s tobit alternative using State, The State Journal 9 (4): 584-592. Cassar, A. Crowley, L. and Wydick, B. 2007. The effect of social capital on group loan repayment: evidence from field experiments, Economic Journal, 117: F58-F106. Cragg, J.C. 1971. Some statistical models for limited dependent variable with application to the demand for durable goods, Econometrica39 (5): 829-844. De Janvry, A., Dustan, A., & Sadoulet, E. 2010. Recent advances in impact analysis methods for ex-post impact assessments of agricultural technology: Options for the CGIAR. SPIA report version 3.1. University of California at Berkeley. Evans, T.G. Adams, A. M. Mohammed, R. and Norris, A.H. 1999. Demystifying Nonparticipation in Microcredit: A Population-Based Analysis, World Development 27, (2): 419-430. FDRE (Federal Democratic Republic of Ethiopia). 2016. Growth and Transformation Plan II (GTP II) (2015/16-2019/20), National Planning Commission, May, 2016, Addis Ababa. FDRE (Federal Democratic Republic of Ethiopia). 2009. Micro-Financing Business Proclamation, Federal Negarit Gazeta, Proclamation No. 626 /2009, Addis Ababa.

21

Gebregziabher, G. Giordano, M. A. Langan, S. J. and Namara, E. R. 2014. Economic analysis of influencing adoption of motor pumps in Ethiopia. Journal of Development and Agricultural Economics, 6(12): 490-500. Gebregziabher, G. Haileslassie, A. Biazin,B. Schmitter, P. Chali, A. Hagos, F. Otoo, M. Lefore,N. Barron, J. and Tegegne, D. 2016. Solar powered water pumping can boost smallholder income: A business model based on action research from LIVES and Africa RISING sites, LIVES-Africa RISING-N2 Africa joint workshop and exhibition DEC., 8-9, 2016, Addis Ababa. Getacher, T. Mesfin, A. Gebregziabher, G. 2013. Adoption and impact of irrigation technology: evidence from household level data in Tigray, Northern Ethiopia, African Journal of Agricultural Research, 8(38): 4766-4772. Ghatak, M. 1999. Group lending, local information and peerSelection, Journal of Development Economics Vol. 60: 27–50. Gebregziabher, G. Hagos, F. Lefore, N. Haileslassie, A. 2017.Economic feasibility of water lifting (WL) technologies in Ethiopia, International Water Management Institute, Mimeo. Gogquin, M. 2004. Microfinance Repayment Performance inBangladesh: How to Improve the Allocation of Loans by MFIs, World Development 32 (11): pp. 1909–1926. Ghosh, J. 2013. Microfinance and the challenge offinancial inclusion for development, Cambridge Journal of Economics, 37, 1203–1219 doi:10.1093/cje/bet042. Hermes, N. Z. and Lensink, R. 2007. The empirics of microfinance: What do we know? Economic Journal, 117: F1–F10. Hagos, F., Jayasinghe, G., Awulachew, S. B., Loulseged, M., Yilma, A.D. 2012. Agricultural Water Management and Poverty in Ethiopia. Agricultural Economics, 43 (Issue Supplement S1):1-13. Hagos, F., Mulugeta, A., Erkossa, E., Lefore, N., Langan, S., and Abebe, Y. 2017. Poverty profiles and nutritional outcomes of using spate irrigation in Ethiopia, Irrigation and Drainage. DOI: 10.1002/ird.2117. Hagos, F. Nakawuka, P. Schmitter, P. Erkossa, T. Tegegne, D. Haileslassie, A. Barron, J. Lefore, N. and Mupangwa, W. T. 2016. Drip irrigation and service provision of irrigation water: New ways to step into affordable small-scale irrigated agriculture, International Water Management Institute, Mimeo. Hailu, B. K. Abrha, B. K. and Weldegiorgis, K. A. 2014. Adoption and impact of agricultural technologies on farm income: Evidence from Southern Tigray, Northern Ethiopia, International Journal of Food and Agricultural Economics, 2 (4): 91-106. Heckman, J., 1979. Sample selection as a specification error, Econometrica 47, 153-161. Jin, S. Yu, W. Jansen, H. G.P. and Muraoka,, R. 2012. The impact of Irrigation on Agricultural Productivity: Evidence from India, Selected Poster prepared for presentation at the International Association of Agricultural Economists (IAAE) Triennial Conference, Foz do Iguaçu, Brazil, 18-24 August, 2012.

22

Karlan, D. 2007. Social connections and gropu banking, Economic Journal, 117: F52-F84. Kassa, Y. 2010. Regulation and Supervision of Microfinance Business in Ethiopia: Achievements, Challenges and Prospects, Paper presented at International Conference on Microfinance Regulation, March 15-17, 2010, Bangladesh, Dhaka, Mimeo. Kereta, B. B. 2007. Outreach and Financial Performance Analysis of Microfinance Institutions in Ethiopia, African Economic Conference United Nations Conference Center (UNCC), Addis Ababa, Ethiopia, Nov., 15-17.2007. Khandker, S.R., Koolwal, G.B., Samad, H.A., 2010. Handbook on impact evaluation:quantitative methods and practices. The World Bank, WashingtonDC. Liverpool, L. S. O. and Winter-Nelson, A. 2010. Poverty Status and the Impact of Formal Credit on Technology Use and Wellbeing among Ethiopian Smallholders, World Development 38 (4): 541–554, doi:10.1016/j.worlddev.2009.11.006. Mal, P. Anik, A.R. Bauer, S. and Schmitz, P.M. 2012. Bt cotton adoption: double-hurdel approach for North Indian farmers, AgBioForum 15 (3): 294-302. Namara, R.E., M. A. Hanjra, G.E. Castillo, H. M. Ravnborg, L. Smith, B. Van Koppen. 2010. Agricultural water management and poverty linkages, Agricultural Water Management 97: 520–527. NBE (National Bank of Ethiopia). 2013. Licensing and supervision of the business of microfinancing institutions: Requirements for Licensing and Renewal of Microfinance Business Directives No. MFI/23/201. Addis Ababa, Ethiopia. Nawai, N. and Shariff, M. N. M. 2012. Factors affecting repayment performance in microfinance programsin Malaysia, Procedia - Social and Behavioral Sciences 62: 806 – 811 RavallionM. 2003. Assessing the Poverty Impact of an Assigned Program. In Francois Bourguignon and Luiz A. Pereira da Silva (eds.) The Impact of Economic Policies on Poverty and Income Distribution: Evaluation Techniques and Tools, Volume 1. New York: OxfordUniversity Press. Sterns, K. 1995. The hidden beast: delinquency in micro enterprise credit programme. ACCION Discussion Thesis Document No.6. Stiglitz, J. E. 1990. Peer monitoring and credit markets, The World Bank Economic Review, 4 (3): 3 5 1 - 3 6 6. Tadesse, A. G. Rashid, S. Borzaga, C. and Getnet, K. 2016. Rural Finance and Agricultural Technology Adoption in Ethiopia: Does the Institutional Design of Lending Organizations Matter? World Development, 84: 235–253. Tucker, M. 2001. Financial Performance of Selected Microfinance Institutions: Benchmarking Progress to Sustainability, Journal of Microfinance 3,(2): 108-123.

23

Wenner, M. 1995. Group credit: a means to improve information transfer and loan repayment performance, Journal of development studies, 32(2): 263-281. Wiedmaier-Pfister, M. Gesesse, D. Amha, W. Mommartz, R. Duflos, E. Steel, W. 2008. Access to finance in Ethiopia Sector assessment study, Volume 2, German Technical Cooperation, Eschborn, Frankfurt am Main, Germany. Wooldridge, M. J., 2010. Econometric analysis of cross section and panel data. MIT, Cambridge, Massachusetts. Zeller, M. 1998. Determinants of Repayment Performance in Credit Groups: The Role of Program Design, Intragroup Risk Pooling, and Social Cohesion, Economic Development and Cultural Change, 46 (3): 599-620.