Determinants of rural households ’ participation in microfinance services : The case of Cheliya District , West Shoa Zone , Oromia National Regional State , Ethiopia

In under developed countries, most of the poor people have limited access to formal financial services, including credit, savings, and insurance. The study was focused on the determinants of the rural households’ participation in microfinance services in the study area. The study was conducted in Cheliya District, Oromia Regional State, Ethiopia. A total of 188 sample households were selected through stratified and simple random sampling techniques and interviewed using a structured questionnaire to elicit data pertaining to participation in microfinance services during the year 2017. The data were analyzed using descriptive statistics and logistic regression model. Logistic regression model was used to analyze determinants of the rural households’ use of service in microfinance services. Accordingly, the outcome of the logistic model regression indicated that household heads’ sex, education level, cultivated land size, livestock holding and frequency of extension contact positively and significantly affected the rural household’s decision to involve in microfinance services; while dependency ratio affected their decision negatively and significantly. It is recommended that the microfinance institutions and other concerning bodies have to arrange the way in which households with high dependency ratio and illiterate can participate in microfinance services. Moreover, attention should be given by microfinance institution staffs and other government bodies to increase female involvement in microfinance services in the study area.

formal commercial banks.The microfinance institutions provide small size of loans, saving, insurance services, money transfer and other relevant services to the target poor people who were excluded by conventional commercial banks due to lack of collateral requirements (Tolosa, 2014).In Ethiopia, many microfinance institutions have been established and have been operating towards resolving the credit access problems of the poor particularly to those participating in the small business (Melese, 2013).
The economy of Ethiopia is predominantly agriculture.The performance of the economy depends on the performance of the agricultural sector.Even though there is a little bit of growth in other economics activities, agriculture is one of the main sector for Ethiopia's economic growth and long-term food security.The stakes are high where 15 to 17% of the Government of Ethiopia's (GOE) expenditures are committed to the sector.Agriculture directly supports 72.7% of the population's livelihoods.It contributes 38.5% of Gross Domestic Product (GDP), and over 80% of export value (NPC, 2016).
The large number population in Ethiopia are rural households, and they have a low level of literacy.Majority of the farm community comprised of subsistence farmers who are not in a position to use high-quality seeds, sufficient fertilizers and improved farm land and limited access to credit.Because of this, small farmers generally characterized by low income, less saving and low capital formation.In line with this, the rural development is hindered due to lack of credits, weak infrastructure, and poor transport systems (Wolday and David, 2010;cited in Simon, 2016).
Lack of finance is one of the basic problems in Ethiopia.It hinders the productivity and income of both rural and urban households.Microfinance institutions are working to solve these problems through providing financial and non-financial services in the country.Moreover, these institutions contribute to reduce poverty and economic growth (Wolday, 2004).The concept of microfinance is not new in Ethiopia but, as an industry, it is a relatively new phenomenon.Traditionally, people have saved with and taken small loans from informal channels for unexpected events from the so-called Iqub, that is, an association of people having a common objective of mobilizing finance and distribute it to members through rotating and Idir, that is, a group or association insurance established and operated by the volunteer community (Bezabih, 2009).
Even though agriculture plays an important role in Ethiopia's economy, recently the sector receives less than 10% of financial services.Moreover, the rural economy of the county was dominated by low distribution of financial services.Although indicators of financial access and inclusion have improved over the past two decades in Ethiopia, recent estimates show that the country is yet to catch up with other developing countries (World Bank, 2014).
Most of the poor people who are living in under developed countries have limited access to formal financial services, including credit, savings, and insurance.They instead rely on the informal financial services providers.This occurred due to the formal financial service providers have not considered the poor as a viable market and penetration rates for formal financial services in developing countries are extremely low.Hence, the inability to acquire formal credit support has constrained poor farmers' capability to expand their production and improve technology adoption, nutrition and health status and their living condition (Bauchet et al., 2011).Feleke (2011) finding result showed that the household's income is positively related to participation in microfinance services.Households participate in microfinance institutions in the expectation that borrowing will increase their earnings, smooth consumption, enhance their food security, sustain self-employment, reduce the risk of vulnerability and increase savings to strengthen the basis for human capital formation.Microfinance also enables households to mobilize and harness their resources and optimally exploit the opportunities available to them.Moreover, microfinance services contribute for the improvement of agricultural productivity by adopting productivity-enhancing inputs and modern farming techniques (Ziaul, 2014).
In Ethiopia, the poor households in the country remain with limited access to formal financial services.The majority of rural people and the poor farmers lack access to credit from modern financial institutions.Besides, formal financial institutions are inefficient and inaccessible in providing credit facilities to the poor (Sileshi, 2014).
In the study area, some studies have been conducted.Kebu (2017) studied focusing determinants of financial performance of microfinance institutions in the study area.Further, Birhanu (2016) investigated on the role of microfinance institutions in reduction of unemployment in the study area.However, these studies did not say anything about determinants of the rural households' participation microfinance services on rural households' in microfinance services of the study area.So that this study was focused on assessing determinants of the rural households' participation in microfinance services in the case of Cheliya district, West Shoa zone, Oromia national regional state, Ethiopia.

Description of the study area
The study was conducted at Chelliya District, West Shewa Zone, Oromia National Regional State.The capital of the district, Gedo town is located at 175 km West of Addis Ababa on the main road to Nekemte.The district has 20 kebeles of which 18 are rural and two urban.The boundaries of the district adjoin MidaKegn district in the north, Jibat and Dano districts in the south, Liban Jawi district in the east and Ilu Gelan and Jimma Rare district in the west.The total population of the district was estimated to be 104,448 of which 52,481 are males and 51,967 are females (Figure 1).Among these, about 89,523 are living in the rural areas, and about 14,925 are urban residents (OoA, 2017).

Types, sources, and methods of data collection
This study was conducted based on cross-sectional data obtained from both primary and secondary sources.Primary data were collected through face-to-face personal interviews using a structured questionnaire.Focus group discussion and key informant interview were also conducted to gather sufficient information and to capture relevant data from beneficiaries.The focus group discussion was carried out with clients of microfinance institutions.
Five focus group discussions involve 7 to 10 members in each group employed.Six key informants were also contacted with the staff members of microfinance institutions to get information about how the institution was operating in the area and about the opinion of the people towards the program intervention.On the other hand, secondary data were collected from secondary sources such as review of books, journal articles, unpublished study documents and other official reports, and internet sources.

Sampling technique and sample size
Cheliya district was selected purposively because of insufficient studies on the impact of microfinance service on rural households' income in the study area.For this study, both simple random and stratified probability sampling techniques were employed to select the sample of respondent households.First, among eighteen rural kebeles of the district, six rural kebeles were selected, using simple random sampling technique through lottery method.Then, households in the sample kebeles stratified into participants and non-participants.Finally, the sample size of the respondents was determined by using Kothari (2004) sampling design formula: where n=sample size; N=total population (4332); Z=95%confidence interval under normal curve (1.96); e=acceptable error term (0.05) and P and q are estimates of the proportion of population to be sampled (P=0.5 and p + q= 1).7% of error term (e=0.07) was used to take representative and cost-effective data for this study.Accordingly, the sample size for the study was determined as follows: Based on this formula, the total sample size was 188 sample household heads.Finally, from a total of 188 sample households, 94 participants and 94 non-participants were selected to get good matching in the propensity score matching estimation.Table 1 shows the households' distribution and sample size.

Methods of data analysis
Both descriptive statistics and econometric analysis were used to analyze the empirical data of the study.These tools are outlined and discussed in the following.

Descriptive statistics
Descriptive statistics such as mean, standard deviation, percentage and frequency of distribution were used to describe the socioeconomic and demographic characteristics of the participant and non-participant groups.Chi-square and t-test were employed to test the statistical significance for both dummy and continuous variables, respectively.

Econometric model
The logit model was used to identify and analyse determinants of the rural household participation in microfinance services in the study area.The mathematical formulation of the logit model is as follows: (1) where P= the probability of participation for i th household and it ranges from 0-1, e = represents the base of natural logarithms (i.e., 2.718…), zi=is a function of n-explanatory variables which is also expressed as: where i = 1, 2, 3, ⋯, n, βo=intercept, βi=regression coefficient to be estimated or logit parameter, U=a disturbance term, Xi=participant households' characteristics, β1, β2 +.ΒnXn=slope of the equation in the model, and Zi=clients' participation.
The probability that a household belongs to non-participant is: (2) Therefore, the odds ratio can be written as: (3) Now, it is simply the odds ratio in favour of participating in microfinance services.It is the ratio of the probability that an individual would participate in the microfinance to the probability that he/she would not participate in the microfinance service.The odds ratio implies the ratio of the probability (Pi) that an individual would choose an alternative to the probability (1-Pi) that he/she would not choose it.Finally, taking the natural logarithm of the Equation 4 and the log of odds ratio can be written as follow: (4) where = is a probability of being participated in microfinance and =is a function of n explanatory variables ( ) which are also expressed as: (5 where βo is an intercept, β2, … , βn are slopes of the equation in the model which is log of the odds ratio, which is not only linear in Xi but also linear in the parameters, Xi = Pre-intervention characteristics of the individual in the study area. If the disturbance term ( ) is introduced, the logit model becomes: (6)

Dependent variable
The dependent variable was participation in microfinance services, which is a dummy variable indicating 1 for participant and 0 for nonparticipant households.

Explanatory variables
The explanatory variables are shown in Table 2.

RESULTS AND DISCUSSION
It shows the descriptive statistical analyses on the demographic, socio-economic and institutional characteristics of sample households.The descriptive analysis further extended to discuss the participant and non-participant households concerning different explanatory variables.It also presents regression analysis using logistic regression to identify determinants of rural households' participation in microfinance services in the study area.

Descriptive statistics results
Household's participation in microfinance services is determined by various household attributes.Among these attribute, demographic and socioeconomic characteristics were the major ones.
Hence, these characteristics are presented and discussed in the following.

Cultivated land size
The mean cultivated land size of the sampled households ranged from 0 to 3 ha.In the study area, the average land size owned by the two groups is 1.35 and 1.10 ha, respectively.The overall average land size of the respondents was 1.22 ha.The result of the t-test depicted that the mean difference between the two sample groups about the size of cultivated land holding was statistically significant at 1% significance level.This implies that the average land size of participant households was higher than nonparticipants.

Livestock holding
The average livestock population held by the sample household was 5.81 in TLU.The mean number of livestock owned by participant and nonparticipant households was 6.73 and 4.88 TLU, respectively.The mean difference between the treated and control groups regarding the size of livestock was positive and statistically significant at 1% level of significance.

Dependency ratio
The result of the finding showed that the mean dependency ratio for the sample households was 0.77.The mean dependency ratio for the participant was 0.65 and 0.89 for non-participants.
There was a significant mean dependency ratio difference between participants and nonparticipants at 1% probability level.The significance mean difference of the computed dependency ration between the two groups implies that the non-participant has more dependent family members (member of family aged under 15 years and aged above 65 years) than the participant.

A frequency of extension contact
As shown in Table 3, the mean frequency of extension contact for the participant and nonparticipant groups was 8.04 and 4.35 per year, respectively.The analysis also indicated that the participant households had better access to extension service than non-participant with the mean difference of 1.6 and which was statistically significant at 1% significance level.

Sex of household head
As shown in Table 4, among the overall sampled households, 124 (66%) were male-headed while 64 (34%) were female.The result also shows that from the participant households, 71 (75.5%) were male-headed households and 23 (24.5%) were female.On the other hand, 53 (56.5%) of non-participant households were headed male households, whereas 41 (43.6%) of nonparticipant households were female.The chi-square value (2=7.68;p=0.008) indicated that there was a statistically significant difference in the sex of household head between participant and non-participant groups at 1% of significance level.This implies that, male was more participated in microfinance service than female in the study area.

Education level of household head
From the selected household heads for the study, 61.2% were literate while 38.8% were illiterate.As shown in Table 4, out of the total sample households, 52.1% of the participants and 35.1% of non-participants received a primary level education.Similarly, 17% of participants and 11.7% of non-participants received secondary school education level.Besides, 4.3% of the participants and 2.1% of the non-participants have acquired certificate and above, education level.The chi-square result indicated that, the education level of the household heads was statistically significant at 1% significance level.This shows that the educated households were more participating in microfinance credit than the illiterate.

Determinants of rural households' participation in microfinance services
The results of logistic regression identify determinants of the rural households' participation in microfinance services in the study area.The marginal effect of logistic regression results in Table 5 shows microfinance participation was significantly influenced by six of the twelve explanatory variables used in the propensity score estimation model.These include sex of household head, education level, cultivated land size, livestock holding, dependency ratio and frequency of extension contact.Accordingly, sex of household head had a positive effect on households' participation in microfinance services and it was statistically significant at 1% significance level.The marginal effect of sex was 0.266.The value of marginal effect indicates that the probability of male-headed households' participation in microfinance services increases by 26.6% more than female-headed households, keeping other variables in the model constant.
Moreover, the result presented the education level of the household head had a positive effect on the probability of participation in microfinance service.But, the significant level was different with different levels of education.Here, from education category, illiterate was taken as the base category.Accordingly, there was a statistically significant difference between illiterate households and those households who have attended primary education level at 1% significance level.The marginal effect of a primary education level was 0.254.The result of marginal effect implies that the probability of those who have acquired a primary level households' participation in microfinance services increases by 25.4% than illiterate household.
Similarly, the finding result indicates that there was a statistically significant difference between illiterate household and those households who have attended secondary education level regarding participation in microfinance services at 5% probability level.In contrast, for occupation, the farmer was taken as a base category, but none of the category was significant.The marginal effect of sex was 0.266.The estimated marginal effect result shows that the probability of male-headed households' participation in microfinance services increases by 26.6% more than female-headed households, keeping other variables in the model constant.
On the other hand, cultivated land size had a positive effect on the rural households' participation in microfinance services and statistically significant at 5% significance level.The finding of the study coincides with Asfaw (2013), who found that landing holding size has a positive and significant effect on households' microfinance participation decision.The marginal effect result shows that a unit increase in livestock holding size increases households' participation in microfinance by 11.1%, keeping other variables in the model constant.
The result of logistic regression showed that the size of livestock positively influenced the probability of participating in the microfinance services at 1% significance level.This result is consistent with the findings of Amine (2016) who found that livestock ownership positively affected the probability of participating in the microfinance services.Moreover, the marginal effect result shows that a unit increase in livestock holding size increases households' participation in microfinance by 11.1%, keeping other variables in the model constant.
Dependency ratio negatively influenced the rural households' participation in microfinance and it was statistically significant at the 5% significance level.This result is similar to that of Feleke (2011), who found that the dependency ratio had a negative and significant influence on the rural households' participation decision in microfinance services.The marginal effect indicated that a unit increase in the dependency ratio decreases the probability of households' participation in microfinance services by 17.7%, keeping other variables constant at their means.
The result of logistic regression indicated that the frequency of extension contact had a positive effect on rural households' participation in microfinance services, and was significant at the 5% significance level.This means that those households getting more extension service have a high probability to participate in microfinance services.The marginal effect of the frequency of extension contact was 0.025.The computed marginal effect result shows that a unit increases in the frequency of extension contact increases the probability of households in microfinance services by 2.5% keeping other variables constant at their means.

Conclusion
Based on the main finding of the study, the following summary and conclusions is drawn.This study has focused on assessing determinants of the rural households' participation in microfinance services at Cheliya District, West Shoa Zone of Oromia National Region State, Ethiopia.In this study, twelve explanatory variables were hypothesized to explain the determinants of the rural households' participation in microfinance services in the study area.These variables were demographic, socio-economic and institutional determinants to explain participation variable.
The descriptive analysis result showed that the mean difference between the two groups regarding the sex of household head, education level, cultivated land size, dependency ratio, livestock holding and frequency of extension contact were statistically significant.However, the two groups have shown a statistically insignificant mean difference regarding age of household head, family size, occupation, distance from microfinance institutions, a household perception of group lending and access to a social network.The estimation result of the marginal effect of the logit model result indicated that among 12 explanatory variables, which were hypothesized, to influence the household heads participation in microfinance services, six variables were statistically significant while the remaining six variables were statistically.The significant variables in the model were sex of household head, education level, livestock holding, cultivated land size and frequency of extension contact are positively and significantly influenced households' participation in microfinance services while dependency ratio is negatively influenced the households' participation in microfinance services in the study area.

RECOMMENDATIONS
Based on the findings of this study, the following recommendations are forwarded.The logistic regression model results indicated that dependency ratio had a negative influence on the probability of households' participation in microfinance service.Therefore, microfinance institutions should encourage rural households those who have a high dependent family member to enhance their involvement in the microfinance service.
As observed from the study, education level had a positive influence on the households' participation in microfinance services.Hence, the microfinance institutions should create awareness of its financial services for those illiterate households to enhance their participation.
In the study area, female participation in microfinance services was less than male.Therefore, the microfinance institutions should give attention to encourage female participation in microfinance services.The study showed that households those have large cultivated land size more participate in microfinance services.Therefore, the microfinance institutions should encourage the households those who have a small cultivated land size to enhance their participation.As the study result indicated households who have small number of livestock less participate in microfinance services.Thus, the microfinance institutions need to introduce its service to the households those who have a small number of a livestock.
According to the findings of the research, frequency of extension contact had a positive effect on households' participation in microfinance services.Therefore, development agents should strengthen their support by providing training and technical support for rural households in order to improve households' participation in microfinance services in the area.

Table 1 .
Distribution of sample households in kebeles.

Table 2 .
Summary of the hypothesis of explanatory variables included in the model Sources: Own Construct (2017).

Table 3 .
Summary of descriptive statistics for continuous variables.

Table 4 .
Descriptive analysis of both dummy and categorical variables.

Table 5 .
Marginal effect estimation of logistic regression for determinants of the rural households' participation in micro finance services.