Getahun Abreham Asefa and Ponguru Chandra S Reddy

Transcription

1 IAARD Journals eissn: International Journal of Economics And Business Management IAARD-International Journal of Economics and Business Management, 2018, 4(1),1-9 Impact of Rural Credit Access on Farmers Income Growth: in the case of Wogera Wereda, North Gondar, Ethiopia Getahun Abreham Asefa 1, Dr.Ponguru Chandra S Reddy 2 Department of Agricultural Economics, College of Agriculture and Rural Transmission University of Gondar. Abstract: The provision of agricultural credit is one of the principal components of rural development, which helps to attain rapid and sustainable growth of agriculture. Rural credit is a temporary substitution for personal savings, which catalysts the process of agricultural production and productivity. To boost agricultural production and productivity farmers have to use improved agricultural technology. However, the adoption of modern technology is relatively expensive and farmers cannot increase their income. Therefore, the study tried to see the impact of rural credit use on farmers income growth, determinants of rural credit use and sources of rural credit in the study area. To achieve this objective, data were composed from 370 households, 251 households from rural credit access users and the rest from non-users, in the study area. PSM (for impact of rural credit use on farmers farm income growth), binary logit models (to identify factors affecting rural credit use) and descriptive statistics (to assess the main sources of rural credit) are used for analyzing the required data. The result showed that the main sources of rural credit are cooperatives and Amara credit and saving institute. Age of the household head and distance of lending institutions from farmers home were found to have a negative and significant influence on farmers rural credit use status. Results of PSM as given by ATT showed that, rural credit access use had no significant impact on farmers gross farm income growth in the study area. Therefore, in order to rural credit has positive impact, lending institutions should open their branches in different rural areas. There should be having road and communication services. Lenders should be interested to give training concerning on loan management for their customers. It is better for rural credit users if the repayment period of loan is relatively not being short. Key words: Impact, PSM, ATT, and Income growth INTRODUCTION Rural credit is defined as any type of lending program or line of credit that is aimed at impacting a rural population in some manner [1]. The accessibility to credit is all aimed at improving the household incomes of the poor people [2]. According tone study, access to agricultural credit enables farmers, who constitute the majority of population in most developing countries to adopt new technologies, and take advantage of new economic opportunities to increase production and income [3]. Agriculture is the major sector in many African countries and the majorities are subsistence farmers who unable to generate enough farm output led by increasing farm income. Access to credit is regarded as one of the key elements in raising agricultural productivity to increase the farm income. Continued population growth, short fall in agricultural production, and wide spread of rural poverty through decreasing the income level of households will force policy makers to continue to promote agricultural development [4]. They also stated that agricultural credit will continue to be a major part of these efforts. Where the environment is conducive to increasing crop productivity through input use, example the high rainfall high lands of Ethiopia, credit appears to contribute significantly to increased income and expenditure on food on the rural farmers. To increase agricultural output and productivity farmers have to use improved agricultural technologies. However, the adoption of modern technology is relatively expensive and farmers cannot increase their income. As a result, the utilization of agricultural technologies is very low and their income has a very slow growth. It is argued that enhanced provision of rural credit would accelerate agricultural production and productivity in addition to the growth of farm income [5]. Since farmers saving are inadequate and have low level of cash to finance various agricultural activities on their farm, they go for credit. Access to rural credit is assumed to improve the agricultural output and farmers income growth. Failure of financial sectors in rural area to provide sufficient agricultural credit to farmers is often viewed as one of the main factors that retard the growth of farmers income on the study area. An impact /evaluation study has found that access to credit (including rural credit) by some poor farmers have large positive effect on their farm income growth [6]. However, other studies have found that income is not increased through microcredit access; poor households simply become poorer through the additional burden of debt. It is extremely important to carefully evaluate whether or not rural credit works in the study area. It is not also well known to what extent the household using access to rural credit are better off than those who do not use in the study are. Page No.1

2 Although a number of studies have been taken on the related to impact of rural credit in the study area, nearly all of them were aimed on poverty reduction. To the best of the researcher s knowledge no study has been made to observe an impact of rural credit access on farmers income growth in the study area. Therefore, does access to rural credit help farmers to increase their income; what factors are found that affect rural credit use and what are the sources of rural credit in the study area do have; are some of the questions which have been addressed in this study. Methods and Materials Data type, sources and method of data collection: During the study both qualitative and quantitative data are used and also both primary and secondary sources of data are used. The data was collected efficiently through a structured questionnaire to collect the required data from selected sample households based on 2013 production year. The questionnaire was first pre-tested and modified before the execution of the survey. Training on methods of data collection and the contents of the questionnaire were conducted. Four enumerators who can understand English and local language were recruited from the study area and have got training for 3 days. The enumerators who are stationed in the survey areas would be administered the structured questionnaires under the continuous supervision of the researcher. The survey was carried out in February Sampling Technique In this study, a multi stage random sampling procedures were used for the selection of sample respondents. In the first stage, from 41 kebeles, 25 kebeles which have access to rural credit are identified in the study area. In this kebeles, there are rural credit access, agricultural extensions, lacks commercial farming activities and relatively similar way of life or standard of living. And also they are not saftnet program users. Therefore, these 25 kebeles are homogenous in terms of the above criteria. Then, three kebeles (Doro Wuha, Gunda Chogi and Yesihak Debir) are randomly selected. In the second stage, the total households who are registered to get credit in 2011 in the three kebeles are stratified in to two strata: rural credit user and non-rural credit user households. The list of total households who are registered to get credit in 2011 in each selected kebeles (in the three selected kebeles) and the list of rural credit user households in these kebeles are taken from rural credit offices of Wereda and rural credit offices of sub District. The households who have registered to take rural credit have almost the same status in terms of their living standard that means they have no big differences in terms of their wealth in general and income level in particular. That is the main reason of the study to select those households to analyze the impact of rural credit access on farmers farm income growth. Therefore, based on the above procedures, representative samples are selected in three kebeles based on probability proportional to sample size. Finally, 370 sample households (251 rural credit and 119 non rural credit users) are selected from the above three kebeles based on simple random sampling technique. For the purpose, four enumerators (three persons who have completed two years college training and working in the rural area as development agent and one person who is completed high school) would be recruited and were trained before the pre-test. This formula is used to calculate the sample sizes of the study [7]. Where, n is the sample size, N total population and e is the level of precision at 95 degree of confidence interval. When this formula is applied to this study the following result is obtained i.e. Table 1: Distribution of sample households by kebeles Kebele No. households Rural credit user sample Non rural credit user sample Total Who have registered Yesihakdebir Gunda Chugi Dorowuha Total Sources: own summary, 2014 Method of data analysis Descriptive statistics: This method of data analysis used to explain the situation of demographic and socioeconomic variables. Descriptive statics mainly used to identify the potential sources of rural credit in the study area. The specific method of data analysis were involved are tabulation, frequency, percentage and computation of descriptive statistics such as mean and standard deviation, t-test and chi square would be used at 1%, 5%, and 10% significance level. The other objective of the study (the impact of rural credit access on farmers income growth) was analyzed by using propensity score matching method while the rest one (identify factors that affect rural credit use) was analyzed through logistic regression model. Page No.2

3 Main sources of rural credit Out of the total borrowers, 62% and 33% of the sample household heads got credit services from cooperative (primary cooperative) and Amhara Credit and Saving Institute (ACSI) respectively. Status of rural credit sources in terms of training and amount of loan able to lend by farmers point of view The status of rural credit sources in terms of size of loan which could give for their customers was not mostly attractively good because farmers could not get the amount of loan what they asked. The lenders mostly gave less than 6000 Birr for their rural farmer customers. This implies that lenders in the study area are weak in terms of loan size. The lenders in the study area were weak in terms of giving training support for their customers. Some respondents view indicated that, even though when lenders gave training, mostly their concentration was attached to political activity rather than loan management and other related activities. Generally, based on respondents point of view, lenders status in terms of giving training for customers in the rural area is poor. Analysis of determinants influencing rural credit use As discussed before, logistic regression was applied to identify the main determinants of rural credit use status. Before running the logistic analysis, the variables which were included in the model were checked for the existence of multicollinarity, heteroscedasticity and omitted variable problems. First, we have checked the multicollinarity problem associated with the explanatory variables. There are two methods to check the multicollinarity problems. For continuous variables, variance inflation (VIF) is used to detect the problems of multicollinarity. The VIF value less than five are believed to have no problems related to multicollinarity.all continuous explanatory variables did not have multicollinarity problems because the value of VIF for each were below five (table 4.8). The second method of detecting multicollinarity problem is through contingency coefficients for high degree of association for discrete variables. The result was shown in table 10. Correlation coefficients with an absolute value higher than are taken as an indicator of multicollinarity. As the table 4.9 showed, there was no serious multicollinarity problem in discrete variables because the contingency coefficients were below Therefore, all of the independent variables were included in the model. In the second, the problem of heteroscedasticity was checked by using the Bruesh Pagan test. The test indicated the existence of heteroscedasticity. To avoid this problem from the model robust standard errors were estimated. In third, the omitted variable problem was tested by using the Ramsey Regression Equation Specification Test (Ramsey RESET). Based on the test, there was no omitted variable in the model. Generally, thirteen explanatory variables were used for logistic regression. Out of thirteen explanatory variables, five variables were continuous and the rest eight variables were discrete. Table 2: Estimation of the coefficients of the logit model variables coefficients Robust S.E Odds ratio Significance Marginal effect constant attrisk gender marstats davistfre.0372** landsiz.1112** fertlandl farexper healthsth.5085* grolendg *** memcomp agehh ** leveleduc.6474*** dileinstn ** Sample size.370 LRχ2 (13) = Prob> chi2 = Log likelihood = Pseudo R2 = *, ** and*** refer to significance at10%, 5% and less than 1% probability levels respectively Source: Model output, 2014 out of the total thirteen explanatory variables hypothesized to influence the agricultural credit use status, seven explanatory variables were significant below 10% probability level. The maximum likelihood estimates of the model showed that development agent visit frequency (DAVISTFRE), size of land holding (LANDSIZ), health status of the house hold head (HEALTHSTH), farmers perception on group lending (GROLENDG), age of the household head (AGEHH), level of education of the household head (LEVELEDUC) Page No.3

4 and distance of lending institutions from the farmers home (DILEINSTN) were the important determinants which influence the rural credit access use status of the rural farmers were presented as follows. Development agent visit frequency (DAVISTFRE): This variable was significant below five %probability level and positively related with the rural credit use status. This result is consistent with [8,9]. The marginal effect of development agent visit frequency shows that the probability of being rural credit user will increases by 0.6% approximately with one unit increase in contacting with development agent. significant relationship with the rural credit use status at significant less than 5% probability level. This implies that age of the household head increases, the probability to use rural credit access will be decreased. The older farmers may accumulate more wealth than the youth ones. In addition to that as age increases farmers able to accumulate their own wealth and decreases to use rural credit access. This result is consistent with [2]. The marginal effect of age of the household head i.e indicates that the probability of being rural credit user decreases by approximately 0.62% with a one year increase in farmers age. This indicates that a farmer whose conduct higher is relatively tends to be rural credit user compared to others. The implication is that development agents play an important role in the introduction, dissemination and adoption of new technologies in agricultural practices. For this purpose, farmers who conduct with development agents have good awareness about rural credit access to use for his/her new agricultural technology adoption. Size of land holding (LANDSIZ): Based on the above model results, coefficient of size of land holding is significant at less than 5% probability level. The marginal effect of the variable i.e shows that the probability of being rural credit user increases by approximately 1.8% with an increase of land size by one hectare. Farmers households who have large size of land are more likely to use rural credit. The explanation for this result is that those farmers who have large size of farm land indicate that a farmer can use this land for different agricultural production purpose. For this reason i.e. to use their land effectively they prefer agricultural credit rather than reducing size of land even when they face shortage of farm inputs to use farm land. This result is consistent with [10]. Health status of the house hold head (HEALTHSTH): This model estimation result shows that the variable health status of the household head has a positive and significant effect at 10% probability level. The above result shows that healthier household head have higher probability to use rural credit access than others. The marginal effect of the variable indicates that the probability of being rural credit user increases by approximately 8% with one unit increase in health status of the household head. Farmers perception on group lending (GROLENDG): This variable is another important variable that is found to have a positive and significant relation with the dependent variable at a probability level of below one percent. It is expected that group formation has a good relation for credit access use. This is because group formation has a good relation for credit access use. This is because group formation enables farmers who have not enough collateral asset especially poor farmers to get rural credit. The marginal effect the variable indicates that the probability of being rural credit user increases by approximately 20% with a unit additional positive farmers perception on group lending. Age of the household head (AGEHH): As expected, the variable age of the household head had a negative and Level of education of the household head (LEVELEDUC): This variable is statistically significant at less than one percent probability level with expected sign. The result predicts that educated farmers are more likely to use rural credit than those who are not educated. Farmers who are relatively well educated can attend in meeting and can get information and knowledge for their agricultural production increment and they can be informed about credit access use functions. This result is consistent with [13]. The marginal effect of the variable level of education of the household head indicates that the probability of being rural credit access user increases by an approximately 10.2% with a one unit change of farmers level of education. Distance of lending institutions from the farmers home (DILEINSTN): The sign of the coefficient of this variable showed a negative relation with rural credit use status and is significant at less than five percent probability level. This indicates that an increase in the distance of lending institutions from farmers home decreases the likelihood for the household to become rural credit user. As a farmer is nearer to the lending institutions, there would be a higher chance to use credit because of access to information about credit. However, if the lending institution is very far the result is the reverse one. The marginal effect of the variable indicates that the probability of being rural credit user about 2.6% with one kilo meter increase in distance of lending institutions. This result is consistent with [5]. Propensity score matching model results Estimation of propensity scores Before conducting the estimation of propensity score, the model of logistic regression was conducted. The dependent variable of the model is gross farm income of the household. Before going to estimation of impact, the variables which are included in the model were checked for the existence of multicollinarity and heteroscedasticity problems. For this purpose, VIF was used to test the presence of strong multicollinarity problem among continuous independent variables and contingency coefficients were applied for discrete explanatory variables. The test result showed that there was no any independent variable dropped because of serious multicollinarity problems. In the same way, the problem of heteroscedasticity was tested by applying the Bruesh Pagan test. The test result showed that the existence of heteroscedasticity which suggest that the need of robust standard error. For this reason, robust standard error was conducted accordingly. Page No.4

5 Table 3: the maximum likelihood estimates of the logit regression used in estimating the propensity scores variables Coefficients Robust S.E Z -value constant landsiz * gender davistfre ** hhsize *** expendfer e ** dismarket memcomp healthsth * farexper agehh ** distextin leveleduc *** N = 370 Pseudo R2 = LR chi2 (9) = 93.86Prob> chi2 = Log likelihood = Sources: own estimation result, 2014 ***, ** and * refers significant at the 1%, 5%, and 10% probability level, respectively. Table 4.11 shows the maximum likelihood estimations of the logit regression used in the propensity score. As shown in the above table (Table 4.11), the pseudo-r 2 value of the model is which is comparatively low. Low pseudo-r 2 value indicates that the distribution of the program has been fairly asymmetric or random. Thus, as the above result shows that treatment respondents do not have diverse characteristics overall and therefore getting good match between treatment and non-treatment respondents becomes simple. According to the results of the estimated logistic regression (Table 4.11), DA visit frequency (davistfre), household size (hhsize), size of land holding (landsiz), expenditure on fertilizer (expendfer), health status of the household head (healthsth), age of the household head (agehh) and level of education (leveleduc) are the factors that affect income of the respondents. Matching users and non-user households Propensity scores of the study were estimated for all the 370 respondents including 251 treated and 119 controlled observations. Among rural credit access users, the predicted propensity score lies between and with its mean value of And, for non-rural credit users, its predicted propensity score ranges from and with a mean of (Table 4). The common support region can be between and For this reason, 23 observations (23 from users and 0 from non-users) were discarded. Table 4: Distribution of estimated propensity scores Groups observation Mean St.Dev min max Treated Control Total Source: own estimation result (2014) The density distribution of the propensity scores for users and non-users were shown in the below figure (Figure 1). The bottom half of the graph show the distribution of propensity scores for the untreated observations while the upper half of the graph shows for treated ones. The vertical axis refers the frequency of the frequency distribution. The below figure shows that all observations, that use rural credit access found a good match among those who did not use rural credit access. Choice of matching algorithm Among the matching algorithms, nearest neighbor and caliper estimators of matching were applied in matching comparison units with treated units fall in the common support region. For selecting the good matching estimator, some criteria such as equal means test (balance test), pseudo-r 2 and matched size (number of matched sample size) were used to select the best one [1]. A matched estimator which produces a large number of Page No.5

6 matched sizes with low pseudo-r 2 and that can balance all explanatory variables is preferable. Based on the results, it has been found that nearest neighbor n (4) is the choice estimator for the data what we have. Therefore, the estimation results and discussion are the direct outcomes of the nearest neighbor matching algorithm with a neighbor of four. Figure 1: Histogram of propensity score Table 5: shows the estimated results of tests of matching quality based on the above three criteria. Matching estimator Performance criteria Balancing test* Pseudo-R 2 Matched sample size NearestNeighbor matching 1 neighbor neighbor neighbor neighbor neighbor Caliper Sources: own estimation, 2014 *Number of explanatory variables with no statistically significant mean differences between the matched groups of treated and control households. Testing the balance of propensity score and covariates Following choosing the good performing algorithm, the following work is checking the balance of propensity score and covariate by using some procedures through applying nearest neighbor matching algorithm. Estimation of propensity scores is for the purpose of balancing the distribution of appropriate variables in both groups rather than obtaining accurate prediction of choosing in to treatment. In this study different measures have taken to test the matching quality in terms of balancing powers of estimation. Mean comparisons between treatment and non-treatment (control) groups (unmatched and matched), standardized bias and overall measures of covariate imbalance were implemented procedures. As shown in the above table (table 6), the mean comparisons between treatment and control households before matching indicated us the presence of significance differences between the two groups in terms of DA visit frequency (davistfre), household size (hhsize), expenditure on fertilizer (expendfer), health status of the household head (healthsth), age of the household head (agehh) and level of education (leveleduc). However, the t-test result after matching shows that there is no any covariate which shows the significant differences between the two groups. These matching procedures suggest that there is a balancing of the covariates in the two groups. The table also shows that the standardized difference of unmatched covariates lies between 91.4 and 1.0 in absolute value and for matched covariates it lies between 14.0 and 1.9. Hence, the procedure of matching gives high degree of covariate balance between the control and treated samples that are ready to apply in the procedure of estimation. The low pseudo-r 2 and the insignificant likelihood ratio tests support the hypothesis that both groups have the same distribution in covariates after matching (Table 7). These result points out that the procedure of matching is capable to balance the characteristics in the treated and the matching comparison groups. For that reason, we can use these results to assess the impact of rural credit access use on farmers income growth among group of households having similar observed characteristics. This can be used us to compare observed outcomes for treatment with those of control groups sharing a common support. In general, as we have seen the above tests, the matching algorithm what we have used is relatively good for this study. Therefore, we can able to estimate ATT for sample observations. Page No.6

7 variable sample mean treated control Table 6: Propensity scores and covariate balance %bias % reduction bias T-test T davistfre unmatched *** matched landsiz unmatched matched expendfer unmatched ** matched healthsth unmatched * matched dismarket unmatched matched leveleduc unmatched *** matched distextin unmatched matched gender unmatched matched agehh unmatched ** matched memcomp unmatched matched farexper unmatched matched hhsize unmatched *** matched Source: own estimation result (2014) ***, ** and * refers significant at the 1%, 5%, and 10% probability level, respectively. P> t Table 7: Chi square test for the joint significance of variables outcome Sample Pseudo R2 LR chi2 p>chi2 Gross farm income growth Raw Matched Source: own estimation result (2014) Treatment effect on treated (ATT) In this sub section, the impact of rural credit access use on the outcome variables (change in total growth farm income, change in income from crop and change in income from livestock) was analyzed for its significant impact on rural credit user households by using propensity score matching model, after the pre intervention differences were contacted. Estimation of average treatment effect (ATT) of growth farm income As in table 8 shown, the estimation of PSM results show that, being rural credit access user had not a significant impact on household growth farm income in terms of total change in growth farm income, change in income from crop and change in income from livestock in the study area. This insignificant result on farmers farm income growth obtained in this study area might be because farmers in this study area use the loan for smooth consumption and loan repayment purposes. That means as some respondents pointed out that, farmers who had large loan from informal lenders were enforced to take credit from formal lenders and this loan was used automatically for loan repayment and smooth consumption purpose rather than being an input for their agricultural activities. In addition to that, period of loan repayment might be one reason. Almost all of rural farmers have got a short term credit access with less than and one year loan repayment period in the study area. This type of loan might affect farmers profitability in their farm production activities negatively because farmers might be pushed to repay their loan without recovery of their income level. Since there was a loan, farmers tried to repay by any means their loan without their profit and consequently they couldn t increase their farm income. The sensitivity analysis on the estimated average treatment effects were not conducted why because there is no any outcome indicator with significant impact of rural credit access use on farmers gross farm income Page No.7

8 growth in our case. For this reason, there is no requiring of conducting sensitivity analysis. Table 8: Average treatment effect on the treated (ATT) Variable Outcome variables Treated Control Difference S.E. T-stat Gross farm income growth Total change far income Change in come livstock Change in crop Change refers income of Source: own estimation result (2014) CONCLUSION AND RECOMMENDATION Conclusion This study analyzed the impact of rural credit access use on farmers gross farm income growth in Wegera Wereda, North Gondar, Ethiopia. The study also identified factors that affect rural credit access use status and assessed the main sources of formal rural credit in the study area. Did access to rural credit help farmers to increase their income? What factors are found that affect rural credit use? And, what are the main sources of rural credit in the study area do have? These were some questions which have been assessed in this study. For that purpose, the primary data were collected from 119 nonrural credit access users and 251 from rural credit access users using structured questionnaire. Propensity score matching (PSM) was employed for analyzing the impact of rural credit access use on farmers income growth while for identifying the main determinants of rural credit use status and for assessing the sources of rural credit and its status in the study area, the binary logistic regression analyses and descriptive statistics with chi2 test were employed respectively. In the binary logistic regression function, the dependent variable, rural credit use status by households was a dichotomous categorical variable talking the value of 1 if the household is being rural credit access user and 0 otherwise. As we have seen in the results of this study, the main sources of rural credit are agricultural office, Amhara credit and saving institute (ACSI) and cooperatives. Out of the total number of rural credit access users, above half of them reported as lenders did not give training support for their customers for loan management and other related activities. Not only this, but also the status of rural credit sources in terms of the size of loan which could give for their customers was not mostly attractive because farmers could not get the amount of loan what they asked to borrow. In the econometrics analysis, the findings indicated that age of the household head, farmers attitudes towards risk for loan repayment and distance of lending institutions from farmers home were found to have a negative and significant influence on farmers rural credit use status. While education level of the household head, farmers attitudes towards group lending, household land size, development agent visit frequency and health status of the household head were found to have a positive influence on farmers rural credit use status in the study area. Results propensity score matching (PSM), as given by average treatment effect on the treated (ATT) showed that rural credit access use had no significant impact on farmers income growth. Recommendation Based on the findings of the study, the following points are recommended to have rural credit use an impact on farmers gross farm income growth in the study area. Distance of lending institutions from farmers home was the main determinants of rural credit use status.when the distance is very long, farmers cost in terms of time will be higher. As a result, farmers are unable to participate in different meetings, training about credit use and management of loan. Therefore, the lending institutions should open their branches in different rural areas. There should be having road and communication services to avoid such problems. Age of the household head and rural credit use status were negatively and significantly related. Therefore, proper attention has to be given to initiate farmers to use rural credit. It is better for rural credit users if the repayment period of loan is relatively not being short; it should be long term. Farmers perception of group lending is found to have positive and significant relation with rural credit use status. This implied that group formation enables farmers who have no enough collateral assets to get credit. Therefore, strengthening group formation of farmers should be requiring high attention. Development agent visit frequency had a positive and significant effect on farmers rural credit use status. Development agents play an important role in the introduction, distribution and implementation of agricultural improved technologies in agricultural activities. Farmers who conduct with development agents have good awareness about credit use status. So, strengthen distribution and contacts of development agent should be promoted to increase the number of rural credit users. Size of land holding and credit use status were positively and significantly related. Accordingly, increasing size of land per household based on efficient land distribution and sustainable land management should be given priority to strength the farmers activities for using rural credit access for increasing their sustainable farm income growth in the study area. Health status of the household head had a positive and a significant effect on rural credit use. Healthier household head had higher probability to use rural credit access than others. Therefore, extension agents and policy makers should provide more Page No.8

9 concentration to have health care centers in rural area and to teaching farmers to keep their environment and themselves to be free from causes of disease as much as possible. REFFERENCES 1. Amare Berhanu (2005). Determinants of formal source of credit loan repayment performance of smallholder farmers: in the Case of North Western Ethiopia, North Gondar, M.Sc. Thesis, Alemaya University, Ethiopia. 2. DuyVuong Quoc (2012). Determinants of household access to formal credit in the rural areas of the Mekong Delta, Vietnam, Research report, Cantho University, Vietnam. 3. Fred (2009). Accessing micro credit, borrowers characteristic and household income in rural areas, M.Sc. Thesis, Makerere University, Pakistan 4. Giang Ho (2004). Rural Credit Markets in Vietnam: Theory and Practice. Research report, Macalester College, Vietnam. 5. K.I. Etonihu, S.A.Rahman and S, Usman, Journal of Development and agricultural Economics, , 5 (5), Lolita Poliquit (2006). Accessibility of rural credit among small farmers in Philips, Thesis, Massey University, New Zealand. 7. Mohamed Ahmed, PoulPreckel and Simon Ehui (2006). Modeling the impact of credit on intensification in mixed crop-livestock systems in Ethiopia: A case study from Ethiopia, Poster paper prepared for presentation at the International Association of Agricultural Economists Conference, Gold Coast, Australia. 8. M.Shah, H.Khan, Jehonzeb andz.khan (2008). Impact of agricultural credit on farm productivity and income of farmers in mountainous agriculture in Northern Pakistan, research report, University of Peshawar, Peshawar, Pakistan. 9. Ololade R.A. &Olagunju F., Determinants of Access to Credit among Rural Farmers in Oyo State, Nigeria, Global Journal of Science Frontier Research, 2-3, 13(2), SisayYehuala (2008).Determinants of smallholder farmers access to formal credit: the case of MetemaWereda, North Gondar, Ethiopia, M.Sc. Thesis, Alemaya University of Agriculture, Ethiopia. 11. Tadele Mamo (2011). Impact of productive safety net program on asset accumulation and sustainable land management practices in the central rift valley: the case of Combolcha and Meskan in Ethiopia, M.Sc. Thesis, Haramaya University, Ethiopia. 12. Yamane, Taro(1967). Statistics: An introductory Analysis, 2 nd Edition, Newyork: Harper and Row Page No.9