Technical efficiency of smallholder producers in barley production in Northern Ethiopia

Abstract Barley is among the major staple cereal crops in Ethiopia. Despite this, the smallholder barley producers experienced a lower level of technical efficiency due to different factors. Thus, this study is designed to estimate the level of technical efficiency and identify factors affecting the level of technical efficiency of smallholder barley producers. The data are collected from 200 randomly selected barley producers through a semi-structured questionnaire. Descriptive statistics and econometrics model are used to analyze data. The stochastic production frontier (Cobb–Douglas production function) and Tobit model were used to estimate the level of technical efficiency and identify factors affecting inefficiency, respectively. The stochastic production frontier model result revealed that there is a space to increase barley output by increasing the use of farm inputs. The mean of barley output for sample households in the 2017/18 production season was 19.5 quintals. The average level of technical efficiency of sample households was 75.1%, which implies that output can be increased by 24.9%. The frontier model indicated that labor, oxen power, fertilizer, land and seed had positive and significant effects on barley production. The Tobit model result shows that technical efficiency was affected by sex, education, livestock number, non-farm activities, credit use and non-input total expenditure. This shows that there is a room to increase technical efficiency of barley production in the study area. Therefore, appropriate policies and strategies directed towards the above-mentioned significant variable should be designed to improve the productivity of barley in the study area.


PUBLIC INTEREST STATEMENT
In Ethiopia, barley is among the major cereal crops, which takes the highest shares from agricultural products in terms of production, area coverage and number of producers.It plays an important role in improving the livelihood of the residents in the country.Besides, it is used as a raw material for industries like beer factory.Although it plays a great role, the barley producers are not efficient, provided that they have lower technical efficiency in barley production due to different reasons.Thus, this research article was designed to estimate the level of technical efficiency and identify the factors affecting the level of efficiency.This will help to improve the level of technical efficiency of the barley producers.Additionally, it will be the input for further researches.
labor, oxen power, fertilizer, land and seed had positive and significant effects on barley production.The Tobit model result shows that technical efficiency was affected by sex, education, livestock number, non-farm activities, credit use and non-input total expenditure.This shows that there is a room to increase technical efficiency of barley production in the study area.Therefore, appropriate policies and strategies directed towards the above-mentioned significant variable should be designed to improve the productivity of barley in the study area.

Introduction
Ethiopian economy is highly dependent on agriculture, which contributes to about 36.3% of gross domestic product (GDP), 73% of employment opportunities, 70% of the foreign exchange earnings and 70% of raw materials for the industries in the country (UNDP United nations development programme, 2018).Therefore, agriculture plays an important role in the economic development of the country.
Cereal crop production takes the highest shares from agricultural products in terms of production, produced by many farmers in the country.Cereals in Ethiopia are the main source of food intake (60%), which are predominantly produced by small landholders (Awika, 2011).Barley is among the major cereal crops in Ethiopia that accounts nearly 25% of the total production in Africa (FAO Food and Agriculture Organization, 2014b).Barley is the predominant cereal in the high altitudes (>2000 m.a.s.l.) and it has a global significance because of its improved traits, including disease tolerance (Bonman et al., 2005;Vavilov, 1951).Barley is the fifth most important crop in Ethiopia after teff, wheat, maize and sorghum in the meher season and the second major cereal crop after maize in terms of area coverage and total production in the belg season.
In 2016/2017 production year, about 0.959 million hectares of land was covered by barley, which contributes to about 20.249 million quintals of production with the yield of 21.11 qt/ha in Ethiopia.Similarly, about 0.324 million ha of land was covered by barley, with the production 6.081 million qt and yield of 18.79 qt/ha in the Amhara Region.In South Wollo zone, land coverage, production and yield of barley crop were 0.0386 million ha, 0.655 million quintals and 16 .97qt/ha, respectively (CSA Central Statistical Agency, 2017).
In Legambo district (study area), barley is the main cereal crop and takes a lion's share in terms of production, food consumption, number of producers and area coverage relative to other cereals grown.It covers 0.0383 million ha of land with the production of 0.779 million quintals (CSA Central Statistical Agency, 2017;LDANRDO Legambo District Agricultural and Natural Resource Development Office, 2018).However, the production efficiency was low and varies for different producers, which needs intervention.Similarly, there is inefficient resourse utilization in smallholder cereal crop production including barley (Chanie, 2011).Efficient production is the source for achieving overall food security and poverty reduction objectives particularly in potential areas of the country producing major food crops (Tolesa et al., 2014).However, farmers are discouraged from producing more because of inefficient agricultural systems and differences in efficiency of production (Mesay et al., 2013).
Different efforts have been made on the use of barley production technology to improve its efficiency.For instance, Endalkachew (2012), Bati (2014) and Gebretsadik (2017) conducted a research on the efficiency of barley production in different areas of the country.Besides, Chanie (2011), Tolesa et al. (2014), Musa et al. (2015), Leggesse (2015), Debebe et al. (2015), and Gebretsadik (2017) studied the efficiency of major crops production.These authors found different levels of efficiency and its factors that show the inefficiency level.However, the authors have not addressed all areas of the country including the study area, and there is limited information on the level of efficiency and its factors.Thus, the efficiency of barley production requires a great concern.However, limited researches are conducted on the technical efficiency of barley production and its factors.Moreover, there are information gaps on it in the study area.Therefore, this study was designed to estimate the level of technical efficiency and identify the factors affecting the level of efficiency in the study area.The main research questions answered in this study were the following: • What is the level of technical efficiency of barley production?
• What are the main factors affecting the level of technical efficiency of barley production?

Description of the study area
Legambo district is 550 km away from Addis Ababa, the capital city of Ethiopia.It is located in the Northern part of Ethiopia, Amhara Region, and South Wollo zone (Figure 1).It is located at Latitude 11°00'00.0"Nand Longitude 39°00'00.0"E.It is bordered on the South by Kelala and Wegde, on the West by Mekane Selam and Amhara Sayint, on the North by Tenta and Mekdela and on the East by Were Ilu, Dessie Zuria and Legahida.Towns in Legambo include Akesta and Embacheber.The district has 35 rural kebeles and 3 small urban Kebeles (LDANRDO Legambo District Agricultural and Natural Resource Development Office, 2018).
The district has a total population of 194,959 of which 98,208 are men and 96,750 are women.A total of 41,176 households were counted in this district.The district has a total area coverage of 108,868 hectares.Based on Ethiopian agro-ecological classification, the study area is categorized as four major agro-ecological zones i.e. wurich, dega, woina dega and kola: 40% wrich (high land), 33% dega (high land), 17% woina dega (mid-altitude) and 10% kola (low land).The annual rainfall ranged from 700 to 1200 ml, and the average temperature is 18°C.The altitude of this district reaches 3000 meters above sea level, with nearly 65% of the area located in the highland, and the livelihood of the community is largely dependent on subsistence agriculture of crop production and livestock, which is highly dependable on rainwater (Figure 1) (LDANRDO Legambo District Agricultural and Natural Resource Development Office, 2018).

Sampling technique and sample size determination
Two-stage random sampling technique was employed to draw the appropriate sample households.In the first stage, four kebeles out of 11 barley-producing kebeles were selected randomly.In the second stage, 200 farmer households were selected randomly from those who were producing barley, taking into account probability proportional to the size of barley producers in each of the four selected kebeles.The sample size was determined based on the following formula given by (Yamane, 1967): where n = sample size, N = total number of households and e = the desired level of accuracy.The total population of barley producer households in the study district with 11 kebeles is 9,350, assuming a 7% level of precision from a total of 200 sample households.

Types, sources and methods of data collection
Both qualitative (sex, crop rotation and non-farm activities) and quantitative (amount of input, amount of output, family size, farm size, education, credit use, livestock ownership, non-input expenditure and distance to the nearest market) types of data from primary and secondary sources of data are used.Primary data are collected by semi-structured questionnaires from sample farmers.Secondary data are collected from related journals, report documents, district agricultural documents, central statistics agency and published (on the internet, graphics in district) and unpublished documents (from the office).

Methods of data analysis
Both descriptive statistics and econometric models (stochastic frontier and two-limit tobit) are used.Descriptive statistics is measured mean, frequency, percentage and standard deviation.A stochastic frontier approach was used to measure the level of technical efficiency.In addition, two-limit tobit model was applied to analyze the factors affecting the level of technical efficiency of smallholder barley producer.

Descriptive analaysis
Descriptive statistics technique is used to describe the demographic, farm-related, socioeconomic and institutional characteristics of smallholder barley producer in the study area.

Econometric model analysis
Most empirical studies on efficiency in Ethiopia were analyzed using stochastic production frontier methodology (Debebe et al., 2015;Mekonnen, 2013;Solomon, 2014).The main reason is that stochastic approach allows for statistical noise such as measurement error and climate change, which are beyond the control of the decision-making unit.Following Aigner et al. (1977), the model is specified as follows: where Y i is the production of the i th farmer, Xi is a vector of inputs used by the i th farmer, β, is a vector of unknown parameters, V i is a random variable assumed to be N (0,δ 2 v ) and independent of the U i , which is a non-negative random variable assumed to account for inefficiency in production.
The appropriate functional form that better fit data is selected after testing null hypotheses using the generalized likelihood ratio test � equals to 21.86 (Table 4).The statistic is distributed with degrees of freedom equal to the number of variables added to the alternative hypothesis; in this case, the degrees of freedom were 15.The calculated value (21.86) is lower than the upper 5% critical value of χ 2 with its respective 15 degrees of freedom (24.99).This implies that Cobb-Douglas production function was adequate in the data set and the more appropriate model for this study.The linear form of Cobb-Douglas production model is defined as: Ln denotes the natural logarithm, j represents the number of inputs that will be used, i represents the i th farmer in the sample, Y i represents the observe barley production of the i th farmer, X ij denotes that j th farmer input variables will be used in barley production of the i th farmer, ß stands for the vector of unknown parameters to be estimated, ε i is a composed disturbance term made up of two elements (v i and u i ), v i accounts for the stochastic effects beyond the farmer's control, measurement errors, as well as other statistical noises and u i captures the inefficiency.Aigner et al. (1977) proposed the log-likelihood function for the model in Equation ( 3), assuming a half-normal distribution for technical inefficiency effects (μ i ).The expressed likelihood function uses λ parameter, where λ is the ratio of the standard errors of the non-symmetric to symmetric error (λ ¼ δ μ =δ ν ).However, there is an association between γ and λ.The reason is that λ could be any non-negative value while γ ranges from zero to one and better measures the distance between the frontier output and the observed level of output resulting from technical inefficiency.According to Bravo-Ureta and Pinheiro (1997), gamma γ ð Þ can be formulated as: Cobb-Douglas production function was preferred over translog based on the generalized likelihood ratio test, as it was considered to be the appropriate functional form that better fit the data.The value of the generalized log-likelihood ratio (LR) statistic to test the hypothesis that all interaction terms, including the square root specification (in the translog functional form), are equal to zero (H 0 ¼ β ij ¼ 0) was calculated as: where: LR = generalized log-likelihood ratio, L (Cd) = log-likelihood value of Cobb-Douglas and L (Tl) = log-likelihood value of translog.
The farm-specific technical efficiency defined in terms of observed output using the existing technology: 2.4.2.1.Tobit model.In order to determine the relationship between demographic, socioeconomic, farm-related and institutional factors affecting the level of technical efficiency, tobit model was applied.Using two-limit censored tobit model on explanatory variables explained variation of efficiency across smallholder barley producers.The two-limit censored tobit model is applied through two tails censored at the minimum and maximum scores with left and right censored, respectively.Following Upadhyaya et al. (1993), the two-limit tobit regression model was estimated as: where i refers to the i th farm in the sample households; j is the number of factors affecting technical efficiency; U i is efficiency score representing the technical inefficiency of the i th farm; U i * is the dormant (latent) variable, β j are unknown parameters to estimate and μ i is a random error term that is independent and normally distributed with mean zero and common variance . Z ij are socio-economic, institutional, farm-related and demographic variables, which are expected to be factors affecting the level of technical efficiency of smallholder barley producers.(Bati, 2014;Endalkachew, 2012;Gebretsadik, 2017).

Description of production
The production function (using five input variables (labor, oxen power, fertilizer, land and seed) and output variable) is used for the level of efficiency.
The production function for the level of technical efficiency estimated using five input variables (labor, oxen power, fertilizer, land and seed) and output is a dependent variable.To draw image about the distribution of input, mean and range of input variables are discussed as follows: Barley output is the dependent variable in the production function and estimated with five significant inputs which are labor, oxen power, fertilizer, land and seed.The mean of barley output for the sample households in the study area, during the 2017/18 production season, was 19.5 quintals with a minimum of 5 qt and a maximum of 46 qt (Table 1).Labor especially family labor had the main role in barley production activities like ploughing, sowing, weeding, plant protection, harvesting and threshing.On average, 40.5 total labor is required for performing all related activities of farming in man-days with a minimum of 7 man-days and a maximum of 75.1 man-days (Table 1).
The average oxen power of 31.9 oxen-days is used by sample households for barley production, and it ranged between a minimum of 10 oxen-days and a maximum of 66 oxen-days (Table 1).The average inorganic fertilizer (Urea/NPS) application for the production of barley among the sample respondents used 35.4 kg with a minimum of zero and a maximum of 100 kg.On average household, farmland allocation was 1.2 ha with a minimum of 0.5 ha and a maximum of 2.5 ha for barley production (Table 1).
Based on a focus group discussion and survey information, there is no use for improved barley seed among the respondents.Therefore, all sample households apply only local barley seed with a mean of 64.5 kg, a minimum of 25 kg and a maximum of 120 kg, which was lower than the predicted way of recommended barley seed rate of 120 kg (Table 1).

Results of econometric model
Prior to analyzing the model, a different econometrics test was checked (Table 2).To estimate the level of efficiency, both Cobb-Douglas and trans-log production function can be used.Likelihood ratio test is applied to choose the appropriate model, and the general likelihood ratio (LR) is calculated on null ðH 0 Þ and alternative ðH a Þ hypotheses.While checking the likelihood ratio test, the null hypothesis states that all coefficients of product term in Cobb-Douglas specifications are equal to zero H 0 ¼ β ij ¼ 0, whereas the alternative hypothesis states that the coefficients of all interaction terms and square specifications in the translog functional forms are different from zero.The first hypothesis is that selected appropriate functional form of a model, which fits the data set using likelihood ratio test.The likelihood ratio test statistic is calculated in the following way: 2) based on Equation 5.The test statistic is distributed with degrees of freedom equal to a number of variables added to the alternative hypothesis with 15 degrees of freedom.
The calculated value (21.86) is lower than the upper 5% critical value χ 2 with its respective 15 degrees of freedom (24.99) (Table 2).Therefore, the null hypothesis is accepted, which states that all coefficients of the product in Cobb-Douglas specifications are equal to zero.This implies that the Cobb-Douglas production function was adequately represented in the data set.
The second test is the null hypothesis of all coefficients explaining that inefficiency is equal to zero and the alternative hypothesis of all coefficients explaining that inefficiency is different from zero.As a result, the null hypothesis is rejected in favor of the alternative hypothesis showing that explanatory variables associated with the inefficiency effect model are simultaneously different from zero, and the null hypothesis of all coefficients explaining that inefficiency is not equal to zero.From the above hypothesis, the value of γ = 0 is rejected, and the value of λ is 1.767 (Table 3).
The third hypothesis determines whether the explanatory variables associated with efficiency effects are simultaneously zero 2).The null hypothesis states a model without explanatory variables of efficiency effects, while the alternative hypothesis states that the full frontier model with explanatory variables is supposed to determine inefficiency .924, which is greater than the critical value of (22.36) at 13 degrees of freedom.Based on the results, the null hypothesis that explanatory variables are simultaneously equal to zero is rejected at 5% level of significance (Table 2).Therefore, explanatory variables of efficiency can together determine the variation of production on barley output in the study area.

Estimation production
The maximum likelihood estimation (MLE) of the parameters of SFPF specified in Equation ( 3) is obtained using the STATA 13 computer program.These results together with the standard Cobb-Douglas frontier estimates of the average production function are presented (Table 3).The returns to scale analysis can serve as a measure of total factor productivity.The coefficients are calculated to be 1.139; this indicates that increasing returns to scale leads to increasing rate because the value of returns to scale is greater than one.This means that there is potential for barley producers   to continue to increase their production because they are in stage I production.If the farmers were to increase 1% in all inputs, it would proportionally increase the total production of barley output by 1.139% (Table 3).Therefore, an increase in all inputs by 1% would increase barley output by more than 1%.This result is consistent with the study of Bati (2014), who estimated the returns to scale to be 1.04% (stage I) on barley production.The finding is consistent with the findings of Beshir (2016) and Ayele (2016) who estimated the returns to scale of 1.33% and 1.38% in the study of technical efficiency of wheat production South Wollo and Hadiya zones, respectively.However, this finding is inconsistent with the study of Gebretsadik (2017) in Meket district, who found a returns to scale to be 0.801, which is a stage II production.

Efficiency score.
The average technical efficiency is found to be 75.1% with a minimum of 41.9% and a maximum of 92.9% (Table 4).It indicates that farmers on average could decrease inputs by 24.9%if they are technically efficient.In other words, it implies that if resources were efficiently utilized, the average technical efficiency of the farmer could increase the current output by 24.9% using the existing resources and level of technology.The result indicates that the farmer with an average level of technical efficiency would enjoy an increase of about 19.16% derived from 1 À 0:751=0:929 ð Þ ½ � � 100 to attain the level of the most efficient farmer.The most technical inefficient farmers would have an efficiency increase of 54.898% derived from 1 À 0:419=0:929 ð Þ ½ � � 100 to attain the level of the most technical efficient farmers (Table 4).

Distribution of efficiency scores
The distribution of technical efficiency scores indicates that the higher distribution groups range from 71% to 90%, which covers 70% of sample households out of the total sample respondents (Figure 2).However, there are also some households whose technical efficiency levels are restricted to the range 41-70%, which covers 27.5% of sample households.
Households in this group have space to enhance their barley production at least by 70% on average.Out of the total sample households, only 2.5% had a technical efficiency use of 91-100%.This implies that 97.5% of households can increase their production at least by 10%.

Factors affecting of technical efficiency of barley producers
After determining the presence of efficiency differences among barley producers and measuring the level of their technical efficiency, identifying the factors affecting the level of efficiency was the next objective of the study.The factors of these efficiency variation estimates from the model regressed on demographic, socioeconomic, farm-related and institutional variables clarify the variations in the level of efficiency across sample households using the tobit model.These are age, sex, family size, education, non-farm activities, farm size, crop rotation, livestock ownership, distance farm to home, credit use, frequency of extension contact, distance to the nearest market and non-input total expenditure, which are expected to be affect the efficiency level.The results of the tobit regression model indicate that five out of thirteen variables have a significant effect on the technical efficiency of sample farmers.These are sex, education, livestock ownership, credit use and total expenditure (Table 5).
The significant variables are discussed separately depending on their siginificance level and marginal effect (Table 6).

Sex of the household head.
The sex of the household head had a positive and statistically significant effect at 1% level of significance on technical efficiency.This indicates that the male household head was better efficient than the female household head.This indicates that the male household head would increase the probability of a farmer to fall under technical efficiency by 3.57% and the expected value of technical efficiency by 17.90%, with an overall increase in the probability and the expected level of efficiency by 18.43% (Table 6).This finding is in line with the study of Mekdes (2011), Bati (2014) and Meftu (2016), who found that sex had a positive and significant impact on the technical efficiency of barley production.

Education level of the household head.
Education level of the household head has positive coefficient and is statistically significant at 10% level of significance related to technical efficiency.This shows that the education status of the household head increases as the technical efficiency of the barley production increases.The educated household heads expected to increase managerial ability and guide to good decisions in farming systems.This indicates that the education level of farmers in years of schooling become one year higher than that of others, technical efficiency of the barley producers would increase the probability of a farmer to fall under technical efficiency 0.07% and the expected value of technical efficiency by 0.29% with an overall increase in the probability and the expected level of technical efficiency by 0.30% (Table 6).This finding is similar to the studies of Solomon (2012), Debebe et al. (2015) and Musa et al. (2015) who found that education level had a positive and significant effect on the technical efficiency of maize production, and Walle (2018), who found that education level has a positive and significant effect on technical efficiency of barley production.

Livestock ownership.
Livestock owned by households had a positive effect on technical efficiency.The result found that farmers having more livestock had a positive and significant effect by 5% level of significance related to technical efficiency.This means that farmers who increased their number of livestock holding by one TLU could increase their technical efficiency.Moreover, the computed mariginal effect of livestock ownership of the farmer showed that a one TLU increase in livestock of the farmer would increase the probability of the farmer being technically efficient by about 0.32% and the expected value of technical efficiency by 1.37%, with an overall increase in the probability and the expected level of technical efficiency by 1.42% (Table 6).This finding is in line with the study of Solomon (2012) who found that livestock ownership had a positive and significant effect on the technical efficiency of wheat seed production, Bati (2014) who found that livestock ownership had a positive and significant effect on the technical efficiency of on barley production and Gebretsadik (2017) who found that livestock ownership had a positive and significant effect on the technical, allocative and economic efficiencies and sources of inefficiencies among large-scale sesame producers.

Credit uses.
The amount of credit used affected the technical efficiency of farmers positively and had statistical significance at a 10% level of significance related to technical efficiency.This indicates that farmers who use more credit have a higher level of technical efficiency.This means that farmers who increased their credit use could increase their technical efficiency.Moreover, the computed mariginal effect of credit use of the farmer showed that a 1-birr increase in credit use of the farmer would increase the probability of the farmer being technically efficient by about 0.05%, and the expected value of technical efficiency by 0.1% with an overall increase in the probability and the expected level of technical efficiency by 1.62% (Table 6).This finding is similar to the studies of Mekonnen (2013) who found that credit use had a positive and significant effect on the technical efficiency of sesame production, Hasen (2013) who found that credit use has a positive effect on the technical efficiency of maize production, Leggesse (2015) who found that credit use has a positive effect on technical efficiency of teff production and Gebretsadik (2017) who revealed that credit use has a positive and significant effect on sesame production.
3.2.3.5.Non-input total expenditure.Non-input total expenditure of households had negative and statistically significant relationships at 5% level of significance with technical efficiency of the farmers.This shows that for the farmers who utilize more of their income in different aspects without agriculture expense, technical efficiency decreases relative to their counterparts.Based on the survey information, for majority of sample respondents, their income is spent more on the consumption of their household feed and social obligations (Table 6).This finding is consistent with the study of Bati (2014), who found that non-input expenditure has a negative and significant effect on technical efficiency of barley production.

Conclusion and recommendation
The study was conducted on smallholder barley producers' use óf resources efficiently in the production of barley production with a mean technical efficiency level of 75.1.This implies that the farmers can increase their barley production on average by 24.9%, if they were technically efficient.
The result of the production function indicates that labor, oxen power, fertilizer, land and seed had positive coefficients of 0.114, 0.153, 0.029, 0.362 and 0.481, respectively.This means that the use of these inputs increased by one unit, and the barley output increased by 11.4, 15.3, 2.9, 36.2 and 48.1%, respectively.The mean of barley output for the sample household in the study area during the 2017/18 production season was 19.5 quintals with a minimum of 5 qt.and a maximum of 46 qt.According to the result of the tobit model, technical efficiency was positively and significantly affected by sex, education, livestock, non-farm activity and credit use; however, non-input total expenditure was affected negatively and significantly.This indicates that male-headed households, higher educated farmers, more number of livestock, participating in the non-farm activity, credit use and lower non-input expenditures has higher technical efficiency.
Based on the results of the study, the following recommendations are drawn: The positive and significant coefficients of labor, oxen power, fertilizer, land and seed indicate the significant inputs to increasing barley production.The local governments should put more emphasis on strengthening the efficient use of labor, oxen power, fertilizer, land and seed for the farmers in the district.
Sex of households' head had a positive and significant effect on technical efficiency.This implies that male-headed households had better efficiency than female-headed households.Therefore, local governments and gender office should focus on how to strengthen female farmers to improve their level of efficiency through experience sharing, giving training on input use and market information, and promoting credit use to improve their agricultural productivity.
Education level had a positive and significant effect on technical efficiency of barley producers.Therefore, the regional and local government should focus on how to deliver sufficient and effective basic educational opportunities for farmers.The district education office should provide youth training centre, practical training, create awareness and knowledge about the application of inputs, technology and different farming systems.
Livestock ownership had a positive and significant effect on technical efficiency.This means farmers having more number of livestock had better technical efficiency for barley production.Hence, designing appropriate policy and strategies for improving livestock production and productivity systems helps to enhance the technical efficiency of barley output.Therefore, the regional and local government should focus on the mixed farming system -livestock and crop production.
Non-farm activities had a positive and significant effect on technical efficiency in the study area.Therefore, local government should focus on how to introduce non-farm activities that enhance the income of households and create an awareness diversification system of non-farm activities, which help with an additional income to facilitate the necessary resource used for barley production.
Credit use had a positive and significant effect on the technical efficiency of barley producers.This means farmers who were using credit were more efficient than those who do not.Therefore, the regional and local governments should intervene to strengthen the operation of rural saving and credit institutions at the village level and create awareness for the farmers.Amhara Credit and Savings Institution should focus on how to provide credit services, create awareness about credit use and create better access for the farmers.
Total non-input expenditure of households excluding production inputs had a negative and significant effect on the technical efficiency of barley producers in the district.The policymakers should have strong work on creating awareness on the use of efficient allocation of resources, mostly on the money resource.

Figure 1 .
Figure 1.Geographic map of the study district.

Table 1 . Summary of descriptive statistics of variables used in the production function
Source: Own survey (2018).

Table 5 . Tobit model results on the source of technical efficiency variation Variables TE Coefficient Std. err.
Source: Based on model output (2018).

Table 6 . Marginal effects of technical efficiency after tobit model Variables The marginal effect of TE
xjMarginal effects computed only for significant variables and values in cell explain