Livestock Research for Rural Development 23 (9) 2011 Notes to Authors LRRD Newsletter

Citation of this paper

Performance of mixed crop-livestock production system: the Data Envelopment approach

Hassen Beshir Hussien

Wollo University, Department of Agricultural Economics, P.O. Box 1145, Ethiopia,
hassenhussien@gmail.com

Abstract

The general objective of the study was to calculate the production efficiency of the mixed crop-livestock farmers in two districts of north eastern Ethiopia. Cross-sectional data were used to analyze the performance of mixed crop and livestock production system and determinants of production efficiencies. For this study, a total of 252 farmers were selected using probability proportional to sample size sampling technique. The non-parametric method DEA was employed to measure production efficiency.

The non-parametric methods of efficiency measurement indicated that most farmers in the study area were not efficient suggesting that efficiency improvement is one of the possible avenues for increasing agricultural production with available resource and technology. The mean TE, AE and EE of the household calculated from non-parametric approach of Data Envelopment Analysis Variable Returns to Scale (DEAVRS) were 55%, 70% and 40%, respectively. The production efficiency of mixed crop-livestock farming was determined by farm size, livestock ownership, labour availability, off/non-farm income participation, total household asset, total household consumption expenditure and improved technology adoption. This study found that improved agricultural technology adoption increased production efficiency of households. Such actions may, in turn, alleviate the current problem of food insecurity and lead in the long run to economic development.

Key words: economic efficiency, non-parametric frontier, technology, technical efficiency


Introduction

Agriculture is the mainstay of the Ethiopian economy and the overall economic growth of the country is highly correlated to the success of the agricultural sector. Agriculture accounts for 43% of the country’s Gross Domestic Product (GDP) and 85% of all exports (coffee, livestock and livestock products and oil seeds). The industrial sector is small in size contributing, on average, to less than 13% of the GDP (FAO 2007; MoFED 2010). Agriculture provided employment for 85% of the population in 2008/09 and raw materials for 70% of the industries in the country (MOFED  2006; MoFED  2010). The bulk of the agricultural GDP for the period 1960-2009 had come from the production of crops (90%) and livestock (10%)  (FAO 2007; MoFED 2010). Especially, the role of agriculture in securing the food needs of the fast growing population is considerable.  

The average annual growth rate of agriculture for the period 1960-1974 was 1.7% and in the 1974-1991 period, it grew to 3.8%.  Moreover, in the 1991-2009 period, the average annual growth rate of the sector reached 5.5%. The reasons for the low growth rate of agriculture and GDP for several decades were mainly severe weather fluctuations, inappropriate economic policies and low adoption of improved agricultural technologies and production efficiency and prolonged civil unrest (Hailu 2008).  

In the north eastern part of Ethiopia, where this study was conducted, crop and livestock productions were the means of livelihood of the people to meet the household consumptions and to generate income. The major crops grown by sample households were improved and local wheat, barley, teff, local and improved horse bean, field pea, maize, local and improved potato, oat, fenugreek, garlic, lentil, chickpea, grass pea, sorghum, haricot bean and linseed. The major crops grown by sample households were improved and local dairy cow, improved and local poultry, local and improved beehives, sheep and goat products. The outputs of crops and livestock were used mainly for home consumption but were rarely used for markets to obtain cash income.  The straws of crops were used for animal feed. Animals like oxen were also used for draft power in plowing and planting. Moreover, the wastes of animal in the form of manure were used for improvement of soil fertility.  The integration of crop and livestock production also serves as a means to cope up with the market and the environmental risks. However, the productivity of crop and livestock agricultural system in the study area is very low.  The poor production and productivity of crop and livestock resulted in food insecurity. Therefore, assessing the factors responsible for low production and productivity of smallholder mixed crop-livestock farmers in Ethiopia in general and in north eastern highlands of Ethiopia in particular was paramount importance. This study aimed at filling this gap. 

The study set out to answer the following questions:

  1. What is the level of efficiency of mixed crop and livestock production of farm households in the study area?
  2. What are the factors that affect the mixed crop and livestock production efficiency of farmers in the study area?

The specific objectives of the study were to:

  1. estimate the farm level efficiency of the mixed crop-livestock production system; and
  2. identify the sources of efficiency differential among the farmers.


The Study Design and Methodology

Description of the study area

This study was carried out in South Wollo. South Wollo is located in the North East part of Ethiopia.  South Wollo is one of the eleven administrative zones of the Amhara National Regional State. It is situated between the Eastern highland plateaus of the region and the North Eastern highland plateaus of Ethiopia. It is divided into 20 administrative districts (weredas) and has two major towns (Kombolcha and Dessie) and 18 rural districts. Among the eighteen rural districts, Dessie Zuria and Kutaber are selected for this study. South Wollo is located between latitudes 10010’N and 11041’N and longitudes 38028’ and 4005’E. According to the Central Statistical Agency’s population census data, in 2007 the total population of South Wollo was 2,519,450 of which 50.5% were females and 88% were rural residents (CSA 2008). The total land area in South Wollo, Dessie Zuria and Kutaber is 1,773,681 hectares, 180,100 hectares and 72,344 hectares, respectively. The cultivated land area accounts for 39%, 20% and 35.3% of the total area of Dessie Zuria, Kutaber and South Wollo, respectively. 

Sample size and sampling procedure

Dessie Zuria and Kutaber districts were selected purposively based on their accessibility and relevance of the study. A multistage random sampling method was used for the selection of the sample respondents. In the first stage of sampling, 6 Farmers’ Associations (FAs) were selected randomly from a total of 54 FAs (3 from Dessie Zuria and 3 from Kutaber). In other words, as the number of Farmers’ Association in Dessie Zuria (28) was equal to that of Kutaber (26), three Farmers’ Associations were selected from each district using simple random sampling procedure. In the second stage, a total of 252 farmers were selected using probability proportional to sample size sampling technique (Table 1).

Table 1. Distribution of sample farm household heads by farmers’ association and district

Name of District

Name of FA

Total household* head

Sample farm household heads

 

 

Female

Male

Total

Male

Female

Number

 

Number

 

Number

 

Dessie Zuria

Tita

686

182

7

 

27

 

34

 

Bilen

1,179

161

8

 

45

 

53

 

Endod Ber

688

102

4

 

27

 

31

 

Kutaber

Boru

490

123

5

 

20

 

25

 

Beshlo

797

201

8

 

32

 

40

 

Alasha

1,297

458

18

 

51

 

69

 

 

Total 

5,137

1,227

50

 

202

 

252

 

Source: *Kebele Administration Office (Personal Communication),

Data collection and sources

A structured questionnaire was designed, pre-tested and refined to collect primary data.  A total of 252 farms were used to analyze their economic efficiency. Data was intensively collected on existing and improved crop and livestock production activities and purposes from July 2009 to November 2009.  Experienced numerators were recruited and trained to facilitate the task of data collection. Farm visit, direct observation and informal interview were undertaken both by the researcher and the enumerators. The secondary data were extracted from studies conducted and information documented at various levels of Central Statistical Agency, Ministry of Agriculture and Rural Development  and Finance and Economic Development Offices in the study area.

Non-parametric Frontier approaches to measure efficiency

The theory and concept of measurement of efficiency has been linked to the use of production functions. Different techniques have been employed to either calculate or estimate the efficient frontiers.  These techniques are classified as parametric and non-parametric methods. Farrell (1957) was the first to formulate a non-parametric frontier method to measure production (economic) efficiency of a firm. According to him, efficiency ratios are estimated from sample observations. He defined technical, allocative and economic efficiencies. Technical efficiency (TE) reflects the ability of a firm to obtain maximum output from a given resources. Allocative efficiency (AE) reflects the ability of a firm to use inputs in optimal proportion given the input prices and production technology. Economic efficiency (EE) is the multiplicative effect of these two efficiencies. 

The single-input/output non-parametric efficiency measure of Farrell (1957) is generalized for the multiple inputs/outputs case and reformulated as a mathematical programming problem by Charnes et al (1978). Charnes et al (1981) named the method introduced in Charnes et al (1978) as Data Envelopment Analysis (DEA). This technique was initially born in operations research for measuring and comparing the relative efficiency of a set of decision-making units (DMUs). Since that seminal paper, numerous theoretical improvements and empirical applications of this technique have appeared in the productive efficiency literature.  

The DEA calculations are designed to maximize the relative efficiency score of each DMU, subject to the constraint that the set of weights obtained in this manner for each DMU must also be feasible for all the other DMUs included in the sample. The technical efficiency score can be calculated using the following mathematical programming formulation (Coelli et al 1998).                       

                                   

Where xi and yi are the input and output levels of the ith farmer; X is an mxn input matrix; Y stands for sxn output matrix representing data for all n farmers in the sample; θ is a scalar, and λ is an nx1 vector of constants. θ is always less than or equal to one. A value of one indicates a point on the frontier and hence the existence of a technically efficient farmer. Assuming identical input prices, economic efficiency is simply calculated by solving the following DEA model:


Where c is a scalar representing cost or budget level and C is a 1xn matrix of observed costs. The problem takes the ith farmer and then seeks the amount by which the input cost; ci can be reduced to remain in the frontier.

                                                                                                                      

Estimation of the determinants of production efficiency

In efficiency analysis, factors that influence efficiency are of paramount importance. Following the quantification of the production efficiency measures, a second stage analysis involved a regression of these measures on several hypothesized socioeconomic, institutional and technological factors that affect the efficiency of the farmers.  The most common procedure is to examine the determinants of efficiency, in that the inefficiency or efficiency index is taken as a dependent variable and is then regressed against a number of other explanatory variables that are hypothesized to affect efficiency levels (Bravo-Ureta and Rieger 1991; Sharma et al 1999; Arega 2003; Jema and Andersson 2006). However, few authors (e.g., Kumbhakar et al 1991; Battese and Coelli 1995) used a specific model that allowed researchers to estimate the efficiency scores and simultaneously to test the effects of explanatory variables. For the former approach, technical, allocative and economic efficiency estimates were regressed, using Tobit model (Sharma et al 1999; Jema and Andersson, 2006) or linear regression model (Sharma et al 1999; Arega 2003) on the farm specific explanatory variables that might explain variations in efficiency across farms. Technical, allocative and economic efficiency estimates derived from DEAVRS were regressed, using a censored Tobit model on the following farm-specific explanatory variables that might explain variations in production efficiencies across farms. The rationale behind using the Tobit model was that there were a number of farms for which efficiency was one and the bounded nature of efficiency between zero and one (Jackson and Fethi 2000). That is, due to large number of fully efficient DEA estimates, the distribution of efficiency measures was censored above from unity. Estimation with OLS regression of the efficiency scores would lead to biased parameter estimates since OLS assumes normal and homoscedastic distribution of the disturbance and the dependent variable (Greene 2003). As the distribution of the estimated efficiencies are censored from above at the value one, Tobit regression (Tobin 1958) is specified as

                                    

Where Ei is an efficiency score, and V~N(0,σ2) and βj are the parameters of interest.  

Description of variables for efficiency measurement
Production function variables

The variables that were used in the Data Envelopment model were defined as follows.

i. Outputs: physical yield of crops and livestock and their respective prices were used to compute the value of output of the farm.  The value of crop and livestock output was derived from output of improved and local wheat, barley, teff, local and improved horse bean, field pea, maize, local and improved potato, oat, fenugreek, garlic, lentil, chickpea, grass pea, sorghum, haricot bean, linseed, milk of improved and local dairy cow milk, improved and local poultry, local and improved beehives, number of sheep and goat products. These outputs were multiplied by their respective market price to obtain the value of crop and livestock output. The respective monthly market prices were collected from South Wollo department of agriculture and rural development office. The averages of these prices were used for computational analysis.

ii. Inputs: these were defined as the major inputs used in the production of crop and livestock. They were:

Land: This represented the physical unit of cultivated land and grazing land in hectares;

Human labour: This was man days worked by family, exchange and hired labour for land preparation, planting, weeding, or cultivation, irrigation, harvesting and rearing livestock;

Oxen labour: This was oxen days worked by the household using oxen labour for land preparation, planting and threshing;

Material inputs: This included the cost of veterinary, feed, organic and chemical fertilizers, improved and local seeds and pesticides used by the farm household.  

Cost function variables:

iii. Input prices:  the input prices of land, human labour and oxen labour needed for deriving the dual cost frontier in the parametric and nonparametric method were collected.  Moreover, the value of the output of crop and livestock was used as computed above and adjusted for statistical noise.  

Variables included in the determinants of efficiency model

The dependent variable was the efficiency score which was computed from non-parametric methods of efficiency measurement.

iv. Efficiency factors: these denoted various factors hypothesized to explain differences in productive efficiency among farmers. These were:

Age: this was the age of the household head in years.

Farm size: it was defined as the total area of cultivated and grazing land in hectare.

Education: it was a continuous variable defined as years of formal schooling;

Labour available:  it was defined as the total active labour available in the family.

Livestock ownership: it was defined as the total livestock available in TLU.

Off/non-farm income: this included income from off-farm and non-farm activities. It was a dummy variable that the variable was 1 if the household earned off/non-farm income and 0 otherwise.

Credit service: it included access to credits for farm inputs and other farm activities from semi-formal sources. It was a dummy variable defined as 1 if the farmers have received credit and 0 otherwise.

Extension service: it was defined as whether the farmer had access to the extension service during the survey year or not. It was a dummy variable defined as 1 if the household had access to extension service and 0 otherwise

Expenditures: it was the total yearly consumption expenditure of the household in goods and services.

Assets: it was defined as the sum of current values of all furniture, farm implements and other equipments and livestock owned by the household.

Technology adoption: this was whether or not the household adopted at least one improved agricultural technology. It was a dummy variable defined as 1 if the farmer had adopted at least one improved technology and 0 otherwise.


Results and Discussion

The descriptive statistics of output and input variables used in DEA methods are summarized in Table 2. As it was explained in the methodology, the total value of crop and livestock outputs were derived from output of improved and local crop and livestock products. These outputs were multiplied by their respective average market price to obtain the value of crop and livestock output. The average value of the output was Birr 10,144 with minimum and maximum values of Birr 205 and Birr 32,201, respectively. The average total land area was 0.72 hectare. The mean of the human and oxen labour for the farm households for which the production function was estimated was 180 man-days and 28 oxen-days, respectively.  The mean of the material inputs applied by the farm households for the sample period was Birr 1,971 with a minimum of Birr 8 and a maximum of Birr 7,079. 

Table 2. Descriptive results of input-output variables

Variable

Mean

SD

Min

Max

Value of output in Birr1

10,144

6,423

205

32,201

Land area in hectare

0.72

0.45

0.02

2.28

Human labour in man days (MD)

180

97

4

652

Oxen labour in oxen days (OD)

29

16

2

78

Material input cost in Birr

1971

1,387

8.2

7,079

Source: Own survey, 2009

1 birr is the local currency which is exchanged about 17 birr for an American dollar

DEA frontier results

The non-parametric DEA models for variable and constant returns to scale were calculated using computer programme DEA 2.1 (Coelli 1996). The frequency distribution and summary statistics of technical (TE), allocative (AE) and economic (EE) efficiency scores from non-parametric methods are presented in Table 3. The average TE, AE and EE score for non-parametric approach using the Data Envelopment Analysis Constant Returns to Scale (DEACRS) were 50%, 64% and 31%, respectively. These imply that there were substantial inefficiencies in production and hence rooms for production gain through efficiency improvement. This suggests that the farm households could reduce their production costs by 50%, 36% and 69% if they could operate at full technical, allocative and economic efficiency levels, respectively. The average TE, AE and EE score for non-parametric approach Data Envelopment Analysis Variable Returns to Scale (DEAVRS) were 55%, 70% and 40%, respectively. They indicate that there was a scope for reducing cost in production and hence obtaining productivity gain through efficiency improvement. This suggests that the farm households can reduce their production costs by 45%, 30% and 60% if they could operate at full technical, allocative and economic efficiency levels, respectively. 

The frequency distribution given in Table 3 showed that there were significant number of farmers whose efficiency scores were less than 50% for non-parametric methods. The number of farmers whose technical, allocative and economic efficiency scores were greater than 90% in DEACRS was 19, 4 and 3, respectively. The number of farmers whose technical, allocative and economic efficiency scores was greater than 90% in DEAVRS were 41, 32 and 16, respectively. The number of farmers whose technical, allocative and economic efficiency scores was greater than 100% in DEAVRS was 30, 13 and 11, respectively. These farmers were operating at the frontier and were best performing farm household in the study area. This implies that the best performing farmers whose efficiency score were 100% should have to be scaled up.  

The standard deviation (SD) of the efficiency scores from non-parametric methods showed that there was variability of efficiency scores across the farmers. For instance, the SD of TE, AE and EE in DEAVRS were 24.02, 17.98 and 23.54, respectively. The minimum and maximum TE in DEAVRS was 16.5% and 100%, respectively. However, the minimum and maximum EE in DEAVRS was 5.8% and 100%, respectively. The results are in agreement with the findings of other studies that have shown the existence of substantial inefficiencies in developing agricultural economies and the consequent implications for agricultural growth possibilities with the existing resource and technology. For example, high technical, allocative and economic inefficiencies had been found to exist among crop producers’ in Ethiopia (Getachew 1995; Arega 2003; Jema and Andersson 2006) suggesting that efficiency improvement is one of the possible avenues for increasing agricultural production with the available resources and technology.   

The DEACRS assumption is appropriate when all firms are operating at optimal scale. However, imperfect information, government regulation and constraints on finance etc may cause a firm to be not operating at optimal scale.  The use of DEAVRS specification permits the computation of Scale Efficiency (SE). SE=TECRS/TEVRS. Where TECRS is technical efficiency due to constant returns to scale assumption and TEVRS is technical efficiency due to variable returns to scale assumption. When SE=1, the firm is at optimal scale or scale efficient. If there is scale inefficiency, firms will operate either under increasing returns to scale (IRS) or decreasing returns to scale (DRS). Therefore, sample farmers were classified under increasing returns to scale (72), decreasing returns to scale (168) and constant returns to scale (12). The number of farmers, who were scale efficient, was 12. For these farmers, the only option to increase productivity is using improved agricultural technologies which will shift the production frontier outwards. 

Table 3. Frequency distribution of technical (TE), allocative (AE) and economic (EE) efficiency from non-parametric methods

Efficiency in %

Non-parametric (N)

DEAVRS

DEACRS

TE

AE

EE

TE

AE

EE

<10

0

0

8

0

0

11

10_20

2

4

33

8

4

42

20_30

24

7

71

24

5

83

30_40

54

6

46

59

6

57

40_50

57

15

28

55

14

37

50_60

35

23

21

40

36

14

60_70

20

69

14

19

107

4

70_80

11

52

6

12

54

0

80_90

8

44

9

16

22

1

>90

41

32

16

19

4

3

Mean (%)

55.1

70.2

39.9

50.1

64.2

31.8

SD (%)

24.0

17.9

23.5

21.2

13.5

14.8

Minimum (%)

16.5

15.

5.80

12.3

15.3

4.65

Maximum (%)

100

100

100

100

100

100

N=number of farmers

The results from this method indicated the existence of considerable inefficiencies in the mixed crop-livestock farmers in the study areas. This means that there are opportunities to increase agricultural outputs without supplying additional inputs, given the existing resource combination and technology. The potential opportunities for improving crop-livestock production system were improving crop management, animal feeding and animal health systems. The results showed that output can be increased given the present level of inputs farmers are using, if policy variables that are determining the level of TE, AE and EE of farmers are identified. The identification of these variables requires estimation of possible policy variables that could have influence on TE, AE and EE. 

Determinants of production efficiency of mixed crop-livestock farm household 

Here, the determinants of efficiencies were identified by incorporating agricultural technology adoption as a covariate. Agricultural technology adoption, was understood to mean, improved crop and livestock technologies used to enhance productivity at a household level. It was hypothesized that the application of improved agricultural technologies, among other socioeconomic factors, affected farm household technical, allocative and economic efficiencies while the improved technology adoption dummy itself is an endogenous variable which can be affected by farmers’ production efficiencies. A simultaneous equation Tobit model was employed to test these hypotheses and to identify the factors that influence household production efficiencies and improved technology adoption.  

Therefore, the model result revealed that the estimated parameter coefficient for the predicted error term (residual) was not statistically different from zero. As a result, the null hypothesis that there is no a simultaneity relationship between household efficiencies and improved technology adoption was accepted. This implies that the efficiencies of the farmers can be modeled by using the normal single equation standard Tobit model and by directly incorporating improved agricultural technologies dummy, along with other explanatory variables.  

Crop and livestock producers’ differences in technical, allocative and economic efficiencies levels were hypothesized to be due to several farm and farmers attributes, mainly reflecting their managerial ability and access to information. The second stage procedure is to identify the determinants of production efficiency among farmers using the Tobit model. This was done by regressing the efficiency levels obtained from non-parametric DEA variable returns to scale method. The Tobit model estimates and the respective marginal effects are provided in Tables 4-6. 

Table 4. Determinants of technical efficiency of smallholder mixed crop and livestock farmers in north eastern highland of Ethiopia 

Variables

Coefficient

Std. Err.

Marginal effect

Age

-0.000795

0.000783

-0.000791

Education

-0.00331

0.00344

-0.00329

Labour available

-0.00732

0.0067

-0.00728

Farm size

0.01247

0.0219

0.01241

Livestock ownership

0.0197***

0.006

0.0196

Off/non-farm income

0.0461**

0.023

0.0458

Household asset 

0.1011***

0.015

0.1005

Household expenditure

-0.1449***

0.024

-0.1441

Extension service

0.0222

0.024

0.0221

Credit service

0.0253***

0.0025

0.0251

Technology adoption

0.0642***

0.024

0.0638

Constant

1.824***

0.056

 

Test statistics

LR χ2*** (11)     =      215

Log –L*** =  116         

***, ** and * implies significant at 1%, 5% and 10% probability level, respectively

The Tobit model results indicated that technical efficiency was positively and significantly affected by size of livestock, off/non-farm income, total household asset, and technology adoption.  Technical efficiency was negatively and significantly related to the total household consumption expenditure. The positive and significant influence of household asset on technical efficiency of the household may be due to the fact that the income is used to improve the experience and human and physical capital of the farm household, serve as additional funding to farm activities and improve managerial skills. The result is in line with the results obtained by Jema and Anderson (2006). 

Table 5. Determinants of allocative efficiency of smallholder mixed crop and livestock farmers in north eastern highland of Ethiopia 

Variables

Coefficient

SE

Marginal effect

Age

-0.00084

0.00082

-0.00083

Education

-0.0043

0.0036

-0.0042

Labour available

0.012*

0.007

0.011

Farm size

0.0736***

0.0229

0.0735

Livestock ownership

0.004

0.006

0.004

Off/non-farm income

0.048*

0.023

0.047

Household asset 

0.0615***

0.0162

0.0614

Household expenditure

-0.0653***

0.0253

-0.0652

Extension service

-0.0086

0.0251

-0.0084

Credit service

0.0068

0.0254

0.0065

Technology adoption

0.0606**

0.0258

0.0603

Constant

0. 852***

0.222

 

Test statistics

LR χ2*** (11)     =      85

Log –L*** =  107        

***, ** and * implies significant at 1%, 5% and 10% probability level, respectively

The allocative efficiency of the farm household was positively and significantly influenced by the available labour, farm size, total household asset and improved technology adoption. It was negatively and significantly affected by household consumption expenditure. The positive and significant effect of man equivalent on allocative efficiency of mixed crop and livestock producers implies that households with more labour force were contributing for efficient utilization of resources.  

This study found out that large farm size was expected to have a significant positive effect on allocative efficiency levels because such farms realize increasing returns to scale. The negative effect of farm size might be related to small farm size (Coelli et al 2002; Getachew 1995; Jema and Andersson 2006). The positive and significant influence of livestock ownership, off/non-farm income and household asset on allocative efficiency of the household might be due to the fact that the income was used to improve the skill and human and physical capital of the farm household, serve as additional funding to farm activities and improve managerial skills. Moreover, farmers with these resources had more information and capacity for optimal allocation of resources.   

The economic efficiency of the farm household was positively and significantly affected by labour force available, farm size, livestock ownership, off/non-farm income, total household asset and improved technology adoption. The rest of the variables, including access to extension and credit service and education had the expected positive signs but insignificant effect on economic efficiency. The statistically significant negative effect on the estimated coefficients on all efficiency scores for the household consumption expenditure may reveal a situation where household that spent excessively on consumption goods were unable to support their agricultural activities. Therefore, these households became less efficient (Jema and Anderson 2006). 

The household asset or wealth significantly and positively affected economic efficiency of the crop and livestock farmers. This means that relatively wealthier farm households were more economically efficient than less wealthy ones. That is, the farmers’ capacity to self-finance may increase as they get wealthier, reducing demand for credit. However, if wealthier farm households expand their farm operations and demand additional external resources, they will be more creditworthy and less rationed in the credit market than the less wealthy farmers.  The positive and significant effect of the total household asset is consistent with the results obtained by Jema and Anderson (2006) and Hussien and Öhlmer (2007).  

The positive and significant effects of farm households’ improved technology adoption on all efficiency scores were related to the fact that the households were adapting improved practice and technology, acquiring and analyzing information. This suggested that better utilization of improved agricultural technologies such as chemical fertilizers, improved dairy cows, forage seeds and wheat varieties improved the technical, allocative and economic efficiencies of mixed crop and livestock farmers.

Table 6. Determinants of economic efficiency of smallholder mixed crop and livestock farmers in north eastern highland of Ethiopia 

Variables

Coefficient

SE

Marginal effect

Age

-0.00083

0.0008

-0.00082

Education

0.0048

0.0038

0.0043

Labour available

0.0124*

0.0073

0.0124

Farm size

0.0424*

0.023

0.0423

Livestock ownership

0.011*

0.006

0.014

Off/non-farm income

0.0438*

0.025

0.0436

Household asset 

0.0739***

0.017

0. 0737

Household expenditure

-0.1176***

0.0265

-0.1175

Extension service

0.0096

0.026

0.0094

Credit service

0.0113

0.0267

0.0112

Technology adoption

0.0541**

0.027

0.0540

Constant

-1.587

0.2326

 

Test statistics

LR χ2***(11)     =      133

Log –L*** =  94      

***, ** and * implies significant at 1%, 5% and 10% probability level, respectively

The marginal effect (0.0735) of farm size for allocative efficiency indicated that, for the sample period and sample households considered an increase in farm size by one hectare, on average, led to an increase of 7.35% in allocative efficiency. The marginal effect for discrete variable like off/non-farm income can be interpreted as, if a farmer participates in off/non-farm income, for example, his allocative efficiency level will increase, on average, by 4.7%. The marginal effect of improved technology adoption, for example, when a farmer adopted improved technology, his economic efficiency level will increase, on average, by 5.4%.  


Conclusions


References

Arega D 2003 Improved production technology and efficiency of smallholder farmers in Ethiopia: Extended parametric and non-parametric approaches to production efficiency analysis. PhD dissertation submitted to the Department of Agricultural Economics, Extension and Rural Development, Faculty of Natural and Agricultural Sciences, University of Pretoria.

Battese G E and Coelli T J 1995 A model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics, 20:325-332.

Bravo-Ureta B E and Rieger L 1991 Dairy farm efficiency measurement using stochastic frontiers and neoclassical duality. American Journal of Agricultural Economics, 73(2): 421-428.

Charnes A Cooper W W and Rhodes E 1978 Measuring the Efficiency of Decision-Making Units. European Journal of Operational Research, 2: 429–444.

Charnes A Cooper W W and Rhodes E 1981 Evaluating program and managerial efficiency: An application of Data Envelopment Analysis to Program Follow Through. Management Science, 27(6): 668–697.

Coelli T 1996 A Data Envelopment Analysis (DEA) Computer program. Center for Efficiency and Productivity Analysis, University of Queens land.

Coelli T Rao D S P and Battese G 1998 An introduction to efficiency and productivity Analysis. Boston: Kluwer Academic Publishers, 275p.

Coelli T J, Sandura R and Colin T 2002 Technical, allocative and cost and scale efficiency in Bangladesh rice production: A non-parametric approach. Journal of Agricultural Economics, 53: 607-626.

CSA (Central Statistical Agency) 2008 The 2007 Population and housing census of Ethiopia. Statistical Report on Population Size and Characteristics, 1(1), Federal Democratic Republic of Ethiopia Central Statistical Agency, Addis Ababa, Ethiopia.

FAO (Food and Agriculture Organization) 2007 Statistical database for Ethiopia, 1994-2005, Food and Agriculture Organization, available on line from the web site at www.FOASTAT.FAO.ORG (Accessed on 4 February 2010).

Farrell M J 1957 The measurement of productive efficiency. Journal of Royal Statistical Society, 120: 253-290.

Getachew A 1995 Production Efficiency Analysis: The case of smallholder farming in the coffee sector of Ethiopia and Kenya. Farming systems and Resource Economics in the Tropics, vol. 23, Germany.

Greene W H 2003 Econometric analysis. 5th  edition. Macmillan, New York, pp 1083.

Hailu B 2008 Adoption of improved teff and wheat production technologies in crop and livestock mixed systems in Northern and Western Shewa zones of Ethiopia, PhD Dissertation, University of Pretoria.

Hussien H and Öhlmer B 2007 Influence of credit constraints on production efficiency: The case of farm households in Southeastern Ethiopia. PhD Dissertation, Swedish University of Agricultural science, Sweden.

Jema H and Andersson H 2006 Determinants of efficiency of vegetable production in smallholder farms: The Case of Ethiopia. Food Economics, 3(3): 125-137.

Kumbhakar S C, Ghosh S and McGuckin J T 1991 A generalized production frontier approach for estimating determinants of inefficiency in US improved forage, improved seed and fertilizer farms. Journal of Business and Economics Statistics, 9: 279-286.

MoFED (Ministry of Finance and Economic Development) 2006 Ethiopia: Building on Progress a Plan for Accelerated and Sustained Development to End Poverty (PASDEP) (2005/06-2009/10).Volume I: Main text, Ministry of Finance and Economic Development, September, 2006, Addis Ababa

MoFED (Ministry of Finance and Economic Development) 2010 National Income accounts, Ministry of Finance and Economic Development (MoFED), available at www.MOFED.database.com (Accessed on 4 February 2010).

Sharma A R, Leung P and Zaleski H M 1999 Technical, allocative and economic efficiencies in swine production in Hawaii: A comparison of parametric and non-parametric approaches. Agricultural Economics, 20: 23-35.

Tobin J 1958 Estimation of Relationships for Limited Dependent Variables. Econometrica, 26: 24–36.



Received 13 July 2011; Accepted 26 August 2011; Published 1 September 2011

Go to top