at SciVerse ScienceDirect

Journal of Cleaner Production 27 (2012) 126e132

Contents lists available

Journal of Cleaner Production

journal homepage: www.elsevier .com/locate/ jc lepro

Assessing the technical efficiency of energy use in different barberryproduction systems

Seyed Hashem Mousavi-Avval a, Ali Mohammadi a,b,*, Shahin Rafiee a, Ahmad Tabatabaeefar a

aDepartment of Agricultural Machinery Engineering, Faculty of Agricultural Engineering and Technology, University of Tehran, Karaj, IranbCenter for Energy and Environmental Sciences (IVEM), University of Groningen, Nijenborgh 4, 9747 AG, Groningen, The Netherlands

a r t i c l e i n f o

Article history:Received 28 April 2011Received in revised form12 January 2012Accepted 13 January 2012Available online 21 January 2012

Keywords:DEAEnergy efficiencyEnergy inputBarberry productionFarm size

* Corresponding author. Department of AgricultuFaculty of Agricultural Engineering and Technology,Iran. Tel.: þ98 2612801011; fax: þ98 2612808138.

E-mail address: mohammadia@ut.ac.ir (A. Moham

0959-6526/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jclepro.2012.01.014

a b s t r a c t

The main objectives of this study were to analyze the technical and scale efficiencies of farmers and toidentify the wasteful uses of energy in different farm sizes of barberry production in Iran. For thesepurposes the data envelopment analysis approach was applied to the data of energy use for barberryproduction in individual farms. The results indicated that total energy input and yield value of smallfarms were higher than those of large farms. Also, energy resources are used more efficiently in smallfarms; technical efficiency of farmers in small and large farms was calculated as 0.66 and 0.50, respec-tively; also, scale efficiency was 0.82 and 0.62 for the respective farms. Total energy input in small andlarge farms could be reduced by 13.2% and 15.2%, respectively; accordingly, total energy requirement intarget conditions was calculated as 20,702.4 and 13,761.2 MJ ha�1. The highest potential improvementwas derived from diesel fuel, followed by electricity and biocides. Improving energy use efficiency ofwater pumping systems, improving timing, amount and reliability of water application, employing theconservation tillage methods and applying integrated pest management technique are suggested forimproving energy use efficiency and reducing the environmental footprints of barberry production.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Global climate change and population growth are placing newpressures on food production systems; demanding increases foodsecurity while safeguarding the natural resources by reducing theenvironmental footprints (Khan et al., 2009). Energy use is of greatconcern within agricultural productions, due to both associatedenvironmental effects and the cost of inputs. Efficient use of energywithin farming systems is critical for reducing the environmentalfootprints of energy inputs in food production, and so, providingsustainable agricultural production. Improving energy use effi-ciency is becoming increasingly important for combating risingenergy costs, depletion of natural resources and environmentaldeterioration (Dovì et al., 2009).

There are several parametric and non-parametric techniques tomeasure the efficiency in agricultural production systems. In somestudies the indicators of output energy to input energy ratio andspecific energy (i.e., input energy to yield ratio) in crop productionsystems have been used to evaluate the performance of farmers

ral Machinery Engineering,University of Tehran, Karaj,

madi).

All rights reserved.

(Meisterlinget al., 2009; Iriarte et al., 2010; Liu et al., 2010; BörjessonandTufvesson, 2011). Similarly in a number of recent researches, theeconometric approach has been used to identify the relationshipbetween energy consumption fromdifferent inputs and yield valuesof crop productions (Mohammadi et al., 2010; Mousavi-Avval et al.,2011a). Kulekci (2010) applied the stochastic frontier analysistechnique in the CobbeDouglas form to determine the technicalefficiency for a sample of 117 randomly selected sunflower farms inTurkey. This method is parametric and requires a pre-specificationof the functional form and an explicit distributional assumptionfor the technical inefficiency term (Hjalmarsson et al., 1996).

On the other hand, Data Envelopment Analysis (DEA) techniqueis a non-parametric linear programming (LP) based technique offrontier estimation for measuring the relative efficiency ofa number of decision making units (DMUs) on the basis of multipleinputs and outputs (Malana and Malano, 2006). In this case theefficiency of a unit is defined as the ratio of weighted sum of itsoutputs to the weighted sum of its inputs and it is measured ona bounded ratio scale. The weights for inputs and outputs aredetermined to the best advantage for each unit so that to maximizeits relative efficiency (Despotis et al., 2010). Due to the highadvantages of DEA, it has been demonstrated to be effectivefor benchmarking in different systems involving complexinputeoutput relationships (Zhu, 2003).

Nomenclature

BCC BankereCharneseCooper (DEA model)CCR CharneseCoopereRhodes (DEA model)CRS constant returns to scaleDEA data envelopment analysisDLP dual linear programmingDMU decision making unitDRS decreasing returns to scaleEMS efficiency measurement systemsESTR energy saving target ratioFYM farmyard manureIRS increasing returns to scaleLP linear programmingSE scale efficiencyTE technical efficiencyVRS variable returns to scale

Table 1Descriptive statistics for energy inputs and output in barberry production in SouthKhorasan, Iran.

Item (unit) Total energyequivalent

Standarddeviation

Correlation coefficientwith output

Inputs (MJ ha�1)Human labor 3670.46 1780 0.758*Machinery 804.20 698 0.368*Diesel fuel 2659.52 1359 0.427*Total fertilizer 7221.32 3693 0.816*Biocide 193.70 150 0.211**Water for irrigation 4633.89 3336 0.574*Electricity 2024.60 2104 �0.191***Total energy input 21,208.04 9176

Output (kg ha�1)Barberry yield 7349.29 3444Land area (ha) 0.65 0.55

* Indicates significant at 1% level.** Indicates significant at 5% level.*** Indicatessignificant at 10% level.

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132 127

In the last decades, there have been numerous applications of DEAto measure the efficiency in agricultural production systems. Abayet al. (2004) applied the DEA technique to analyze the efficiency ofinput use in tobacco production in Turkey. They used data obtainedfrom 300 farmers in different regions of Turkey. According to theirresults,mean technical efficiencywas found tobe0.456 for all regions,and Eastern and Southeastern Anatolia were relatively moresuccessful regions in terms of input use. In another study, DEA wasapplied to investigate the efficiency of individual farmers and toidentify theefficientunits in citrusproduction inSpain (Reig-Martínezand Picazo-Tadeo, 2004). From this study, they suggested the DEA asan appropriate analytical tool to explore thepossibilities of short-termviability of individual farms, after eliminating the inefficient practices.In another study by Nassiri and Singh (2009), the DEA techniquewassubjected to the data of energy use for paddy production; also, tech-nical efficiency of farmers in different farm sizes was analyzed.Banaeian et al. (2010) applied DEA technique to benchmark theproductive efficiencies of farmers with respect to energy use forwalnut production in Iran. The inputswere energy consumption fromhuman labor, farmyard manure, fertilizers and transportation, andyield as output. Also, they analyzed energy efficiencies of farmers indifferent farm sizes. In another study byMousavi-Avval et al. (2011a),the DEA technique was subjected to the data of energy use for appleproduction in Iran. In this study, the technical, pure technical and scaleefficiencies of farmers were estimated and the productivity perfor-mance of apple producers based on the amount of various energyinputs and output of apple yield was analyzed.

Seedless barberry (Berberis vulgaris (L.)) is a perennial shrubplant belongs to the Berberidaceae family and grows in Asia andEurope; it is a well knownmedicinal plant in Iran and has also beenused as a food additive. Barberry fruit is used in medicine to cureliver, neck and stomach cancer, blood purification and mouth scent(Duke, 1991; Aghbashlo et al., 2008). Barberry cultivation in Iran isconcentrated in the South Khorasan provincewhere environmentalcondition (i.e. hot weather, low relative humidity, water shortageand soil condition) is unfavorable for the growing of other horti-cultural crops. Total barberry fruit production in Iran was about37,000 tones from 11,000 ha land area, in 2008, from which about99% was produced in this province (Anonymous, 2008).

Qaen region is the main center of barberry fruit production inSouth Khorasan province. Barberry production has much greatersocio-economic significance for the study area as compared toother regions of South Khorasan province, since it is a region withsmall-scale farms, high labor potential, and limited alternativeincome sources.

Due to the indispensable function of barberry production in thisregion, efficiency studies play an important role in determiningalternative policies. So, the main objectives of the present studywere to analyze the technical and scale efficiencies of barberryproducers and to compare small and large farms from an energyefficiency point of view. The study has also aimed to segregateefficient farmers from inefficient ones, identify wasteful uses ofenergy from different sources by inefficient farmers and to suggestreasonable savings in energy uses from different inputs.

2. Materials and methods

2.1. Sampling design

Data used in this study were collected from 144 barberryproducers from Qaen region in South Khorasan province, Iran. Inthis study South Khorasan province was chosen as a representativeof the Iranian barberry production enterprises. This province islocated in the east of Iran, within 30� 320 and 34� 500 north latitudeand 57� 570 and 60� 570 east longitude.

For collecting the data, an interview was conducted in theproduction year of 2008/2009. For sampling, stratified randomsampling method was used. The sample size was calculated usingthe Neyman method (Yamane, 1967). So, the research sample sizewas found to be 144; then the 144 barberry producers from 11villages from the target region were randomly selected.

2.2. Energy analysis

A standard procedure was used to convert each agriculturalinput and output into energy equivalent. The inputs were in theform of chemicals, chemical fertilizers, farmyard manure (FYM),diesel fuel, electricity, water for irrigation, human labor or machinepower. The energy equivalent may thus be defined as the energyinput taking into account all forms of energy in agriculturalproduction. The energy equivalents were computed for all inputsand outputs using the conversion factors presented in the previousstudy (Mousavi-Avval et al., 2011a). Multiplying the quantity of theinputs used per hectare with their conversion factors gave theenergy equivalents.

2.3. Descriptive statistics of inputs and output

Descriptive statistics for energy inputs and output in barberryproduction are presented in Table 1. An initial analysis of thesurveyed results demonstrates the substantial variation in the

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132128

total energy input (21,208 MJ ha�1, with the standard deviation of9176) and the production yield (7349 kg ha�1, with the standarddeviation of 3444); indicating that there is ample scope forimproving the efficiency of energy consumption for barberryproduction in the region. Also, the average land area of barberryproduction was found to be 0.65 ha, and it varied from 0.05 to 3 hawith the standard deviation of 0.55. Due to sloping land in theregion and intensive labor requirements, barberry production ismostly done on small areas. However by considering the data, it isevident that, between the farmers, there is a wide variation inenergy use from inputs, the yield level and also the land area ofcultivation. Therefore, in order to analyze the efficiency of farmersin the homogenous groups, graphical distribution of the farm sizeswas considered, and the farms were divided into two size groups(Esengun et al., 2007). The farm size of the first group was lessthan 1 ha (small farms), while, the second group had the farms ofmore and equal to 1 ha (large farms). Subsequently, the non-parametric method of DEA was applied to analyze the efficiency ofbarberry producers in each group. So, energy consumption fromdifferent energy sources including human labor, machinery, dieselfuel, total fertilizers, biocides, water for irrigation and electricity(MJ ha�1), was defined as input variables and the yield value ofbarberry (kg ha�1) was defined as output; also each farmer calleda DMU.

For investigating the correlation between different variables thePearson correlation test was applied. The correlation coefficients ofinput variables with output are presented in the last column ofTable 1. The results revealed that the output variable of barberryyield was correlated with all of the input variables at 10% signifi-cance level. The analysis furthermore revealed that, there was nostatistically significant correlation between input variables.

2.4. Data envelopment analysis (DEA)

In DEA application, a unit can be made efficient either byreducing the input levels and getting the same output (inputorientation), or symmetrically, by increasing the output level withthe same input level (output orientation). The input-orientedanalysis is becoming more common in DEA applications becauseprofitability depends on the efficiency of the operations. In thispaper, we adopt an input-oriented DEA approach for efficiencyestimation. This approach was deemed to be more appropriatebecause there is only one output while multiple inputs are used;also as a recommendation in agricultural production, inputconservation for given outputs seems to be a more reasonable logic(Zhou et al., 2008); so, barberry production yield is hold fixed andthe quantity of source wise energy inputs can be reduced.

In DEA, efficiency is defined as technical efficiency, pure tech-nical efficiency and scale efficiency which can be described asfollow:

2.4.1. Technical efficiency (TE)The technical efficiency of a DMU (e.g. a farmer) is a comparative

measure of how well it actually processes inputs to achieve itsoutputs, as compared to its maximum potential for doing so, asrepresented by its production possibility frontier. The TE score (qc)in the presence of multiple-input and output factor can be calcu-lated by the ratio of sum of weighted outputs to the sum ofweighted inputs as follows (Cooper et al., 2004):

qc ¼ u1y1j þ u2y2j þ/þ usysjv1x1j þ v2x2j þ/þ vmxmj

¼Ps

r¼1 uryrjPmi¼1 vixij

(1)

Considering the DMUj to be evaluated on any trial be desig-nated as DMUo (o¼ 1, 2, ., n), to measure the relative efficiency

of a DMUo based on a series of n DMUs, the model is formulatedas a fractional programming problem as follows (Cooper et al.,2006):

Max : qc ¼Ps

r¼1 uryroPmi¼1 vixio

(2)

S:t: :Ps

r¼1 uryrjPmi¼1 vixij

� 1; j ¼ 1;2;.;n

ur � 0; vi � 0

where n is the number of DMUs in the comparison, s and m thenumber of outputs and inputs, ur (r¼ 1, 2, ., s) the weighting ofoutput yr in the comparison, vi (i¼ 1, 2, ., m) the weighting ofinput xi, and yrj and xij represent the values of the outputs andinputs yj and xi for DMUj, respectively. Eq. (2) can equivalently bereformulated into a linear programming (LP) problem as follows(Cooper et al., 2006):

Max : qc ¼Xs

r¼1

uryro (3)

S:t: :Ps

r¼1uryrj �

Pm

i¼1vixij � 0; j ¼ 1;2;.;n

Pm

i¼1vixio ¼ 1

ur � 0; vi � 0

In reality, the dual linear programming (DLP) problem, due tofewer constraints, is simpler to solve than Eq. (3). Mathematically,the DLP is written in vectorematrix notation (Cooper et al., 2006):

Min : qc (4)

S:t: :Yl � yoXl� qcxo ¼ 0l � 0

where yo is the s� 1 vector of the value of original outputs producedand xo is them� 1 vector of the value of original inputs used by theoth DMU. Y is the s� nmatrix of outputs and X is them� nmatrix ofinputs of all n units included in the sample. l is a n� 1 vector ofweights and qc is a scalar with boundaries of one and zero whichdetermines the technical efficiency score of each DMU. Model (4) isknown as the input-oriented CCR DEA model. It assumes constantreturns to scale (CRS), implying that a given increase in inputswould result in a proportionate increase in outputs.

2.4.2. Pure technical efficiency (PTE)The TE derived from CCR model, comprehends both the tech-

nical and scale efficiencies. So, Banker et al. (1984) developeda model in DEA, which was called BCC model to calculate the PTEscores (qv). The BCC model is provided by adding a restriction of(l¼ 1) in CCR model in DLP form (model (4)), resulted to nocondition on the allowable returns to scale. This model assumesvariable returns to scale (VRS), indicating that a change in inputs isexpected to result in a disproportionate change in outputs. Becausethe VRS analysis is more flexible and envelops the data in a tighterway than the CRS analysis, the PTE score (qv) is generally equal to orgreater than TE score (qc).

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132 129

2.4.3. Scale efficiency (SE)SE relates to the most efficient scale of operations in the sense of

maximizing the average productivity. A scale efficient farmer hasthe same level of technical and pure technical efficiency scores. Itcan be calculated by the relationship between technical and puretechnical efficiency scores as follow (Mousavi-Avval et al., 2011a)

qs ¼ Technical efficiencyPure technical efficiency

¼ qcqv

(5)

where qs denote the scale efficiency score. SE gives quantitativeinformation of scale characteristics. It is the potential productivitygained from achieving optimum size of a DMU. An SE score of 1indicates scale efficient or CRS conditions; however, scale ineffi-ciency is due to the presence of either increasing returns to scale(IRS) or decreasing returns to scale (DRS) conditions.

In order to calculate the efficiencies of farmers and discriminatebetween efficient and inefficient ones, the Microsoft Excel spreadsheet, SPSS 17.0 and EMS software (Scheel, 2000) were applied.

3. Results and discussions

3.1. Inputs and output in different farm sizes

This study was conducted in order to analyze the efficiency ofenergy consumption for barberry production in South Khorasanprovince of Iran. The population investigated was divided into twogroups of small farms (66 farms) and large farms (78 farms). Theamount of inputs and output for small and large farms is compar-atively presented in Table 2. The results revealed that human laborwas used about 1962 and 907 h ha�1 in small and large farms,respectively. Moreover machinery requirements in small farmswere higher than those of large farms. The higher use of humanlabor in small farms was mainly due to the lower level of mecha-nization in these farms. Moreover, higher use of machinery anddiesel fuel can be interpreted by lower field efficiency of agricul-tural machinery in these farms. The use of chemical fertilizers andfarmyard manure in small farms was higher compared to the largefarms. On the other hand, the yield level of barberry fruit in smalland large farms was considered as 7531 and 7008 (kg ha�1),respectively. Totally, the results revealed that, the small farms usedthe inputs more intensive than the large farms. Similar result was

Table 2Quantity of inputs and output in barberry production in different farm sizes in Iran.

Inputs (unit) Small farms Large farms Weighted mean

InputsHuman labor (h) 1962.14a 907.22b 1595.85Machinery (h) 14.42a 9.89b 12.85Diesel fuel (L) 51.75a 38.74b 47.23Chemicals (kg)Herbicides 0.72 0.30 0.58Insecticides 0.36 0.91 0.55

Total fertilizer (kg)Nitrogen 50.56 40.33 47.01Phosphate (P2O5) 100.84 72.11 90.86Potassium (K2O) 0.44 0.25 0.38Sulfur (S) 0.16 0.09 0.14Farmyard manure (kg) 13,957.59 6027.74 11,204.17

Water for irrigation (m3) 5079.63a 3534.23b 4543.03Electricity (kWh) 489.20a 699.98a 562.39

OutputBarberry fruit (kg ha�1) 7530.98a 7007.71b 7349.29Land area (ha) 0.12 1.13 0.65

a,bDifferent letters show significant difference of means at 5% level.

reported in a previous study on energy analysis in corn silage in Iran(Pishgar Komleh et al., 2011).

3.2. Efficiency estimation

The results obtained by the application of input-orientated CCRand BCC DEA models are illustrated in Table 3. It is evident thatwhen the CCR model is applied, 9.6% from small farms and 4% fromlarge farms were found to be efficient; also, 4.3% of farmers had anefficiency score of less than 0.4, while it was found to be 28% forlarge farms. Moreover, the results revealed that, when the BCCmodel is applied, 18.1% from small farms and 30% from large farmshad a pure technical efficiency score of one. The higher percentageof efficient farmers under variable returns to scale assumption (BCCmodel) reflects the fact that, under variable returns to scale, inef-ficient farms are only compared to efficient farms of a similar scalesize; therefore, the technical efficiency is either less than or equal tothe pure technical efficiency of every farm. For this reason, morefarms are efficient under the variable returns to scale formulation.From Table 3 it is evident that, there is awider variation in technicalefficiency of farmers compared to pure technical efficiency, indi-cating that the majority of farmers don’t operate at the mostproductive scale size.

Descriptive statistics for three estimated measure of efficiencyfor small and large farms are presented in Table 4. The resultsrevealed that the average technical (global) efficiency of smallfarms (0.66) was higher than that of large farms (0.50). The widevariation in the technical efficiency of farmers also implies that allthe farmers were not fully aware of the right production techniquesor did not apply them at the proper time in the optimum quantity.

The results also revealed that, the large farms had higher puretechnical efficiency score (0.81 vs. 0.80). Moreover, the mean puretechnical efficiency of both farm groups was considerably higherthan technical efficiency. The high difference between technicaland pure technical efficiency scores indicates their disadvantageousconditions of scale size; so there is not an efficient scale size for thebarberry production and there is a potential productivity earned byachieving the optimal size of barberry production under the study.This is obvious by low level of scale efficiency which was found tobe 0.82 and 0.62 in small and large farms, respectively. Nassiri andSingh (2009) investigated energy efficiency of Indian paddyproducers by applying DEA. They reported that, in small and largefarms, the technical efficiency was 0.74 and 0.62; while the puretechnical efficiency was calculated as 0.82 and 0.79, respectively. Inanother study by Iráizoz et al. (2003), for tomato and asparagusproduction in Spain, the technical, pure technical and scale effi-ciencies were found to be 0.75, 0.80 and 0.94 for tomato productionand 0.81, 0.89 and 0.91 for asparagus production, respectively. Orenand Alemdar (2006) estimated the technical efficiencies of tobacco

Table 3Frequency distribution of technical and pure technical efficiency in different farmsizes of barberry production.

Efficiency(decimal)

Small farms (%) Large farms (%)

Technicalefficiency

Pure technicalefficiency

Technicalefficiency

Pure technicalefficiency

Equal to 1 9.6 18.1 4 300.9 to <1 4.3 6.4 2 20.8 to <0.9 10.6 23.4 e 180.7 to <0.8 10.6 25.5 e 180.6 to <0.7 22.3 21.3 14 300.5 to <0.6 21.3 5.3 18 20.4 to <0.5 17.0 e 34 e

Less than 0.4 4.3 e 28 e

Table 6Comparison analysis of energy inputs and barberry yield for efficient and inefficientfarmers.

Item Superior efficientfarmers

Inefficientfarmers

Difference(%)

Inputs (MJ ha�1)Human labor 2436.16 3707.78 52.2Machinery 686.39 1538.64 124.16Diesel fuel 1397.43 2951.99 111.24

Total fertilizer 3982.09 7160.34 79.81Nitrogen 2521.80 2773.18 9.97Phosphate (P2O5) 1459.47 977.85 �33Potassium (K2O) 0.32 3.30 931.25Sulfur 0.50 0.15 �70Farmyard manure 3218.75 3405.86 5.81

Biocides 7.54 192.53 2453.45Herbicides 5.23 136.36 2507.27Insecticides 2.31 56.17 2331.6

Water for irrigation 2896.66 4397.07 51.8Electricity 650.66 2591.24 298.25Total energy input 15,275.68 22,539.59 47.55

Output (kg ha�1)Barberry yield 9543.33 6732.55 �29.45Land area (ha) 0.58 0.69 15.94

Table 4Descriptive statistics of efficiency scores in different farm sizes of barberryproduction.

Particular Small farms Large farms

Mean SD Mean SD

Technical efficiency 0.66a 0.18 0.50b 0.16Pure technical efficiency 0.80a 0.14 0.81a 0.15Scale efficiency 0.82a 0.16 0.62b 0.19

a,bDifferent letters show significant difference of means at 5% level.

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132130

farms in Southeastern Anatolia using non-parametric method ofDEA. They reported the mean efficiency of tobacco farmers as 0.45and 0.56 for constant and variable returns to scale assumptions,respectively.

3.3. Ranking analysis

The results of standard DEA models divide the DMUs into twosets of efficient and inefficient units; the inefficient units can beranked according to their efficiency scores; while, DEA lacks thecapacity to discriminate among efficient units. A number ofmethods are in use to enhance the discriminating capacity of DEA(Adler et al., 2002). In this study, the benchmarking method wasapplied to overcome this problem. In this method, an efficient unitwhich is chosen as a useful target for many inefficient DMUs and soappears frequently in the reference sets, is highly ranked. So, theefficient DMUs are ranked on the basis of counting the number oftimes they appears in a referent set (Adler et al., 2002). Each set isformed by the efficient DMUs that are similar to the input andoutput levels of inefficient DMUs. Those efficient DMUs that appearmore frequently in the referent set of inefficient DMUs, areconsidered superior because they are not only efficient but are alsoclose to inputeoutput levels of inefficient DMUs in the sample.Considering the results obtained by the study, DMUs 72, 75, 93, 122and 56 appear 102, 56, 40, 26 and 25 times in the referent set,respectively. The results of ranking 10 superior efficient farmers areshown in Table 5. By using these farmers as benchmarks, inefficientfarms can determine which changes in resource usage are neces-sary in order to establish the best practice management andimprove their performance from an energy use efficiency point ofview.

3.4. Energy use by efficient and inefficient farmers

The input use pattern and yield obtained by superior efficientand inefficient ones are compared in Table 6. The results revealedthat the use of all inputs (except for phosphate and sulfur fertil-izers) for superior efficient farmers was less than that of inefficientones. Also the main difference between efficient and inefficientfarmers was in electricity and biocide usage. The use of machinery

Table 5Ranking the superior efficient farmers base on the results of BCC model.

Ranking Farmerno.

Frequency inreferent set

Energy inputs (MJ ha�1) andoutput (kg ha�1)

Total energy input Yield

1 72 102 9911.8 18002 75 56 22,132.2 10,0003 93 40 37,316.3 20,0004 122 26 58,965.2 56,0005 56 25 3479.9 8006 43 21 22,757.3 75007 22 16 5408.8 13338 138 14 7952.0 20009 113 13 4102.3 100010 81 11 10,194.2 1000

and diesel fuel for inefficient farmers was found to be 124.16% and111.24% less than that of efficient farmers, respectively. Also inef-ficient farmers used fertilizers 79.81% greater than efficient ones.

Totally, it is observed that the inefficient farmers used higheramounts of energy from all sources, and thereby, the total energyinput of inefficient farmers was 47.55% more than that of superiorefficient farmers. On the other hand, the yield obtained by ineffi-cient farmers was about 29% less than that of the efficient farmers.This implies that inefficient farmers did not use the resourcesefficiently.

3.5. Setting realistic input levels for inefficient producers

The pure technical efficiency score of a producer that is less thanone indicates that, at present, he/she is using more energy thanrequired from the different sources. Therefore, it is desired tosuggest realistic levels of energy to be used from each source forevery inefficient farmer in order to preventwastage of energywhileholding the same level of output. This can be done by using thevalue of slacks.

In Table 7 the present use of energy, optimum energy require-ment and saving energy from various inputs for small and largefarms based on the results of BCC model are given. As it is seen, thetotal fertilizer and water for irrigation energy requirements werefound to be 8373.8 and 5181.2 MJ ha�1, in small farms, and 5054.5and 3604.9 MJ ha�1 in large farms, respectively, which had thehighest share from total energy requirement in present conditions.Energy consumption by human labor and machinery inputs wasalso calculated as 4512.9 and 903.9 MJ ha�1, in small farms, and2086.6 and 620.4 MJ ha�1 in large farms, respectively. Moreover,total energy input in present condition for small farms(23,854.8 MJ ha�1) was found to be higher than that of large farms(16,232.1 MJ ha�1); while, in target conditions, total energyrequirement was calculated as 20,702.4 and 13,761.2 (MJ ha�1) forsmall and large farms of barberry productions, respectively.Accordingly, total energy saving was found to be 3152.36 and2470.93 MJ ha�1 which contributed to the total energy input by13.2% and 15.2%, for the respective small and large farms.

Singh et al. (2004) concluded that the existing level of produc-tivity could be achieved by 22.3%, 20.8%, 9.8%, 7.1% and 15.9%

Table 7Energy consumption in present and target conditions for different farm sizes ofbarberry production.

Inputs Small farms Large farms

Present(MJ ha�1)

Target(MJ ha�1)

Saving(%)

Present(MJ ha�1)

Target(MJ ha�1)

Saving(%)

Human labor 4512.9 3987.8 11.6 2086.6 2085.8 0.04Machinery 903.9 827.6 8.4 620.4 618.0 0.4Diesel fuel 2913.8 1821.9 37.5 2181.4 1019.0 53.3Biocides 208.1 98.9 52.4 164.4 84.0 48.9Total fertilizer 8373.8 8231.9 1.7 5054.5 4765.8 5.7Water for irrigation 5181.2 4904.6 5.3 3604.9 3419.0 5.2Electricity 1761.1 829.7 52.9 2519.9 1769.7 29.8Total energy 23,854.8 20,702.4 13.2 16,232.1 13,761.2 15.2

Table 8Improvement of energy indices for different farm sizes of barberry production.

Items Unit Small farms Large farms

Presentcondition

Targetcondition

Presentcondition

Targetcondition

Energyproductivity

kgMJ�1 0.32 0.36 0.43 0.51

Specific energy MJ kg�1 3.17 2.75 2.32 1.96

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132 131

reducing the energy input over the actual energy input for wheatproduction in zones 1e5 of Punjab, respectively.

From the results it is evident that, electricity, diesel fuel andbiocide energies had the highest percent saving in both the farmgroups; indicating that the use of these inputs by farmers had thehighest inefficiency. High percentage of electricity, diesel fuel andbiocides energies can be interpreted by the low prices and freelyavailability of these inputs in the surveyed region.

The shares of various sources from total energy saving areillustrated in Fig. 1. It is evident that the highest contribution to thetotal energy saving in small and large farms was 34.6% and 47% bydiesel fuel energy, followed by 29.5% and 30.4% by electricity. Thisindicates that in the case of diesel fuel and electricity inputs there isa great potential to improve the efficiency of energy consumptionin barberry production in surveyed region. Apart from diesel fueland electricity, human labor had the highest potential improve-ment of energy input in small farms; while, in large farms it wasfound to be biocide energy. Mousavi-Avval et al. (2011b) reportedthat about 11.29% from total energy consumption for appleproduction in present condition could be saved; from which thecontributions of electricity and biocide energies were relativelyhigh. Also, Chauhan et al. (2006) reported that about 11.6% fromtotal energy input in paddy production could be saved; also, thefertilizer, diesel fuel and human labor had the highest contributionsto the total energy savings.

Improving energy use efficiency of water pumping systems,employing new irrigation systems and leveling farms properly canbe suggested to avert from electrical energy wastage by inefficientfarmers, and so, improve the energy use efficiency and reduce the

Fig. 1. Distribution of energy saving for different farm sizes of barberry production.

environmental footprints of energy use in barberry production inthe region. Also, applying a better machinery management tech-nique, employing the conservation tillage methods and techno-logical upgrade to substitute fossil fuel inputs with renewableenergy resources may help to reduce diesel fuel and machineryenergies in barberry production.

3.6. Energy productivity improvement

The effect of optimization of energy inputs for barberryproduction on energy indices including energy productivity (i.e.yield obtained to output energy ratio) and specific energy (i.e.output energy to yield ratio) was investigated. The results arepresented in Table 8. As it is seen, energy productivity in presentand target conditions was calculated as 0.32 and 0.36 (kgMJ�1) insmall farms and 0.43 and 0.51 (kgMJ�1) in large farms, respectively.Also specific energy was found to be 3.17 and 2.32 (MJ kg�1) inpresent conditions and 2.75 and 1.96 (kgMJ�1) in target conditions,for small and large farms, respectively.

To sum it up, barberry is a crop with relatively high require-ments on energy inputs. Its fertilizer and irrigation energyrequirements are high and it needs a high amount of diesel fuelconsumption. Efficient use of the energy resources is necessary interms of increasing crop production, productivity of resource usage,competitiveness of agriculture, and environmental sustainability.There is a need to reduce dependency on increasingly scarce energyresources, and understand the energy use efficiency in alternatefarming systems. The methodology presented in this studydemonstrates how energy use efficiency in barberry productionmay improve by applying the operational management tools toassess the performance of farmers. On an average, considerablesavings in energy inputs may be obtained by adopting the bestpractices of high-performing ones in crop production process.Adoption of more energy-efficient cultivation systems would helpin energy conservation and better resource allocation. Accordingly,new policies and programs to promote sustainable barberryproduction should be developed and implemented. Some strategiessuch as use of advanced technologies (technologies for applying therenewable energies such as wind or solar, energy-efficient farmmachinery and irrigation systems, low or no till faming, gravity-feddrip and sprinkler irrigation, precise application practices such aslevel basins and surge irrigation, canal lining or piped waterdeliveries, etc.), reduced allocation to farmers, and water savingirrigation practices such as alternative wet and dry irrigation,improving timing, amount, and reliability of water applications, toincrease yield level and to improve the quality of product, should bedeveloped in order to increase the technical efficiency of barberryproduction in the region. Moreover, the farmers should be trainedwith regard to the optimal use of inputs, especially, fertilizers,chemicals and irrigation water as well as employing the newproduction technologies. The local agricultural institutes in theregion have an important role in these cases to establish the moreenergy efficient and environmentally healthy barberry productionsystems in the region.

S.H. Mousavi-Avval et al. / Journal of Cleaner Production 27 (2012) 126e132132

4. Conclusions

In this study the input-oriented CCR and BCC DEA models weresubjected to the data of energy use in different barberry productionenterprises. The average farm size of studied farms was 0.65 ha;accordingly, the efficiencies of farmers in small and large farm sizeswere analyzed. Based on the results of the study the followingconclusions are drawn:

1. Total energy input in small and large farms was found to be23,854.8 and 16,232 MJ ha�1, respectively. Also, the small farmshad the higher yield level compared to the large farms.

2. The results showed substantial production inefficiency forfarmers; so that, only 18.1% from small farms and 30% fromlarge farms were recognized as efficient based on the variablereturns to scale assumption. Also, the technical efficiencydecreases, as farm size increases.

3. Large farms showed higher inefficiency in energy uses; so thatfrom total energy input in small and large farms, the reductionpotentials of 13.2% and 15.2% could be achieved withoutaffecting the yield level, provided that all of the farmers oper-ated efficiently and assuming no other constraints on thisadjustment.

4. Improving energy use efficiency of water pumping systems,applying a better machinery management technique, employ-ing the conservation tillage methods and adoption of inte-grated pest management technique may be also the pathwaysto made barberry production systems more energy efficientand environmentally healthy.

5. Totally, DEA is very suitable to analyze these data and extractmany distinctive features of their practices. It helped in findingthe wasteful uses of energy by inefficient farmers and sug-gested necessary quantities of different inputs to be used byeach inefficient farmer from every energy source.

Acknowledgment

The financial support provided by the University of Tehran, Iran,is gratefully acknowledged.

References

Abay, C., Miran, B., Gunden, C., 2004. An analysis of input use efficiency in tobaccoproduction with respect to sustainability: the case study of Turkey. Journal ofSustainable Agriculture 24, 123e143.

Adler, N., Friedman, L., Sinuany-Stern, Z., 2002. Review of ranking methods in thedata envelopment analysis context. European Journal of Operational Research140, 249e265.

Aghbashlo, M., Kianmehr, M.H., Samimi-Akhijahani, H., 2008. Influence of dryingconditions on the effective moisture diffusivity, energy of activation and energyconsumption during the thin-layer drying of berberis fruit (Berberidaceae).Energy Conversion and Management 49, 2865e2871.

Anonymous, 2008. Annual Agricultural Statistics. Ministry of Jihad-e-Agriculture ofIran. www.maj.ir.

Banaeian, N., Zangeneh, M., Omid, M., 2010. Energy use efficiency for walnutproducers using data envelopment analysis (DEA). Australian Journal of CropScience 4, 359e362.

Banker, R., Charnes,A., Cooper,W.,1984. Somemodels forestimating technicalandscaleinefficiencies in data envelopment analysis. Management Science 30, 1078e1092.

Börjesson, P., Tufvesson, L.M., 2011. Agricultural crop-based biofuels e resourceefficiency and environmental performance including direct land use changes.Journal of Cleaner Production 19, 108e120.

Chauhan, N.S., Mohapatra, P.K.J., Pandey, K.P., 2006. Improving energy productivityin paddy production through benchmarking e an application of data envel-opment analysis. Energy Conversion and Management 47, 1063e1085.

Cooper, W.W., Seiford, L.M., Tone, K., 2006. Introduction to Data EnvelopmentAnalysis and its Uses. Springer, New York.

Cooper, W.W., Seiford, L.M., Tone, K., 2004. Data Envelopment Analysis, AComprehensive Text with Models, Applications, References and DEA-SolverSoftware. Kluwer Academic Publishers, Massachusetts, USA.

Despotis, D.K., Stamati, L.V., Smirlis, Y.G., 2010. Data envelopment analysis withnonlinear virtual inputs and outputs. European Journal of Operational Research202, 604e613.

Dovì, V.G., Friedler, F., Huisingh, D., Klemes, J.J., 2009. Cleaner energy for sustainablefuture. Journal of Cleaner Production 17, 889e895.

Duke, J.A., 1991. Handbook of Medicinal Herbs, eighth ed. CRC Press Inc., Florida.Esengun, K., Gündüz, O., Erdal, G., 2007. Inputeoutput energy analysis in dry apricot

production of Turkey. Energy Conversion and Management 48, 592e598.Hjalmarsson, L., Kumbhakar, S., Hesmati, A., 1996. DEA, DFA and SFA: a comparison.

Journal of Productivity Analysis 7, 303e327.Iráizoz, B., Rapún, M., Zabaleta, I., 2003. Assessing the technical efficiency of

horticultural production in Navarra, Spain. Agricultural Systems 78, 387e403.Iriarte, A., Rieradevall, J., Gabarrell, X., 2010. Life cycle assessment of sunflower and

rapeseed as energy crops under Chilean conditions. Journal of CleanerProduction 18, 336e345.

Khan, S., Khan, M.A., Hanjra, M.A., Mu, J., 2009. Pathways to reduce the environ-mental footprints of water and energy inputs in food production. Food Policy34, 141e149.

Kulekci, M., 2010. Technical efficiency analysis for oilseed sunflower farms: a casestudy in Erzurum, Turkey. Journal of the Science of Food and Agriculture 90,1508e1512.

Liu, Y., Høgh-Jensen, H., Egelyng, H., Langer, V., 2010. Energy efficiency of organicpear production in greenhouses in China. Renewable Agriculture and FoodSystems 25, 196e203.

Malana, N.M., Malano, H.M., 2006. Benchmarking productive efficiency of selectedwheat areas in Pakistan and India using data envelopment analysis. Irrigationand Drainage 55, 383e394.

Meisterling, K., Samaras, C., Schweizer, V., 2009. Decisions to reduce greenhousegases from agriculture and product transport: LCA case study of organic andconventional wheat. Journal of Cleaner Production 17, 222e230.

Mohammadi, A., Rafiee, S., Mohtasebi, S.S., Rafiee, H., 2010. Energy inputseyieldrelationship and cost analysis of kiwifruit production in Iran. Renewable Energy35, 1071e1075.

Mousavi-Avval, S.H., Rafiee, S., Jafari, A., Mohammadi, A., 2011a. Energy flowmodeling and sensitivity analysis of inputs for canola production in Iran. Journalof Cleaner Production 19, 1464e1470.

Mousavi-Avval, S.H., Rafiee, S., Mohammadi, A., 2011b. Optimization of energyconsumption and input costs for apple production in Iran using data envelop-ment analysis. Energy 36, 909e916.

Nassiri, S.M., Singh, S., 2009. Study on energy use efficiency for paddy crop usingdata envelopment analysis (DEA) technique. Applied Energy 86, 1320e1325.

Oren, M.N., Alemdar, T., 2006. Technical efficiency analysis of tobacco farming inSoutheastern Anatolia. Turkish Journal of Agriculture and Forestry 30, 165e172.

Pishgar Komleh, S.H., Keyhani, A., Rafiee, S., Sefeedpary, P., 2011. Energy use andeconomic analysis of corn silage production under three cultivated area levelsin Tehran province of Iran. Energy 36, 3335e3341.

Reig-Martínez, E., Picazo-Tadeo, A.J., 2004. Analysing farming systems with dataenvelopment analysis: citrus farming in Spain. Agricultural Systems 82, 17e30.

Scheel, H., 2000. EMS: Efficiency Measurement System Users Manual, Ver. 1.3,Operations Research and Wirtschaftsinformatik. University of Dortmund,Dortmund, Germany. http://www.wiso.uni-dortmund.de/lsfg/or/scheel/ems.

Singh, G., Singh, S., Singh, J., 2004. Optimization of energy inputs for wheat crop inPunjab. Energy Conversion and Management 45, 453e465.

Yamane, T., 1967. Elementary Sampling Theory. Prentice Hall Inc., Englewood Cliffs,NJ, USA.

Zhou, P., Ang, B.W., Poh, K.L., 2008. A survey of data envelopment analysis in energyand environmental studies. European Journal of Operational Research 189,1e18.

Zhu, J., 2003. Quantitative Models for Performance Evaluation and Benchmarking:Data Envelopment Analysis with Spreadsheets and DEA Excel Solver. KluwerAcademic Publishers, Boston.