The Spread of Pesticide Practices Among Cost-Efficient Farmers

Download The Spread of Pesticide Practices Among Cost-Efficient Farmers

Post on 20-Jan-2017




2 download

Embed Size (px)


<ul><li><p>The Spread of Pesticide Practices Among Cost-EfficientFarmers</p><p>Jean-Philippe Boussemart &amp; Herv Leleu &amp; Oluwaseun Ojo</p><p>Received: 19 December 2011 /Accepted: 3 March 2013 /Published online: 15 March 2013# Springer Science+Business Media Dordrecht 2013</p><p>Abstract The purpose of this paper is to analyse the spreadof pesticide practices with respect to crop production(wheat, barley, and rapeseed) in French agriculture and toevaluate the potential pesticide reductions for cost-efficientfarmers. This is made possible by conducting a double-stepanalysis based on non-parametric robust technology. First,from a panel data of 650 farms over a 12-year period locatedin the French department of Meuse, we selected the cost-efficient farms thanks to a robust free disposal hull (RFDH)technology. A second RFDH frontier analysis was run onlyon the selected cost-efficient farms thus enabling us toreveal the units which minimize the pesticide use per hectarewhile maintaining constant yields. Therefore, all the differ-ent total cost-efficient practices among farmers were evalu-ated in terms of pesticide per hectare and the minimum useswere selected. Our main conclusion is that the pesticidereductions per hectare for the cost-efficient farms couldreach 24 %.</p><p>Keywords Pesticide . Cost efficiency . Agriculture .</p><p>Environmental performance . Robust free disposal hull</p><p>1 Introduction</p><p>As a major provider of environmental services, agricultureplays important roles in carbon sequestration, flood control,groundwater recharge, soil conservation, biodiversitypreservation, open space, scenic vistas, isolation from con-gestion, and purifying water, soil, and air. These coveralmost all ecological services provided by natural ecosys-tems, including provisioning services, regulating services,supporting services, and cultural services [1]. Unfortunately,most are not recognized and are unremunerated. On theother hand, unlike natural ecosystems that produce positiveecological services only, agro-ecosystems also contribute tonegative environmental externalities: greenhouse gas(GHG) emissions, soil erosion, reduction in biodiversity,wildlife habitat destruction, less attractive rural landscapesfrom specialized crop cultivation, nutrient, and pesticiderunoff [2, 3].</p><p>From the economic point of view, the use of pesticides isbased on the three-legged supports of efficiency in produc-tion: the increase in production of crops, the increase inquality of production, and the reduction in agriculturallabour and energy expenses [4]. It is important to note thatover the last 60 years, farmers and growers have changedthe way they produce food in order to meet the expectationsof consumers, governments, and more recently, food pro-cessors and retailers. In doing so, they have made manychanges to the way they farm, including the intensive use ofpesticides. This has been done principally in order to pre-vent or reduce agricultural losses to pests, resulting in im-proved yield and greater availability of food at a reasonableprice, all year round [5]. This belief is still widely shared byfarmers, although society, environmentalists, consumers,and public health professionals increasingly debate itsserious social, environmental, and health impacts [6]. It issimply due to the fact that excess pesticide use forms a</p><p>J.-P. Boussemart (*)LEM-ISEG School of Management and University of Lille,3 rue de la Digue,59000 Lille, Francee-mail:</p><p>H. Leleu :O. OjoCNRS-LEM and ISEG School of Management,3 rue de la Digue,59000 Lille, France</p><p>H. Leleue-mail:</p><p>O. Ojoe-mail:</p><p>Environ Model Assess (2013) 18:523532DOI 10.1007/s10666-013-9363-5</p></li><li><p>typical case of negative externality, where one or more pro-ducers are the sources, and one or more individuals are thereceivers of the externalities [7, 8]. Irrespective of the envi-ronmental disadvantages attached to lots of pesticide usage,farmers may probably use it since they hope that pesticideuse will improve their revenue. This best economical use ofpesticide does not necessarily converge to the interest of thesociety which entails the minimization of pesticide perhectare (per ha in the remaining text).</p><p>Therefore in taking a decision as to the quantity of aproduct to apply, normally a farmer makes the evaluationin relation to the marginal productivity and the privatemarginal cost of using it. However, this may not be the bestresult from the perspective of social and even individualwell-being in the long term, since the individual marginalcost or marginal benefit may ignore effects to human healthand that of the ecosystems, as well as the impacts of these onthe health system and on society as a whole. Hence, themarginal cost of the use of pesticides by the farmer, whichcomprises items such as the price of raw material, cost oflabour of the person applying the pesticide, and the materialused in the application, does not frequently include thedamage done to fauna and flora, to water and soil quality,and more importantly, to human health [9].</p><p>Thus, in order to satisfy continued growth in food de-mand, it requires a sustainable use of capital and land whilemeeting productivity goals. This agrees with the context ofthe agreement of about 50 % reduction in pesticide usesaccording to the accords du Grenelle de lenvironnementin France. More precisely, the objective of this research is toevaluate the differences in pesticide practices amongfarmers in order to select the best practice of pesticide use.This brings to mind that it is very possible for farmers to betotally cost-efficient with either more or less pesticide use,dependent on the substitution possibilities between land andchemical inputs (pesticides, fertilizers). It is worthy of noteto mention here that less intensive pesticide use is often arational strategy when sufficient land is available but incontrast, its more intensive use is likely when productiveland is absolutely or relatively scarce [10]. This corroboratesthe fact that although European farmers are constrained bythe European legislation, there are still some ways to reducepesticide use thanks to good agricultural practices (GAPs)adoption.</p><p>It is therefore now admitted that severe and long-termpressure exerted by government regulations such as thecurrent EU legislation on the Directive 2009/128/EC asregards the sustainable use of pesticide and the new regula-tion (EC 1107/2009) represent a significant incentive toreduce pollution. The European Commission (EC) is pro-moting low pesticide-input farming in Member States andindividual governments will be expected to create the nec-essary conditions for farmers to implement Integrated Pest</p><p>Management (IPM). IPM relies on minimizing pesticide usethrough the complementary adoption of alternative methodsto control pests, diseases, and weeds. Community-widestandards for IPM are being developed and this will becomemandatory across the EU from 2014. The aim of theEuropean Parliament in the short to medium term is thatthe use of pesticides in farming should follow a decliningtrend. The percentage of land cultivated with reduced or lowpesticide-input cropping systems, sometimes called integratedproduction, is therefore expected to increase very significantlyas noted by the Thematic Strategy on the Sustainable Use ofPesticides [11].</p><p>Most of this research work has therefore been done on theenvironmental external costs of pesticide use in Germany,the Netherlands, the Philippines, Italy, France, Denmark, theUK, the US and China [12]. As there are no standardframeworks and methods for assessment, the results cannoteasily be compared. To analyse technologies and cost effi-ciencies, a variety of alternative methods have also beendeveloped in the literature. In addition to deterministic andstochastic parametric frontiers, several non-parametric ref-erence technologies have been suggested, including dataenvelopment analysis (DEA; see, for example, [13]) andthe non-convex free disposal hull (FDH) reference technol-ogy introduced by Deprins et al. [14].</p><p>Not surprisingly, several recent studies have used thesemethodologies to analyse the efficiencies of different orga-nizations [1518]. However, most of these researches havebeen based on either stochastic frontier approaches or non-parametric methods such as DEA or FDH. Based on theimportance of the underlying reference technology, the pur-pose of this paper is to add to the evolving literature onpesticide practices evaluation by studying the cost efficiencyof French farmers that produce wheat, barley, and rapeseedon 650 farms. This entails the use of panel data from laMeuse (a French department) over a 12-year period (19922003). Both temporal and spatial dimensions of the sampleallow us to test the robustness of the empirical results. Incontrast to DEA framework, our study gives priority to therobust free disposal hull (RFDH) approach which presentsthe advantage to compare an evaluated farmer to a realobserved practice by relaxing the convexity assumption ofthe production frontier.</p><p>In view of this, we conduct a double-step analysis basedon RFDH technology devised by Cazals, Florens, and Simar[19]. It is well known that deterministic approaches likeFDH or DEA are very sensitive to outliers that may beselected as referents for estimating efficiency. RFDH con-siders the potential presence of outliers thanks to MonteCarlo simulations which allow a multiple comparison of afarm to a large number of randomized referent sub-samplesinstead of a single comparison to the whole sample as in theusual FDH approach. This prevents the possibility of</p><p>524 J.-P. Boussemart et al.</p></li><li><p>comparing a farm to an outlier. The final efficiency score isestimated by the average of the sub-samples scores. In thefirst step, the cost-efficient farms are selected using thisRFDH technology. In the second step, another RFDH fron-tier analysis is run but only on the cost-efficient farms thatwere selected from the first step, thus revealing the units thatminimize the pesticide use per ha while maintaining con-stant yields. Therefore, in terms of pesticide use per ha, allthe different total cost-efficient practices among farmers areevaluated with the best ones selected.</p><p>The remaining part of this paper is therefore organized asfollows. In the next section, we give the methodology forRFDH and state its relevance to this paper while section 3gives the details on the computation of cost efficiencymeasures for our empirical applications. Lastly, section 4summarizes our conclusions.</p><p>2 The Robust Free Disposal Hull Model</p><p>The methodology used in this paper is introduced sequen-tially. First we develop the FDH cost frontier which aims atselecting the cost-efficient farms. Second the technical fron-tier which selects the best practice of pesticide uses per haamong cost-efficient farms is revealed. Lastly, we state theRFDH framework which circumvents the sensitivity prob-lem of the frontier to outliers which is the main drawback ofthe traditional FDH.</p><p>2.1 The Cost Free Disposable Hull (FDH) Frontier: Aimsat Selecting the Cost-Efficient Farms</p><p>Free disposable hull (FDH) is a well-known empirical ap-proximation of the production possibility set, which is basedon minimal assumptions concerning the properties of thetrue but unobservable production set. In contrast to thepopular DEA model, FDH is not restricted to convex tech-nology but only compares evaluated decision making units(DMUs) to others by rejecting both additivity and divisibil-ity assumptions of the production possibility set. This isparticularly convenient since it is frequently difficult to finda good theoretical or empirical justification for convexity1</p><p>(see e.g. [20]).Since production technologies are not always known,</p><p>inefficiencies must be measured relative to some cost fron-tier which is estimated from the data. Thus, measurementsof inefficiency are really measures of the deviations of costsor input usage away from some minimal levels found in thedata rather than from any true technologically based</p><p>minima. The differences among techniques found in theefficiency literature largely reflect differing maintained as-sumptions used in estimating the frontiers.</p><p>Let us consider that K DMUs are observed and we denotethe associated index set by K 1; . . . ;Kf g. We also assumethat DMUs face a production process with M outputs andN inputs and we define the respective index sets of out-puts and inputs as M 1; . . . ;Mf g and 1; . . . ;Nf gwhere y = (y1, , yM) Z R+</p><p>M x = (x1, , xN) Z R+N and</p><p>w = (w1, , wN) Z R+N are respectively the vector of</p><p>output quantities, input quantities and input prices. Theproduction cost is equal to C=wxT where the superscript Tdenotes a transposed vector.</p><p>We begin by introducing the assumptions on the produc-tion possibility set (PPS) of all feasible output vectors with acost C and which is defined as follows:</p><p>PPS C; y 2 R1M : y can be produced at cost C </p><p>1Now, we suppose that the technology obeys the follow-</p><p>ing axioms of the FDH:</p><p>A1: 0; 0 2 PPS; 0; y 2 PPS ) y 0 , that is, no freelunch or no production of output(s) without input(s);</p><p>A2: the set AC u; y 2 PPS : u Cf g of dominatingobservations is bounded by C Z R+, that is infiniteoutputs cannot be obtained from a finite cost level;</p><p>A3: PPS is closed;A4: for all (C, y) PPS, and all (u, v) Z R+1+M, we have</p><p>C;y u;v ) u; v 2 PPS (free disposabilityof input-cost and outputs). In words, if it costs C toproduce y then it is feasible to produce less than y atthe same level of cost C or to produce an equal outputamount y at a higher cost than C. Intuitively, wastesare always feasible and so producers can freely dis-pose of their productions.</p><p>We now introduce the distance function to compute theefficiency scores as the distance to any DMU in the PPS tothe FDH frontier. We select an input-cost-oriented radialefficiency measure defined by:</p><p>D!</p><p>FDH C; y Min d 2 R : dC; y 2 PPSf g 2The optimization program in (2) can be solved using</p><p>alternative approaches. Traditionally, following Deprins etal. [14], a mixed-integer program (MIP) is solved to com-pute FDH efficiency scores. However we prefer to followAgrell and Tind [21] and Leleu [22] to derive LinearPrograms that will be used in solving (2). Indeed LP ismuch more efficient than MIP to solve the optimizationprogram in (2). While FDH models are generally consideredas non-convex models they could however be solved withtraditional LP solvers which also give a dual economic</p><p>1 The convexity assumption has often been questioned because thedivisibility of inputs and outputs are not always possible especially inagriculture.</p><p>The Spread of Pesticide Practices Among Cost-Efficient Farmers 525</p></li><li><p>interpretation to the FDH technology in terms of shadowprices. Following Leleu [22], the input cost inefficiency fora DMU j with a production plan (Cj, yj) is computed via thefollowing LP program:</p><p>minhk ;zk</p><p>d Pk2K</p><p>hk</p><p>s:t: zk ymk ymj </p><p> 0 8m 2 M; 8k 2 KzkCk hkCj 8k 2 KPk2K</p><p>zk 1zk 0 8k 2 Khk 0 8k 2 K</p><p>3</p><p>The...</p></li></ul>