An assessmentof the total external costs ofUK agricultureJ.N.Prettya,*, C. Brettb, D. Geec, R.E. Hinea, C.F. Masond,J.I.L. Morisond, H. Ravene, M.D.Raymentf, G. van der BijlgaCentre for Environment and Society, University of Essex, Colchester CO4 35Q, UKbDepartments of Economics, University of Essex, UKcEuropean Environment Agency, Kongens Nytorv 6, DK-1050, Copenhagen,DenmarkdDepartment of Biological Sciences, University of Essex, UKeAratornish, Morvern, by Oban, Argyll, PA34 5UZ, UKfRoyal Society for the Protection of Birds, The Lodge, Sandy, Beds SG19 2DL, UKgCentre for Agriculture and the Environment (CLM), PO Box 10015, 3505 AA Utrecht, NetherlandsReceived16February2000;receivedinrevisedform12May2000;accepted30June2000AbstractThis trans-disciplinary study assesses total external environmental andhealth costs ofmodernagricultureintheUK. Awiderangeof datasetshavebeenanalysedtoassesscostdistributionacrosssectors.WecalculatetheannualtotalexternalcostsofUKagriculturein1996 to be 2343 m (range for 19901996: 11493907 m), equivalent to 208/ha of arable andpermanentpasture. Signicantcostsarisefromcontaminationofdrinkingwaterwithpesti-cides (120 m/year), nitrate (16 m), Cryptosporidium (23 m) and phosphate and soil (55 m),from damage to wildlife, habitats, hedgerows and drystone walls (125 m), from emissions ofgases(1113m),fromsoilerosionandorganiccarbonlosses(106m),fromfoodpoisoning(169m),andfrombovinespongiformencephalopathy(BSE)(607m).Thisstudyhasonlyestimatedthoseexternalitiesthatgiverisetonancialcosts,andsoislikelytounderestimatethe total negative impacts of modern agriculture. These data help to identify policy priorities,particularlyoverthemost ecient waytointernalisetheseexternal costs intoprices. Thiswould imply a redirection of public subsidies towards encouraging those positive externalitiesunder-provided in the market place, combined with a mix of advisory and institutionalmechanisms, regulatoryandlegal measures, andeconomicinstruments tocorrect negative0308-521X/00/$-seefrontmatter # 2000ElsevierScienceLtd.Allrightsreserved.PI I : S0308- 521X( 00) 00031- 7AgriculturalSystems65(2000)113136www.elsevier.com/locate/agsy* Corresponding author. Fax: +44-1206-87-34-16.E-mail address: jpretty@essex.ac.uk (J.N. Pretty).externalities. Further work examining the marginal costs and benets of UK agriculture wouldhelp to inform future policy development. #2000 Elsevier Science Ltd. All rights reserved.Keywords:Externalities; Agriculture; Waterpollution; Health; Pesticides; Biodiversity; Foodpoisoning;Policies1. Denitionandconcept of externalitiesFarmersintheUKhavebeenhighlysuccessfulatincreasingfoodproductioninthe 20th century. Compared with 1950, per hectare yields of wheat, barley, potatoesand sugar beet have tripled, while milk yields per cow have more than doubled. Buttheseremarkableachievementshavealsobroughtcostlyenvironmental,healthandsocialproblems(ConwayandPretty,1991;Pretty,1995,1998;Mason,1996;EEA,1998;Krebs etal.,1999).Most economic activities aect the environment, either through the use of naturalresources as an input or by using the `clean' environment as a sink for pollution. Thecostsofusingtheenvironmentinthiswayarecalledexternalities,becausetheyareside eects of the economic activity and their costs are not part of the prices paid byproducers or consumers. Whensuchexternalities arenot includedinprices, theydistort themarket byencouragingactivities that arecostlytosocietyevenif theprivatebenetsaresubstantial(BaumolandOates,1988;PearceandTurner,1990;Lewis, 1996;EEA,1998; Brouwer,1999; Pretty et al.,1999).An externality is any action that aects the welfare of or opportunities available toanindividualorgroupwithoutdirectpaymentorcompensation,andmaybeposi-tiveor negative. The types of externalities encounteredintheagricultural sectorhave ve features: (1) their costs are often neglected; (2) they often occur with a timelag;(3)theyoftendamagegroupswhoseinterestsarenotrepresented;(4)theiden-tityoftheproduceroftheexternalityisnotalwaysknown; and(5)theyresult insub-optimaleconomic and policy solutions.2. Costingnegativeagricultural externalitiesAlthough several attempts have been made to put a cost on some of the pollutionarising fromagriculture in the USA andEurope,it has proven dicult to do. First,it isnecessarytoknowabout thevalueof nature'sgoodsandservices, andwhathappenswhenthesearelost. Thecurrent systemofeconomiccalculationsgrosslyunderestimatesthecurrentandfuturevalueofnatural capital (Abramovitz, 1997;Costanza et al.,1997; Daily, 1997).Second is the diculty of putting a value on non-market goods. How do we value,forexample, skylarkssingingonasummer'sday, oralandscapewithhedgerowsandtrees, orawatershedproducingcleanwater?Environmental economistshavedeveloped methods for assessing people's stated preferences for environmentalgoods through hypothetical markets (Willis et al., 1993; Hanley et al., 1998;114 J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136Brouwer, 1999). The value of nature's goods and services is represented to an extentbypeople'swillingness-to-pay(WTP)forthem,andthispermitsestimationofben-etsforegone,thecorrecteconomicwelfaremeasure.However,inviewofscienticandeconomic uncertainties about many agriculturally relatedproblems, we usenancial coststohelptoovercomeuncertaintieswithvaluation. Thisrestrictstherangeofimpactsweareabletovalue,thoughitcanhelptoovercometheproblemofscienticuncertaintywhereexpenditureisincurredinrelationtoaspecicagri-culturalissue.Thisapproachdoesnot,therefore,actuallyvaluetheexternality,butusesas a proxy the expenditure which societyincurs in dealing withthatexternality(Bailey et al.,1999; Hanley and Oglethorpe,1999;Hill and Crabtree, 2000).Therehavebeenseveral studies ontheexternal costs of modernagricultureinGermany,Netherlands, UK and the USA (Pimentel et al.,1992, 1995; Evans, 1995,1996;Steineretal.,1995;Davisonetal.,1996;FleischerandWaibel,1998;WaibelandFleischer, 1998; Baileyet al., 1999; Ribaudoet al., 1999). Thesesuggest thattotal external costsaresome$81117/haofarableandpermanentpastureinGer-many(onlypesticidesandgaseousemissionscosted)andtheUSA,risingto$112274forarablelandonly(Prettyet al., 2000). Forseveral reasons, however, thesedata are not wholly comparable in their original form, and methodological concernshave been raised about some studies (Bowles and Webster, 1995; Crosson, 1995; vander Bijl and Bleumink, 1997;Pearceand Tinch,1998)1.The need for estimates of externality costs occurs at two levels. The rst is the levelat whichnational andinternational policystrategiesaredeveloped. Here, thereisneed forestimates of totalcosts,bothoverall and bytypeof externality.Theseesti-mates provide a broad guide for policy emphases. Areas in which costs appear to begreatest are then obviouscandidates for policyemphasesand furtheranalyses.The secondlevel is that of particular policies, programmes, or projects. Here,estimates of social costs and benets, in the form of cost-benet or cost-eectivenessstudies, can help guide decisions, for example, about which agri-environmentalinitiatives are best suited to reducing externalities. Hanley et al. (1999)recentlyreviewedandsummarisedadozensuchcost-benet studies of UKagri-environmental schemes.Suchstudiesareextremelyuseful inhelpingtojudgewhe-therthecostsassociatedwithparticularpoliciesorprogrammesarewarrantedintermsoftheirsocial benets. However, asWhitby(2000)hasrecentlynoted, mostevaluations of UK agri-environmental schemeshave not actually been able to valuebenetsatthemargin. Inotherwords, evenifaschemehasproducedmoresocialbenetsthancostsinaparticulararea,itisstillunknownhowreducingorexpand-ing the scheme would aect benets and costs. Oglethorpe et al. (2000) have recently1Some critiques of earlier studies on externalities have noted that several eects could not beassessedinmonetaryterms,whilstothershaveappearedtobemorearbitrary(e.g.thecostofbirddeathsintheUSA($2billion)isarrivedatbymultiplying67millionlossesby$30abird;Pimenteletal.,1992).TheDavisonet al. (1996) studyonNetherlandsagriculturewasevenmorearbitrary, asit addedthecostsfarmers would incur to reach stated policy objectives, and these were based on predicted yield reductionsof1025%arisingfromneithercheapnorpreferabletechnologies, whichledtoalargeoverestimateofenvironmental damage (c.f. van der Bijl and Bleumink, 1997).J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136 115demonstratedhowatruemarginalanalysiscanbeusedtoaidpolicydevelopment,usingthe example of a case studyin Scotland.Thisstudyisfocusedoninformationforpoliciesat therst level, wherebroadpolicystrategies are developed. We address the policyimplications of this studybrieyattheendofthearticle,andinmoredepthinacompanionarticle(Prettyetal., 2000). Thereis needfor better informationat bothlevels, however, toguideeectivelythefutureroundsof agricultural policyreformintheUK, throughoutEurope,and elsewherein industrialisedcountries.3. Frameworkfor assessingthenegative externalities of UKagricultureInthis study, we use aframeworkof sevencost categories toassess the totalenvironmentalandhealthcostsofUKagriculture.Twotypesofdamagecosthavebeen estimated: (1) the treatment or prevention costs (those incurred to clean up theenvironment and restore human health to comply with legislation or to return thesetoanundamagedstate); and(2) the administrationandmonitoringcosts (thoseincurred by public authorities and agencies for monitoring environmental, food andhealthparameters). Wehaveestimatedonlythoseexternalities whichgiverisetonancial costs.Thisframeworkincludesonlyexternalcosts,i.e.thecostsincurredbytherestofsocietyfortheactionsoffarmers.Additionalprivatecostsbornebyfarmersthem-selves arenot included, suchas fromincreasedpest or weedresistancefromtheoveruseofpesticides, orfortrainingintheuse, storageanddisposal ofpesticides.However, thereremaindistributional problems, e.g. insect outbreaksarisingfrompesticide overuse can aect all farmers, even thosenot using pesticides.We have also not yet measured the positive externalities (the benecial side-eects)createdbyfarmingandencouragedbycertainpolicies(OECD, 1997a; LobleyandPotter, 1998; HanleyandOglethorpe, 1999; vanHuyelenbroekandWhitby, 1999;DarlingandTopp, 2000). Theseinclude: landscapeandaestheticvalue; recreationandamenity; wateraccumulationandsupply; nutrient recyclingandxation; soilformation; wildlife, includingagriculturallybenecial organisms; stormprotectionand ood control; and carbon sequestration by trees and soils. These are likely to besubstantial (Daily,1997; Smith et al.,1998; Hanley and Oglethorpe, 1999).Table1summarisestheannual total external environmental andhealthcostsofUK agriculture. A conservative estimate puts these at 2343 m for 1996 alone (rangefor 19901996: 11493907 m). These agricultural externalities can be expressed in avariety of ways:1. externalities comprise 89%of average net farmincome (2.62billion), and13% of averagegross farm returns (17.46billion)for the 1990s;2. externalitiesarisingfromall 11.28mhaofarablelandandpermanentgrass-land (butnot rough grazings) average208/ha/year;3. externalitiesarisingfromarablefarmingalone(1048m)resultinanaveragecost of 229/ha of arable land (4.58 m ha) arable farming costs are taken to116 J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136Table1The annual total external costs of UK agriculture, 1996 (range values for 19901996)aCost category UK( million)Rangeb( million)1. Damage to natural capital water1a. Pesticides in sources of drinking water 120 841291b. Nitrate in sources of drinking water 16 8331c. Phosphate and soil in sources of drinking water 55 22901d. Zoonoses (esp. Cryptosporidium) in sources of drinking water 23 15301e. Eutrophication and pollution incidents (fertilisers, animal wastes,sheep dips)6 471f. Monitoring and advice on pesticides and nutrients 11 8112. Damage to natural capital air2a. Emissions of methane 280 2483762b. Emissions of ammonia 48 23722c. Emissions of nitrous oxide 738 41817002d. Emissions of carbon dioxide 47 35853. Damage to natural capital soil3a. O-site damage caused by erosionc14 8303b. Organic matter and carbon dioxide losses from soils 82 591404. Damage to natural capital biodiversity and landscape4a. Biodiversity/wildlife losses (habitats and species) 25 10354b. Hedgerows and drystone walls 99 731224c. Bee colony losses 2 124d. Agricultural biodiversity +d+5. Damage to human health pesticides5a. Acute eects 1 0.41.65b. Chronic eects + +6. Damage to human health nitrate 0 07. Damage to human health: microorganisms and other disease agents7a. Bacterial and viral outbreaks in food 169 1002437b. Antibiotic resistance + +7c. BSEeand nvCJD 607 33800Total 2343 11493907aThis table does not include private costs borne by farmers themselves.bTherangesforcostsdonotrepresent formal standarddeviationsofthedataasthisisimpossiblegiven the huge variation in types of data and contexts. The ranges represent best estimates for higher andlowerquartilesforcostsincurredannuallyduringthe1990s. Therangevaluesfortheexternal costsincategory 2arecalculated from therangesstated instudiesofexternal costs ofeachof thesegases, ratherthan the variation of emissions during the 1990s.cTheositedamagecausedbyerosionincategory3adoesnotincludethecostsofremovingsoils/sediments from drinking water (these are in cost category 1c).d+, Not yet able to calculate costs.eBSE costs are an average for 1996 and 1997.J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136 117be80%of1a, 2c; 50%of1b-c, 1e-f, 2a, 2d, 3a-b, 4a-4d, 5a; and25%of7a(the total costs arising fromlivestock systems are higher than for arablelargely because of the one-o bovine spongiformencephalopathy [BSE]problem);and4. pesticideexternalitiesaloneaverage8.6/kgofactiveingredient(a.i.)usedinagriculture (22.5 m kg), or 33/ha of land receiving pesticides (on average 3.84kg a.i./ha) (pesticide externalities are taken to be 100% of 1a, 5; 50% of 1e-f, 4:which equals193 m).Modernfarmingclearlyresults insubstantial external costs per hectare andperkilogramofnon-renewableinput. Theseperhectarecostsaresubstantiallygreaterthanthoseestimatedinotherstudies, probablyreectingthemorecomprehensivenatureoftheframeworkandrangeofimpactsmeasured. Nonetheless, webelievethemto be a conservativeestimateof the truecosts.4. Whyestimates of externalities arelikelytobeconservativeThis study attempts to estimate the total external costs of UKagriculture.Giventhemultifacetedanddynamicnatureofagricultureanditsimpactonenvir-onment andhumanhealth, we have hadtomake manyassumptions about thedata. Inmost cases, these result inconservative estimates of costs, thoughone(theBSEcrisis)hasinatedthecostsabovethelong-termannualaverages,assum-ingthattheexpectederadicationofBSEdoesoccur.1. Some costs are known to be substantial underestimates (e.g. acute and chronicpesticidepoisoningofhumans;monitoringcosts;eutrophicationofreservoirs;restoration of all hedgerow losses), to be limited to certain geographic areas ofthe UK (water company returns are for England and Wales only), or currentlycannotbecalculated(e.g. dredgingtomaintainnavigablewater; ooddefen-ces; marineeutrophication;poisoning of domestic pets).2. We do not generally calculate the costs of returning the environment or humanhealthtopristineconditions. Pesticidesindrinkingwater, forexample, mustnotexceedamaximumof0.5 mg/lforallcompounds.YetBSErepresentsanexample of the cost of complete eradication of a problem. If all cost categorieswereestimatedforsuchrestoration,thenthetotalexternalitieswouldbesub-stantiallygreaterthan estimated in this study.3. Treatmentandpreventioncostsmaybeunderestimatesofthetruecosts.Thissignicantlyunderestimates thetruecosts of biodiversityloss, sinceinmostcasesspeciesandhabitatplansaimtorestoreonlyasmallproportionofpre-vious biodiversity losses. Agriculture's eect on biodiversity is estimatedaccordingtothecostofimplementingplanstoreturnspeciesandhabitatstoacceptablelevelsforsociety(afteraccountingfornon-agriculturalimpactsonbiodiversity).Butthisunderestimatesthenon-uservalues,whichmaybesub-stantial. We have not estimated people's WTP for option values (the option of118 J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136enjoyingsomethinginthefuture); bequest values(ensuringthat descendantswill beabletoderivevalue); andexistencevalues(thevalueofknowingthatsomethingsimplyexists). Ingeneral, thisstudyaccountsforusevalues, andunderestimatestherest(e.g.archaeologicalremainsmanyhavehighexistencevaluetosomepeople,yetmanyhavebeendestroyedbycultivation).Inaddi-tion, byfocusingoncosts,thisstudyunderestimateshowmuchpeoplemightbewillingtopaytoseepositiveexternalitiescreated(c.f. Williset al., 1993;Darling and Topp, 2000).4. This studydoes not account for time lags betweenthe cause of aproblemand its expression as a cost. Some costs incurred in the 1990s may havearisenfromactivitiesortechnologieslongsincestopped(e.g. organochlorineresiduescarriedintoriversinCornwallbyerodedsoillongafterbannedfromuseRCEP, 1996). Othersmaynot yet beexpressed(e.g. accuratepredic-tions about the future number of cases of newvariant Creutzveldt-JakobDisease [nvCJD] are impossible; and the role of pesticides as endocrinedisruptors).5. Wehavenotincludedthecostofresearchinthisstudy, asitisimpossibletodisaggregateresearchbudgets onthebasis of allocations for addressingtheproblems of negativeexternalities.6. Wedonotincludetheverysubstantial publicsubsidyforfarming(3billionannuallyduringthe1990s), asthereisnoacceptedrelationshipbetweenpro-vision of subsidies and creation of negative externalities. Subsidies are atransfer from taxpayers to farmers (whether or not external costs are induced).Someofthissupportisusedtocreatepositivebenets,directlythroughagri-environmentalprogrammes(approximately100mannuallyintheUKtothelate1990s,andrisingunderCAPreforms),andtheremovalofpublicsupportcould,indeed, leadtogreaternegativeexternalities, suchasviafarmamalga-mation, removal of hedgerows, and increased pesticideuse.7. Thisstudyhassoughttoestimateonlythosecostsarisingfromthefarm.Wehavenot includedthemanyenvironmental andsocial costs associatedwithgettingfoodfromthefarmgatetoconsumers' plates. Transportexternalitiesare likely to be signicant (Raven and Lang, 1995; ECMT, 1998). We have alsonot includedanassessment of thecostscausedbymodernfarmingonruralcommunities.5. Datasources andkeyassumptionsThis studyhasdrawnon17 datasetscollectedand maintainedbya wide rangeofagencies and authorities in the UKand rest of Europe:1. Oce of the Director General of Water Services (Ofwat, 19921998) dataset of28 water companies in Englandand Wales;2. British Crop Protection Association (formerly the British AgrochemicalsAssociation) data(BAA,1998);J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136 1193. DepartmentoftheEnvironment, TransportandtheRegions(DETR, 1998a)data on water pollution;4. Environment Agency(EA) dataonbotheutrophicationandpollutioninci-dents (EA, 1998a, b);5. EA data on monitoring of ground and surfacewater(EA, 1998c);6. MinistryofAgriculture,FisheriesandFood (MAFF)monitoringdata,Pesti-cidesSafetyDirectorate(PSD),andVeterinaryMedicinesDirectorate(VMD,1997, 1998;WPPR,1997);7. DETRemissionsinventory, National AtmosphericEmissionsInventory, andtheEuropeanEnvironment Agency(EEA) EuropeanCommunityinventory(DETR, 1998b,c; EEA, 1999);8. EEA ExternE study for external costs of gases (c.f. Eyre et al., 1997; Holland etal., 1999);9. DETRNational Soils Inventory, andlocal authoritydataonaccidents andclean-up costs (Evans, 1995;RCEP,1996;DETR,1997a; Smith et al.,1998);10. EEAdata on soil erosion and organic matter losses (EEA, 1998);11. DETRandInstituteofTerrestrialEcologydatasetsonspecieslosses(DETR,1997a, b, 1998a, d; EA, 1998c);12. UKBiodiversity Steering Group(1995, 1998, 1999) by costings for Biodi-versity and Habitat Action Plans;13. DETR data on hedgerow and stonewall losses (DETR, 1997c, 1998d);14. Rothamsteddataonbeecolonies andlosses since1940s (CarreckandWil-liams, 1998);15. Health and Safety Executive (HSE) data on pesticide poisoning (HSE,1998a, b);16. Public Health Laboratory Service (PHLS) data on food poisoning (Wall et al.,1996; Evans etal., 1998; PHLS, 1999); and17. National AuditOce (NAO) and MAFFdatafor BSE(NAO,1998).5.1. Costcategory1: damage to naturalcapital waterPesticides, nutrients (nitrogen and phosphorus), soil, farmwastes and micro-organismsescapefromfarmstopollutegroundandsurfacewater.Costsareincur-redbywaterdeliverycompanies(andpassedontotheircustomers)tocomplywithdrinkingwaterstandardssetoutinEuropeanUnion(EU)legislationforpesticidesandnitrates(standardsforpesticidesare0.1mg/l forasingleproductand0.5 mg/lfortotalpesticides; fornitratesthemaximumisto 45 mgnitrate/l or 10 mgnitrate-N/l),toremovepathogens,particularlyCryptosporidium,topayforrestoringwatercoursesfollowingpollutionincidentsandeutrophication, andtoremovesoil fromwater. It is important tonotethat somecosts wouldbebornewhatever typeofagricultural systemwasinuse. Thekeypolicyquestionis, aretheretypesofagri-culture that could substantially reduce these costs? We also do not assess all externalcosts, suchastheavoidancecostsofwaterconsumersswitchingtobottledwater.Companies incur bothcapital andoperatingexpenditurefor water qualitytreat-ment. These costs are reported annually by each of the 28 water companies in England120 J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136andWalestoOfwat.2Weusethese, togetherwithvariousDETRdatasets, toesti-mate external costs at 231 m/year for damage to water. These costs would be muchgreater if the policygoalwere completeremoval of all residues/contaminants.5.1.1. Pesticides insources of drinking water (1a)Annual capital expenditure bywater companies onpesticide removal between1992and1997was 124.9m/year after depreciation(at 1996prices). Operatingexpenditurebythe28companiesisgenerallymuchhigherintheeastofEngland,but is not always reportedseparately. Basedonreturns toOfwat, we estimateannual operatingcosts for pesticideremoval tobe9.5m. As 89%of pesticides(some22.5mkg)areappliedinfarming(BAA,1998),thetotalpesticidecostsaris-ing from farming are 119.6m/year.5.1.2. Nitrate in sources of drinking water(1b)Nitrateentersdrinkingwatersourcesfromfertilisersandlivestockwastes, frommineralisationof organicnitrogeninthesoil, fromatmosphericdepositions, andfromhumansewage. Capital expenditurebywatercompaniesonnitrateremovalbetween1992and1997was18.8m/year,andoperatingexpenditures1.7m(botharehigherintheeast).Weestimatethat80%ofnitrogenisfromagriculturalsour-ces,putting annual external costs at 16.4m/year.5.1.3. Phosphate and soil insourcesof drinking water(1c)Phosphatealsocontaminateswater, withsome43%inwaterestimatedtocomefrom agriculture (29% from livestock;14% from fertilizers), mostly attached to soilparticles from eroded land, the remainder coming from point sources (human waste,detergents, industry; RCEP, 1996; EA, 1998a; EEA, 1998; Withers and Jarvis, 1998).Capital expenditure on both phosphate and soil particle removal was 68.8 mbetween1992and1997,withafurther4.3mforoperatingexpenditure.Assuming2The government's Oce of the Director General of Water Services sets industry price levels each veyears,whichdetermineboththemaximumlevelsofwaterbillsandspeciesinvestmentsinwaterqualitytreatment. During the 1990s, water industry undertook pesticide and nitrate removal schemes, resulting intheconstructionof 120plantsfor pesticideremoval and30for nitrateremoval (Ofwat, 1998). Ofwatestimates that water companies will spend a further 600 m between 2000 and 2005 on capital expenditurealoneduetocontinuingdeteriorationof `rawwater' qualityduetoall factors. Ofwat predictscapitalexpenditure for pesticides tofall to88 m/year at the end of the1990s/early 2000s; and for nitrate to fallto 8.3 m/year.Although Ofwat has sought to standardise reporting, individual companies report water treatment costsindierent ways. Most dodistinguishtreatment for pesticides, nitrate, Cryptosporidium, andseveralmetals(iron,manganese,lead).Theremainingtreatmentcostsforphosphorus,soilremoval,arsenicandother metals, appear under a category labelled `other'. Of the 28 water companies in England and Wales,threereportnoexpenditureontreatmentwhatsoever;andafurtherthreedonotdisaggregatetreatmentcosts, with all appearing under `other'. Twenty companies report expenditure on removal of pesticides, 11onnitrates, and10onCryptosporidium. Itisimpossibletotell fromtherecordswhetherastatedzeroexpenditureisactuallyzero, orwhetherthishasbeenplacedinthe`other' category. UsingOfwatandwater companies' returns, we estimate that 50%of expenditure under the `other' categoryrefers toremoval of agriculturally related materials.J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136 121halfisforsoilremoval,and43%oftheremainingphosphoruscostsarefromagri-culture, then total external costs for both phosphorus and soils are 52.3 m.5.1.4. Zoonosesin sources of drinking water (particularlyCryptosporidium) (1d)Water companies must remove some 50zoonoses derivedfromlivestock, wildanimalsandhumansewage.ThemostimportantagriculturalcontaminantisCryp-tosporidium, aprotozoancausingseverediarrhoeal illness. It isresistant tochlor-inationandcanonlyberemovedbyltration. Expenditurewas lowintheearly1990s(27.6mfor19921997),butbytheendofthe1990shadrisento8m/year(DETR,1998e).Ofwat,however,hasindicatedthat66m/yearwillberequiredtomeet legal compliances. Wetakealowmidrangeannual gureof 25m.3Untilrecently, it was assumed that all Cryptosporidium arose from animals low levels ofinfectionarecommonincattle, andtherearenopracticablemeasuresforeradica-tion. Butrecentresearchindicatesthatsomestrainsoccuronlyinhumans, andsowe assume that 90% of the treatment costs are agricultural externalities, i.e. 22.5 m(Kempetal., 1995; Sturdeeetal., 1998). Type1strainsofCryptosporidiumaectonlyhumans,whilstType2aectbothanimalsandhumans,andmaybefoundinboth livestock wastes and sewage (ENDS, 1999a,b).5.1.5. Eutrophication and pollution incidentsfrom agriculture (1e)Farmwastesfurtherdisrupt watersystems: cattleandpigslurry, silageeuent,dairy wastesthatcauseeutrophication, andtoxic products (suchas sheepdips)thatkill aquaticlife. Ineachof 19961997, therewere31,00032,000reportsof waterpollutioninEnglandandWalesfromallsources,ofwhichabout20,000weresub-stantiated(EA, 1998b). Agricultureaccountedanaverage2600incidentsperyearduringthe1990s(ofwhichabout50areinCategory1,250inCategory2,2300inCategory3). UsingthecostsincurredbytheEAforrestockingriverswithshtorestore themto theirpre-incidentcondition,weestimatethe totalcostoftheseinci-dents to be 1.142.35m/year.43Expenditure on Cryptosporidium removal is likely to grow: Ofwat assumes that some 1000 m will beinvestedonimprovedltrationandforCryptosporidiumtreatment over19982013(dwarngpreviousrequirements for pesticide and nitrate removal); DETR assumes only 120 m, but bases this on only 10 of28 companies requiring expenditure on Cryptosporidium removal.4Pollution incidents are sorted bythe EAinto three categories.Category 1incidentsare themost ser-ious andmayinvolveoneor moreof thefollowing: theclosureof asourceof water abstraction; anextensive sh kill; a potential or actual persistent eect on water quality or aquatic life; a major eect ontheamenityvalueofthereceivingwater; orthesubsequentneedforextensiveremedial measurestobetaken.Category2incidentsaresignicantbutlesssevere,andmayinvolvethenecessitytonotifydown-stream abstractors; result in a signicant sh kill; render water unt for livestock; have a measurable eectonanimal lifeinthewater; contaminate thebedoftheriverorcanal;orreducetheamenityvalue ofthewater totheirowners or tothegeneral public. Category 3incidentsarerelatively minor and haveno sig-nicantorlastingeectonthereceivingwater.TheEAbuysshatacostof5001000for10001-yearold sh, and 9001900 for 1000 2-year olds. We estimate the 50 Category 1 incidents kill 40005000 shperincident; the250Category2incidentseachtokill 2550%of thistotal; andthe2300Category3incidents to kill 510%. These costs, however, include only the restocking of the sh that were lost and donot account for the losses of invertebrates and other aquatic life. The cost of river aeration and labour arealso not included.122 J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136Eutrophicationaects water supply(algaecanblocklters, stimulatebacterialgrowth, andgivedrinkingwateranunpleasant taste), irrigation, sheries, naviga-tion, water sports and angling (Mason,1996; EA, 1998a; Withers and Jarvis, 1998).Between1989and1997, some3000freshwaterbodieshavebeenaectedbyalgalblooms, some of which have caused human health problems, the deaths of sheep anddogs, andthe closure of water bodies for recreation. Most reservoirs ineasternEnglandsuerperiodicallyfromaninabilitytotreat waterowingtoahighcon-centrationofplankton,leadingtocostlyclosureoftreatmentworksforperiodsofupto26months(EA).Therearenonationaldataonthecostsofeutrophication,thoughtheremedial costsinreservoirsalonehavebeenestimatedtobe4m/year(RCEP, 1996). Marine eutrophication, comprising algal blooms, red tides andanoxiaindeepwaters, isalsoincreasinglycommon, butcostsarenotestimatedinthis study.5.1.6. Monitoring and adviceon pesticides and nutrients (1f)CostsareincurredbythePSD, VMD, andtheEAformonitoringof pesticideresidues andnutrients infoodandthe environment, andfor administration ofschemes andgrants toreduce pollutionandfor advisoryservices. The costs are5.4mforpesticidemonitoringinfoodandlivestock,and4.75mformonitoringpesticidesat2500surfaceandgroundwatersites.5Nodataareavailableonresiduemonitoringcarriedoutbysupermarketsandfoodprocessors.TheUKgovernmentalso incur 0.55 m costs in providing advice to farmers to encourage the adoption ofmoresustainablenutrientandcatchmentmanagement. Althoughweincludethesecostsintheanalysis, itcouldbearguedthatanalternative, moresustainableagri-culture, would still incursuch monitoring costs.5.2. Costcategory2: damage to naturalcapital airAgriculturecontributes toatmosphericpollutionthroughtheemissions of fourgases: methanefromlivestock, nitrousoxidefromfertilisers, ammoniafromlive-stockwastesandsomefertilisers, andcarbondioxidefromenergy/fossil fuel con-sumption and loss of soil carbon. According to the National Atmospheric EmissionsInventoryandtheEEAinventory(DETR, 1999b, c; EEA, 1999), UKagricultureannuallyemits1.064mtonnesofmethane,0.098mtonnesofN2O,0.278mtonnesof ammonia, and 0.75 m tonnes of CO2-C (these data do not include a factor for soilcarbonloss). These inturncontribute toatmospheric warming(CH4, N2Oand5ThePSDpesticideresiduesmonitoringprogrammetook3000samplesin1997(downfrom4000in1996) fromretail outlets, whichweresubjectedto80,000analyses, at acost of 1.7m(2min1996;WPPR, 1997). The VMD tested 39,152 samples of domestic and imported livestock foodstus for residuesthrougha statutoryprogrammetoimplementEUlegislation,withannual costsof2.7m,andinanon-statutoryprogrammefundedbyMAFFconducted16,767analyseson2547samplesatacostof1m(VMD,1997,1998; MAFF,1998).TheEAsubjectswaterto200,000pesticideanalyses;at8%ofsites,Environmental QualityStandardswereexceededbyatleastonepesticide(EA, 1998c). Thecostisnotrecorded by the EA, but assuming similar unit costs as for the PSD, we put the annual cost at 4.75 m.J.N. Pretty et al. / Agricultural Systems 65 (2000) 113136 123CO2), ozonelossinthestratosphere(N2O), acidicationofsoilsandwater(NH3)and eutrophication (NH3; DETR, 1998b, c;EEA,1999).Extensiveresearchisunderwayontheexternalcostsofthesegases(Pearceetal.,1996; Eyreet al., 1997; Hollandet al., 1999). TheExternEstudyontheexternaleects of climate change gases includes analysis of impacts on climate change,health, parasiticandvectorbornediseases, sea-level rise, wateravailability, biodi-versity, andstorm, oodanddroughtincidence(Eyreetal.). Itputsthemarginalcostsofmethaneat263/tonne(range239353); ofnitrousoxideat7530/tonne(range426717,333);andofcarbondioxide(asC)at63/tonne(range47113).6Theexternalcostsofammoniahavebeencalculatedbyidenticationofemissions,changes inexposure/impacts, quanticationof impacts, andvaluationbasedonWTP, using the more conservative VOLY (value of a life year) concept, rather thantheVOSL(valueofastatisticallife),andindicatinganexternalcostof171/tonne(range 83259) (Holland et al.). These studies indicate that annual external costs ofthese gases arising from UK agriculture are 280 m for CH4, 738 m for N2O, 47 mforCO2, and 48 m for NH3.5.3. Costcategory3: damage to naturalcapital soil5.3.1. O-site damage causedby soil erosion (3a)Ahealthysoil isvital foragriculture, but modernfarminghasacceleratedero-sion, primarilythroughthecultivationofwintercereals, theconversionofpasturetoarable, theremoval ofeldboundariesandhedgerows, andovergrazingofani-malsongrasslands(Evans, 1995, 1996; RCEP, 1996; DETR, 1997a; EEA, 1998).Soil erosioncausesbothon- ando-farmproblems.7Wedonot includeinternalcosts,eventhoughlossofsoilfertilityrepresentsalossofpublicgoodinthelong-run. O-site costs arise when soil carried o farms by water or wind blocksditchesandroads,damagesproperty,inducestracaccidents,increasestheriskofoods,andpolluteswaterthroughsedimentsandassociatednitrate,phosphateandpesticides. Evans, using data fromlocal authorities, estimates that the nationalexternal costs topropertyandroads alonetobe13.77m(4mfor damagetoroads and property; 0.1 m for trac accidents; 1.19 m for footpath loss; 8.47 forchannel degradation), but not countingwater companycosts (seecategory1) orlossestosheries.6The data in the Open Framework and FUND models take account of dierences in discount rate, areweightedaccordingtowealthdierencesinaectedcountries, andtakeaccountof`social contingency'(thecapacity ofregions/countriestoadapttochange).Thismeansthatuncertaintyisstillverylarge(c.f.Eyreetal.,1997).Weadoptaconservativegurefordamagecostsbasedonaquarterofthedierencebetween the lowest and highest estimates contained in the Open and FUND models, and according to twodierent discount rates (1 and 3%).7Costs incurred byfarmers themselves from soilerosion arisefrom: (1)loss oforganic matter leadingtodecreasedwaterholdingcapacityofsoilsandincreasedrun-o; (2) lossoforganicmatter-richsoilsreduces yields as crops are slower to germinate; and (3) loss of nutrients and crops themselves in water andwind erosion. Evans (1995, 1996) estimates the annual on-farm costs of soil erosion to be 1011 m (1996prices).124 J.N. Pretty et al. / Agricultural Systems 65 (2000) 1131365.3.2. Organicmatter (OM)and carbondioxidelosses (3b)Soils in England, Wales and Scotland contain some 21.78 billion tonnes of carbon(Howardet al., 1995; DETR, 1997a; Smithet al., 1998). Arablesoilscontainonaverage 162 tonnes C/ha, permanent pasture 207 tonnes C/ha, and soils under semi-natural vegetation350tonnes C/ha. Most carbonintheUKis inScottishpeats(16.4 billion tonnes alone). Carbon accumulates in soil when arable land isconverted to grassland or forest, and when sustainable farmpractices lead toOMincorporation,butitisrapidlylostwhenpasturesareploughedorwhenagri-cultural land is intensively cultivated (Ka tterer and Andre n, 1999). The OM contentinUKsoilshasdeclinedinrecentyears(DETR; EEA, 1998; butseeSkinnerandTodd, 1998, for another view). Soils lose CO2 when OM is lost, and using estimatesof soil OM loss, we calculate that 20% of arable soils (0.92 m ha) have lost 1.7% ormoreofOMsince1980, amountingto1.42tonnesC/ha/year.8Givenanexternalcost for CO2of 63/tonne(range47113), this putstheannual cost at 82.3m(range59140).95.4. Costcategory4: damage to naturalcapital biodiversityand landscape5.4.1. Biodiversityand wildlife (habitats and species) (4a)Modernfarminghas hadasevere impact onwildlife: 170native species havebecomeextinctthiscentury,including7%ofdragonies,5%ofbutteriesand2%ofshandmammals.Inaddition,95%ofwildower-richmeadowshavebeenlostsince1945;3050%ofancientlowlandwoods;50%ofheathland;50%oflowlandfens, valleyandbasinmires;and40%ofhedgerows(DETR,1998d;Pretty,1998).Species diversity is also declining in the farmed habitat itself. Draining and fertilizershavereplacedoristicallyrichmeadows withgrass monocultures, overgrazingofuplands has reducedspecies diversity, andherbicides havecut diversityinarableelds. Farmland birds have particularly suered: the populations of nine species fellbymorethanahalfbetween1970and1995(CampbellandCooke,1997;PainandPienkowski,1997; Mason, 1998).Inthisstudy, weusethecostsofrestoringspeciesandhabitatsundertheBiodi-versityActionPlans(BAPs)asaproxyofthecostsofwildlifeandhabitatlosses.Theseplanscontaincostedtargetsandactionplansfor406priorityspeciesand388AccordingtotheNationalSoilsInventory(DETR,1997a),therewasanincreasebetween1980and1995 in the proportion of arable topsoils found to have OM concentrations of 7% OM (from 22 to 11%)8. Thuswith 11% of samples showing OM declines from 7 to5.3% (mid-range) and a further 11% showingOMdecline from 5.3 to