Reality of Food Losses: A New Measurement Methodology

Abstract

Measuring food loss, identifying where in the food system it occurs, and developing effective policies along every stage of the value chain are essential first steps in addressing the problem of food loss and waste in developing countries. Food loss has been defined in many ways, and disagreement remains regarding proper terminology and measurement methodology. Although the terms " post‐harvest loss, " " food loss, " " food waste, " and " food loss and waste " are frequently used interchangeably, they do not refer consistently to the same aspects of the problem. In addition, none of these classifications includes pre‐harvest losses. Consequently, and despite the presumed importance of food loss, figures regarding food loss remain highly inconsistent, precise causes of food loss remain undetected, and success stories of decreasing food loss remain few. We improve over this measurement gap on food losses by developing and testing the methodology traditionally used with three new methodologies that aim to reduce the measurement error and that allow us to assess the magnitude of food loss. The methods account for losses from the pre‐harvest stage through product distribution and include both quantity loss and quality deterioration. We apply the instrument to producers, middlemen, and processors in seven staple food value chains in five developing countries. Loss figures across all value chains fluctuate between 6 and 25 percent of total production and of the total produced value; these figures are consistently largest at the producer level and smallest at the middleman level. The identified losses are in addition to the existing yield gaps identified across the different commodities studied which are in the range of 50 to 80%. Throughout the different estimation methodologies, losses at the producer level represent between 60 and 80 percent of total value chain losses, while the average loss at the middleman and processor level lies around 7 and 19 percent, respectively. Differences across methodologies are salient, especially at the producer level. While the estimation results from the three new methods implemented are close and the differences are mostly not statistically significant, the aggregate self‐reported method reports systematically lower loss figures. Finally, our results show the major reasons behind the losses identified for each commodity and country. Specifically, we find that they included pests and diseases and lack of rainfall. When looking at the produce left in the field, the major reason for the loss is a lack of appropriate harvesting techniques. Finally, the loss reported at the post‐harvest level is due mostly to damage done during selection, as a result of workers' lack of training and experience in selecting the produce. Therefore, technology, improved seeds and the proper soil management techniques together with better market access could help to substantially reduce the losses at the producer level.

Full text

1


TheRealityofFoodLosses:ANewMeasurementMethodology1
LucianaDelgado2,MonicaSchuster3,andMaximoTorero4
Abstract
Measuringfoodloss,identifyingwhereinthefoodsystemthatlossoccurs,anddevelopingeffective
loss‐reductionpoliciesalongeverystageofthevaluechainareessentialfirststepsinaddressingthe
problemoffoodlossandwasteindevelopingcountries.Foodlosshasbeendefinedinmanyways,and
disagreementremainsregardingproperterminologyandmeasurementmethodology.Consequently,
and despite the presumed importance of food loss, figures regarding food loss remain highly
inconsistent,precisecausesoffoodlossremainundetected,andsuccessstories ofdecreasing food
lossremainfew.Weaimtofillthismeasurementgapbydevelopingthreenewmethodologiesthat
aimtoreducethemeasurementerrorandthatallowustoassessthemagnitudeoffoodloss;wealso
testtheseagainstthemethodologytraditionallyused.Ournewmethodsaccountforlossesfromthe
pre‐harvest stage through product distribution and include both quantity loss and quality
deterioration.Weapplytheinstrumenttoproducers,middlemen,andprocessorsineightstaplefood
valuechains insixdeveloping countries.Lossfiguresacrossallvaluechainsfluctuatebetween 6and
25percentoftotalproductionandofthetotalproducedvalue;thesefiguresareconsistentlylargest
attheproducerlevelandsmallestatthemiddlemanlevel.Theidentifiedlossesareinadditiontothe
existingyieldgapsidentifiedacrossthedifferentcommoditiesstudied,whichareintherangeof50to
80percent.Throughoutthedifferentestimationmethodologies,lossesattheproducerlevelrepresent
between60and80percentoftotalvaluechainlosses,whiletheaveragelossatthemiddlemanand
processorlevelslies ataround7and19 percent,respectively. Differences acrossmethodologiesare
salient,especially at the producerlevel. While theestimation results from thethreenew methods
implementedarecloseandthedifferencesaremostlynotstatisticallysignificant,theaggregateself‐
reportedmethodreportssystematicallylowerlossfigures.Finally,ourresultsshowthemajorreasons
behind the losses identified for each commodity and country; these reasons included pests and
diseasesandlackofrainfall.Whenlookingattheproduceleftinthefield,themajorreasonfortheloss
isalackofappropriateharvestingtechniques.Finally,thelossreportedatthepost‐harvestlevelisdue
mostly to damage done during selection, as a result of workers’ lack of training and experience in
selecting the produce. Therefore, improved techno logy, improvedseeds,propersoilmanagement
techniques,andbettermarketaccess,couldsubstantiallyreducelossesattheproducerlevel.


1ThisworkwasundertakenaspartoftheCGIARResearchProgramonPolicies,Institutions,andMarkets,which
isledbyIFPRIandfundedbytheCGIARFundDonors.ThispaperhasnotundergoneIFPRI’sstandardpeer‐review
process.WearethankfultoFranciscoOlivet forhisincrediblesupportontheimplementationofthesurveysin
GuatemalaandHondurasaswellastotheTeamsoftheCIPandICARDAfortheirsupportinPeru,Ecuadorand
Ethiopia.Theopinionsexpressed herebelong tothe authorsandnotnecessarilyreflectthoseofPIM,IFPRI,or
CGIAR.Anyremainingerrorsarethesoleresponsibilityoftheauthors.
2InternationalFoodPolicyResearch,Luciana.Delgado@cgiar.org
3InstituteofDevelopmentPolicy(IOB)UniversityofAntwerpMonica.Schuster@uantwerpen.be
4ExecutiveDirector,WorldBank,mtorero@worldbank.org

2


1. Introduction
Food loss and food waste have become an increasingly important topic in the development
community.Infact, theUnitedNationsincludedthe issue of foodlossandwasteintheSustainable
DevelopmentGoaltarget12.3,whichaimsto “halve per capita global foodwaste at the retailand
consumerlevels and reducefoodlossesalong production and supplychains, includingpost‐harvest
losses” by 2030. Food loss and food waste have caught the attention of both researchers and
policymakersforseveralreasons.First,growingpopulationsandchangingdietsassociatedwithgreater
wealthareincreasingthe pressureontheworld’savailableland,constitutingseriousthreatstofood
security, especially in developing co untries. Policies to reverse this situation have mainly aimed at
increasingagricultural yields and productivity, butthese efforts are often cost‐ and time‐intensive.
Second,thelossofmarketablefoodcanreduceproducers’incomeandincreaseconsumers’expenses,
likelyhavinglargerimpactsondisadvantagedsegmentsofthepopulation.Third,foodlossandwaste
entailunnecessarygreenhousegasemissionsandexcessiveuseofscarceresources.
Foodlossandwasteoccuratdifferentstagesofthefoodvaluechain(VC):production,post‐production
procedures,processing,distribution,andconsumption(FAO,2011;HLPE,2014;Lipinskietal.,2013).
Figure1showsthestagesofthevaluechainatwhichfoodlossoccurs,aswellasthedimensionsthat
arepotentially responsible for lossat eachstage.Thedistributionof lossandwastealongthe food
chainisdifferentdependingonthecommodityandthegeographicallocationinquestion,butfood
loss and waste are commonly the result of underlying inefficien t, unequal, and unsustainable food
systems.
Byreducingfoodlossandwaste,wecanimprovefoodavailabilityandfoodaccesswithoutincreasing
theuseofagriculturalinputs, scarcenaturalresources,or improvedtechnologiesontheproduction
side.Recent reports,however,highlightthat successstoriesofdecreasingfoodwaste(WRAP,2009)
andfoodloss(WorldBank,2011)arenotmany,andfiguresonfoodlossandfoodwasteremainhighly
inconsistent.Thus,whilevariousgovernmental,research,andcivilsocietyinitiativeshavebeen
launchedtoaddressthisimportantissue,largeresultsareyettobeseen.
The implementation of a strategy to reduce food loss faces three important challenges. First, no
accurateinformationexistsabouttheextentoftheproblem(especiallyindevelopingcountries).The
availableestimatessuggestthatfoodlossisalarminglyhighandmayaccountforatleastone‐thirdof
total global food production. For the most part, calculations of food loss hinge upon accounting
exercisesthatuseaggregatedatafromfoodbalancesheetsprovidedbynationalorlocalauthorities.
These“macro”estimationsaresubjecttoconsiderablemeasurementerror,relyonpoorqualitydata,
orarenotbasedonrepresentativesamples.Moreover,theyonlyquantifythevolumeoffoodthatis
lostand do nottakeintoaccountpotentialdeteriorationofqualityorreductionsofeconomicvalue
thatalsoaffectfarmersandconsumers.
Morerecently,effortshavebeenmadetousemicrodatatoestimatefoodloss.Theseestimationsrely
onsurveyscollectedamongdifferentactorsacrossthefoodvaluechain.However,theytend to be
basedoncasestudiesthatarenotrepresentativeofacountry’slargerpopulations.Additionally,these
studiesusedifferentdefinitionsoffoodloss,hamperingcomparisonsacrossdifferentareasandcrops.
Duetotheirlackofrepresentativeness and differences in their methodologies, the available micro‐
basedestimatesarewidelyvariableandyieldinconclusiveevidenceabouttheextentoffoodloss.
Thesecondchallengeisthescarceevidenceregardingthesourceoffoodloss.Foodlossisassociated
with a wide array of factors (e.g., poor agricultural management skills and techniques, inadequ ate

3

storage,deficientinfrastructure,inefficientprocessing,lackofcoordinationinmarketingsystems,etc.)
and can occur in different stages of the value chain (i.e., production, harvesting, po st‐production,
processing,distribution,orconsumption).Becauseoftheaggregatenatureoftheirdata,macrostudies
areunabletocapturethecriticalstagesatwhichfoodlossoccurs.Arguablyduetothecostofprimary
datacollection,most microstudieshavenot incorporateddetailedinformationregardingsourcesof
foodloss intheirsurveyinstruments.Mostofthesestudiesaimtocapturetotal food lossbasedon
farmers’self‐reportedestimatesbutdonotaimtodisentangletherelevantproductionphasesinwhich
losses are generated. For example, studies using the nationally representative Living Standard
Measurement Surveys – Integrated S urveys on Agriculture (LSMS –ISA)askfarmerstoassessthe
proportionoftheircropslosttorodents,pests,insects,flooding,rotting,theft,orotherreasons;these
studiescanonlyprovideglobalestimates.Afewstudieshavecollected more comprehensive
informationabouttheparticularstagesin whichlossesoccur;however,thesestudiesarebasedon
smallsamplesinparticularlocations,makingtheirresultsdifficulttoextrapolate.
Third,thereislittleevidenceregardinghowtosuccessfullyreducefoodlossacrossthevaluechain.
Therehavebeeneffortsto introduceparticulartechnologiesalongspecificstagesofthevaluechain
(e.g.,silosforgrainstorage,triplebaggingforcowpeastorage,ormechanizedharvestingandcleaning
equipment for wheat and maize). However, little evidence exists regarding adoption rates or the
economicsustainabilityof these efforts. Inparticular,there is a needtobetterunderstand how to
introduce economic incentives for actors from farm‐to‐fork, taking into account the upstream and
downstreamlinkagesacrossthevaluechain.
Thispaperaimstoresolvethefirsttwoaforementionedchallenges.Ourobjectiveistoimprovehow
foodlossisquantified5andtocharacterizethenatureoffoodlossacrossthevaluechainfordifferent
commoditiesinawidearrayofcountries6.Forthispurpose,wedesignedasetofsurveystomeasure
theextentoffoodloss.Whilethe surveys were tailored to specificcountries and commoditiesand
commodityvarieties(forexample,whilemaizeinHondurasandGuatemalahavethesameattributes,
wheatinChinahasdifferentattributesthanwheatinMexico),theyprovideaconsistentmeasurement
offoodlossacrossdifferentagentsinthevaluechain(i.e.,farmers,middlemen,andprocessors).The
surveyscapturedetailedinformationaboutthese agents’ different processesandquantify foodloss
alongeachproductionstagebycollectingself‐reportedmeasuresofthevolumesandvaluesoffood
lossesincurredduringdifferentprocesses(harvesting,threshing,milling,shelling,winnowing,drying,
packaging,transporting,sorting,picking,transforming,etc.).Inaddition,weestimatelossesbasedon
commoditydamageby collectingdetailed data from farmers, middlemen,andprocessorsregarding
thequality (based ondamage coefficients) ofagricultural commodities that theyuseasinputsand
outputs.Thisallowsustoquantifyfoodlossintermsofthequalityattributabletoeachagentacross
the value chain. Finally, we also estimate food loss based on commodity attributes by capturing
informationaboutdifferenttypesofcommodityattributes(e.g.,size,impurities,brokengrain,etc.)
andascertainingthepricepenaltythateachofthesetypesofcropdamageentails.Inthisline,weare
abletoidentifyparticularfactorsthatdiminishcommodities’valuesandthusareabletoquantifyfood
qualitylossbasedonmarketconditions.
Thesurveys implemented allowus to quantifythe extent of foodloss across thevalue chain using
consistentapproachesthatarecomparableacrosscommoditiesandregions. Theyalsoenableusto
characterizethenatureoffoodloss;specifically,weareabletoascertaintheproductionstagesacross

5WefollowthedeMelet.al(2009)frameworkbyexploringdifferentwaystomeasurefoodlossesinorderto
reconcilehowfarwecanreconcileself‐reportedfoodlossesthroughmoredetailquestionsacrossthedifferent
stagesofthevaluechain.
6ItisimportanttomentionthatthispaperdoesnotmeasurefoodwasteasperBellemareet.al(2017).

4

thevalue chain andtheparticularprocesses in which lossesareincurred.Theresultswilltherefore
informusabouttheparticularareasthatrequireinvestmentstoreducefoodloss.
Thepaperisdividedasfollows.Thefirstsectionlooksat differentissuesregardingthedefinitionof
foodlossacrossthevaluechain.Wethenconductareviewoftheexistingworkonvaluechainsand
identify the major problems and gaps in the literature. In the third section, we present our
methodological approach, followed by our key findings for China, Ethiopia, Ecuador, Honduras,
Guatemala,andPeru.Finally,weexaminethemajorreasonsfortheidentifiedlosses,usingdetailed
regressionanalysis.Thepaperendswithconclusionsandpolicyrecommendations.

2. Divergenceinterminologyanddefinitions
Theliteraturecommonlyagreesaboutvaluechainstages(Figure1),aswellasaboutthefactthatfood
loss occurs at each stage (e.g., FAO, 2011; Lipinski et al., 201 3; Parfitt, Barthel, and Macnaughton,
2010).However,noagreementexistsregardingfurtherclassificationoffoodlossandfoodwaste.The
terms‘Post‐HarvestLosses’(PHL),‘FoodLoss’(FL),‘FoodWaste’ (FW), and ‘Food Loss and Waste’
(FLW) are frequently used interchangeably, but they hardly everreferconsistentlytothesame
concept.Forsomeauthors,thedistinctionislinkedtothestagesatwhichthelossoccurs.Forothers,
thedistinctionisbasedonthecause of the food loss and whether itwasintentional. Some recent
publicationshavetriedtocreatemoreclarity(FAO,2014;HLPE,2014;Lipinski etal.,2013).Inthese
studies,FL refers tounintentionalreductionsinfoodquantityorqualitybeforeconsumption;these
lossesusuallyoccurintheearlierstagesofthefoodvaluechain,fromproductiontodistribution.PHL
is a sub‐section of FL, excluding losses at the production level (although losses during harvest ar e
sometimesmisleadinglyincludedintheconcept;e.g.Affognon,2014;APHLIS,2014).FWreferstofood
thatisfitforhumanconsumptionbutthatisdeliberatelydiscarded;thisismostcommonattheend
ofthevaluechain.Thetotalityoflossesandwastealongthevaluechainwithrespecttototalharvested
productionareencompassedinFLW(FAO,2014);however,thisdefinitiondoesnotincludecropslost
before harvest because of pests and diseases or left in the field, crops lost dueto poor harvesting
techniquesorsharppricedrops,orfoodthatwasnotproducedbecauseofalackofproperagricultural
inputs.Toincludethesepre‐harvestlosses,weproposeamoreexpansivedefinitionthatwillcapture
alllossesacrossthevaluechain(seeFigure1).Itisimportanttonotethatinthispaper,wedonotlook
atwasteattheend of the value chain.Thisis because, from an integrated valuechainperspective,
pre‐harvest conditions have direct imp acts on eventual losses at later stages of the chain, due to
products’differentquality,storage‐andshelf‐life,andtransportsuitability.
ThereisalsonoagreementintheliteratureregardingthedefinitionoffoodlosswithineachVCstage.
To give just one example of differing definitions: losses across the value chain can originate from
reductions in both food quantity and food quality and can thus describe either weight, caloric,
nutritional,and/oreconomiclosses.Duetoestimationdifficulties,productseasonality,andmarkets’
sensitivitytofoodquality,moststudiesanalyzequantitativelosses,descr ibinglo ssesint ermsofweight
reductions(e.g.,APHLIS,2014;HLPE, 2014);thesereductions sometimestranslateintocaloricterms
(e.g.Kummuetal.,2012;Lipinskietal.,2013),buttheystilldonotcapturequalitativedimensionssuch
asnutritionalcontentandphysicalappearance(seeAffognonetal.(2014)foraliteraturereview).The
choice of definition used dependson a stakeholders’ priorities, as well as on the data available;
however,thatchoicehasimportantimplicationsfortheestimationmethodologyusedtoexaminefood
loss,aswellasontheinterpretationofresults.

5

Figure1:Levelsatwhichfoodlossoccurs



Production
 Quantitativeloss:
Production‘leftinthefield’
 Qualitativeloss:
Qualitydeteriorationof
harvestedproduction
Post–harvest,onfarm
 Quantitativeloss:Production
totallylostduringon‐farmpost‐
harvestactivities
 Qualitativeloss:Quality
deteriorationduringon‐farm
post‐harvestactivities
PRODUCER‘ON–FARM’LOSS MIDDLEMAN
Sales
Post–harvest,offfarm
 Quantitativeloss:Production
totallylostduringoff‐farmpost‐
harvestactivities
 Qualitativeloss:Quality
deteriorationduringoff‐farm
post‐harvestactivities
PROCESSOR
Transformation
Sales
 Quantitativeloss:Production
totallylostduringproduct
transformationactivities
 Qualitativeloss:Quality
deteriorationduringproduct
transformationactivities



3. Howlosshasbeenmeasured
Twomainestimationmethodologieshavebeenusedtostudyfoodlossacrossthevaluechain:amacro
approach, using aggregated data from national or local authorities and large companies, and a micr o
approach,usingdataregardingspecificactorsinthedifferentvaluechain stages (Figure 2). Themacro
approachreliesonmassorenergybalancesinwhichrawmaterialinputs,ineitherweightorcaloricterms,
arecomparedtoproduceoutputs.Thismethodprovidesacost‐effectiveindicationoftheoveralllosses
alongtheentirevaluechainandwasusedbyGustavssonetal.(FAO,2011),thestudythathasbeenmost
quotedandusedasareferenceforfoodlossatthegloballevel.ByusingFAOStat’sFoodBalanceSheets,
thisstudyestimates that around 32 percent of global foodproduction, across allproductionsectors,is
lostalongtheentirefoodvaluechain.Kummuetal(2012)andLipinskietal.(2013)usethesamerawdata
andfindthatthistranslatesintoa24percentdecreaseincaloricterms.Incountry‐specificstudies,macro
energybalancesshowthat48 percentofthetotalcaloriesproducedarelostacrossthewholefoodVC
(Beretta et al. 2013; Switzerland), while mass balance data series from USDA data, using alternative
assumptions, show that 28.7 percent of the harvested product is lost between post‐production and
consumption (Venkat et al., 2011; US) and that  31 percent of the available food supply is lost at
distributionandconsumption(Buzbyetal.2014,US).Onedisadvantageofthismethodisthedemand
forrepresentativeandgoodqualityproduction,loss,andwastedata.Datagapsareparticularlyapparent
forcertainregionsoftheworld,suchaslow‐andmiddle‐incomecountries,andspecificstagesoftheVC,
suchasprimaryproduction,processing,andretail(Stuart,2009).Themethodisalsonotrepresentative
ofsmallerregional units, preventing identificationofthevaluechainstagesatwhichthelossesoccur;
thesechallengesthe appropriatetargeting oflossreductioninterventions.Finally,the aggregated data
usedformassbalancesareoftenincapableof differentiatingbetweennaturalloss(e.g.,moistureloss)
andunnaturalweightloss(forexample,duetospoilage),aswellasedibleandinedibleloss.
Themicroapproach,ontheotherhand,usessampledataregardingspecificvaluechainactors.Dataare
obtained through different methods: structured questionnaires andinterviews,foodlossandwaste
diaries compiled directly by the VC actor, direct measurements by the researcher, and food scanning
methods,whichcanbeusedindevelopedretailmarkets.Thesemethodsarehighlyregion‐andcontext‐
specific,aremoreusefulindisentanglingtheoriginoflossalongthevaluechain,andtendtoprovidemore
insightsintocausesandpreventionpossibilities.Themostfamousestimatefordevelopingcountriesis
given by the African Postharvest  Losses Information System, which provides post‐harvest weight loss
estimates for cereal crops in Africasouth of the Sahara (APHLIS, 2014). According to APHLIS, FL from
productionandpost‐productionforcerealsliesbetween14.3and15.8percentoftotalproduction.Kader
(2009) reviews previous estimates of losses in both developingand developed countries and finds an
averageof32percentlossforfruitsandvegetables.OfficialEurostatdataareusedinthestudybyMonier
etal.(2010)toquantifylossalongdifferentstagesoftheVCfor27EUmemberstates;byexcludingwaste
attheagriculturalproductionlevel,Eurostatestimatesanannualaverageof89milliontonsofwaste(i.e.
179Kgpercapita).AstudybyWRAP(2010)analyzeswastefromtheUKfoodanddrinksupplychainand
findsthatacross processing, distribution, andconsumption, 18.4Miotonsoftotalfoodanddrinkare
wastedannuallyintheUK;householdsareresponsibleforthelargestshare,wasting22percentoftheir
purchases(WRAP,2009).
The main challenges for the use of these micro methods to estimatefoodlossiscostandtimeto
implement the studies, as well as the difficulty in getting a large enough proportion of responses to
representanentireVCorregion.Inaddition,resultsarehardtocomparebecausestudiesareadaptedto



their specific objective, focus only on specific stages of the VC, and use different data collection and
estimationmethodologies.
Figure 2 summarizes the two approaches to PFLW estimation, highlighting their advantages and
drawbacks.Figure3providesanoverviewofglobalPFLWmagnitudesfromrecentstudies,distinguishing
thetwoestimationapproaches.
7


Figure2:PFWLestimationmethodologies




7
Thisdoesnotintendtobeacompleteliteraturereview,butmerelyprovidesreferenceonestimatesfromprevious
research. We selected studies encompassingmorethanoneleveland/orcommodityofthevaluechain.Fora
completeliteraturereview,pleaseseeAffognon,2015;Fusions,2013;orKader,2009
Macro
approach



Figure3:OverviewofglobalPFLWmagnitudesfromrecentstudies




4. Proposedapproach
One main barrier to dealing with food loss and waste is the lack of clear knowledge regarding the
magnitudeoftheproblem(Lipinskietal.,2013).Uniformestimationmethodstoprovideconsistentloss
figures are necessary, but they alone will not be sufficient to identify the underlying causes of and
potentialsolutionstofoodlossortooutlineprioritiesforactionandmonitorspecificprogressonloss
reductiontargets.
First,astandarddefinitionandterminologyforfoodlossandwasteiscruciallyneeded.Thisdefinition
mustadoptavalue‐chainapproach,accountingforthefactthatconditionsatonestageofthechainlikely
affect losses and waste at later chain st ages. Specifically, this definition need s to include pre‐harvest
losses,astheirexclusioncouldleadtofoodlossreductioninterventionsthatdonottacklethesourceof
theproblem.Thisnewdefinitionmustincludebothquantitativeandqualitativereductioncriteria,exclude
natural, inedible, and unavoidable loss, and be able t o be measured in economic, caloric, or quality‐
adjustedweightterms.
Second,lossassessmentmustprioritizeanalysesthatidentifytheVCstagesatwhichlossesarecreated,
ratherthananalysesthatidentifyanexactoverallfigure.Lossmeasurementmustalsotakeintoaccount
theoriginoffoodreductionsalongthevaluechain,aswellastheirgeographicaldistribution.
Weproposeadevelopingcountrymethodologythatcanmeasurelossesatdifferentstagesofthevalue
chain and that can be applied across crops and regions. Specifically, we propose three alternative
methodologiesratherthanthetraditionally usedmethodologyofaggregate self‐reported measuresof
loss.Theanalysiswillbelimitedtolossesbetweentheproductionandprocessingstages,asthisiswhere
inefficiencies are largest in developing countries.Information willbe collectedthrough representative
surveysoffarmers,middlemen,andprocessors.Thesesurveyswillallowforthecharacterizationofinputs,
harvesting, storage, handling, and processing practices for each of these agents and will estimate the
quantities,quality,andpricesoftheproductionasittravelsalongthevaluechain.
Ourmethodologycapturesbothquantitativeandqualitativelosses,aswellasdiscretionarylossesamong
the processing, large distribution,andretailsectors.Foodwaste and household waste are more
challengingto capture, and dataneedto be collected onrepresentativesamples. This will require the
developmentofawidely`acceptedsamplingandmeasurementframework,whichwilllikelybecomposed
ofamixture ofmethods(e.g. wastecomposition analysis,questionnaires,interviews,orwaste diaries;
seeWRAP,2013).Thispaperdoesnotlookatfoodwaste.

5. Methodology
Wetestdifferentmethodologiestoestimatefoodlossalongthevaluechainbydrawingontheliterature
andeconomictheory.Ourmethodologiesareappliedtotheproducer,middleman,andprocessorlevelof
thevaluechaintocoverthemainstagesatwhichlossmightoccur.Duetotheheterogeneityofthecrop
transformationprocessesatlaterstagesinthevaluechain,atthewholesalelevel,onlytheaggregate‘self‐
reported’foodlossmeasurementmethodmightbeused.Allmethodologiesestimateboththetotalfood
thatislost(quantitativeloss)andtheproductthat,albeitnotbeingcompletelylost,isaffectedbyquality
deterioration(qualitativeloss).Thereferenceperiodisthelastcroppingseasonattheproducerlevel;for
themiddlemenandtheprocessors,itisadefinedtimeperiod(dependingonthecountry).



Self‐reportedmethod
Theaggregate‘self‐reportedmethod’(S‐method)isbasedonreportingbytheproducers,middlemen,and
processorsregardingthefoodlossesthey eachincurred.Self‐reportinghas beenwidelyusedinrecent
studiesonfoodloss(e.g.,KaminskiandChristiansen,2014;Mintenetal.,2016a;Mintenetal.,2016b).
Directsurveyquestionsinquireeachactorabouttheirquantitativeandqualitativelosses.Attheproducer
level,the survey instrumentincludesquestions about pre‐harvestandpost‐harvestlosses. Middlemen
andprocessorsareasked about losses atdifferentstagesofpost‐harvestactivities and transformation
processes.TableA1intheAppendixprovidesinsightsabouttheexactsurveyquestionsusedinthethree
surveyinstruments.Theresponsestothequestionsareaddeduptoobtainthetotallossfiguresinweight
andvaluesatthelevelofthethreevaluechainactors.
Categorymethod
The‘categorymethod’(C‐method)isbasedontheevaluationofacropandtheclassificationofthatcrop
intoqualitycategories.Themethodbuildsonthe‘VisualScale Method’, developed by Compton and
Sherington(1999)torapidlyestimatequantitativeandqualitativegrainloss.TheC‐methodclassifieseach
productintoitsenduse,i.e.suitableforexport,theformalmarket,theinformalmarket,animalfeed,etc.
Eachcategoryisassociatedwithacropdamagecoefficient,indicatingthepercentageofthecropthatis
damagedwithineach category. The categoriesareestablished prior to datacollection in collaboration
withcommodityspecialists,localexperts,andvaluechainactorsandvarybetweenfourandsix,according
tothecommodityandcountry.Inaddition,anextensivepilotwasconductedtovalidatethecategories.
By means of the described categories and damage coefficients, farmersareaskedtoevaluatetheir
production at harvest and after post‐harvest activities, while middlemen are asked to evaluate their
productatpurchaseandsales.Bothfarmersandmiddlemenindicateatwhichpricetheyselltheproduce
inthedifferentcategories,aswellasasalespricefor ideal produce inthehighandlowseason.At the
producer level, the quantitative and qualitative loss in weightandinvaluearegivenbyeq.1and2,
respectively:

(1)

(2)

whereciisthedamagecoefficientforcategoryI(wherethetotalnumberofcategoriesareI),
isthe
sampleaveragesalespriceforanidealproduct8,
isthesampleaveragesalespriceforaproductin
categoryi, andisthequantityineachcategoryafterpost‐harvest.andarerespectively
thequantityandvalueofallproduceafterpost‐harvest,whileandarethequantityandvalue
ofallproduceafterproduction.Thedifferenceinquantitiesorvalues(thesecondtermsofequation1and
2)provideuswiththetotalquantityorvaluelostbetweenproductionandpost‐harvestactivities.

8Averageacrossthelowandhighseason



 ∗
 
 





 

∗ 
 



Atthemiddlemanlevel,thequantitativeandqualitativelossinweightandinvaluearegivenbyeq.3and
4,respectively:

(3)

(4)

whereciisthesamedamagecoefficientasintheproducers’survey,
and
aretheaveragesale
priceforanidealproductandsalepriceforaproductincategoryiatthemiddlemenlevel,and
andarethequantitiesineachcategoryatpurchaseandatsale.Togetthefullquantitative
and qualitative loss measure, we add the weight (or value) of thequantitythatwastotallylost,i.e.
disappearedfromthevalue chain. This figure isideallyobtainedfromthedifferencebetween thetotal
purchaseandtotalsaleswithinagivenperiodoftime.Practically,middlemenareoftenunabletoindicate
theseexact quantities,as the purchasedcropismixedwithproductin storage.We therefore usethe
informationfromthedirectsurveyquestionregardingtheweightandvaluetotallylostatthemiddleman
level,i.e.productthatcompletelydisappearedfromthevaluechain.
Attributemethod
The‘attributemethod’(A‐method)isbasedontheevaluationofacropaccordingtoinferiorvisual,tactile,
andolfactoryproductcharacteristics.Theseattributesareidentifiedpriortothesurveyimplementation
and in collaboration with commodity experts, local experts, and value chain actors. In addition, an
extensivepilotwasimplementedtovalidatetheattributes9.Thenumberofattributesvariesbetween10
and14,accordingtothecommodityandcountry.Atthetimeofthesurvey,theproducerevaluateshisor
herproductionandestablishestheshareoftotalproductionthatisaffectedbytheattributes,bothafter
harvest and after post‐harvest. Middlemen evaluate their product from the previous month at both
purchaseandsale.Theproducerandthemiddlemendeclarehowmuchtheirrespectivebuyerspunish
themforinferiorproductattributesbypayingalowerprice.Thepricepunishmentinformationforeach
productattributeisusedtoestimatethevalueloss.Attheproducerlevel,thequantitativeandqualitative
lossinweightandinvaluearegivenbyeq.5and6,respectively:
(5)

(6)

whereistheshare ofproductaffectedbyattributejand




istheaveragepricepunishment foran
inferiorproductattributeatsale.Asbefore,andarerespectivelythequantityandvalueofall

9It isimportanttomention thatincertain countries;theattributesare definedaslegalstandards forthespecific
commodity.






∗
 
  


 






∗
  



 ∗
 









 ∗ 
 



produce after post‐harvest, while andare the quantity and value of all produce after
production.Whilethefirsttermsofeq.5and6provideuswiththequantityaffectedbyaloss(qualitative
loss), the second terms provide us with the total quantity or value lost (quantitative loss) between
productionandpost‐harvestactivities.
Atthemiddlemanlevel,thequantitativeandqualitativelossinweightandinvaluearegivenbyeq.7and
8,respectively:
(7)


(8)

where, and,arethequantitiesineachattributesold andpurchasedwithacertain
damageattribute,,and,arethevaluesatsalesandpurchasethatarelostdueto
a damage attribute (these are obtained by multiplying the previous quantities by the average price
punishment).Theweight(orvalue)ofthequantitythatwastotallylost(i.e. disappearedfromthevalue
chain)providesuswiththefullquantitativeandqualitativelossmeasure.
Pricemethod
The‘pricemethod’(P‐method)isbasedonthereasoningthathigher(lower)v aluesofacommo dityr eflect
higher(lower)quality.Adecreaseinprice,allelseequal,isthusaproxyforadeteriorationinquality.Data
regardingproducers’andmiddlemen’sidealsalevalueareusedandcomparedtothevalueoftheiractual
production,purchase, andsales.Thefollowingequationsprovide uswiththetotalloss attheproducer
level:
(9)
whereisobtainedbymultiplyingfarmers’productionbytheaverageidealsales’priceandis
thetotalvalueofthefarmers’productionafterpost‐harvest,asassessedbythefarmerhimself.Thevalue
losscanbetranslatedintoaweightlossbydividingitbytheidealsalesprice:

(10)
Forthe middlemen, wetakethedifferencebetweenthe value (orweight)affectedbylossatsalesand
thevalue(orweight)affectedbylossatpurchasetoestimatethetotalvalue(weight)affectedbylossat
thislevelofthechain.Thevalue(orweight)affectedbythelossatpurchaseorsaleisestimatedbytaking
thedifferencebetweenthesale(purchase)valueofanidealproductandtheactualsale(purchase)value.



 





, 
,

  WeightLost

, 
,

 ValueTotLost



Weaddtheweight(orvalue)ofthequantitythatwastotallylost(i.e.disappearedfromthevaluechain)
togetthefullquantitativeandqualitativelossmeasure.Thistranslatesintothefollowingtwoequations:

;;
; 
;  (11)


6. Data
Asmentionedinourliteraturereview,therehaverecentlybeeneffortstousemicrodatatoestimatefood
loss. These estimations rely on surveys collected among different actors along the food value chain;
however,theyarebasedoncasestudiesthatarenotrepresentativeofacountry’sbroaderpopulation.
Additionally, these studies use different definitions of food loss,whichhamperscomparisonsacross
different areas and crops. Due to this lack of representativeness, as well as to differences in their
methodologies, available micro‐based food loss estimates are widely variable and yield inconclusive
evidenceregardingtheextentoffoodloss.

Wehavedevelopeddetailedsurveysacrossthedifferentcomponentsofthefoodvaluechainandspecific
todifferentcommodities.Thesesurveysallowustoquantifytheextentoffoodlossacrossthevaluechain
usingconsistentapproachesthatarecomparableacrosscommoditiesandregions.Theyalsoenableusto
characterizethe nature offood loss, specifically theproduction stages and theparticular processes at
whichlossisincurred.

Our survey instruments quantify food loss along the value chain before consumption (food waste by
consumersis excluded fromthe calculations). The richnessof thedataallowsustoprovideestimates
using alternative methodologies. We first calculate aggregate self‐reported measuresofloss:weask
farmers,middlemen, and processorsaboutthe quantities (andthe corresponding monetaryvalues)of
cropsdiscardedduringtheprocessesthattheyperform(e.g.,winnowing,threshing,grading,transporting,
packaging, etc.). This methodologyis,ingeneral,consistentwith the basic elements in the available
literature on the measurement of food loss. Our surveys, however, include a more disaggregated
descriptionofthestagesandprocesses at whichlossoccurs.Theproducer,middlemen,andprocessor
surveysweredesignedtohavedifferentmodulestomeasurelossacrossthevaluechain.

Theproducersurveyhasthreemodules.Thefirstmoduleasksaboutthequantityofthecropleftinthe
field,thetotalproductionharvested,andthequalities,attributes,andpricesoftheharvest.Thesecond
moduleasks about the post‐harvestactivities conductedbytheproducers(e.g., winnowing, threshing,
grading,transporting,packaging,etc.);foreachoftheseactivities,theproducerisaskedforthequantity
ofaffected product 10 andthequantity totally lost.11 The third modulerecordsthe destinationof the
product(i.eforconsumption,forsale,fordonation,etc.),aswellastheattributesandcategoriesforthe
quantityforsale.

10Affectedproduct:Productthatlowersqualitybutcanstillbeused.
11Totallylost:Productthatiscompletelylostandcannotbeused.

; 
;
; 
; )(12)



The middlemen survey has three modules. The first module asks about the quantity, quality, and
attributesofthetotalproductpurchasedinadefinedtimeperiod(dependingonthecountry).Thesecond
moduleasksmiddlementoreportthequantity,quality,andotherattributesofthetotalproductsoldina
definedtimeperiod(dependingonthecountry).Thethirdmoduleasksquestionsaboutthepost‐harvest
processing activities conducted by the middlemen (e.g., winnowing, threshing, grading, transporting,
packaging,etc.);ineachoftheseactivities,thequantityofaffectedproductandthequantityoftotalloss
arereportedforeachcrop.

Theprocessorsurveyhastwomodules.Thefirstmoduleasksforthequantity,quality,andattributesof
thetotalproductpurchasedinaspecifictime‐period(dependingonthecountry).Thesecondmoduleasks
aboutthespecificstepsrequiredtoobtainthefinalproductforconsumerconsumption.

Withineach survey,we categorize thecrop damageand crop attributesforeachcropandcountry.In
ordertocategorizethedamageforeachcrop,wecreatedadamagecoefficient,measuredbycategorizing
thetotalamountofeachcropintodegreesofquality.Inoursurveys, each crop has its own damage
coefficient,whichwasdeterminedusingtheinternationalclassificationincollaborationwithlocalexperts.
FormaizeandbeansinHondurasandGuatemala,therearefivecategories,withcategory1classifiedas
having1‐2percentofdamagedgrain(grainwithnoproblems)andcategory5classifiedashavingmore
than25percentofdamagedgrain(grainthatisunusable).InEthiopia, the fivecategories range from
category1(undamagedgrain)tocategory5(morethan80percentofdamagedgrain).InEcuadorand
Peru,thecategoriesarerelatedtothecaliber12oft hetuber; crops cate gori zedascali ber1 havea diam eter
biggerthan10cm(CategoryExtra),whilecategory5consistsoftuberswithadiameteraround6cm,which
isusedtofeedanimals.InChina,eightcategoriesareconstructedbasedonthecrops’degreeofimpurity
(≤1percent and >1 percent)and degree ofsoundness of thekernel (<=6 percent;>6 percent and <=8
percent;>8percentand<=10percent;>10percent).

Theattributessectionofthesurveyevaluatesthecropsaccordingtophysicalorchemicalcharacteristics
tosee whether they have inferior visual, tactile,andolfactorycharacteristics.Thesecharacteristicsare
specifictoeachcountryandcrop.Inoursurveys,wemeasurethedamagetoeachcropbytexture,size,
moisture,and the presence of fungus or insects,etc.Theseattributecategorieswerecreatedwiththe
collaborationoflocalexperts.

One drawback to the aggregateself‐reported method is that it is reported by the farmers in a more
‘aggregate way’ through a direct question (see Appendix Table A1), which does not allow for the
identificationofwherealongthevaluechainthelossesoccurandthedifferentiationofwhichlossesare
ofqualityandwhichareofquantity.Whilefoodisnotnecessarilydiscardedcompletelyalongdifferent
processes,qualitydowngradesatdifferentstagesofthevaluechaincanaffectfood’seconomicvalue.Our
surveyinstrumentsimproveupon thesetraditionalmeasuresbyallowingustoquantifyqualitativeloss
usingtwoalternativemethods.First,weestimatethesharesoftotalfoodproductionateachstageofthe
valuechainthatwasdamagedandissubjecttoqualitativeloss(basedondamagecoefficients).Second,

12Caliber:Sizeofinternaldiameterofthetuber



wecollectinformationaboutdifferenttypesofcommodityattributes(e.g.,size,impurities,discoloration,
etc.) and ascertain the price penalty that each of these types of crop damage entails (i.e., attribute
penalties).Wearethusabletoidentifyspecificfactorsthatdiminishcommodities’valuesandtoquantify
foodqualitylossesbasedonmarketconditions.

Valuechainsanddescriptivestatistics
Forallcountries,wechoseoursamplebasedonapre‐censusoftheproducersofthespecificcropof
interest;thisformsourbaseline.Selectedproducersmusthaveproducedcropsinthelastseason.
PotatoesareessentialtotheEcuadoriandiet,witheachpersonconsumingaround30kgperyear
(MAGAP,2014).Thecropisthetenthmostconsumedproductsinthecountryandisoneofthetopeight
produced crops. Ecuador produces 397,521 tons of potatoes annually, with the province of Carchi
producing36.48 percentofthenationalvolume (ESPAC,2015).OursurveysinEcuadorwere organized
betweenJuneandOctober2016foreachsegmentofthepotatovaluechain.Allproducersinthesurvey
camefromtheprovinceofElCarch i,whilethemiddlemenwerefromtheprovincesofElCarchi,Imbabura,
andPichinchaandtheprocessorswerefromtheprovinceofPichincha.

Potatoeshavealsobeenessentialto the diet of Peruvians formillennia.Peru’sannualconsumptionof
potatoesisaround89kgperperson(MINAGRI,2016).Thecroprankssecondforthemostcultivatedcrop
areainPeru,with318,380hectaresplantedtopotatoand4,704,987metrictonsofpotatoesproducedin
2014(FAOSTAT).ThetwoprincipalprovidersofpotatoestotheLimamarketarethedepartmentsofJunín
and Ayacucho, which provide around 60 percent of the potatoes that go to the wholesale market
(EMMSA).OursurveysinPeruwereorganizedbetweenSeptemberandDecember2016foreachsegment
ofthepotatovaluechain.TheproducersinthesurveywerefromthedepartmentsofJunínandAyacucho,
whilethemiddlemenandprocessorswerefromthedepartmentofLima.
FortheCentralAmericanregion,maizeand beancropscomposestaplesforavarietyofreasons.These
cropsformthefundamentalbasisoffoodsecurityformuchofthepopulation,and they contribute to
householdandnationaleconomiesthroughemploymentgenerationandincomegeneration.
InHonduras,maizeisoneofthemostimportantbasicgrains,butthedomesticmaizesupplyonlycovers
42percentofthecountry’sdemand(SAG/UPEG,2015).TheannualconsumptionofmaizeinHondurasin
2013wasaround77.96kgper person, while the production of maizein2014was609,312metrictons
overanareaof263,343hectares(FAOSTAT).Thethreeprincipalproductiondepartmentsofwhitemaize
inHondurasareOlancho,ElParaíso,andComayagua.
BeansarethesecondmostimportantbasicgraininHonduras,bothinareaplantedandinproductionfor
consumption.In2014,the annual consumption of beans inHonduras was12.05kgper person and an
averageof132,659hectareswereplantedwithbeans.Beanproductionin2014was105,812metrictons
(FAOSTAT).ThethreeprincipalproductiondepartmentsforbeansinHondurasareOlancho,El Paraíso,
andYoro.
Oursurveys for Honduraswereorganized between JulyandSeptember 2016 foreachsegmentofthe
maizeandbeanvaluechains. The producers, middlemen, and processorsinthesurvey were from the
departmentsofCholuteca,Copan,ElParaiso,FranciscoMorazán,Intibucá,LaPaz,Lempira,Ocotepeque,
Olancho,SantaBarbara,andValle.



InGuatemala ,ma izeisthemo stw ide lyc ultivatedc ropandisoneofthemostvaluableandrootedsymbols
ofGuatemalanculture.In2014,theareacultivatedtomaizewas871,593hectares,withaproductionof
1,847,214 metric tons. Per capita consumption for 2013 was around 87.25 kg per person per year
(FAOSTAT). The three principal production departments of white  maize in Guatemala are Petén (18.5
percent),AltaVerapaz(9.4percent),andJutiapa(7.3percent)(MAGA,2016).
BeansarethesecondmostimportantbasicgraininGuatemala,both inareaplantedand inproduction
forconsumption.In2014,theconsumptionofbeansinGuatemalawas12.12kgperpersonperyear;area
planted to beans covered an average of 250,414 hectares, with production at 235,029 metric tons
(FAOSTAT).ThethreeprincipalproductiondepartmentsforbeansinGuatemalaarePetén(27percent),
Jutiapa (13 percent), and Chiquimula (10 percent) (MAGA, 2016). Our surveys in Guatemala were
organizedbetweenSeptemberandDecember2016foreachsegmentofthemaizeandbeanvaluechains.
The producers, middlemen, and processors were from the departments of Chimaltenango, Escuintla,
Guatemala,Quetzaltenango,Sacatepéquez,SanMarcos,Sololá,andTotonicapán.
TeffconstitutesamajorcropinEthiopia,intermsofbothproduction and consumption. Teff is the
dominantcerealcropfortotalareaplanted(3,760,000hectaresin2012/2013;FAS,2014)andsecondin
production and consumption, with 3,769,000 metric tons (Berhane, Paulos, Tafere and Tamru, 2011;
EthiopianAgriculturalTransformationAgency[EATA], 2013).AccordingtoBerhane,et al.(2011),based
on national data fro m the Household Income, Consumption and Expenditure Survey (HICES, 2011), in
2001‐2007,urbanconsumptionofteffpercapitawasashighas61kgperyear,whileruralconsumption
was20kgpercapitaperyear.TeffisgrownmainlyinAmharaandOromia,whichtogetheraccountedfor
84 and 86 percent of the total cultivated area and production in 2011. Our surveys in Ethiopia were
organizedbetweenAugustandOctober2016inthezonesofOromiaandAmhara;however,thesesurveys
forcoveredtheproducerchainonly,giventhatinthecaseofteff,therearenoimportantintermediaries
andprocessors.
WheatisthesecondmostimportantfoodcropinChinafollowingrice.Itisthedominantstaplefoodin
the northern part of the country, where it is used mainly to produce noodles and steamed bread.
(CIMMYT).In2014,Chinaproducedabout120millionmetrictonsofwheateachyearonapproximately
24millionhectaresofland(FAOSTAT).AccordingtoFAOSTATin2013,theannualconsumptionofwheat
percapitainChinawasaround63.1kgpercapita.MostofChina’swheatproductioncomesfromNorth
China;threenorthernprovinces–specificallyHenan,Shandong,andHebei‐collectivelyaccountforover
50percentofChina’swheatoutput(China'sStateStatisticalYearbook,2001).OursurveysinChinawere
organized between August and October 2016 for each segment of the value chain. The producers,
middlemen,andprocessorswerefromtheprovincesofHenanandShandong.
Weadaptedourinstrumentforthespecificationsofeachcropandcountry.Forexample,inEcuadorand
Peru,weworkwithpotatovaluechains;inthesecases,theinstrumenthassixdifferentcategoriesand
ninedifferentattributes.InGuatemalaandHonduras,wherewework with themaize and beanvalue
chains,theinstrumenthasfivedifferentcategoriesand12differentattributes.InEthiopia,weworkwith
the teff value chain, in which the instrument has five different categories and 12 different attributes.
Finally,inChina,weworkwithwheatvaluechain,andtheinstrumentshaseightdifferentcategoriesand
sixdifferentattributes.




Theformulausedforcalculatingtherepresentativerandomsampleforallthecountriesis:

 (13)

wheren=thesamplesizerequiredandwhichisstatisticallyrepresentative,N=thetargetpopulation
size,e=toleratedmarginoferror(forexample,wewanttoknowtherealproportionwithin5percent),
Z=levelofconfidenceaccordingtothestandardnormaldistribution(foralevelofconfidenceof95
percent,z=1.96,foralevelofconfidenceof99percent,z=2.575),andp=estimatedproportionofthe
populationthatpresentsthecharacteristic(whenunknownweusep=0.5)

Inastratifiedrandomset‐up,wesampledamoderatenumberofactorspersegmentineachcountry.At
theend,thesampleconsistedof:

Table1:Samplesize 

  Ecuador  Peru  Honduras  Guatemala  Ethiopia China

     
Producer 302 411 1209 1155 1203 1114
Middlemen 182 85 325 365 ‐‐‐ 140
Processor 147 139 224 245 ‐‐‐ 53

Total  631  594  1758  1765  1203 1307

Specifically,inthecaseofteffinEthiopia,weonlysurveyproducersbecausemostoftheproducerswill
bringtheirtefftomillerswhoworkmostlyon a fee‐for‐servicebasis,returning milled teff flour tothe
producerswithoutanymajorintermediationofmiddlemen.
Tables2‐4pro videdescriptivestatisticsofthesamp leofeach differentcropineachcountryforproducers,
middlemen,andprocessors,respectively.
InTable2,wecanseethatforallcountries,themajorityofproducersaremaleandhavereachedatleast
aprimarylevelofeducation.TeffproducersfromEthiopiaaretheyoungestonaverage,whileChinese
wheatproducersaretheoldestandhavethemostyearsofexperienceworkingwiththeircrop.Morethan
70percentofproducersfromEthiopiaandChinausedimprovedseedsinthelastcropseason(forteffand
wheat,respectively);43percentof producers usedimproved seedsinPeru,whiletheuseofimproved
seedsislessthan20percentinEcuador,Honduras,andGuatemala.PotatoesinPeruandEcuadorwere
storedforshorterperiodsoftimecomparedtograinsinalloftheotherstudycountries.

InTable3,wecanseethatforallcountries,around60percentofmiddlemenaremale,withanaverage
agebetween 40 and 50years.The average numberofyears that middlemenhave beeninbusiness is
higherfor middlemenbuyingandsellingpotatoesinEcuadorandPeruthanformiddlemen buyingand
sellingmaizeandbeansinGuatemala,Honduras,andChina



Across all countries, middlemen purchased more commodities from producers than from other
middlemen.Thiscouldbeduetothefactthatpricesfromproducersmaybecheaperandproducersmay
bemorelikelytoseekoutmiddlemeninthebigcities.
InTable4,we canseethatthemajorityofpro cessorsinPeru andEcuadoraremale,andthemainproducts
tradedareFrenchfries.InChina,almostallprocessorsaremale,andthemainproductsarenoodlesand
steamedbread.InHondurasandGuatemala,themajorityofprocessorsarefemale,andthemainproducts
tradedaremaizetortillasandpackagedbeans.Forallcountries,theaverageageofprocessorsis40years.

In Peru and Ecuador, all of the potato processors’ businesses are formal (legal) and in China, a large
majorityareformal;however,formaizeandbeanprocessorsfromGuatemalaandHonduras,somewhat
lessthan40and60percent,respectively,areinformal.




Table2:Producercharacteristics


Note:aThisincludesfertilizers,insecticides,herbicides,andfungicides;bThisincludesactivitiessuchasirrigation,trimming,andpruning;cMachine‐driven,insteadofmanual,
includeactivitiessuchassoil preparation, sowing, pest control, fertilizer application, weeding, mulching, cutting and harvest;d Thisincludes activitiessuch asselection,
classification, drying, etc. e This includes activities such as chemical fumigation, natural fumigation, and ventilation; f storage summary statistics are obtainedfrom the
restrictedsample offarmersstoring grains;gThesevariablesarenot mutually exclusive,asfarmers canhavemorethanonesaleslocation andtypeof buyer.Theofficial
exchangerateintheyearofthesurveyare0.04492USD/Birr;0.1305USD/Quetzal;0.0411USD/Lempiras;0.297USD/Soles;0.155USD/Yuan(www.oanda.com)



Table3:Middlemancharacteristics


Note:Theofficialexchangerateintheyearofthesurveyare0.04492USD/Birr;0.1305USD/Quetzal;0.0411USD/Lempiras;0.297USD/Soles;0.155USD/Yuan
(www.oanda.com)




Table4:Processorcharacteristics


Note:Theofficialexchangerateintheyearofthesurveyare0.04492USD/Birr;0.1305USD/Quetzal;0.0411USD/Lempiras;0.297USD/Soles;0.155USD/Yuan
(www.oanda.com)













7. Results
AsshowninTable5,weestimate loss levels at the producer, middlemen, and processor levels
separatelyandalternativelyapplythefourestimationmethodologies,i.e.subjective(S),category(C),
attributes(A),andpricemethod(P).Weusethelossfiguresestimatedwiththeattributemethod(A‐
measure)asourdependentvariableandadduplossesateachlevelt oobtainlossfigure sfort heentire
valuechain.13Someobservationsarelostduetomissingvalues and outliers.14 Loss figuresinclude
both the quantitative loss, i.e. the product entirely disappeared from the value chain, and the
qualitativeloss,i.e.theproductaffectedbyqualitydeteriorations.Lossesarealternativelyexpressed
inweightandvalues,withthelatterprovidinginformationregardingtheeconomicdamagecausedby
theloss.AppendixApresentsadetaileddecompositionofallthemethodsbycommodityandcountry
attheproducerlevel.
Lossfiguresacrossallvaluechainsfluctuatebetween6and25percentoftotalproductionandofthe
totalproducedvalue.Loss figures areconsistentlylargestattheproducerlevelandsmallest at the
middlemanlevel.Acrossthedifferentestimationmethodologies,lossattheproducerlevelrepresents
between60and80percentofthetotalvaluechainloss,whiletheaveragelossatthemiddlemanand
processorlevelsliesaround7and19percent,respectively.Itisimportanttomentionthattheselosses
donot include yieldgaps,whichcouldvarybetween 50and80percent.Theseyield gapsrepresent
thedistance to the production possibilityfrontier, defined asthedistanceofthe sale quantities or
pricesandthefrontier(seeDelgadoet.al2017forfurtherdetails).
Differencesacrossmethodologiesare salient,especiallyattheproducerlevel.Whiletheestimation
resultsfromtheC‐,A‐,andP‐methodsarecloseanddifferencesaremostlynotstatisticallysignificant,
theaggregateself‐reportedmethodreports systematically lower lossfigures.As shown in Table5,
thesegapsare largest in thebeansvalue chain in Hondurasand the potato value chaininPeru,in
which self‐reported loss estimates are between 10 and 15 percentage points lower than those
estimatedwithanyoftheothermethods.DifferencesacrossmethodsaresmallestintheEthiopian
teffvaluechain,butestimatesfromtheC‐,A‐,andP‐methodsremainsignificantlylargerthanthose
estimatedwiththeS‐method.
Percentagelossesexpressed invaluetendtobe slightly smaller thanthoseexpressedinweightfor
theS‐method;however,thisdifferenceis found particularlyintheA‐method,indicatingthatsome
quality degradations at the farm‐level do not seem to be punished by the market. The category‐
methodleadstoresultswhicharemoresimilarintermsofweightandvalueloss.
TablesA2–A9intheAppendixsplitlossfiguresattheproducerlevelintoquantitiesleftinthefield,
(i.e., good quality product which is not harvested), quantities affected by quality deterioration
previous to harvest, and quantitiestotallylostoraffectedby quality deteriorations during post‐
harvestactivitiesonthefarm.Thelattercanincludecleaning,winnowing,threshing,drying,storage,
transportactivities,etc.,dependingonthevaluechainandcountry.Thequantitiesleftinthefieldare
fairlysmall,ataround1percentoftotalproduction,orareeven negligible in the caseof teff and
wheat.Thepercentagevalueoftheunharvestedproductintermsofthetotalproducedvalueiseven
smaller,indicatingthattheproductleftinthefieldtendstobeoflowerqualitythantheharvested
product.Overall,thequantityaffectedbylossatpre‐harvestisconsiderablylargerthanthequantities

13Forthemiddlemenandprocessors,weassumethatthepercentagelostontheirpurchaseinthemonthprior
tothesurveycorrespondstotheaveragemiddlemanandprocessorlossinthevaluechain
14Weusea‘‘winsorizing”technique,replacingextremeoutliersbeyondthe99thpercentilewithmissingvalues
undertheassumptionthatallextremevaluesareduetomeasurementerror



totallylostoraffectedby a loss during post‐harvest activities. This indicatesthatthe largest losses
occurinthefieldorduringharvestactivities.
WiththeexceptionofthebeanvaluechaininHonduras,lossfiguresacrossmethodologiesaresimilar
andnotstatisticallydifferentformiddlemen.Atthewholesalelevel,lossesfluctuatebetween2and3
percent.
Causesbehindtheloss
Figure4(a‐h)presentsthemajorreasonsreportedbyfarmersastheexplanationfortheirpre‐harvest
loss,theircropleftinthefield,andtheirpost‐harvestloss.Inthespecificcaseofpre‐harvestloss,the
major reasons reported by farmers included pests and diseases andlackofrainfall;teffwasthe
exception,withlackofrainfallbeingthemajorreportedreasonforpre‐harvestloss.Whenlookingat
the produce left in the field, the major reason for the loss is a lack of appropriate harvesting
techniques.Finally,thelossreportedatthepost‐harvestlevelisduemostlyto damagedoneduring
selection,asaresultofworkers’lackoftrainingandexperienceinselectingtheproduce.
Tables 6‐10 try to control for the heterogeneity among farmer characteristics through regression
analysis.Theresultsshow that education and experiencetendtobecorrelatedwith areductionin
losses.Inparticular,educationissignificantforthepotatovaluechaininEcuadorandPeru,themaize
valuechaininHonduras,andthewheatvaluechaininChina.Thenumberofyearsinwhichaproducer
hasbeeninvolvedintheproductionofaspecificcropsignificantlycorrelateswithareductioninlosses
inthepotatovaluechaininEcuadorandPeru,themaizevaluechaininGuatemala,andtheteffvalue
chaininEthiopia.Whileweonlyhavefarmers’incomedataforPeruandEcuador,wefindthatwhen
a producer’s main income stems from an agricultural activity, it is correlated with a statistically
significantlowerloss;thisresultisinlinewiththeeffectswefindforcropcultivationexperience.
The large majority of farmers are men, but there is no clear gender pattern in food loss across
countries.Forexample,beingamalefarmertendstobecorrelatedwithadecreaseinbeansloss,but
itincreasesmaizelossinGuatemala.Nogendereffectisdetectedintheothercommoditychains.
Costs to reach markets are significantly correlated with increased losses in Peru, Guatemala, and
Ethiopia,indicatingthattheabsenceofmarketscanrepresentimportantlimitationsforfarmers.This
directlysupportspreviouswork,whichshowstheimportanceofaccesstobetterroadstoreducefood
lossacrossthevaluechain(see,forexample,Rosegrantet.al,2015).
Technology and improved seeds also matter. The more resistant pests andweather‘unica’ potato
varietyreducelossinEcuadorcomparedtothe‘capiro’and‘superchola’varieties.Similarly,theuse
of improved seeds is correlated with a decrease in losses in the maize and bean value chains in
Honduras.In potato valuechains,the harvesting toolusedconsiderably impactsloss;for example,
traditionalhoesbreakthe potato during theharvest.InPeru, new (mechanized) toolsareusedto
reducethisdamage.Boththetractorandthe‘lampa’arecorrelatedwithasignificantreductionofthe
shareof potatothatislost duringharvest.Thepotatovalue chaininEcuador,ontheotherhand,is
moretraditional,withveryfewmechanicaltoolsused.InEcuador,noalternativetoolstothehoewere
mentionedbythesurveyedfarmers.InEcuador,anincreasednumberofactivitiesto‘takecareofthe
crop’(suchasirrigationandplanttrimming)andalargerlaborforceareshowntoreducethelikelihood
oflossinthismoretraditionalpotatovaluechain.
Inthemaize,bean,andteffvaluechainsunderanalysis,productionactivitiesareshowntohavelittle
impact on food loss. The exception is the bean value chain in Guatemala, where mechanical
productionactivitiesareshowntobepositivelycorrelatedwithincreasedloss;mechanicalharvesting



techniqueslikelydamagethe crop and/or leavecropsinthefield(especiallyif the machinesareof
poorquality).
Whenanalyzinghowthetypeandnumberofpost‐harvestactivitiescarriedoutbythefarmersaffect
loss, we found that both the overall number of post‐harvest activities and the increased
mechanizationinsomecommoditychainscanhaveoppositeeffects.Thetotalnumberofpost‐harvest
activities,includingactivitiessuchaswinnowing,threshing,drying,puttinginbags,transporting,etc.,
decreaseslossintheGuatemalanbeanandtheChinesewheatvaluechains,butincreaseslossinthe
Guatemalan maize value chain and the Ethiopian teff value chain. In both the latter cases, the
increasedlossoriginatesmainlyfrompost‐harvestwinnowingandpackagingactivities.
Mechanicalpost‐harvestactivitiesarenotverywidespread,withmechanicaldrying,winnowing,and
threshing activities only being observed in the maize and bean value chains in Honduras and
Guatemala. Post‐harvest mechanization has no effecton maize valuechains in either Hondurasor
Guatemala. In the bean value chain, on the other hand, increased mechanization of drying and
winnowingactivitiesreduceslossinGuatemala,butmechanicalthreshingincreaseslossinHonduras.
Farmers likely cause grain damage, cracks, and lesions when mechanically (instead of manually)
stripping the grain fromthe plant; thismakes the grainmore vulnerable to insects, aswell as less
visuallyappealing.Onlyaveryfewfarmers(6percentofoursample)engageinmechanicalthreshing
in Honduras (and no pr oducers do so in Guatemala). Mechan ical transport with a car significantly
increaseslossinGuatemalaandEcuador,pointingtoimportantlossesduringtransport,especially if
largerdistancesaretraveled.
PotatofarmersinPeruandEcuadorrarelystoretheirproduct,buttheoppositeistruefortheother
commoditychains.Storagesignificantlyincreaseslossinthebean value chains in Honduras and
Guatemala,aswellasinthemaizevaluechain inHondurasandthewheatvaluechaininChina.For
beansinHondurasandwheatinChina,storagedurationissignificantly correlated withincreasesin
losses. These storage losses are shown t o be mitigated by improved storage techniques (silos) in
Honduras,Guatemala, andChina,the use of ‘pits’ratherthan othertraditionalstoragefacilitiesin
Ethiopia(nomodernstoragetechniquesareusedforteffinEthiopia),and‘bag’versus‘bulk’storage
in China. Storage conservation activiti es, such as chemical or natural fumigation and/or increased
ventilation,arecorrelatedwithdecreasedstoragelossesinHonduras.
Finally,unfavorableclimaticconditionsandpestanddiseasesarementionedmostoftenasproblems
facedbyfarmersduringproduction.Farmersmostoftenmentionedlimitedknowledgeandaccessto
equipment,credit,andmarketsasachallengetoincreasedproductionofhigherqualityproducts.All
ofthesefactorsarealsoshowntoaffectfoodlosses.
8. Conclusions
Improvingthemethodologyusedtomeasurefoodlossacrossfoodvaluechains,aswellasidentifying
the causes and costs of loss across value chains, is critical to promoting food loss reduction
interventionsandsettingprioritiesforaction.
Weaddresstheexistingmeasurementgapbydevelopingandtestingthreenewmethodologiesthat
aimtoreducemeasurementerrorandthatallowustoassessthemagnitudeoffoodloss.The methods 
accountfor lossfrompre‐harvesttoproduct distributionandincludebothquantitylossand quality
deterioration.Weapplytheinstrumenttoproducers,middlemen,andprocessorsinsevenstaplefood
valuechainsinfivedevelopingcountries.Comparativeresultssuggestthatlosses arehighestatthe
producerlevelandthatmostproductdeteriorationoccurspriortoharvest.Self‐reportedmeasures,
whichhavebeenfrequentlyusedintheliterature,seemtoconsistentlyunderestimatefoodloss.Loss



figuresacrossallvaluechainsfluctuatebetween6and25percentoftotalproductionandofthetotal
produced value. Loss figures are consistently largest at the producerlevelandsmallestatthe
middleman level. Across the different estimation methodologies, losses at the producer level
representbetween60and80percentofthetotalvalue chain losses, while the average loss at the
middlemanandprocessorlevelsliesaround7and19percent,respectively.
Differencesacrossmethodologiesare salient,especiallyattheproducerlevel.Whiletheestimation
resultsfromthethreenewmethodsweimplementarecloseandthe differences are mostly not
statisticallysignificant,theaggregateself‐reportedmethodreportssystematicallylowerlossfigures.
Inaddition,ourfiguresarelargerthanthoserecentlyobtained byKaminskiandChristiansen(2014)
andMintenetal.(2016aandb). These differences are due to the inclusionof qualitative loss(not
previouslyconsidered)andtothefactthatwealsoincludequalityandquantityeffects.
Addressingfoodlossacrossthevaluechainfirstrequiresacommonunderstandingoftheconceptby
allactors,15aswellasacollaborativeefforttocollectbettermicro‐dataacrossdifferentcommodities
andcontexts.Thepresenceofpests,lackofrainfall,andlackofappropriatepost‐harvesttechnologies
seemtobethemajorfactorsbehindthelossesidentifiedinourstudy.Alackofappropriatestorage
facilities(FAO,2011;Liu,2014)andefficienttransportsystems(Rolle,2006)arealsoconsideredtobe
importantmicro‐causesoffoodloss;however,othercauses,rangingfromcropvarietychoices,pre‐
harvestpests,andprocessingandretaildecisions,arealsoimportant.Micro‐causescanbelinkedto
broadermeso‐causes,overarchingdifferentstagesofthevaluechain;forexample,theHLPE report
(2013) sees credit constraints as one of the main bottlenecks to the successful adoption of
technologiestoreducefoodlossandwaste.LikeKaminskiandChristiaensen(2014),wealsoidentify
alackofeducationasanimportantbottleneck.
Finally,policymakersandvaluechainactorsneedtotranslatetheseinsightsintoaction.International
organizations have the power to bring the important topic of foodlosstothetableandcreate
platformsforinformation exchange;atthesame time,individualstatesplaya keyroleincreating a
successfulenablingenvironment.Allpublicandprivatevaluechain actorsneedtoworktogetherto
transformtheoryintoconcretePWLFreductioninterventions.



15A good step in this directionhasbeenmadebythe multi‐stakeholder“FoodLossandWasteStandardand
Protocol”initiative,althoughthisinitiativedoesexcludepre‐harvestlossfromitsdefinition.



Figure4:Self‐ReportedCausesofofPre‐HarvestLosses
Figure4.a:PotatoEcuador

Figure4.b:PotatoPeru

55.87%
25.7%
11.17%
7.263%
Other pest ; disease; anima ls Little ra in
Lack or ex cess of inputs Freeze
Source:own data collecti on from 302 producers in 2016
Ecuador, Potato - Reason for Pre-Harvest Loss
22.45%
36.73%
24.49%
16.33%
Bad harvest technique Small or bad quality potato
Lack or cos tly labor Low price
Source:own d ata collection from 302 produc ers in 2016
Ecuador, Potato - Reason for product left in the field
62.78%
9.444%
14.44%
10%
3.33%
Laborer da mages at harvest Laborer damages at selection/cla
Climate, too much sun or r ain Transpor t
Plagues, rodents, animals
Source:own data collecti on from 302 producers in 2016
Ecuador, Potato - Reason for loss at Post-Harvest
34.82%
32.09%
10.07%
23.02%
Other pest , disease, anima ls Little r ain
Lack or ex cess of inputs Freeze
Source:own data collecti on from 411 producers in 2016
Perú, Potato - Reason for Pre-Harvest Loss
43.63%
10.78%
35.29%
10.29%
Bad harvest technique Small or bad quality potato
Lack or cos tly labor Low price
Source:own d ata collection from 411 produc ers in 2016
Perú, Potato - Reason for product left in the field
34.02%
12.2%
13.23%
25.43%
12.2%
2.921%
Laborer da mages at harvest Laborer damages a t selection/cla
Climate, too much sun or r ain Transpo rt
Plagues, rodents, animals Lack of labor
Source:own data collecti on from 302 producers in 2016
Perú, Potato - Reason for loss at Post-Harvest



Figure4.c:BeansGuatemala


Figure4.d:BeansHonduras

21.59%
27.58%
48.89%
Excess rain Little rain
Other pest; disease; animals Excess chemicals
Wind Stolen
Source:own data collecti on from 450 producers in 2016
Guatemala, Beans - Reason for Pre-Harvest Loss
81.45%
18.55%
Bad harvest technique Lack or costly labor
Source:own data collecti on from 450 producers in 2016
Guatemala, Beans - Reason for product left in the field
22.56%
44.78%
18.86%
8.75%
5.05%
Plagues, rodents, animals Laborer damages at selection/cla
Laborer da mages at harvest Climate, too much sun or ra in
Storage
Source:own data collecti on from 450 producers in 2016
Guatemala, Beans - Reason for loss at Post-Harvest
14.83%
35.24%
48.75%
Excess rain Little rain
Other pest; disease; an imals Exce ss chemicals
Wind
Freeze
Stolen
Source:own d ata collection from 685 produc ers in 2016
Honduras, Beans - Reason for Pre-Harvest Loss
98.86%
1.14%
bad harve st technique lack or costly labo r
Source:own data collecti on from 685 producers in 2016
Honduras, beans - Reason for product left in the field
7.03%
65.69%
21.41%
2.93%
2.93%
Plagues, rodents, animals Laborer damages at selection/cla
Laborer da mages at harvest Climate, too much sun or ra in
Storage
Source:own data collecti on from 685 producers in 2016
Honduras, Beans - Reason for loss at Post-Harvest



Figure4.e:MaizeGuatemala


Figure4.f:MaizeHonduras

19.61%
27.65%
43.09%
7.87%
Excess rain Little rain
Other pest; disease; animals Excess chemicals
Wind Freeze
Stolen
Source:own data collecti on from 922 producers in 2016
Guatemala, Maize - Reason for Pre-Harvest Loss
81.9%
17.65%
Bad harvest technique Lack or costly labor
Low price
Source:own data collecti on from 922 producers in 2016
Guatemala, Maize - Reason for product left in the field
36.06%
35.62%
12.85%
9.19%
5.10%
Plagues, rodents, animals Laborer damages at selection/cla
Laborer da mages at harvest Climate, too much sun or ra in
Storage Transport
Source:own data collecti on from 922 producers in 2016
Guatemala, Maize - Reason for loss at Post-Harvest
9.51%
40.52%
46.62%
Excess rain Little rain
Other pest; disease; animals Excess chemicals
Wind Stolen
Source:own data collection from 1024 producers in 2016
Honduras, Maize - Reason for Pre-Harvest Loss
95.33%
4.67%
Bad harvest technique Lack or costly labor
Source:own data collection from 1024 producers in 2016
Honduras, Maize - Reason for product left in the field
10.13%
64.35%
9.56%
4.69%
9.38%
Plagues, rodents, animals Laborer damages at selection/cla
Laborer da mages at harvest Climate, too much sun or ra in
Storage Transport
Source:own data collection from 1024 producers in 2016
Honduras, Maize - Reason for loss at Post-Harvest



Figure4.g:TeffEthiopia









10.14%
50.29%
13.63%
2.23%
19.36%
Excess rain Little rain
Other pest; disease; animals Crop lodging
Crop shattering Weeds
Excess of chemicals Freeze
Source:own data collection from 1203 producers in 2016
Ethiopia, Teff - Reason for Pre-Harvest Loss
3.08%
16.01%
27.05%
53.86%
Laborer damages at harvest Climate, too much sun or rain
Storage Blow out
Source:own data collection from 1203 producers in 2016
Ethiopia, Teff - Reason for loss at Post-Harvest



Figure4.h:WheatChina












2.93%
35.15%
37.93%
10.45%
3.08%
5.78%
4.67%
Excess rain Little rain
Other pest; disease; animals Crop lodging
Weeds Freeze
Crop shattering
Source:own data collection from 1114 producers in 2016
China, Wheat - Reason for Pre-Harvest Loss
55.56%
44.44%
Bad harvest technique Wheater
Source:own data collection from 1114 producers in 2016
China, Wheat - Reason for product left in the field
29.48%
6.07%
2.18%
7.72%
40.57%
4.30%
8.60%
Laborer da mages at harvest Laborer dam ages at hulling
Climate, too much sun or rain Storage
Plagues, rodents, animals Machine damage
Transport Blow away / spilled
Source:own data collection from 1114 producers in 2016
China, Wheat - Reason for loss at Post-Harvest



Table5:Quantitativeandqualitativefoodlossesalongthevaluechain,estimatedwithfourmethodologies

Note:S=Self‐reportedmethod,C=Categorymethod;A=Attributemethod;P=Pricemethod;^Dataareimputedfromthe'Self‐reportedmethod’
QuantitativeLoss==Totalloss(productdisappeared);QualitativeLoss=Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)
Theofficialexchangerateintheyearofthesurveyare0.04492USD/Birr;0.1305USD/Quetzal;0.0411USD/Lempiras;0.297USD/Soles;0.155USD/Yuan(www.oanda.com)



SCAPSCAP SCAPS CAPSCAP SCAPSCAP SCAP
Nbofobs ervati ons
Kglost 1,498 5,926 4,982 4,146 3,548 9,216 11,523 7,998 7.47 16.01 24.79 26.59 55.67 137.74 194.93 191.24 26.47 66.96 114.16 129.69 78.61 186.08 198.19 284.26 27. 44 57.02 127.90 47.59 303 934 1,092 1,033
%oftotalproductionthat
islost 8.11% 12.82% 12.17% 11.84% 9.38% 15.99% 19.62% 19.84% 9.77% 12.80% 19.67% 16.72% 9.84% 14.58% 20.46% 15.27% 6.25% 13.27% 19.77% 17.39% 9.95% 16.69% 15.95% 17.41% 6.88% 8.67% 19.76% 8.69% 6.56% 10.96% 11.89% 11.48%
Value lost( USD) 269 1,543 1,007 990 454 2,116 2,202 1,805 8.24 26.31 32.64 38.42 18.37 60.20 55.68 82.88 18.56 73.73 90.01 116.53 23.30 65.43 65.19 99.12 40.24 97.98 91.03 73.91 104 316 360 349
%ofvalueoftotal
productionthatislost 6.22% 13.78% 10.03% 11.84% 5.58% 16.73% 16.13% 19.84% 7.72% 12.95% 17.97% 16.72% 7.57% 15.04% 13.42% 15.27% 5.23% 15.34% 17.56% 17.39% 8.87% 16.64% 15.41% 17.41% 6.26% 9.49% 9.02% 8.69% 6.11% 10.96% 12.37% 11.48%
Nbo
f
observations
Kglost 952 541 2,893 1,222 2,048 1,392 5,777 5,575 2.44 2.59 2.48 2.28 9.15 8.47 6.90 6.46 12.64 8.63 19.32 19.31 14.04 19.30 23.92 21.07 35,155 33,363 32,434 35,290
%oftotalpurchasethatis
lost 1.70% 0.91% 1.77% 1.52% 1.22% 1.60% 3.72% 2.05% 0.63% 0.66% 0.58% 0.57% 0.80% 0.54% 0.50% 0.55% 0.74% 0.55% 0.93% 1.57% 0.60% 0.59% 0.29% 0.65% 1.93% 2.09% 1.75% 1.92%
Value lost (USD) 232 284 685 518 517 49 2 1,266 2,704 3.99 3.64 3.69 3. 15 3.77 3.20 2.28 2.40 8.75 12.93 20.10 20.86 7.16 5.31 8.27 8.13 5,726 5,167 5,269 5,228
%ofvalueoftotal
purchaseth atislo st 1.36% 1.65% 1.55% 1.91% 1.34% 1.49% 2.89% 2.83% 0.78% 0.67% 0.67% 0.62% 0.83% 0.50% 0.45% 0.60% 0.45% 1.08% 1.58% 1.83% 0.63% 0.41% 0.31% 0.72% 1.62% 1.79% 1.54% 1.42%
Nbofobs ervati ons
K
g
lost 0.83 0.83^ 0.83^ 0.83^ 59.31 59.31^ 59.31^ 59.31^ 2.44 2.44^ 2.44^ 2.44^ 24.76 24.76^ 24.76^ 24.76^ 2.43 2.43^ 2.43^ 2.43^ 21.40 21.40^ 21.40^ 21.40^ 128
,
889 128
,
889^128
,
889^128
,
889^
%oftotalpurchasethatis
lost 2.45% 2.45%^ 2.45%^ 2.45%^ 2.27% 2.27%^ 2.27%^ 2.27% ^ 2.94% 2.94% ^ 2.94%^ 2.94%^ 3.50% 3.50%^ 3.50%^ 3.50%^ 3.67% 3.67%^ 3.67%^ 3.67%^ 3.82% 3.82%^ 3.82%^ 3.82% ^ 3.15% 3.15%^ 3.15%^ 3.15% ^
Value lost (USD) 14.59 14.59^ 14.59^ 14.59^ 41.22 41.22^ 41. 22^ 41.22^ 3.62 3.62^ 3.62^ 3.62^ 9.38 9.38^ 9.38^ 9.38^ 1.09 1.09^ 1.09^ 1.09^ 6.84 6.84^ 6.84^ 6.84^ 42,133 42,133^ 42,133^ 42,133^
%ofvalueoftotal
purchaseth atislo st 2.27% 2.27%^ 2.27% ^ 2.27%^ 3.31% 3.31%^ 3.31%^ 3.31%^ 3.42% 3.42%^ 3.42%^ 3.42%^ 2.88% 2.88%^ 2.88%^ 2.88%^ 1.96% 1.96%^ 1.96%^ 1.96% ^ 3.75% 3.75% ^ 3.75%^ 3.75%^ 3.06% 3.06%^ 3.06% ^ 3.06%^
%oftotalproductionthat
islost 11.50% 16.18% 16.39% 15.80% 12.87% 19.86% 25.62% 24.17% 13.34% 16.40% 23.19% 20.23% 13.53% 17.99% 23.84% 18.70% 8.95% 17.49% 24.37% 22.63% 14.37% 21.10% 20.06% 21.88% 6.88% 8.67% 19.76% 8.69% 11.64% 16.21% 16.79% 16.55%
%ofvalueoftotal
productionthatislost 9.86% 17.71% 13.85% 16.02% 10.23% 21.53% 22.32% 25.97% 11.93% 17.05% 22.06% 20.76% 11.88% 19.07% 17.42% 19.32% 7.65% 18.39% 21.11% 21.18% 13.24% 20.81% 19.47% 21.88% 6.26% 9.49% 9.02% 8.69% 10.79% 15.82% 16.97% 15.96%
Entirevalue
chain
Ethiopia:teffGuatemala:maizeGuatemala:beans
Producer
Midd lemen
Processor
Ecuador:potato
287
176
146
225
121
988
121
118
Honduras:beansPeru:potato
355
81
152
431 884
150162
120 104
650
China:wheat
1,099
137
47
Honduras:maize
‐
‐
1,186



Table6:DeterminantsoflossesinthepotatovaluechainsinEcuadorandPeru(GLMmodel);
Dependentvariable:shareofproductlostatpre‐harvestandpost‐harvest(A‐measure)

Note:MarginaleffectsfromGLMmodelsare reported.Standard errorsinparenthesisclusteredattheprovince levelforPeru
andatthecantonlevelforEcuador.aThisincludesfertilizers,insecticides,herbicidesandfungicides;bThisincludesirrigation,
'aporque' and corte del yuyo;cMachinedriven,insteadofmanual,activitiesinclude:soilpreparation,sowing, pestcontrol,
fertilizerapplication,weeding,'aporque','cortedelyuyo',harvest;dThisreferstoselection,classification,drying,andtransport
afterdrying
Maleproducer
0.000 0.002 0.002 0.005 0.011 0.012
(0.039) (0.026) (0.023) (0.033) (0.023) (0.025)
Ageofproducer(in10years)
0.021* 0.020 0.019 0.002 0.004 ‐0.001
(0.012) (0.014) (0.014) (0.025) (0.026) (0.027)
Education:Primary(vsnoEducation)
‐0.102** ‐0.076* ‐0.068* ‐0.032*** ‐0.007 ‐0.018
(0.044) (0.045) (0.042) (0.007) (0.017) (0.016)
‐0.057 ‐0.031 ‐0.022 ‐0.061 ‐0.011 ‐0.03
(0.037) (0.035) (0.046) (0.077) (0.057) (0.045)
‐0.088*** ‐0.115*** ‐0.102*** ‐0.015 ‐0.01 ‐0.006
(0.013) (0.036) (0.030) (0.034) (0.032) (0.030)
‐0.015*** ‐0.007* ‐0.009** ‐0.089** ‐ 0.048 ‐0.049
(0.005) (0.004) (0.004) (0.042) (0.035) (0.037)
‐0.004 ‐0.006 ‐0.007 1.448* * 1.150** 0.983
(0.005) (0.005) (0.005) (0.568) (0.537) (0.628)
log(Totalproductionpotato)
‐0.009 ‐0.008 ‐0.021 ‐0.022*
(0.006) 0.006 (0.013) (0.012)
Improvedseeds(dummy)
0.037 0.031 0.008 0.000
(0.065) 0.07 (0.030) (0.025)
Resistantpotatovariety
‐0.039** ‐0.038** ‐0.001 0.004
(0.018) 0.017 (0.041) (0.039)
Numberofdifferentinputsapplied
a
0.007 ‐0.005 ‐0.03 ‐0.01
(0.032) 0.026 (0.070) (0.080)
‐0.010** ‐0.010* 0.003 0.003
(0.005) 0.006 ( 0.013) (0.014)
0.014 0.017 ‐0.029* ‐0.026**
(0.045) 0.038 ( 0.016) (0.012)
Harvesttechnique:tractorvsazadon
‐0.165*** ‐0.166***
(0.017) (0.018)
Harvesttechnique:lampavsazadon
‐0.177*** ‐0.173***
(0.014) (0.017)
Hiredlaborforharvest
‐0.071*** ‐0.072*** ‐0.037 ‐0.012
(0.007) (0.009) (0.026) (0.032)
Storagedummy
0.019 0.013 ‐0.002 ‐0.003
(0.015) (0.015) (0.034) (0.037)
Nbofpost‐harvestactivities
d
‐0.046 ‐0.045 ‐0.002 ‐0.01
(0.063) (0.050) (0.003) (0.007)
0.017** 0.023** 0.011 0.025
(0.007) 0.012 ( 0.042) (0.022)
Climate
0.033** (0.020)
(0.016) (0.026)
Pests
‐0.005 0.063**
(0.015) (0.029)
Limitedknowledge
0.032*** ‐0.019
(0.007) (0.026)
Limitedequipment
‐0.012 0.118***
(0.013) (0.036)
Limitedmarketaccess
0.035 ‐0.011
(0.042) (0.040)
Limited creditaccess
‐0.019 0.055*
(0.025) (0.032)
parroqui a parroqui a parroqui a di stri ct dis tri ct dis tric t
yes yes yes yes yes yes
No.ofObs. 287 287 287 369 369 369
Production
problems&
limitationsto
producehigh
quality(as
perceivedby
thep rodu cer)
PeruEcuado
r
Mainincomefromagriculture(vsnon‐agric)
Production
Numberofproductionactivitiesdonemechanically
c
Numberofdifferentfieldmaintenanceactivities
b
Socio‐
economic
variables Education:Secondaryorhigher(vsnoEdu)
Experienceincultivationofpotato(in10years)
Market Costtoreachmarket(USD/Kg)
Locationfixedeffects
Post‐harvest
Agroecologicalzonedummies
Mechanicaltransport(notsoldonplot)



Table7:DeterminantsoflossesinthebeanvaluechainsinGuatemalaandHonduras(GLMmodel);
Dependentvariable:shareofproductlostatpre‐harvestandpost‐harvest(A‐measure)

Note:MarginaleffectsfromGLMmodelsare reported.Standarderrorsinparenthesisclusteredat thedepartmentlevelfor
Honduras and Guatemala.
a
 This includes fertilizers, insecticides, herbicides and fungic ides;
b
Thisincludesirrigationand
'chapeo';
c
 Machine driven, instead of manual, production activities i nclude: cleaning, sowing, herbicide application, pest
control,fertilizerapplication,andharvest;
d
Thisreferstowinnowing(sopla),threshing(desgrane),drying,puttinginbags,and
transport;eThisincludeschemicalfumigation,naturalfumigation,andventilation





Table8:DeterminantsoflossesinthemaizevaluechainsinGuatemalaandHonduras(GLMmodel);
Dependentvariable:shareofproductlostatpre‐harvestandpost‐harvest(A‐measure)


Note:MarginaleffectsfromGLMmodelsarereported.Standarderrorsinparenthesisclusteredatthedepartment
levelforHondurasandGuatemala.
a
Thisincludesfertilizers,insecticides,herbicidesandfungicides;
b
Thisincludes
irrigation and 'chapeo';
c
 Machine driven, instead of manual, production activities include: cleaning, sowing,
herbicideapplication,pestcontrol,fertilizerapplication,andharvest;
d
Thisreferstowinnowing(sopla),threshing
(desgrane), drying, putting in bags, and transport;
e
 This includeschemical fumigation, natural fumigation, and
ventilation





Table9:DeterminantsoflossesintheteffvaluechaininEthiopia(GLMmodel);
Dependentvariable:shareofproductlostatpre‐harvestandpost‐harvest(A‐measure)

Note:MarginaleffectsfromGLMmodelsarereported.Standarderrorsinparenthesisclusteredatthedistrictlevel.
a
Thisincludesfertilizers,insecticides,herbicidesandfungicides;
b
Thisincludesmechanicalherbicideandpesticide
application,andplowing;
c
Thisreferstocutting,drying,piling,threshing,winnowing,packaging,andtransportto
piling,threshing,and/orstorage;
d
Thisincludescleaningprevioustostorageandpreparationofstoragesite



Table10:DeterminantsoflossesinthewheatvaluechaininChina(GLMmodel);
Dependentvariable:shareofproductlostatpre‐harvestandpost‐harvest(A‐measure)


Note: Marginal effects from GLM models are reported. Standar d errors in parenthesis clustered at the
countylevel.
a
Thisincludes fertilizers, insecticides, herbicidesand fungicides;
b
This includesmechanical
land preparation, planting, fertilizer application, chemical application and harvesting;
c
Thisrefersto
cutting,bundling,strewing,hulling,packing,transport,dryingandcleaning





References
Affognon,H., Mutungia C.,Sangingac, P. and C.Borgemeistera. 2014. “UnpackingPostharvest Losses in
Sub‐SaharanAfrica:AMeta‐Analysis”.WorldDevelopment.66:49‐68.doi:10.1016/j.worlddev.2014.08.002
APHLIS.2014.“UnderstandingAphlis‐ TheAfricanPostharvestLossesInformationSystem.”Availableat:
http://www.aphlis.net/downloads/Understanding%20APHLIS%20ver%20%202.2%20May%2014.pdf
Bellemare, Marc; Cakir, Meting; Peterson, Hikaru Hanawa; Novak, Lindsey; Rudi, Jeta; (2017). On the
MeasurementofFoodWaste.Amer.J.Agr.Econ.0(0):1–20.
Beretta,C., Stoessel, F., Baier,U.,andS.Hellweg.2013. “Quantifying Food Losses and the Potential for
ReductioninSwitzerland.”WasteManagement.33(3):764–73.doi:10.1016/j.wasman.2012.11.007.
Buzby,J.C.,Wells,H.F.andHyman,J.2014."Theestimatedamount,value,andcaloriesofpostharvestfood
losses at the retail and consumer levels in the United States". EIB‐121, US Department of Agriculture,
EconomicResearchService
CalRecycle‐CaliforniaDepartmentofResourcesRecyclingandRecovery.2009."FoodWasteComposting
Regulations White Paper California Integrated Waste Management Board"
http://www.calrecycle.ca.gov/LEA/Regs/Review/FoodWastComp/FoodWastcomp.pdf
Compton, J. A. F. and J. Sherington. (1999). “Rapid assessment methods forstored maizecobs: Weight
lossesduetoinsectpests.”JournalofStoredProductsResearch35(1):77–87.
Consultation,PublicElectronic.2014.“HLPEReportonFoodLossesand Wasteinthe Conte xtofS ustainable
FoodSystems.”No.April2013:1–6.
DeMel,Suresh;McKenzie,David;andWoodruff,Crhistopher;(2009).Measuringmicroenterpriseprofits:
Mustweaskhowthesausageismade?JournalofDevelopmentEconomics88(2009)19‐31.
Delgado,L;Schuster,M; and Torero, M; (2017).Including thepotential lossesacrossthevaluechainsas
partofthedefinitionoffoodlosses.IFPRI,Mimeo.
EMMSA.EmpresaMunicipaldeMercadosS.A.http://www.emmsa.com.pe/.Lima.Peru.
ESPAC. 2015. Encuesta de Superficie y Produccion Agropecuaria C ontinua.Dirección Responsable de la
información estadística y contenidos: Dirección de estadísticas agropecuarias y ambientales. Quito,
Ecuador.
FAO.2011."GlobalFoodLossesandFoodWaste.Extent,CausesandPrevention".Rome:UNFAO.
______.2013."Toolkit:reducingthefoodwastagefootprint".Rome:UNFAO.
______.2014.“WorkingPaperDefinitional Framework ofFoodLoss‐GlobalInitiativeonFoodLossand
WasteReduction.”Rome:UNFAO.
FAOSTAT.FoodandAgricultureData.http://www.fao.org/faostat/en/#home.FAO.Rome.
Fonseca, J. and D. Njie. 2009. "Addressing food losses due to non‐compliance with quality and safety
requirementsinexportmarkets:thecaseoffruitsandvegetablesfromtheLatinAmericaandtheCaribbean
region".Rome:UNFAO
Fusions.2013."Reportonreviewof(food)wastereportingmethodologyandpractice".FusionsEUProject.
Availableat:http://www.eu‐fusions.org/index.php/publications
G20AgricultureMinistersMeeting.2015.FinalCommuniqué.7‐8
th May, Istanbul. Available at:
https://g20.org/wp‐content/uploads/2015/05/G20‐Agriculture‐Ministers‐Final‐Communique.pdf



HLPE.2013."Biofuelsandfoodsecurity.AreportbytheHighLevelPanelofExpertsonFoodSecurityand
NutritionoftheCommitteeonWorldFoodSecurity".Rome.
HLPE.2014."Foodlossesandwasteinthecontextofsustainablefoodsystems.AreportbytheHighLevel
PanelofExpertsonFoodSecurityandNutrition".CFS‐CommitteeonWorldFoodSecurity.
Hodges,R.2010."Postharvestweightlossestimates for cereal supply calculationsin East andSouthern
Africa".NaturalResourcesInstitute:Chatham,UK.
Kader, A. 2005. "Increasing food availability by reducing postharvest losses of fresh produce". Acta
Horticulturae682:2169–2175.
Kader, A. 2009. "Postharvest Losses of Fruits and Vegetables inDevelopingCountries:AReviewofthe
Literature".PresentationatUCDavis(November10th).
Kaminski,J., & Christiaensen,L.(2014). Post‐Harvest Loss in Sub‐SaharanAfrica:WhatDoFarmersSay?
GlobalFoodSecurity3:149–158.
Kummu,M.,H.deMoel,M.Porkka,S.Siebert,O.Varis,andPJ.Ward.2012."Lostfood,wastedresources:
Globalfoodsupplychainlossesandtheirimpactsonfreshwater,croplandandfertilizeruse".Scienceofthe
totalenvironment438:477‐489.
Lipinski,B.,C.Hanson,J.Lomax,L.Kitinoja,R.Waite,andT.Searchinger.2013.“ReducingFoodLossand
Waste.”WorkingPaper,Installment2ofCreatingaSustainableFood Future. Washington, DC: World
ResourcesInstitute.Availableathttp://www.worldresourcesreport.org.
Liu,G.2014."FoodlossesandfoodwasteinChina:afirstestimate".OECDFood,AgricultureandFisheries
Papers.No.66.OECDPublishing(http://dx.doi.org/10.1787/5jz5sq5173lq‐en)
Lundqvist,J.,C.deFraiture,andD. Molden. 2008. "Savingwater:fromfieldtofork—curbinglossesand
wastageinthefoodchain".InSIWIPolicyBrief.Stockholm,Sweden:SIWI.
Ministerio de Agricultura y Ganaderia (MAGAP). 2014. Boletin Situacional Papa.
http://sinagap.agricultura.gob.ec/phocadownloadpap/cultivo/2014/kboletin‐situacional‐de‐papa‐2014‐
actualizado.pdf.
MINAGRI 2016. http://minagri.gob.pe/portal/monitoreo‐agroclimatico/papa‐2016.The Ministry of
AgricultureandIrrigation(MINAGRI).Lima.Peru
Minten,B,T.Reardon,S.DasGupta,H.Dinghuan,andK.Murshid.2016a.“WastageinFoodValueChains
inDevelopingCountries: Evidence from thePotatoSectorinAsia.”In Food Securityina Food Abundant
World.http://dx.doi.org/10.1108/S1574‐871520150000016010
Minten,B,E.Engida,andT.Seneshaw.2016b.“Howbigarepost‐harvestlossesinEthiopia?Evidencefrom
teff.”ESSPWORKINGPAPER93.WashingtonDC:IFPRI.
Monier,V., M. Shailendra, V.Escalon,C.O’Connor,T.Gibon,G. Anderson, M.Hortense,ndH.Reisinger.
2010."PreparatoryStudyonFoodWasteacrossEU27".European Commission (DG ENV) DirectorateC‐
Industry.2010.FinalReport.ISBN:978‐92‐79‐22138‐5
Parfitt,J.,M.Barthel,andS.Macnaughton.2010.“FoodWastewithinFoodSupplyChains:Quantification
and Potential for Change to 2050.” Philosophical Transactions of the Royal Society of London. Series B,
BiologicalSciences365(1554):3065–81.doi:10.1098/rstb.2010.0126.
Rolle,R.S.ed.2006."ImprovingpostharvestmanagementandmarketingintheAsia‐Pacificregion:issues
andchallengestrends".InPostharvestmanagementoffruitandvegetablesintheAsia‐Pacificregion,23–
31.R.S.Rolle,ed.Tokyo,AsianProductivityOrganization.



Rosegrant,M.,E.Magalhaes,R.almonte‐Santos,andD.Mason.2015."ReturnstoInvestmentinReducing
Postharvest Food Losses and Increasing Agricultural Productivity Growth ‐ Post‐2015 Consensus".
CopenhagenConsensusCenter
SAG/UPEG 2015. http://sisem.sag.gob.hn/. Secretaria de Agricultura y Ganaderia y la Unidad de
PlaneamientoyEvaluacióndelaGestion.Tegucigalpa,Honduras.
Stuart,T.2009.Waste:UncoveringtheGlobalFoodScandal.London:W.W.NortonCo.
Tefera,T.,F.Kanampiu,H.DeGroote,J.Hellin,S.Mugo,S.Kimenju,Y.Beyene ,P. M.B odd upa lli,B. Shi feraw,
andM.Banziger.2011.“TheMetalSilo:AnEffectiveGrainStorageTechnologyforReducingPost‐Harvest
InsectandPathogen Losses inMaizewhileImproving SmallholderFarmers’FoodSecurity in Developing
Countries.”CropProtection30:240‐245.
Venkat, K. 2011. “The Climate Change and Economic Impacts of Food Waste in the United States”.
InternationalJournalonFoodSystemDynamics2:431‐446.
Waarts,Y.,M.Eppink,E.Oosterkamp,S.Hiller,A.VanDerSluis,andA.Timmermans.2011."Reducingfood
waste:obstaclesandexperiencesinlegislationandregulations".RapportLEI2011‐059.128
WorldBank.2011."Missingfood:ThecaseofpostharvestgrainlossesinSub‐SaharanAfrica".Washington,
DC:TheWorldBank.
WRAP. 2009. "Household Food and Drink Waste in the UK". Banbury: WRAP. Available at:
http://www.wrap.org.uk/sites/files/wrap/Household_food_and_drink_waste_in_the_UK_‐_report.pdf
______. 2010. "Waste arisings in the supply of food and drink to households in the UK". Av ailable at:
http://www.wrap.org.uk/sites/files/wrap/Waste%20arisings%20in%20the%20supply%20of%20food%20a
nd%20drink%20toUK%20households,%20Nov%202011.pdf
______.2013."MethodsUsedforHouseholdFoodandDrinkWasteintheUK2012."






AppendixA
TableA1:Surveyquestionstoestimatefoodlosseswiththe‘Self‐reportedmethod’






TableA2:ProducerlossesalongthepotatovaluechaininEcuador

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)







TableA3:ProducerlossesalongthepotatovaluechaininPeru

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)





TableA4:ProducerlossesalongthebeanvaluechaininGuatemala

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)




TableA5:ProducerlossesalongthemaizevaluechaininGuatemala

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)






TableA6:ProducerlossesalongthebeanvaluechaininHonduras

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)




TableA7:ProducerlossesalongthemaizevaluechaininHonduras

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)



TableA8:ProducerlossesalongtheteffvaluechaininEthiopia

Note:P=Pricemethod;S=Self‐reportedmethod,C=Categorymethod;A= Attributemethod;^ Dataare
imputedfromthe'Smeasurement';QuantitativeLoss=Totalloss(productdisappeared);QualitativeLoss=
Productaffectedbyqualitydeteriorations(productdidnotentirelydisappearbutqualityisreduced)




TableA9:ProducerlossesalongthewheatvaluechaininChina


























AppendixB
Thecountriesinwhichweworkandthedistributionofthesurveyswere:
1. Ecuador:Wecollected631surveys(302farmers,182middlemen,and147wholesalebuyers)inthe
provincesofCarchi,ImbaburaandPichincha;thefollowingmapshowsdistribution.


2. Peru: We collected 534 surveys (411 farmers, 77 middlemen, and 139 wholesale buyers) in the
departmentsofAyacucho,JunínandLima;thefollowingmapshowsthedistribution.



3.Honduras:Wecollected1777surveys(1155 farmers,377 middlemen,and245wholesalebuyers)inthe
departmentsof Choluteca,ElParaiso,FranciscoMorazán,Intibucá,LaPaz, Lempira,Ocotepeque,Olancho,
SantaBarbara,Valle,Cortes,CopanandYoro;thefollowingmapshowsthedistribution.

4.Guatemala:Wecollected1758surveys(1209farmers,325middlemen,and224wholesalebuyers)inthe
departments of Solola, Quetzaltenango, Totonicapan, San Marcos, Guatemala, Sacatepequez,
ChimaltenangoandEscuintla;thefollowingmapshowsthedistribution.



5. Ethiopia: We collected data from 1203 surveys for farmers intheregionsofOromiaandAmhara;the
followingmapshowsthedistribution.

6.China:Wecollecteddatafrom1307surveys(1114farmers,140middlemen,and53wholesalebuyers)in
theprovincesofHenanandShandong;thefollowingmapshowsthedistribution


