Why the numbers don’t add up: A review of estimates and predictions of people displaced by environmental changes

Why

the

numbers

don’t

add

up:

A

review

of

estimates

and

predictions

of

people

displaced

by

environmental

changes

Franc¸ois

Gemenne

*

a_{InstituteforSustainableDevelopmentandInternationalRelations(IDDRI),SciencesPo,27RueSaintGuillaume,75337ParisCedex07,France} b

CenterforEthnicandMigrationStudies(CEDEM),UniversityofLiege,BoulevardduRectorat7/45,BuildingB31,4000Liege,Belgium

1. Introduction

Estimatesandpredictionsofpeopledisplacedby environmen-talchangeshavebeenhighlyinstrumentalintheever-increasing attentiongiventoenvironmentalmigrationinthemedia.Figures rangingfromtensofmillionpeopletoabillionweretobefound not onlyin the headlinesof themainstream press, but alsoin officialcommunicationsandresearchreports.Thecurrentinterest for the topic is not only due to the specific nature of these migrationflows,butalsoduetotheirpotentialmagnitude.

Yetnoconsensualestimateexists,letaloneacommonlyagreed methodology.Asaresult,predictionsandestimateshavebecome oneofthemostcontentiousissuesinthedebateson environmen-tal migration. Numerous authors have criticised the existing estimates as artificially inflated, excessively alarmist, or ‘gues-stimates’(Kolmannskog,2008). Crispnotesthat‘whileallofthe standardworksonrefugeesarerepletewithnumbers,feweven begintoquestionthesourceoraccuracyofthosestatistics’(1999): theremarkappliestomanyworksonenvironmentalmigration,as mostofthemreproduce previousstatistics, withoutconducting anycriticalassessmentofthesedata.

Thisarticleaimstofillinthisgapbypresentingkeyestimates and predictions existing in the literature, as well as the methodologies theyare builton. It wasinitially commissioned as a methodological reviewfor the ForesightProject on Global EnvironmentalMigration,aprojectoftheUKGovernmentOffice forScienceaimedatassemblingandanalysingthelatestevidence andresearchonglobalenvironmentalmigration.1

For each estimate and prediction, the article provides an assessment of therobustnessof its methodology,itspredictive value,aswellasitsimpacts inthepublicdebates.Thepaperis dividedintofoursections:thefirstsectionreviewstheestimatesof peoplecurrentlydisplacedbyenvironmentalchangethatexistin boththeacademicandgreyliteratures–thelatterbeingdefinedas theworksemanating fromcivilsociety andadvocacy organisa-tions.Asecondsectionexaminesthepredictionsthathavebeen maderegardingthenumberofpeoplewhocouldbe–orwouldbe –displacedbyenvironmentalchangesinthefuture.Theseinclude predictionsmadeinthepastwithregardtoatimehorizonthathas sincebeenpassed.Athirdsectionsynthesisessomeofthecommon difficulties, weaknessesand shortcomingsof the estimates and predictions,andsuggestssomepossiblewaysforward.Finally,a fourthandfinalsectionoutlinessomepossiblewaysforwardin ordertoprovidemorerobustfigures.Itshouldbenotedthatthe review focuseson global estimatesand forecasts,and does not

ARTICLE INFO Articlehistory: Received10June2011

Receivedinrevisedform2September2011 Accepted7September2011 Availableonlinexxx Keywords: Estimates Predictions Methods Environmentalmigration ABSTRACT

Estimatesandpredictionsofpeopledisplacedbyenvironmentalchangeshavebeenhighlyinstrumental intheever-increasingattentiongivento environmentalmigrationinthemedia.Yetnoconsensual estimateexists,letaloneacommonlyagreedmethodology.Asaresult,predictionsandestimateshave becomeoneofthemostcontentiousissuesinthedebatesonenvironmentalmigration.Thisarticleseeks toreviewthekeyestimatesandpredictionsexistingintheliterature,aswellasthemethodologiesthey arebuilton,andtheproblemsandcaveatstheyarefraughtwith.Thefirstpartreviewsthefiguresrelated tocurrentestimatesofpeopledisplacedbyenvironmentalchanges,whileasecondpartexaminesthe predictionsforfuturedisplacements.Thenextsectionsynthesisesthekeymethodologicaldifficulties andafinalsectionsuggestssomepossibleavenuesforimprovement.

ß2011ElsevierLtd.Allrightsreserved.

*Correspondenceaddress:InstituteforSustainableDevelopmentand Interna-tionalRelations(IDDRI),SciencesPo,27RueSaintGuillaume,75337ParisCedex07, France.Tel.:+33145497679;fax:+33145497685.

E-mailaddresses:Francois.Gemenne@iddri.org,F.Gemenne@ulg.ac.be.

1

Moreinformationontheprojectcanbefoundonlineathttp://www.bis.gov.uk/ foresight/our-work/projects/current-projects/global-environmental-migration.

ContentslistsavailableatSciVerseScienceDirect

Global

Environmental

Change

j ou rna l hom e pa ge : w w w. e l s e v i e r. c om/ l o ca t e / gl oe n v cha

0959-3780/$–seefrontmatterß2011ElsevierLtd.Allrightsreserved. doi:10.1016/j.gloenvcha.2011.09.005

accountforfiguresprovidedwithregardtoaspecificregionora specificevent.

2. Estimatesofpastandcurrentmigrationflows

There are farfewerworks that attempt to counttheactual numberofpeopledisplacedbyenvironmentalchangethanworks thataimtopredictfuturemigrationflows.Oneofthemostcited worksonthesubject,areportcommissionedbyUNEPin1985to EgyptianscholarHassamEl-Hinnawi,gavebirthtothefigureof 30millionof displacedpeople (Milan, 2004;El-Hinnawi,1985). This figure, along with others provided by the Red Cross and environmental scientist Norman Myers, will be repeated in numerousreportsandarticles,sometimesinaccurately.

El-Hinnawi defined ‘environmental refugees’ as those people whohavebeenforcedtoleavetheirtraditionalhabitat,temporarilyor permanently,becauseofamarkedenvironmentaldisruption(natural and/ortriggeredbypeople) that jeopardisedtheirexistenceand/or seriouslyaffectedthequalityoftheirlife(1985:4).Thisverywide definitionencompassesbothinternalandinternationalmigration, aswell as temporary and permanent migration. Thedefinition however,focuses onforceddisplacement,and doesnotaddress mobilityassuch.Itaddressesalltypesofenvironmentalchanges, andnotonlythoseinducedbyclimatechange.Thedefinitionwas vividlycriticisedbecauseofitslargescope:amongstothers,Bates notesthatthisdefinitionmakesnodistinctionbetweenrefugeeswho fleevolcaniceruptionsandthosewhograduallyleavetheirhomesas soil quality declines. So many people can be classified under the umbrella of ‘environmental refugees’ that critics question the usefulnessoftheconcept(2002:466).

Ina2004interview,El-Hinnawinotesthat30millionpeople werein this situation (Milan, 2004). This number, however, is providedwithoutamentionofthemethodologyusedtoproduceit. Onecanthereforecastseriousdoubtsabouttherobustnessofthe estimateandquestionitsscientificvalidity.Thoughthenumberis aglobalestimateanddoesn’tdistinguishbetweendifferenttypes ofenvironmentalmigration,El-Hinnawihadpreviouslyattempted toputforwardatypologyofenvironmentalmigration,comprising thefollowingcategories:

-Those who had been temporarily displaced because of an environmentalstress(usuallyanaturaldisaster).

-Thosewhohadtobepermanentlydisplacedandresettledina newarea.

-Finally,thosewhomigratedwithintheir ownnational bound-ariesbecause‘theresourcebaseintheiroriginalhabitat(had) deterioratedtosuchadegreethatit(could)nolongermeettheir basicneeds’(1985:4).

Thesethreetypesofmigrantsnolongermaintaintheclassical distinctionbetweenforcedandvoluntarymigrants,thusallowing foraninflationofthenumbersofso-called‘refugees’.

This estimatewas followedby another one,produced three yearslaterbyJodiJacobseninapaperoftheWorldWatchInstitute, aninfluentialWashington-basedenvironmentalthink-tank,ledby environmentalist Lester Brown(Jacobson, 1988). Jacobson pro-posed a typology similar to that put forward by El-Hinnawi, distinguishingbetweentemporarydisplacementsassociatedwith temporaryenvironmentalstress,permanentdisplacements asso-ciatedwithpermanentenvironmental stress,and temporary or permanent displacement due to progressive environmental change(Jacobson,1988:5).Thistypology,however,wasflawed byundueemphasisonthetimeframeofthedisplacement,atthe expense of other important variables, such as the degree of coercionofthemigration,orthedistancetravelledfromthepoint oforigin.Therewaslittledistinction madebetween forcedand

voluntarymigration.InafashionsimilartoEl-Hinnawi,Jacobson vaguelydefined‘environmentalrefugees’as‘peoplefleeingfrom environmental decline’ (1988: 6). Her report identified some regionswhereenvironmentaldisplacementsweretotake place, sub-Saharan Africa and coastal regions of South-East Asia in particular.Sheestimatedthenumberofpeopledisplacedbecause ofenvironmentalreasonsat10millionworldwide,withmorethan half of this number beingdisplaced in sub-Saharan Africa. She asserted that ‘environmental refugees (had)become the single largestclassofdisplacedpersonsintheworld’.Thisfigureappears tobebasedonthenumberofpeoplewhowereinternallydisplaced –whateverthereason–intheregionsidentifiedbyJacobson.A similarmethodologywasusedbyWestingin1992:hecompared thenumberofpeoplebeingforciblydisplacedworldwide,either internationallyorinternally,between1986(26.4millionpeople) and 1992 (41.5 million people). He then speculated that the increase of internally displaced people (IDPs) could simply be attributedtoadditionalflowsof‘environmentalrefugees’( West-ing,1992).IfonefollowsWestingreasoning,thiswouldmeanthat therehadbeen15.1millionpeopleforciblydisplacedbecauseof environmentalchangesbetween1986and1992.Thoseestimates were made without a serious examination of the cause of migrationandattributedtheriseinforcedmigrationtoasingly proxy–environmentalchanges.

BothEl-Hinnawi’sand Jacobson’sreportswerereceivedwith greatinterestbythemediaandcivilsociety,butattractedharsh criticismfromscholars:theyhada‘short-livedshock-effectonthe publicdebatebutwererejectedasunseriousbyscholars’(Suhrke, 1993:6).El-Hinnawi’sreporthadbeencommissionedbytheUN EnvironmentProgramme,whileJacobsonwasaseniorfellowat theWorldWatchInstitute:thereportswerethereforeperceivedas an attempt to use forced migration to draw attention to environmental problems. The estimates they provided proved instrumentalinthisendeavour.

The most frequently cited figure, however, comes from environmentalscientistNormanMyers,whowroteseveralpapers inthe1990s andearly2000stoalert policy-makersandpublic opinionaboutthe‘growing phenomenon’represented by ‘envi-ronmentalrefugees’(Myers,1993,1997,2002;MyersandKent, 1995).Allofthesepaperscontainedestimatesandpredictionsof thenumberofpeopledisplacedbyenvironmentalchange.Asthe conceptofenvironmentalmigrationgainedcurrency,theybecame themostinfluentialandcontroversialfiguresdiscussedinpublic debates.To date,theyremainthemostelaboratedestimates of environmentalmigration.

Myersdefines ‘environmentalrefugees’aspeoplewhocanno longergainasecurelivelihoodintheirhomelandsbecauseofdrought, soilerosion, desertification, deforestationandother environmental problems, together with the associated problems of population pressuresandprofound poverty(Myers, 2002). Like El-Hinnawi’s definition,this onealsoencompassesbothinternaland interna-tionalmigration,aswellastemporaryandpermanentmovement. Thoughthedistinctionbetweenforcedandvoluntarymigrationis notexplicitlystated,itappearsclearlythatthedefinitionfocuses on displacement rather than mobility. In his most elaborated reportonthesubject,Myers–along withhisassistantJennifer Kent – contended that there were in 1995 25 million people displaced because of environmental changes (Myers and Kent, 1995). This number appears to be based on a compilation of differentregionalorcountryreports,someofthemconductedby Myers, somebyotherresearchers.Thustheglobal figureof25 millionisbrokendownindifferentregions:

-5million intheSahel,‘where a full10 millionhad fled from recentdroughts,onlyhalfreturninghome’(Myers,2002); -7millioninotherpartsofAfrica,seekingrelieffood;

-6millionofinternalmigrantsinChina–Myersexplainedthat6 million outofChina’s 120millioninternalmigrants ‘couldbe characterised as environmental refugees’, because of the shortageofagriculturalplots(Myers,2002);

-2millioninMexicancities–Myers(2002)addedthattherewere onemillionnewenvironmentalrefugeeseachyearinMexico; -1 million people who had been displaced by public works

projectsandremainedinarefugee-likesituation.

Thoughtheadditionofthesedifferentnumbersfallsslightly shortof25million,Myerscontendedlaterthathisestimateof1995 was ‘cautious and conservative’ (Myers, 2002). This estimate warrantssomeobservations:

-First,theestimatedoesn’tdescribeaflow,buta stock:people whohavebeendisplacedbecauseofenvironmentalfactorsbut have returned or resettled elsewhere are not included in the count.

-Second, the estimate does not distinguish between different types of environmental changes as migration drivers, and includesdevelopmentprojectsamongstthese.

-Finally,theestimateisbasedonotherworks,whose methodolo-gieshavenotbeendescribed,butareunlikelytobesimilar–this castsseriousdoubtsaboutthescientificaccuracyoftheestimate. Overall, Myers’ estimate of the number of ‘environmental refugees’ does not rely on any specific methodology: for each regionoftheworld,thenumberofinternallydisplacedpeopleis considered.Onthebasisofthesefigures,Myersmakesanestimate of the proportion that could have been displaced because of environmentaldisruptions.Thisestimateisbasedonreportsand observations of environmental degradation in the considered region,butnoattentionisgiventoanexaminationofthelinkages between environmental changeandmigration behaviour. Inan essentialistfashion,Myersassumesthatallpeopledisplacedinan areaaffectedbyenvironmentalchangeshavebeendisplacedsolely becauseofthesechanges.Anotherinterestingpointtoconsideris thatMyersrules outthepossibilitythat somecouldhave been displacedoutsideoftheircountry–internationalmigrationisnot consideredinhisestimate.

Despitetheselimitations,thefigureof25millionmadealasting impactinthepublicdebates,andwaspickedupinanumberof reports and documents aimed at raising awareness about the plight of ‘environmental refugees’. Simms and Conisbee, for example, made ampleuse of the figurein an advocacy report requesting that people displaced by environmental changes be granted a refugee status (Conisbee and Simms, 2003). The credibilityofthefigurewasreinforcedaftertheRedCrossstressed inits2001WorldDisastersReportthatmorepeoplewereforcedto leavetheirhomes becauseofenvironmentaldisasters thanwar (InternationalFederationofRedCrossandRedCrescentSocieties, 2001).ThisclaimwasbasedonoperationalreportsfromtheRed Crossactivitiesinthefield.Thefigureof25millionwasalsoquoted in thefamously controversial reportof the NGO ChristianAid, ‘‘HumanTide’(ChristianAid,2007).Inthislatestreporthowever, thefigureis nottakenfromMyers, but ratherfroma factsheet issued in 2006 by UNHCR with regard to internally displaced people(IDPs).Aftercarefulexaminationofthesource,itappears thatthenumberof25millionpeoplerelatestothenumberofIDPs worldwide,andnottothenumberofpeopledisplacedbecauseof environmental change. It therefore appears that the UNHCR factsheet has been misquoted in the Christian Aid report. Notwithstandingthis mistake, thereport asserts that thereare currently 163 million people forcibly displaced worldwide. Amongst these people are 25 million displaced by natural

disasters,asstatedabove,butalso105millionpeopledisplaced by development projects. This figure is solely based on an interview with Prof. M. Cernea, a specialist of population displacementsinducedbydevelopmentprojects.

Finally,inaninterviewwiththeGuardianandarelatedpress briefing, Prof. J. Bogardi, then Director of the Institute for the EnvironmentandHumanSecurityoftheUnitedNationsUniversity (UNU-EHS, Bonn), claimed that there were 10 million people displaced every year because of environmental deterioration (Adam,2005).Thisfiguredoesn’t seemrelated toanyempirical studyorstatisticalinformation,butappearstobederivedfromthe projection,madeinthesameinterview,thattherewouldbe50 million‘environmentalrefugees’by2010.Giventhattheinterview wasmadein2005,onecanassumethatProf.Bogardimadethe calculationthat10millionpeople neededtobedisplacedevery yearinordertoreachthenumberof50millionby2010.Inany case,thisfigureseemstohavebeenprovidedforthesolepurpose offeedingthemediaappetitefornumbers.

At this stage, it seems relevant to consider also how the literatureon migrationatlarge hasaddressedthis issue.Piguet (2009)rightlyrecallsthatenvironmentalconditionswereinitially consideredasfactorsthatdrovepeopletomigrateintheworksof thepioneersofmigrationstudies,beforebeingovershadowedby economic and political factorsand eventuallyforgotten. Onlya few,morerecentworks,suchasthoseofHunter(2005)orHenry and Piche´ (2004), address environmental migration as part of broader migration dynamics. Piguet makes the case for the environment to be ‘re-embedded’ into migration theories and models. Though it has indeed been little considered in most migration models, it is worth noting that some authors have proposed alternative methodologies to measure environmental migration, someofthem directlyinspiredby demographical or geographicalmeasurementsofmigration.AfifiandWarner(2008)

proposedtousea gravitymodel,directlyinspiredbypush–pull theories of migration.The modelintegrated differentmigration variables, both environmental and non-environmental, such as landdegradationortheoccurrenceoffloods,butalsothedistance betweencountriesortheGDPpercapita.Themodelpersewasnot abletoproduceanyestimateorprediction,butcontrollingfor non-environmental variables allowed the authors to conclude that environmentalfactorshadasignificantimpactonmigrationflows betweencountries.Amajorcaveatofthemodelwashoweverthat itappliedonlytointernationalmigrationandthusproved short-lived, since it had long been established that environmental migration is generally restricted to internal movement (Boano etal.,2008;Ja¨geretal.,2009).

ScholarsfromtheUniversityofSussexhavealsosuggestedthat methodsborrowedfromdemographicandgeographicstudieson migrationcouldproveinsightfulforestimatingandpredictingthe number of people displaced by environmental change, such as agent-based modelling (Kniveton et al., 2008). Agent-based modelling is a method that simulates the behaviours and responsesofactorstostimuli,inthiscaseenvironmentalchange, aswellastheinteractionsbetweenthesedifferentactors.Though themethodhasbeensporadicallyappliedinsomecase-studies,no globalestimateorpredictionhasbeenconductedyet.

Finally, estimates on the number of people displaced by catastrophic eventssuchasnatural disastersseem morerobust andlesscontroversial.Thelatestandmostaccuratestudyisthe onereleasedinJune2011bytheInternalDisplacementMonitoring CenterandtheNorwegianRefugeeCouncil(Yenotani,2011).The study reviews comprehensively the natural hazard-induced disastersthatoccurredin2009and2010,andconcludesthat17 millionpeopleweredisplacedbysuchdisastersin2009,and42 millionin2010,mostoftheminAsia.Thesenumbershavenot attractedcontroversysofar,and relyonasimple, four-pronged

methodology:first,abaselineofeventsisidentified,thankstothe EM-DATdatabase.2Thensomeeventsareprioritised–thosewhere morethan50,000peoplewereaffected–inordertocreateacore dataset. For each event, data is collected from various and organisationsaboutthenumberofdisplacedpeople.Anestimate oftotaldisplacementisthenmadeforeachevent,andthenumbers areanalysed.

Otherestimatesofpeopledisplacedbynaturaldisastershave alsobeenconductedinthepastbyvariousorganisations,including theFederationofRedCrossandRedCrescentSocietiesortheUN OfficefortheCoordinationofHumanitarianAffairs(OCHA).These reportshavebeenrelativelyconsensual,astheyallrelyonasimilar methodology– indeed, thekey methodological challenges and caveatsliewithslow-onsetenvironmentalchanges.

3. Predictionsandforecastsoffuturemigrationflows Manyoftheauthorswhohadprovidedestimatesofcurrentand pastnumbersofpeopledisplacedbyenvironmentalchangealso provided forecasts for future displacements, at different time horizons.Mostofthepredictionswereaffectedbythesameflaws and biases as were current estimates: a weak or inexistent methodology,andatendencytousethenumbersinordertoraise awareness.

Inlinewithhiscurrentestimate,El-Hinnawiwarnedthat‘this number(would) increasewithdeterioratingenvironmental and economic conditions in parts of the developing world’ (Milan, 2004). One can therefore assume that the predictions also incorporate,atleastpartly,peopledisplacedbecauseofeconomic conditions.In 1989, UNEP Executive Director Mostafa K. Tolba echoedthese figures, and warned that ‘as many as 50 million people could become environmental refugees’ by 2010 (Tolba, 1989). The forecast was also reinstated later by then UNEP Executive Director Klaus To¨pfer3 _{in numerous statements and} interviews,whichgavethefigureacertainprominenceinpublic debatesovertheissue.Inearly2011,differentblogsandinternet websites noted that the ‘50 million refugees’ had failed to materialise,which gaverisetoa controversydocumentedinan articlebyTheNewScientist(Pearce,2011).Butevenmorethanthe figureof50million,itisaforecastof150millionpeopledisplaced by2050thatcapturedthepublic’sattention.

ThefigurewasmentionedbyMyers,outliningthemethodology thatallowedhimtocomeupwiththeforecast(MyersandKent, 1995; Myers, 2002). Myers alsocited El-Hinnawi’s and UNEP’s estimates,thoughitisnotclearwhetherheusedthesenumbers and addeda methodological apparatus orwhether he used El-Hinnawi’smethodologyandcameupwiththesameconclusions. Inanycase,Myerscontendedthattherisewasmostlytheresultof threeconcomitantandinterrelatedfactors:environmentaldecline, spreading poverty and demographic expansion (Myers, 2002). AccordingtoMyers,thetwolatterproblemsweredirectlyrelated tothefirstone.Itappearsthereforethattheprojectionsforecast notonly thepeople who would bedisplaced solely becauseof environmentalchanges, butalso thepeopledisplaced forother reasons,thoughthesereasonswouldbelinkedtoenvironmental factors.Therisefrom25million‘environmentalrefugees’in1995 to50millionin2010caneasilybeexplainedbythefactthatsuch displacementsstartedin1980,accordingtoMyers.Heconcludes that at least the same number of people would be displaced between1995and2010thanbetween1980and1995.Theforecast of 150 million displacees by 2050 takes into account the demographic expansion of developing countries as well as the

deteriorating environmental conditions. Amongst these, Myers considersprincipallysea-levelriseanddesertification.Hisforecast isbasedonthenumberofpeoplethatareexpectedtoliveinthe regionsatriskby2050:inadeterministicmanner,Myersassumes thatallthesepeoplewillbeforcedtoleave.

Contrary toa widely held belief,this figure hasnever been reproducedintheAssessmentReportsoftheIntergovernmental PanelonClimateChange(IPCC).Ithasbeenpickedup,however,in another report on climate change that received high media exposure: theStern Review onthe Economics of Climate Change (Stern, 2007). The review, however, notes cautiously that ‘the estimatehasnotbeenrigorouslytested,butitremainsinlinewith the evidence presented throughout this chapter that climate changewill lead tohundreds of millionsmore people without sufficient water or food tosurviveor threatenedby dangerous floodsandincreaseddisease’.

Inasubsequentpaper,Myersupdatedhis2050prediction,and warnedthatwhenglobalwarmingtakeshold,‘therecouldbeas many as 200 million people4 _{overtaken by sea-level rise and} coastal flooding, by disruption of monsoon systems and other rainfallregimes,andbydroughtsofunprecedentedseverityand duration’(Myers,2002).Thepeopleatriskoffloodingwouldbe distributedasfollows:

-Bangladesh:26million -Egypt:12million -China:73million -India:20million

-Otherpartsoftheworld,includingsmallislandstates:31million. Inadditiontothese162millionpeople,Myerswarnsthat at least50millionareatriskbecauseofdroughtsandotherclimate dislocations.

Throughfrequentrepetitions,anddespiteitshighlyspeculative natureandquestionablescientificaccuracy,thislatestfigureof200 million people displaced by 2050 has beentaken as empirical evidence,andhasbeenfrequentlycitedinmediareportsandother studies.Suchreportsandstudiesincludeworksand communica-tionsbytheUNU-EHS(Renaudetal.,2007),FriendsoftheEarth (FriendsoftheEarthAustralia,2007)ortheGeneva-basedGlobal HumanitarianForum(GlobalHumanitarianForum,2009). These works garneredlarge attention,especially in therun-up tothe 2009CopenhagenConference,contributingtothespreadingofthe forecastinpublicdebates.

Myerscancertainlybecreditedwithdrawingworldwideattention tothetopicofenvironmentalmigration.However,hisworkislargely basedonspeculativecommonsenseratherthanonactualfiguresand estimates – a point thathas been vigorouslycriticised by many scholars,includingBlack(2001)andCastles(2002).

Anotherhighly controversialestimatecame fromthe above-mentioned Christian Aid report (Christian Aid, 2007), which predicted that as many as one billionpeople couldbe forcibly displaced by 2050. Out of these 1 billion people, the report predictedthat300millionwouldbedirectlydisplacedasaresult ofenvironmentaldisruptions:

-50millionpeople displacedbynaturaldisasters,atarateof1 millionpeopledisplacedperyear;

-250millionpeoplepermanentlydisplacedbecauseof climate-changerelatedphenomenasuchasdroughts,famines,floodsand hurricanes.

Thislast figureisbasedonaninterviewwithProf.N.Myers, reflecting upon his previous estimates. In addition, the report mentionsthat645millionpeoplewillbedisplacedasaresultof

2

TheEM-DATdatabase,hostedintheUniversityofLouvain(Belgium),iswidely consideredasthemostcomprehensivepublicdatabaseofnaturaldisasters.

3

MrTo¨pfer’stermasExecutiveDirectorofUNEPwas1998–2006. 4

developmentprojects,atarateof15millionayear,andconcludes that‘withouturgentaction,climatechangewillmaketheforced displacementcrisisthebiggestthreatfacingdevelopingcountries over the next 50 years’. The report by Christian Aid attracted considerablemedia attention,but was harshlyrebuffed by the scholarlycommunityasunseriousandoverlyalarmist.

BiermannandBoas(2010)alsosoughttopredictthenumbers and origins of ‘climate refugees’, recognising that ‘all current predictionsarefraughtwithnumerousmethodologicalproblems andcaveats’.Inordertocircumventthesedifficulties,theyattempt todistinguishthenumberofpeopleaffectedbydifferentclimate impacts (extreme weather events, sea-level rise, drought and waterstress)indifferentregionsoftheworld.Inordertodothis, theyusedifferentstudiesassessingthenumberofpeopleatriskof theseimpacts,andconcludeprudentlythat‘thetotalnumberof peopleatriskofbecomingclimaterefugeesby2050couldwellbe aroundorover200million(...)’.Akeycaveatofthisassessment, thoughtheauthorsadmitit’sonlyaroughestimate,isthatitis based on secondary sources – including the works of Myers– whosemethodologiesarenotconsistentwitheachother.Despite thiscaveat,thepaperappearsmorerobustthanitspredecessors, becauseitattemptstobreakdownthisglobal200millionfigureby regions and climate change impacts. In particular, the authors calculatedthenumberofpeoplewhowouldbeaffectedeachyear by2080sbystormsurgeswithasea-levelriseofabout38cmin differentregions.Thoughthissea-levelrisecouldbedisputedfor beingunderestimated(Rahmstorf,2010),theycameupwiththe figureof82millionpeopleaffectedinAfrica,141millionforAsia, and2.5millionforsmallislandstates.Interestingly,theydeparted fromtheoften-assumeddeterministicperspectiveandcalculated how these numbers could be reduced thanks to evolving protection mechanisms. In that case, the numbers would 14 millionfor Africa, 76 millionfor Asiaand 1.2 million forsmall island states. For each continent, they provided examples of estimateddisplacement induced bysea-level rise,including 12 million‘climaterefugees’inEgypt,26millioninBangladesh,73 millioninChinaand20millioninIndia,allbytheyear2050.They repeatedtheexercisewithotherclimateimpactssuchastropical cyclones and water stress, but provided figures that were considerablyless detailed.For example,people atrisk ofwater stressinAfricaby2085duetoatemperatureincreaseof2–38C, dependingonpopulationlevel,wereestimatedtobebetween197 millionand1.65billion–aneightfoldvariation.

Morerecently,a number of studiesof studiesdiscussed the possibleeffectsofclimatechangeintheeventwhentemperature increasewouldreach48C+bytheendofthecentury.Inthatcase,

Nichollsetal.(2011)statedthatupto187millionpeoplecouldbe forciblydisplacedbecauseofsea-levelriseoverthecentury.This figureischaracterisedbyanexcessivedeterministicbias,astheit reflectsthenumberofpeopleatriskofcoastalfloodingbutdoes nottakeintoaccountotherfactorsthatwillweighuponmigration behaviours,suchasadaptationstrategiesorpoverty.

Figuresthatforecastthenumberofpeoplewhocouldbeatrisk becauseoftheadverseeffectsofclimatechangebutdidnotdiscuss theireventualdisplacementprovedfarlesscontroversial.Nicholls predictedthatupto30millionpeopleayearcouldbeatriskof coastalfloodingbecauseofsea-levelriseby2080(Nicholls,2004), whiletheSecretariatoftheUnitedNationsConventiontoCombat Desertification(UNCCD)warnedthat135millionpeoplewereat riskofbeingdisplacedbecauseofdesertificationby2030,with60 million people expected to move from sub-Saharan Africa to NorthernAfricaandEurope(UnitedNationsConventiontoCombat Desertification,2010).

Thesedifferentforecastssharesomecommoncharacteristics:in particular,theywerereceivedwithgreatscepticisminthescholarly community, and great interest in the media and policy-making community. Thus,atthesametimethatthey werefacingharsh criticism,theyalsomadeasignificantimpactinpublicdebates,as whistle-blowers of an upcoming humanitarian catastrophe. The mediaappetitefornumbersiswellknown:‘startlingstatisticsshape our thinkingaboutsocialissues’anddefine the importanceand seriousnessoftheseissues(Best,2001).Forecastsof‘environmental refugees’perfectlyfitthatbill,andindicatedanissueoftremendous magnitude andimportance.Doingso,theyalsoshapedhowthe publicandpolicy-makerswouldthinkabouttheissue.

Mostoftheseforecastswerepublishedinthegreyliterature, which might explain why little attention was devoted to methodological issues. Only a few reports were published in academicjournals,meaningthatthemethodologyofmostworks wasneverreviewedexternally.

Finally, most forecasts use the same timeframes for the estimates and predictions:2010 and 2050.Thechoiceof these timeframeshasbeenheavilyinfluencedbythetimeframesusedby theIPCCforitsAssessment Reports– andindeed,theforecasts dependheavilyuponclimatechangeprojections.Wearenowin 2011,oneyearaftertheyearthatwasusuallychosenasthefirst yearofreferencefortheforecasts.Thusitwouldseemrelatively easy to check the validity of these forecasts against reality. Unfortunately,intheabsenceofanystatisticalinformation,it is currentlyimpossibletoassesstheaccuracy ofthesepredictions against empirical realities. Furthermore, the absence of a consensual definition of ‘environmental migration’ means that differentnumberscanrefertodifferentrealities–thisdefinitional problemwillbeaddressedinthenextsection.

Table 1 synthesises the principal and comparable figures providedinthedifferentestimatesandforecasts.

4. Difficultiesandobstacles

Predictionsandestimatesremainoneofthemostcontentious issues in public debates on environmental migration. Many scholarssuspectthatthesefigureslackanyempiricalbasis, and areputforwardwiththesolepurposeofraisingawarenessand garneringmediaattentionaroundtheissueofclimatechange,or

Table1

Estimatesandforecastsofthenumberofpeopledisplacedbyenvironmentalchanges.

Source Estimatesatthetimeofpublicationofthereport Predictionsby2010 Predictionsby2050

UNEP(Tolba,1989;Milan,2004) 30million 50million –

Jacobson(1988) 10million – –

Myers(1993,2002) 25million – 150,then200million

MyersandKent(1995) 25million 50million 212million

RedCross(2001) Morethanpeopledisplacedbywar – –

UNU-EHS(Adam2005;Renaudetal.,2007) 10million/year 50million 200million

Stern(2007) – – 150–200million

ChristianAid(2007) 25million – 300million

environmentaldegradationatlarge.Yetthefiguresareimportant, astheyareoftenrequestedbypolicy-makersinordertoassessthe importanceofthephenomenonanddeveloppolicyresponses.The roleofscienceandstatisticsinthepolicyprocesshasbeenwidely documented in the literature (Sabatier, 1987; Majone, 1989). Predictionsandestimateshavebeenabundantlyusedtoadvocate forthedevelopmentofnewpoliciesandmechanismsofprotection to assist the people displaced by environmental changes. Yet whereasback-of-the-envelopefigurecansufficetogeneratepublic awareness,thedesignandfundingofappropriatepolicyresponses willoftenrequiremoredetailedandrobustfigures.

Despitetheirflawsandbiases,predictionsandestimatesmade alastingimpactinbothpolicyandscholarlydebates:theyarenow oftenquotedasscientific truth,withoutanyquestioningof the methodologyused.Themulti-causalityofdisplacement,aswellas theconfusionbetween forcedandvoluntarymigration,make it difficulttoidentifyapreciseandrigorousnumberof environmen-tally displaced persons. Given the lack of comprehensive methodologyand empirical studies, the field is wide open for guessesanddoom-ladenestimatesratherthanactualnumbers.

Differentfactorsimpedeuponthecollectionanddevelopment ofaccuratedataandforecasts.Someofthesefactorsarecommon toalldisplacement-relatedissues;othersarespecificto environ-mentalmigration.

First, thequest for numbersis hamperedby thedebates on controversiesovertheconceptandverydefinitionof environmen-talmigration. Without cleardefinition,providing accuratedata seemsadauntingtask.AsnotedbyBiermannandBoas(2010),‘all studiesoperatewithdifferentterminologyanddefinitions,which makesitdifficulttocompareresults’.Differentproblemsimpede upon the search for a common definition, including tensions between scholars who would like to establish ‘environmental migration’asadistinctmigrationcategoryandscholarswhoinsist onthemulti-causalityofmigration,aswellastensionsbetween international agencies over the perimeter of their respective mandates(DunandGemenne,2008).Itisworthnoting,however, thata similarproblem is encounteredwhen countingrefugees, eventhoughalegaldefinitionexists.In thewordsofCrisp,‘any formofenumerationexercisemustbebaseduponaclearlydefined unit of measurement if it is to produce reliable, usable and comparabledata.Inthecaseofrefugeestatistics,however,such claritydoesnotalwaysexist’(1999).

Mostofdisplacementstriggeredbyenvironmentalfactorsare intra-national, rather than international. This poses a further statistical problem,since countinginternal movementsis much moredifficult than cross-border movements: ‘the machinery to collectdataonthesemovementssimplydoesnotyetexist’(Brown, 2008).Crispprovidesalistofquestionsthatareleftunanswered whenoneisattemptingtocountinternallydisplacedpeople(IDPs): In theabsenceofaclearcriterionsuchasthecrossingofan internationalborder,howfardoesapersonhavetomovetobe considered ‘internally displaced’? When do internally (dis-placed)peopleceasetowarrantthatstatus:whentheyreturn totheiroriginalplaceofresidence,orwhentheyhaveachieved acertaindegreeofphysicalandsocio-economicsecurityinthe placetowhichtheyhavefled?Giventhatalargeproportionof theworld’sIDPsarethoughttoliveintownsandcities,howcan they be differentiated from other rural-to-urban migrants? (1999)

Some furtherproblemsarespecifictothenatureof environ-mental migration. When it comes to predictions, figures are usuallybasedonthenumberofpeoplelivinginregionsatrisk,and not on the number of people actually expected to migrate. Estimatesdonotaccountforadaptationstrategies,nordifferent

levels of vulnerability to change. This confusion between the number ofpeople living inaffected regionsand thenumber of migrants is rooted in a deterministic perspective, and is a key reasonfortheinflationofnumbers.

Predictionsalsoneedtotakeintoaccountdemographicchanges that areexpectedtooccur overthenextdecades.Whereas the globalpopulationgrowthcanbepredictedtoacertainextent,its geographic distribution remains uncertain. In recent years, migrationtourbanareashasaccelerated.Anincreasingnumber of people live in regions highly exposed to natural risks and hazards, resulting in an increased number of disasters. These demographic shifts are particularly significant in developing countries, and they make estimates of potential migrants a difficulttask.

Also,itisexpectedthatalargenumberoffuturedisplacements willbeinducedbytheimpactsofclimatechange.Itispossibleto mitigate these impacts through adaptation strategies, and to mitigate climate changeitself by reducing our greenhouse gas emissions.Thusthefutureimpactsofclimatechangeonsocieties willgreatlydependonfuturelevelsofgreenhousegasemissions, aswellasontheadaptationstrategiesthatwillbeimplementedin vulnerable regions. Some impacts can be avoided through significantemissions cuts,while adaptationstrategies can help populationscopewiththeimpactsthatcannotbeavoided.Ina nutshell,futuredisplacementsdependtoalargeextentuponthe climatepoliciesthatareimplementedtoday.Becauseoftheinertia oftheclimatesystem,theimpactsofglobalwarminguntil2050are largelypre-determinedbyourpastemissions(Hansenetal.,2006). Impactsbeyond 2050dependheavilyonourcurrentemissions, andthisisthereasonwhyonlyfewpredictionsgobeyond2050. Though the impacts of climate change until 2050 are largely known, thanks to the IPCC reports, adaptation strategies can minimise how these impact affect societies. Hence future displacementsalsodepend,toagreatextent,upontheadaptation strategiesthatareimplementedtoday–andpro-activemigration mightactuallybepartofthem.Overall,publicpoliciesplayagreat roleinshapingmigrationdynamics(Zolberg,1989).Withregardto environmentalchanges,theymightencourageordetermovement, or provideadaptationschemesthat will sustainpeople’s liveli-hoodsintheirregionoforigin.Yettheroleofpublicpolicies–be they environmental or migration policies, or both – is often unaccountedforintheestimatesandforecasts.

Finally,theimportanceoftimeframesisofcrucialimportance: dothepredictednumbersdescribeanaccumulationofmigration flowsoveracertainperiodoftime,ordotheyaccountforastockof environmentalmigrantsatacertainperiodoftime?Arethosewho havebeenabletoreturnhomeincludedinthepredictionsornot? Thoughthesequestionsareofcrucialimportance,theyarerarely addressedandfuelthesuspicionthatthenumbersareartificially inflated.

Despite these difficulties, getting the numbers rights is an importantstepfordevelopingadequatepolicies.Theprogramming of assistance and mobilisation of resources, including funding, depends on accuratenumbers (Crisp, 1999). In the absence of reliablestatistics,numberscanbeeasilyinflatedandmanipulated, inordertoattractattentiononsomepopulations,sometimesatthe expenseofotherpopulationsinneedofprotection.

5. Waysforward

In order to moveforward,three different avenuescould be explored.First,arightframeworkforresearchneedstobeset.This impliesthedesignofacommonworkingdefinition,aswellasthe developmentofstatisticalsystemsandlocalresearchcapacities. Second,quantitativeresearchinthefieldneedstoberescaled,both withregardtotheirtimeframeandtheirspatialdimension.Third

andlast,newmethodsandtechniqueshavetobetested,suchas longitudinal studies, scenarios, or agent-based modelling. The followingparagraphsdetailafew suggestionsalong thesethree avenues, which could pave the way towards more accurate estimatesandprojections.

5.1. Settingtherightframeworkforresearch

First,thedefinitional issueseemsparamount.Aspointedby Suhrke, ‘broad categorizations invite large numbers’ (Suhrke, 1993): many figures rely on very broad, loosed definition of ‘environmentalmigration’,inordertoencompassasmanypeople as possible. International organisations and agencies also use differentdefinitionsandterminologies,reflectingtheperimeterof their mandate. These differentdefinitions and typologies make estimatesandprojectionshardlycomparable,andanyattemptto come up with more robust figures will be hampered by this definitionalissue.Ahigh-levelgroupcomprisedofscholarsand representativesfrominternationalorganisationscouldbetasked with providing a working definition and typology that would becomea referential.The definition would needto distinguish betweenforcedandvoluntarymovement,ascurrentfiguresoften confusethesetwotypesofmigration,thoughtheyreflectreactions toenvironmentalchangesthatarefundamentallydifferentfrom each other. Though the line between forced and voluntary migration hasbecomeincreasingly blurred in recent years and bothcanhardlybedefinedasclear-cutcategories(Hugo,1996), figuresshouldstatemoreclearlywhethertheyconsiderforcedor voluntarymovements,orboth.Acommontypologywouldbemost helpfulforthis.

Statistical systemsalso needto befurther developed. Many developingcountriessimplydon’t havethecapacitytomonitor migration flows, especially within their own borders. The development of country profiles could be a first step in this endeavour,whichwillrequiresignificantresourcesand capacity-building.Theinclusionofquestionsrelatedtotheperceptionof environmentalchangesandinternalmobilityinnationalsurveys andcensuswouldanothermajorstepinthisdirection.

Researchonenvironmentalmigrationismostlyconductedfrom industrialisedcountries.Yetmostofthesemovementstakeplace in developing countries, where local expertise is insufficiently mobilised. Too many estimates and predictions are produced withoutanysupportingevidencefromempiricalstudies. Quanti-tativeresearchneedstobecomplementedwithandtestedagainst qualitativeresearch,andlocalresearchcapacitiesurgentlyneedto bedeveloped.Thesettingupoflocalobservatories,forexample, couldprovide a monitoring of environmental migration in the long-run.

5.2. Rescalingestimatesandpredictions

Despiteafewexceptions,estimatesandpredictionstendtobe producedon thegloballevel. Asa result,theyoftenhavelittle policyrelevance,astheyaggregatedifferenttypesofmovement, requiring different types of policy intervention. Boano et al. suggestthat,insteadofdraftingglobalestimates,‘amorevaluable routetounderstanding thepotentialscaleofdisplacement,and thusthescopeofpolicyintervention’,wouldconsistingettinga morenuancedunderstanding ofthedifferentformsof environ-mental migration, and developing a much more empirically groundedapproachtotheissue(2008).Theysuggestthatresearch should focus on the identification and mapping of potential environmental‘hotspots’,themonitoringofchangingconditions, theexaminationof‘tippingpoints’andthetrackingofmigration trends. Kniveton et al. concur that a first step is to ‘try to understandhowpeoplecopewiththedifferenttypesofgradual

stressesandsuddenshocksbroughtaboutbyclimatechangeand variability’(2008).Asamatteroffact,themostrobustquantitative studiessofarwereconductedonthelocalornationallevel.They includetheworksofMasseyetal.inNepal(2007)ortheworksof Henryetal.inBurkinaFaso(2003;2004).Studiesthatfocusonthe locallevelallowforacrisperunderstandingofmigrationbehaviour and a more accurate measurement of migration trends. In the currentstateofourunderstandingofenvironmentalmigration,it is probablyimpossibletoprovideareliableandaccurateglobal estimate.

Anotherscalingissueconcernsthetimeframeofestimatesand predictions,whichisoftenunclear:forexample,whenonestates thattherewillbe150million‘environmentalrefugees’by2050, doesthatmeanthat150millionpeoplewillbedisplacedbetween nowand2050,orthat150millionpeoplewillbedisplacedinthe year2050?Whataboutthemigrantswhowillreturntotheirplace of origin? Does one speak about permanent or temporary migration?Moreclarityisneededwithregardtothetimescales thatareused.

Finally,anotherrescalingeffort isneededwithregard tothe expectedmagnitudeofclimatechange.Mostforecastsarebasedon theassumptionthatglobalwarmingwillbelimitedto28Coverthe courseofthecentury,whichistheofficialobjectiveadoptedduring the Copenhagen and Cancu´n conferences, in 2009 and 2010 respectively.Yetmanyscientistsandobserversbelievethatthis objectivewillnotbeachieved,andthataverageglobaltemperature increaseislikelytobearound48Cbytheendofthecentury(Betts etal.,2011).A48Ctemperaturerisecouldincreasemigrationflows insomeplaces,butcouldalsolimitmigrationpossibilitiesforsome populations(Gemenne,2011). Predictionsshouldberescaled in ordertoaccountforthepossibilityofa48Ctemperaturerisebythe endofthecentury,whichappearsasapossibilitythatoughttobe consideredseriously,giventhecurrentcommitmentstoreduced greenhouse gasemissions. Predictionsalso tendto confusethe movementsthatcouldbeinducedbyclimatechangeimpactsand otherenvironmentaldisruptionssuchasnatural disasters.More effortshouldbemadeinordertoidentifytheadditionalmigration volumesthatcanbeattributedtodifferenttypesofenvironmental changes:thiswouldallowtoassesstherelativeweightofglobal warmingonmigrationtrendsovertime.

5.3. Exploringnewmethods

Piguethasconductedausefulreviewofthedifferentmethods thatcouldbeusedinordertomeasureenvironmentalmigration (2010).Whilesomemethodstendtobefrequentlyused,others havebeenlittleexplored,despitepromisingprospects.

Kniveton et al. (2008) suggest that agent-based modelling shouldbeusedtosimulatefuturemigrationpatternsonthebasis of current migration behaviours. An agent-based model is a computationalmodelthatsimulatesthebehavioursofagentsin reactiontosomestimuli.Themodelhasbeensuccessfullyusedto study migration trends, but has not yet been applied to environmental migration. Though the method remains quite controversial,itcouldprovidevaluableinsightsfortheprediction offutureflowsofenvironmentalmigration.

Multi-levelanalysishasbeenlittle usedaswell,thoughthis methodiswidelyrecognisedasoneofthemostrobustmethods,as it combines different methods on different levels, including householdsurveysandtimeseries.Thisallowsforthemonitoring ofmigrationtrendsoverlongerperiodsoftime:currentstudiesof environmentalmigrationareusuallysnapshotsthatfailtocapture long-term migration dynamics. As environmental changes are often incremental over time, it is essential to track migration trendsoverlongerperiodsoftime.Multi-levelanalysisshouldthus beusedmorefrequentlythanithasbeensofar,wheneverpossible.

AnexampleofsuchstudiesincludetheworksofHenryetal.in BurkinaFaso (2003; 2004), which tracked themigration beha-vioursofthousandsofhouseholdsoverseveral yearsand cross-checkedthemagainstlongitudinalenvironmentaldata.

Finally, thespeculativecharacter of predictions needs tobe acknowledged. A way to remedy this problem is to resort to differentpossibleevolutionscenarios.

The EACH-FOR project (‘EnvironmentalChanges and Forced MigrationScenarios’,www.each-for.eu)developedsucha model-ling exercise, based on scenarios relating to different policy directions(Ja¨geretal., 2009). Theexerciseprovidessomeways forwardinordertotakeintoaccountthedifferentvariablesthat willimpactuponthenumberofpeopledisplacedby environmen-talchanges.However, inordertoproviderobustscenarios,one needstobeprovidedwithconsistentstatisticalindicators,which areabsentinmostdevelopingcountries.TheEACH-FORscenarios wereonlyrealisedforalimitednumberofcountries,andwithout robustquantitativedata.Inthatsense,theyneedtoberecognised asamethodologicalexerciseratherthananaccurateprojectionof thepeoplewhocouldbedisplacedbyenvironmentalchange.They highlight,however,theneedtobettertakeintoaccounttheroleof public policies as a determining condition of the nature and magnitudeoffuturedisplacementsandmigration.Asmentioned earlier,mostforecastsandpredictionsignorecompletelytherole ofpublicpoliciesinthegenerationofenvironmentalmigration. 6. Conclusion

As they stand today, estimates and predictions regarding environmentaldisplacementarenotsatisfactory.Many ofthem appear to have been put forward in order to generate media attentionratherthantoprovideempiricallygroundedestimates and predictions. In particular, they lack robust methodological foundations, and are generally grounded in a deterministic perspective,assumingthatallpeopleimpactedbyenvironmental changeswillmoveawayfromtheirhomes.Furthermore,theyare stronglybiasedtowardsclimatechange,oftenattheexpenseof other environmental changes. Overall, it seems that a key methodologicaldebatereliesontherelative weightof environ-mental factorsamongst thevarious drivers of migration. Since earlyon, scholars have been divided between minimalists and maximalists(Suhrke,1994),thelatterassumingthat environmen-tal change is a major driver of migration and displacement, whereas the former tend to insist on the multi-causality of migration.Asoftoday,thisdebateremainsextremelyfierce.

Yetitseemsdifficult,atthemoment,toprovidemoreaccurate data.Many developingcountries lack thestatistical capacity to monitor migration movements. In the absence of commonly agreed definition, robust empirical studies and a statistical apparatus,itseemsdifficulttoenvisionhowaccuratedatacould beprovided.Thisdoesn’tmeanhoweverthatweshouldabandon the idea of building quantitative data, as these are of central importance for the development of policy responses. Current estimatesandforecastsneedtoberescaledondifferentlevels,and innovative methods such as scenarios, multi-level analysis or agent-based modelling can provide some interesting ways forward. This will require however that the current quest for globalfiguresisabandoned,inordertomakewayformorefocused studies,as thesecouldprovide numbersthat would be policy-relevantandyetscientificallysound.

Acknowledgements

This article hasbeencommissionedas partof theForesight ProjectonGlobalEnvironmentalMigration.Theauthorisgrateful

totheproject’steam forsupportand advice,aswellastotwo anonymousreviewersforveryhelpfulandinsightfulcomments.

References

Adam,D.,2005.50mEnvironmentalRefugeesbyEndofDecade,UNWarns.The Guardian,London.

Afifi,T.,Warner,K.,2008.TheImpactofEnvironmentalDegradationonMigration FlowsacrossCountries.WorkingPaperNo.5.UNU-EHS,Bonn.

Bates,D.C.,2002.EnvironmentalRefugees?ClassifyingHumanMigrationsCaused byEnvironmentalChange. PopulationandEnvironment23(5),465–477. Best,J.,2001.DamnedLiesandStatistics.UntanglingNumbersfromtheMedia,

Politicans, and Activists. University of California Press, Berkeleyand Los Angeles.

Betts, R.A.,Collins, M., etal.,2011.Whencouldglobalwarming reach48C? PhilosophicalTransactionsoftheRoyalSocietyA369(1934),67–84. Biermann,F.,Boas, I.,2010. Preparingfora warmerworld. Towardsa global

governancesystemtoprotectclimaterefugees.GlobalEnvironmentalPolitics 10(1),60–88.

Black,R.,2001.Environmentalrefugees:mythorreality?NewIssuesinRefugee Research,No.34.UNHCR,Geneva.

Boano,C.,Zetter,R.,etal.,2008.Environmentallydisplacedpeople:understanding thelinkagesbetweenenvironmentalchange,livelihoodsandforcedmigration. ForcedMigrationPolicyBriefing.RefugeeStudiesCentre,Oxford.

Brown,O.,2008.Migrationandclimatechange.IOMMigrationResearchSeries,No. 31.IOM,Geneva.

Castles,S.,2002.Environmentalchangeandforcedmigration:makingsenseofthe debate.NewIssuesinRefugeeResearch,No.70.UnitedNationsHigh Commis-sionerforRefugees,Geneva.

ChristianAid,2007.HumanTide:TheRealMigrationCrisis.ChristianAid,London. Conisbee,M.,Simms,A.,2003.EnvironmentalRefugees.TheCaseforRecognition.

NewEconomicsFoundation,London.

Crisp,J.,1999.Whohascountedtherefugees?UNHCRandthepoliticsofnumbers. NewIssuesinRefugeeResearch,No.12.UNHCR,Geneva.

Dun,O.,Gemenne,F.,2008.Defining‘environmentalmigration’.ForcedMigration Review31,10–11.

El-Hinnawi,E.,1985.EnvironmentalRefugees.UNEP,Nairobi.

FriendsoftheEarthAustralia,2007.ACitizen’sGuidetoClimateRefugees.Friends oftheEarthAustralia,Fitzroy,Australia.

Gemenne,F.,2011.Climate-inducedpopulationdisplacementsina48C+world. PhilosophicalTransactionsoftheRoyalSocietyA369(1934),182–195. GlobalHumanitarianForum,2009.TheAnatomyofASilentCrisis.Global

Humani-tarianForum,Geneva.

Hansen,J.,Sato,M., etal.,2006.Globaltemperaturechange.Proceedingsofthe NationalAcademyofSciences103(39),14288–14293.

Henry,S.,Boyle,P.,etal.,2003.Modellingtheinfluenceofthenaturalenvironment oninter-provincialmigrationinBurkinaFaso,WestAfrica.AppliedGeography 23(115-136).

Henry,S.,Piche´,V., etal.,2004.Descriptiveanalysisoftheindividualmigratory pathwaysaccordingtoenvironmentaltypologies.PopulationandEnvironment 25(5),397–422.

Hugo,G.,1996.Environmentalconcernsandinternationalmigration.International MigrationReview30(1),105–131.

Hunter,L.,2005.Migrationandenvironmentalhazards.Populationand Environ-ment26(4),273–302.

InternationalFederationofRedCrossandRedCrescentSocieties,2001.World DisastersReport.FocusonRecovery.InternationalFederationofRedCrossand RedCrescentSocieties,Geneva.

Jacobson,J.,1988.EnvironmentalRefugees:AYardstickofHabitability. World-WatchPaper86.WorldWatchInstitute,Washington,DC.

Ja¨ger,J.,Fru¨hmann,J., etal.,2009.EACH-FORSynthesisReport.EACH-FOR, Buda-pest.

Kniveton,D.,Schmidt-Verkerk,K.,etal.,2008.Climatechangeandmigration: improvingmethodologiestoestimateflows.IOMMigrationResearchSeries, No.33.IOM,Geneva.

Kolmannskog,V.O., 2008.Future Floodsof Refugees. ACommentonClimate Change,ConflictandForcedMigration.NorwegianRefugeeCouncil,Oslo. Majone,G.,1989.Evidence,Argument,andPersuasioninthePolicyProcess.Yale

UniversityPress,NewHaven,CT.

Massey,D.S.,Axinn,W.,etal.,2007.EnvironmentalChangeandOut-Migration: EvidencefromNepal.ResearchReports.UniversityofMichigan,Population StudiesCenter,AnnArbor.

Milan,S.,2004.SearchingforaPlaceUndertheSun.InterPressService,Rieti. Myers,N.,1993.Environmentalrefugeesinagloballywarmedworld.BioScience43

(11),752–761.

Myers,N.,1997.Environmentalrefugees.PopulationandEnvironment19(2),167– 182.

Myers,N.,2002.Environmentalrefugees: agrowingphenomenonofthe21st century.PhilosophicalTransactionsoftheRoyalSocietyB357(1420),609–613. Myers,N.,Kent,J.,1995.EnvironmentalExodus:AnEmergentCrisisintheGlobal

Arena.ClimateInstitute,Washington,DC.

Nicholls,R.J.,2004.Coastalfloodingandwetlandlossinthe21stcentury:changes undertheSRESclimateandsocio-economicscenarios.GlobalEnvironmental Change14(1),69–86.

Nicholls,R.J.,Marinova,N.,etal.,2011.Sea-levelriseanditspossibleimpactsgiven a‘beyond48Cworld’inthetwenty-firstcentury.PhilosophicalTransactionsof theRoyalSocietyA369(1934),161–181.

Pearce,F.2011.Searchingfortheclimaterefugees.TheNewScientist.2810. Piguet,E.,2009.Re-embeddingtheenvironmentintomigrationtheory.In:

Confer-ence:RemakingMigrationintoMigrationTheory: Intersectionsand Cross-Fertilisations, UniversityofBrightonandUniversityofSussex.

Piguet,E.,2010.Linkingclimatechange,environmentaldegradation,andmigration: amethodologicaloverview.WileyInterdisciplinaryReviews:ClimateChange1 (4),517–524.

Rahmstorf,S.,2010.Anewviewonsealevelrise.NatureReports4,44–45. Renaud,F.,Bogardi,J.J.,etal.,2007.Control,AdaptorFlee.HowtoFace

Environ-mentalMigration?InterSecTions,No.5.UnitedNationsUniversityInstitutefor EnvironmentandHumanSecurity(UNU-EHS),Bonn.

Sabatier,P.A.,1987.Knowledge,policy-orientedlearning,andpolicychange:an advocacycoalitionframework.ScienceCommunication8(4),649–692.

Stern,N.,2007.TheEconomicsofClimateChange.TheSternReview.Cambridge UniversityPress,Cambridge.

Suhrke, A., 1993.Pressure Points: EnvironmentalDegradation, Migration and Conflict.AmericanAcademyofArtandScience,Cambridge,MA.

Suhrke, A.,1994.Environmentaldegradationandpopulation flows.Journalof InternationalAffairs47(2),473–496.

Tolba,M.,1989.Ourbiologicalheritageundersiege.BioScience39,725–728. UnitedNations Conventionto CombatDesertification,2010. FrequentlyAsked

Questions.Retrieved10June,2010.From http://unccd.int/library/menu.php?-newch=l81.

Westing, A.H.,1992.Environmentalrefugees:agrowingcategory ofdisplaced persons.EnvironmentalConservation19(3),201–207.

Yenotani,M.,2011.DisplacementduetoNaturalHazard-InducedDisasters.Global Estimatesfor2009and2010.IDMC&NRC,Oslo.

Zolberg, A.R.,1989.Thenext waves: migration theoryfor achangingworld. InternationalMigrationReview23(3),403–430.