Secondary displacementIn some cases, people fleeing a naturalhazard or disaster were already livingin displacement before it struck. If it wasclear, for example, that people alreadydisplaced by conflict were then forcedto flee again in 2014 by an event such asthe flooding of their camp, the new movementswere recorded as new incidencesof displacements related to a naturalhazard. It should be noted that the veryfew events of this type that we recordprobably under-represent the frequencyof secondary displacement related tonatural hazards as information acrossmost situations is hard to identify.Sources of informationWe regularly review the types of informationdifferent sources release onthe number, needs and characteristics ofdisplaced people, primarily by gatheringand monitoring secondary reports. Wesystematically seek a range of sourcesfor each country and each disaster. Forour 2014 estimates, we increased our researchcapacity and analysed data fromsources including government reportsand national disaster loss databases, IOM,IFRC’s disaster management informationsystem, UN Office for the Coordinationof Humanitarian Affairs (OCHA) andother UN agencies, humanitarian clusterreports, the Asian Disaster ReductionCentre’s global unique disaster identifierwebsite and NGO reports.We also used reputable media outletsas a source of quotes from governmentofficials and local authorities, and localmedia reports provided an additionalsource for small events. IOM country officesprovided field data and/or gave usaccess to official sources that we incorporatedinto our data for 37 countries.If the original source of quantified informationwas unidentifiable, mostly in thecase of figures quoted in the media, werecorded it as “unspecified” rather thanreferencing the publisher.Selection and calculation of estimatesby event or disasterIn providing our global estimates, weaim to arrive at the best approximationof the total number of people displacedby a specific event or disaster, measuringthe incidence of displacement ratherthan the evolution of the number of peopledisplaced and their movements and situationsover time. Our analysis and interpretationof information from multiple sourcesincludes the cross-checking of reportedlocations and dates to ensure that figuresare associated with the same disaster andtime period, and that double counting isavoided or minimised. All new incidences ofdisplacement during a given event or disasterperiod are recorded, which requiresthe analysis of reporting dates and theconsideration of series of situation reports.The estimate per event is selectedaccording to the most accurate and reliablefigure provided or calculated basedon a single source, or combined sourceswhen it is clear that overlap and doublecounting can be avoided. The number oforiginal sources available (disregardingthose that re-publish original informationfrom elsewhere) varies from one or twofor smaller events to more than four forlarger events. Disasters widely coveredby the media or which continue for longperiods of time also tend to have moresources from which to draw.If displacement was clearly reportedbut no explicit figures were available onwhich to base an estimate, we did not includethe event in our data. When figureswere only provided in generic terms andmore precise data was not available, weapplied the following rule: “hundreds” =200; “thousands” = 2,000; “hundreds ofthousands” = 200,000Reporting terms that identifydisplacementA wide range of methods, definitionsand terms - such as evacuated, homeless,damaged and destroyed housing,fled, relocated and affected - are used forcollecting and reporting figures, and differentsources use them in different ways.Such variations arise in part because organisationshave different reasons forcollecting and reporting data in the firstplace. In operational settings, the term“displaced” is often applied more narrowlythan our definition. It may be used only forpeople staying in official collective sitesor camps, or those displaced a certaindistance from their homes.In some cases evacuees who move toshort-term evacuation centres are countedseparately from IDPs in camp-likesettings. In others, they are considered asubset of the displaced population. IDPsare sometimes considered a subset ofthe affected population, and sometimesadditional to it. Information describing thecontext and point in time at which displacementis reported, knowledge of typicalpatterns observed in similar contextsand the quality and reliability of differentsources are also taken into account.We interpret the data we collect usingthe same broad and inclusive definitionof displaced people across all eventsworldwide. Our definition assumes theyare part of the population affected by adisaster, but this does not imply that allthose affected have been displaced. Weconsider evacuees to be IDPs whetheror not their evacuation was pre-emptive,and we define people whose homes arerendered uninhabitable as displaced,regardless of how near or far from theirhomes they move and whether or not theyare able to return.We recognise that different situationscreate different needs, but our researchdoes not suggest that IDPs who fleefurther necessarily have greater needsor are more vulnerable. On the contrary,displacement over short distances maybe a better indicator of vulnerability, particularlyif it is recurrent, given that thoseaffected may be unable to make their wayto safer locations or places where theyhave better access to assistance.Evacuation dataIn addition to direct reports of peoplehaving been displaced, fled or been forcedto leave their homes, we often use dataabout mandatory evacuations and peoplestaying in official evacuation centres toestimate event-based displacement.On the one hand, the number of peoplecounted in evacuation centres mayunderestimate the total number of evacuees,given that some may take refuge atunofficial sites or with family and friends.On the other, the number of people orderedto evacuate may overstate theirtrue number, given that some will usuallynot heed the order. The potential for suchdiscrepancies is much greater when authoritiesadvise rather than order evacuation,and as a result we do not incorporatesuch figures into our estimates.82 Global <strong>Estimates</strong> 2015
Data on housing rendereduninhabitable and people madehomelessData on people rendered homelesspoints to a severe situation and the riskof prolonged displacement. Areas wherehomes and infrastructure have been severelydamaged or destroyed are unlikelyto be able to support early safe returns.The number of people made homelessmay be reported directly, but if not we inferit from the number of homes reportedas severely damaged or destroyed, multipliedby average household size. We donot use data on homes reported simplyas damaged, because the term tends tobe too broad to determine whether or notthey have been made uninhabitable.A new formula for calculating averagehousehold sizeNearly half of our event-based estimatesrely in part on calculations basedon houses rendered uninhabitable multipliedby average household size, butstandardised and up-to-date informationis not available for all countries. In its absenceit is not an easy parameter to enumerate,but given its importance we haveimproved this aspect of our methodology.We previously applied a rough butconsistent calculation by adding an“adults in household” constant of two tothe modelled fertility rate for 2010 to 2015.Our new formula, applied to our 2013 and2014 datasets, is as follows:1. If household data is available fromthe 2013 UN Statistics Division(UNSD) household size dataset,these figures are used - total population/totalhouseholds - otherwise:2. If household data is available fromthe 1995 UNSD household dataset,these figures are used - 1995 totalpopulation/total households - otherwise:3. If living quarter data is available fromthe 1995 UNSD household dataset,these figures are used - 1995 totalpopulation/total living quarters - otherwise:4. If fertility rate data for 2000 to 2005is available from the UN PopulationDivision is available, these figures areused - adding an “adults in household”constant of 1.7, otherwise:5. The previous formula using the modelledfertility rate for 2010 to 2015 plusan “adults in household” constant oftwo is applied.The process is honed further by prioritisingselected events and countries for additionalresearch to glean the most recentstatistics available from the websites ofnational statistics offices. For 2014, thiswas done for all disasters that displacedmore than 100,000 people and all countrieswith three or more event estimatesthat relied on household-level data.Where time permitted, the same researchwas done for countries with significantevents that displaced fewer than100,000 people but which met the secondcriterion. Official average household sizesfor Bangladesh, Bolivia, Cambodia, China,India, Indonesia, Japan, Malaysia, Pakistan,the Philippines, Sri Lanka, Sudanand the US were calculated in this way.Full technical notes on this part of ourmethodology are available on request.Reporting biasThere are a number of causes of biasin our source information and methodologythat should be noted:It is often difficult to determine whetherdisplacement data is reliable and comprehensive.Global reporting tendsto emphasise large events in a smallnumber of countries where internationalagencies, donors and media have a substantialpresence, or where there is astrong national commitment to, and capacityfor disaster risk and informationmanagement.Data on smaller-scale disasters is farmore scarce and on the whole significantlyunder-represented. The effectsof disasters on isolated and insecure areasalso tend to go relatively unreportedbecause access and communicationsare limited.There tends to be significantly moreinformation available on displacedpeople in official or managed collectivesites than there is on those livingwith host families and communities orin other dispersed settings. Given thatthe majority of IDPs usually fall into thesecond category, figures based on datafor collective sites only are likely to besubstantial underestimates.Reporting tends to be more frequent butless reliable in the most acute and highlydynamic phases of a disaster, whenpeak levels of displacement are likely tobe reached. It becomes more accurateonce there has been time to make morereliable assessments. This means thatestimates based on later evaluations ofseverely damaged or destroyed housingwill be more reliable, but they are alsolikely to understate the peak level ofdisplacement, given that they will notinclude people whose homes escapedsevere damage but who fled for otherreasons.Reporting bodies may have an interestin manipulating the number of peopledisplaced. It may be to maximise theamount of external assistance received,downplay the scale of a disaster if thegovernment may be held accountableor because international attention isdeemed politically undesirable.Improvements in the systematic collectionand sharing of reliable informationon displacement are essential if we areto continue to improve the quality of ourreporting and monitoring - a critical firststep in identifying needs, prioritising assistanceand informing longer-term solutions.Quality assuranceAhead of this year’s report, our eventbaseddatasets for 2008 to 2014 underwentfurther improvements in terms ofnormalisation and standardisation, whichhas increased the type and quality of analyticswe can run. Background informationcollected for all displacement events isnow archived, and preliminary estimatesfor all events that displaced 100 peopleor more were reviewed by a minimum oftwo in-house researchers. The thresholdof 100 was chosen to ensure that as muchdata as possible was of the best qualitywithin time constraints that did not permitthe review of every event.Reviewers checked the data recordedfor each event against its source documents,recorded discrepancies and commentedon the quality of the overall estimate.Revisions were made by the personwho originally entered the data. Eventsthat displaced 100,000 people or morewere reviewed for a third time by a differentreviewer to minimise the potentialAnnexes83
- Page 1:
Global Estimates 2015People displac
- Page 4 and 5:
IDMC core project teamCoordinator/l
- Page 7 and 8:
5. Mind your assumptions: Protracte
- Page 9 and 10:
FIGURES, TABLES AND MAPSFigure 2.1:
- Page 11 and 12:
Modelled global displacement trend
- Page 13 and 14:
1INTRODUCTIONSince 2008, an average
- Page 15 and 16:
2CONCEPTUALISINGDISPLACEMENTin the
- Page 17 and 18:
Box 2.1: Slow-onset hazards and gra
- Page 19 and 20:
Repeated and frequent displacementR
- Page 21 and 22:
3THE GLOBAL PICTUREScale, patterns
- Page 23 and 24:
Figure 3.2: Global displacement by
- Page 25 and 26:
Box 3.1: “Super” El Niño and d
- Page 27 and 28:
While more resilient families may b
- Page 30 and 31:
28 Global Estimates 2015A woman sta
- Page 32 and 33:
4.1 RegionsAs in previous years, As
- Page 34 and 35: 4.2 CountriesThe number of people d
- Page 36 and 37: forward, the municipalities of Pasc
- Page 38 and 39: 4.4 The big three: China, India and
- Page 40 and 41: Map 4.1: Philippines regions affect
- Page 42 and 43: Displacement caused by seasonalfloo
- Page 44 and 45: 4.6 Multiple hazards in fragile and
- Page 46 and 47: SPOTLIGHTAFGHANISTANBlurred lines b
- Page 48 and 49: 46 Global Estimates 2015A displaced
- Page 50 and 51: The global data presented in this r
- Page 52 and 53: The international experts we interv
- Page 54 and 55: perstorm Sandy in 2009, and indigen
- Page 56 and 57: SPOTLIGHTPAPUA NEW GUINEAManam isla
- Page 58 and 59: SPOTLIGHTINDONESIASidoarjo mudflow
- Page 60 and 61: SPOTLIGHTBANGLADESHSix years after
- Page 62 and 63: Figure 5.4: Movement of IDPs from a
- Page 64 and 65: SPOTLIGHTHAITIChronic vulnerability
- Page 66 and 67: Conditions have continued to deteri
- Page 68 and 69: Map 5.3: Ongoing displacement in th
- Page 70 and 71: The mental and physical health ofID
- Page 72 and 73: SPOTLIGHTUNITED STATESDisplaced peo
- Page 74: assistance available and implementa
- Page 77 and 78: 6THE POST-2015GLOBAL POLICYAGENDAKe
- Page 79 and 80: A landslide devastated villages in
- Page 81 and 82: ANNEX AMethodologyIDMC’s annual G
- Page 83: protracted displacement situations
- Page 87 and 88: Modelling and calibration usingthe
- Page 89 and 90: We also asked each interviewee toth
- Page 91 and 92: RankingCountry Hazard Affected area
- Page 93 and 94: * Text in parentheses indicates the
- Page 95 and 96: Country DisasterStartdateTotal no.
- Page 98 and 99: Country DisasterStartdateTotal no.
- Page 100 and 101: Country DisasterStartdateTotal no.
- Page 102 and 103: REFERENCESSection 21. UNISDR, Termi
- Page 104 and 105: 34. NDRRMC, SitRep No.27 re Effects
- Page 106 and 107: nation Group, October 2009, availab
- Page 108: digidepo_8841940_po_076709.pdf?cont