Page - 74 - in Loss and Damage from Climate Change - Concepts, Methods and Policy Options

74 L.M.Bouwer noted that there is insufficient evidence7 toconclude that severeconvectiveweather hasalreadybecomemore frequent. Otherweather hazards forwhich positive loss trends are found include studies on tropical storms.WhileNordhaus (2010) foundapositive trend forUShurricane losses after normalisation, Bouwer andBotzen (2011) found no trend using other loss recordsandalternativenormalisationof the sameevents.The studybyEstrada etal. (2015)usedanalternativeformulation,assuminganon-linearrelationbetween changesinexposureandlosses,whichhasnotyetbeenconfirmedbyotherstudies,nor hastherebeenasufficientexplanationforthecauseoftheremainingincreaseinlosses (Hallegatte 2015). Schmidt et al. (2009) foundan increasing trend inUShurricane losses after normalisation since 1970, but this is likely due to natural variability (Bouwer 2011a); in this case the low frequency of landfalling hurricanes in the NorthAtlanticandCaribbeaninthe1970s,andthesubsequent increase in the1990s and early 2000s. For river flooding some studies find increases after normalisation (Fengqinget al. 2005;Changet al. 2009), but theseare relatively short-lived, and it is unclearwhether these increases are related to changes infloodhazard driven by naturalvariabilityor anthropogenicclimatechange. Finally,therearesomestudiesthatindicateincreasinglossesafternormalisationat theglobal level, forseveral typesofweatherextremes(Milleretal.2008;Neumayer andBarthel2011),but these trendsarealsoover recent times(over30yearsor less), andhereit isalsounclearwhetheranyrelatedchangesinhazardaredrivenbynatural variabilityoranthropogenicclimatechange. 3.3.3 InterpretationofDriversofLosses As shownabove, few studies find signals in losses beyond the driver of increasing exposure.Less isknownabout theroleofvulnerabilitychanges thatpotentiallymay playanimportantrole.Associetiesbecomewealthier, theyarelikelytostart toinvest moreinriskreductionandadaptation,therebyreducingimpactsfromweatherrelated hazards.Thismayresult in reduced lossesover time.Fornormalisationstudies, this may imply that accounting for increases inexposureonly,woulddownplay the role ofanyothercontributing factors, includinganthropogenicclimatechange (Nicholls 2011). Indeed, therearestudies that showthatespecially lossof lifeandalsomone- tary losseshavedecreased,despite increasingexposure(MechlerandBouwer2015; Bahinipati and Patnaik 2015; Kreibich et al. 2017; Bouwer and Jonkman 2018). Jongmanet al. (2015) for instance stress that despite the fact that total losses from riverfloodinghaveincreased,fatalitiesandmonetisedlossesasashareofpopulation andGDP, have fallen over past decades. However, the question is how significant these changes in vulnerability are, compared to the very rapid increase in expo- sure (Bouwer2011b).While lossof life clearlyhasbenefitted fromimprovedearly 7“There is lowconfidence inobserved trends in small spatial-scale phenomena such as tornadoes andhailbecauseofdata inhomogeneitiesandinadequacies inmonitoringsystems”(IPCC2012:8).