An Overview of the Issues

download An Overview of the Issues

of 11

Transcript of An Overview of the Issues

  • 7/27/2019 An Overview of the Issues

    1/11

    AnOverviewoftheIssues

    The theoretically correct measure ofeconomic impactsfrom a natural disasterwouldbe the

    change in welfare that occurred asa result ofthe event. Welfare can be evaluated ex post,

    asthe compensation required to avoid loss, or ex ante, which accountsforuncertainty

    (Rose 2!2". #lthough thin$ing in termsofhypotheticalwelfare measurescan be

    instructive, a complete welfare analysisisusually quite difficult empirically and would

    require ma$ing a numberof assumptionsand simplificationsin analysis. %fsociety were

    ris$&neutral, ex ante welfare could be evaluated with the expected economic loss(Rose

    2!2". 'cholarsinterested in empirical estimates(asopposed to modeling results, which

    can be useful in estimating welfare calculations"have attempted to measure observable

    disasterdamagesand follow&on economic impactsasa rough approximation ofthe neteconomic costsofa disaster.

    ariouslistsand typologiesofdisasterimpactshave been created. )ost scholarsof

    disastershave generally classified disasterimpactsinto direct and indirect impacts. *irect

    impactshave been described asthe physical destruction from a disaster, and indirect impacts

    (some authorspreferthe term higher-orderimpacts"are considered the consequencesofthat

    destruction (+ational Research ouncil !---". %n thisway, direct damagesreferto damages

    to structures, contents, and infrastructure that occurasa direct result ofexperiencing the

    haard. *irect impactsalso include mortality and in/ury caused directly by the haard.

    %ndirect damagesreferto lost economic activity, such aslossofpotential production,

    increased costsofproduction, lossin expected income, and otherwelfare losses, which occur

    asa result ofthe initial damage. *irect and indirect damagesinclude nonmar$et impacts,

    such asdeclinesin the quality oflife, environmental degradation, orlost recreational

    amenities, forexample. %n theory, it should be possible to sum up all direct and indirect

    lossesto generate a measure ofthe total economic costsofa disaster. %n practice, thisisquite

    difficult, forseveral reasonsdiscussed in thissection.

    The approach ofdividing disasterimpactsinto direct and indirect damagesand

    summing them appearsto be a theoretically straightforward accounting method forestimating

    the total economic impact ofnatural disasters0 however, the difficulty in practice hasled much

    ofthe literature to focusinstead on the impact ofdisasterson macroeconomic variables. This

    approach isanotherlensthrough which to viewdisastersand isclearly not additive with direct

    and indirect damages (1# 23" The assumption is that direct and indirect effects would

  • 7/27/2019 An Overview of the Issues

    2/11

    be reflected in macroeconomic accountsifthe disasterwassignificant. The focuson

    macroeconomic variables isprobably due in part to the fact that good data are available on

    them, but the ease ofdata availability doesnot imply that macroeconomic variablesare the

    best measure ofdisaster impacts, as%discussfurtherbelow.

    DirectandIndirectImpactsfromaDisaster

    %ndirect lossesinclude businessinterruption coststo those businessesthat did not

    sustain direct damage but may not be able to operate because, forexample, theirsupplier

    wasdamaged, theirwor$ersevacuated, orthey lost power. %t also includesthe multiplier

    effectsfrom reductionsin demand orsupply (more on these below". %n addition to causing

    businessinterruption, lossofinfrastructure orotherlifelines(e.g., power, sewage, orwater"

    can lead to utility lossto householdsin termsofa diminished quality oflife orcould cause

    both householdsand businessesto adopt costly measures(such asincreased commuting

    time asa result ofdamaged roadsorthe extra costsofrunning a private generatorwhen the

    electricity isout".

    Twomain complicationsarise when trying to measure the full economic costsin each

    ofthe categoriesin Table !. 4irst, it isnecessary to be very clearabout the spatial and

    temporal scale being examined because a different drawing ofthe boundariesofthe analysis

    can lead to different results. 4orexample, considerthe economic costsofa disasterfrom the

    point ofviewofa homeownerwho lost herhome. 'ome direct losses, such asthe home, are

    reimbursed by insurance oraid from government orothergroups, and some lossesare un

    reimbursed and borne fully by the victims. That is, assetsthat are replaced (some may not be"

    are eitherfrom inter temporal transfersofthe individual orinterpersonal transfersfrom one

    person to another(+ational Research ouncil 25". %fthe individual receivesdisasteraid

    from the government, the economic cost ofthe disasterto that person will be the value ofthe

    lost home minusthe amount ofthe aid. 4rom the perspective ofsociety, however, the aid is

    /ust a transferfrom one taxpayerto anotherand thusshould not be added orsubtracted from

    the direct damage to the home.

    Temporal boundariescan also matter. #san example, it hasbeen shown (see below"

    that construction sectorscan experience a boom right aftera disasteraspeople rebuild. #couple ofyearsafterward, however, they may face a lull because people will underta$e

  • 7/27/2019 An Overview of the Issues

    3/11

    upgradesduring the post&disasterreconstruction that they would have deferred until further

    into the future. oo$ing only one yearpost&disastermay suggest a benefit to the construction

    sector, but loo$ing overthree yearsmight diminish thisbenefit. #nd to highlight again the

    point made in the preceding paragraph, although the construction sectormay get a benefit,

    had the disasternot occurred, the fundsspent on rebuilding would have been spent elsewhere

    in the economy, with a higherutility to the homeowner0 thus, post&disasterspending should

    not simply be counted asa benefit to the disaster.

    The second challenge isthat it isquite easy to double&count losses. 4orexample,

    assume a machine isdamaged irreparably in a flood. The value ofthat machine isthe net

    present value ofthe future returnsfrom itsoperation. Thusthe value ofthe machine and the

    lost production ofit should not both be counted asa loss(Rose 26". #sanotherexample,

    one would not want to count both the aid disbursed by government and the rebuilding costsbecause much ofthe aid isprobably used forthe rebuilding.

    'o, given these difficulties, what would be the most preferable measure ofdirect and

    indirect damages7 The next three sectionsdiscussin more detail the challengesconfronting

    an analyst, who must comprehensively assessthe economic lossfrom a natural disaster, in

    arriving at a total estimate ofthe disasterimpacts.

    DirectDamages

    The economic cost that usually first comesto mind when thin$ing about natural

    disastersisdamage to buildingsand contents. Though seemingly straightforward to measure,

    getting the precise economic costsofthisimpact isnot theoretically trivial. onsidera house

    that iscompletely destroyed. The economic losscould be measured aseitherthe mar$et value

    ofthe house right before the disasterhit orthe replacement cost to rebuild it. The most

    appropriate measure isthe mar$et value at the time ofdisasterimpact. The replacement cost

    could be higherorlowerforseveral reasons. 8ost&disaster, some materialsmay be in short

    supply and more expensive substitutesused orhigherpricescharged, forexample, orlabor

    may be in short supply and thuswageshigher, driving the cost ofrebuilding above what it

    would have been before the disaster(9lsen and 8orter2:". Thisisoften referred to as

    demandsurge. #lthough these highercostsare a lossto the homeowner, they are a gain to the

    suppliersand builders0 therefore, from the point ofviewofsociety asa whole, they are

    ;/usttransfers. 9n the flip side, if businessinterruption issevere and more laborersare

    loo$ing fortemporary wor$, rebuildingcostscould be lower. Thisagain wouldbe a savings

    to the homeownerthat, from the perspective ofsociety asa whole, would offset the lossto the

    wor$er

  • 7/27/2019 An Overview of the Issues

    4/11

    Thispicture iscomplicated by government disasteraid payments. 4orthe individual,

    aid will lessen the economic impact ofa disaster. 4rom the point ofviewofsociety, the

    government aid isa transferfrom one taxpayerto another, asstated above. The dead eight

    lossoftaxation ispositive, however, and the marginal opportunity cost ofa dollarof

    government spending isin most casesli$ely to be greaterthan

    to include thiscost ofgovernment spending. =owever, it isnot necessarily the case that

    disasteraid will require newtaxation because fundsmay instead be diverted from anotheruse.

    %n such a case, it ispossible that thisdiversion could lessen the dead eight lossoftaxation if

    the aid waslessdistortionary than the fundsin theirnon disasteruse. %fthe fundswere from

    increased government borrowing, thiscost offederal borrowing would need to be included.

    The homeownercould also receive insurance payoutsifhe orshe had disaster

    insurance. Thiswould again lessen the negative wealth shoc$ to the homeowner. #ssuming

    ris$&based pricing ofinsurance and well&diversified companies, claimspayoutsshould not be

    considered a cost ofa disaster. They are often used asa proxy foreconomic costs, however,

    asthey should theoretically be closely correlated with the lost value ofthe homesand

    structures>at least in areaswith high ta$e&up ratesofinsurance. 4urther, insurance

    companiesusually $eep extremely good recordsand so are an excellent data source.

    %n addition to the cost ofthe lost home, otherdirect lossesto the homeownerinclude

    the time lost to the rebuilding effort, emotional trauma orstress, and lossofnonmar$et items

    ofvalue, such asbaby photographsorfamily $eepsa$es. These lossesare rarely included in

    disasterdamage estimates.

    *estruction to the buildings, contents, inventory, and capital offirmscan be similarly

    analyed. 4ordestroyed capital, depreciation must also be considered, and the correct

    measure ofeconomic lossisthe depreciated value ofthe lost asset. %fproduction islost from

    the delay in replacing damaged capital, then the lost production from delay should also be

    counted asan economic loss. The literature examining possible positive impactsofdisastersis

    premised on the notion that replaced capital could be more productive than the capital

    destroyed ifthere hasbeen technological change (discussed in 'ection ? below". This

    productivity bump could not be so high asto ma$e the firm betteroff, orelse they would

    have already upgraded the capital. 'till, the productivity increase will offset some ofthe

    economic loss. %fthe firm receivesdisasteraid such that the upgrade is, in a sense, free to the

    firm then it could in theory be better off post disaster if the productivity bump is great

  • 7/27/2019 An Overview of the Issues

    5/11

    enough. #gain, though, from the point ofviewofsociety, the aid issimply a transfer.

    %nfrastructure damage isanothercategory ofdirect lossfrom a natural disaster. #gain,

    the depreciated value isthe correct measure ofeconomic loss. *elaysin repairand rebuilding

    can triggerindirect costs, discussed next, through an interruption in use orservice.

    1specially in the developing world, lossoflife and in/ury from disasterscan be large,

    and these are direct costsofa disaster. #n enormousdebate centerson howto value lossof

    life and in/ury, and %will not rehash that here, except to note that a value&of&a&statistical life

    ('"estimate based on disasterris$ explicitly would be the best measure. To my

    $nowledge, very few, ifany, ' estimateshave loo$ed explicitly at natural disasterris$,

    although one comparative stated preference study findsthat willingnessto pay (WT8"toreduce mortality ris$ isgreaterforterrorism than fornatural disastersand that reducing the

    mortality ris$ from natural disastersisvalued about the same asthat from traffic accidents,

    even though the latterisa much higherris$ (iscusi 2-". %n/ury and illnesscan be measured

    in quality&ad/usted life&yearsorsimilarmeasures.

    *irect damagescan also include environmental degradation. 4orsuch nonmar$et

    losses, an estimate ofsociety@stotal WT8to have avoided the lossex ante isa measure oftheeconomic loss. #gain, a large literature addressesnonmar$et valuation techniquesthat can be

    applied to obtain such estimates, which will notbe discussed here (see, forexample, 4reeman

    23".

    4inally, emergency response and debrisclean&up could also be considered direct costs

    of a disaster. Thiswould include the opportunity cost ofpeople@stime spent hauling away

    debris, forexample, orthe costsofevacuation. )any ofthese costsare borne by

    governments, and ifthe interest isin total economic impactsto society asa whole, care must

    be ta$en in correctly estimating these costs. %ssuesthat must be considered include whether

    newtaxation was required, ifgovernment indebtednessincreased, andAorwhat the funds

    would have been spent on in the absence ofa disaster.

    #lthough thispaperisfocused exclusively on economic impacts, previouswor$ has

    examined broaderimpacts, including demographic shiftspost&disaster. 4orexample, it was

    found that after=urricane #ndrew, low&income groupsmoved into areasthat had been

    damaged (potentially because these areaswere cheaper", the proportion ofmiddle&income

    groups in damaged areas declined and the wealthy remained (perhaps because insurance and

  • 7/27/2019 An Overview of the Issues

    6/11

    self&protectionwere more affordable forthisgroup0 ('mith etal.25". 'uch changescould

    have welfare effects.

    IndirectDamages

    *isasterscan be viewed asa negative capital shoc$ to a region. Thishasfollow&on

    economic consequencesin addition to the value ofthe lost assets. 4irst, economic lossesarenot exclusive to firmsorhouseholdsthat sustain direct physical damage. %felectricity or

    wateris lost, forinstance, it can cause businessinterruption to even firmsthat are not

    themselvesdirectly damaged. Thisisa widely recognied disastercost. 'imilarly, the lossof

    such servicescan lead to a decline in the quality oflife forhouseholds, and thusa utility loss,

    and could also lead to the need forcostly measuresto compensate, although thisisrarely

    discussed in the literature. 'uch compensating actionscould involve longertravel timesdue

    to a road outage orthe purchase ofbattery&powered lighting in response to a lossof

    electricity, forinstance. These are indirect damagesto include in estimatesoftotal costs.

    #ttention in the literature hasfocused on possible multipliereffectspost&disaster.

    onsumerdemand post&disastermay be higherforsome sectors>such asconstruction,

    particularly ifaid orinsurance isfunneled to rebuilding>and lowerforothersasconsumers

    forgo some expendituresto use theirfundsforrebuilding. These typesofexpenditure changes

    could have economic multipliereffectswithin the community (positive ornegative". #

    similarstory can be told forbusinessinterruption. Thiscould decrease demand forinputsand

    reduce outputs, having negative ripple effectsin the supply chain0 although here, aid and

    insurance could mute these impactsifsuch fundsallowfora fasterresumption ofnormalbusinessactivity.

    4rom the perspective ofthe whole economy, however, multipliereffectsmay well be ero,

    with positive andnegative impactscancelling out (+ational Research ouncil !---". 4or

    instance, ifa firm failsto produce an output, itscustomermay simply purchase the good

    elsewhere. Thisisa lossforthe disaster&impacted firm but a gain fora competitor, and thusa

    wash from the point ofviewofthe whole economy. #sanotherexample, touristsmay avoid a

    hurricane&stric$en coast, but instead ofnot traveling, they may /ust frequent anotherarea.

    #gain, thismay be a wash from the point ofviewofthe entire economy, but clearly thedistributional impactscould be quite large and could have significant consequencesfor

    individuals, firms, orcommunities.

    %fa government doesma$e changesto taxation orresource allocation post&disaster,

    thiscould have indirect economic effects. 4orhard&hit countries, particularly small orpoor

    countries, thisisa distinct possibility and would need to be evaluated. ountriescan also

    receive international assistance (which, again, would be a transferfrom a global perspective".

    ase study evidence suggeststhat donorsdo not necessarily provide additional aid aftera

    disaster, but simply reallocate aid budgets(Benson and lay 26".

    )ortality or illness could also occur as a result not of the haard but of the initial

  • 7/27/2019 An Overview of the Issues

    7/11

    damage. 4orinstance, ifwaterbecomescontaminated asa result ofthe shutdown ofa

    treatment plant and thisleadsto illness, it would be an indirect cost ofthe disaster. #fter

    hurricane Catrina, an increase in mortality rateswasobserved because the storm destroyed

    much ofthe health infrastructure ofthe city ('tephens2D". These would be deaths

    classified asan indirect cost.

    4inally, disasterscould causepeople to altertheirris$ perceptions. Thiscould then

    induce behavioral responsesand a reallocation ofresources. These could have

    economic consequences, such aswor$ersrequiring a ris$ premium post&disaster(on

    thispoint in the context ofterrorism, seeE Rose 2!2", but are not furtherconsidered in

    thisreview. 'imilarly, utility functionsmay be state dependent and may change aftera

    disaster, such that ex ante valuationsare not the same asex post valuations.!#

    complete welfare assessment would need to considerthese possibilities.+otably,

    positive utility gainscould occurvia public aid post&disasterifpeople feel good abouthelping those in need and reassured that ifthey are victims, aid will be forthcoming.

    i$ewise, utility lossescould be associated with any increasesin fear(or other

    negative emotions", which #dler(26"arguesshould be measured and included in

    regulatory costFbenefit analysiswhen relevant.

    1conomistsexhibit some hesitancy regarding estimationsofhigher&ordereffects, and

    the complicationsdiscussed above hint at why. Rose (26"notesthe following concernsE

    indirect effectsare hard to verify, modeling them can be difficult, the sie ofthe impactscan

    very substantiallydepending on the resiliency ofthe economy and pace ofrecovery, and themodeling ofsuch effectscould be manipulated forpolitical purposes(e.g., inflating the

    multiplier". 'till, when calculated carefully, they are a true cost ofthe disasterand should be

    included in any complete accounting. )ost estimatesoftheirmagnitude have been done

    through modeling ratherthan empirical analysis.

    MacroeconomicApproaches

    The ma/ority ofeconomic studies, instead ofattempting to estimate direct or

    indirect costs, evaluate the impact ofnatural disasterson macroeconomic indicators,

    primarily grossdomestic product (G*8". %t isthusworth saying a word about howthese

    estimatesofdisasterimpactsrelate to estimatesofdirect and indirect economic effects.

    %t ispossible that the direct andindirect effectsofa disastercould be large enough to

    have macroeconomic effects, includingimpactson economic growth, balance ofpayments,

    fiscal revenues, levelsofindebtedness, and investment rates(1# 23". %fdamagesare

    severe, output could decline. 9utput could also increase from post&disasterreconstruction. %t is

    unclear, on net, howthese effectswould balance out. *amagesto firmscould alterimports

    and exports. Government spending foremergency response, ifhigh enough, could change

    indebtedness. Tax revenue could be impacted. %fseriousprice increasesresult from the

    disaster this could fuel inflation 4oreign direct investment could fall if companies see too

  • 7/27/2019 An Overview of the Issues

    8/11

    great a ris$ ortoo much damage.

    'ome ofthese impactsare essentially indirect economic impactsthat should be counted in

    total economic impact estimates. )ore often, however, macroeconomic variablesare used as

    aproxy forthe direct and indirect impacts/ust discussed. 4orexample, government spending

    isoften used asa measure ofthe damagesfrom a disasterbut need not be directly related to

    economic losses. 'imilarly, G*8isoften used to capture total economic impacts. %t isworth

    stressing, however, that G*8issimply a measure ofeconomic activity, not ofwealth or

    welfare. The usual argumentson thispoint extend to the case ofnatural disasters. The

    literature on the G*8impactsofdisastersisreported here, but with thisstrong note ofcaution

    that it isa poorproxy foreithertotal economic costsorwelfare impactsofa disasterevent.

    MeasurementProblems

    The thorny theoretical problemsinvolved in estimating the economic consequencesofdisastersare coupled with extreme data limitationsthat ma$e actual estimatesfarfrom

    what would be the hypothetical ;truedisastercosts. )any scholarshave stressed the need

    for reliable, comprehensive, systematicallycollected disasterlossdata (e.g., Thomas2!".

    Good data on disasterlossesare needed fora range ofpurposes, including costFbenefit

    analysisofmitigation measures, government preparednessplanning, calibration ofloss

    models, and ris$ analysisforinsurersand otherentities. 1ven in highly developed countries

    with generally good record&$eeping, comprehensive disasterlossdata are difficult to come

    by. The Hnited 'tates doesnot $eep systematic recordsin one location oflossesassociated

    with natural haards. )any expertshave called forsuch a database to be developed and

    maintained by the federal government (e.g.,+ational Research ouncil !---", but thusfarit

    hasnot occurred.

    %n general, the data available on disasterimpactsare on those thingsthat are easily

    observable ex post. 4orinstance, depreciated replacement value isthe best estimate of

    damage to structuresand assets, but isnot available in most data sets()ileti !---"0

    replacement cost ismore li$ely to be observed. Because public expendituresare fairly easy to

    observe and often are already measured in some form in most countries, they are frequently

    used asa proxy forlosses0 but, asalready mentioned, public expenditureson reconstruction

    may not match the magnitude oflossesto capital (avallo and+oy 2!". )ost disasterdata

    setsdo not include indirect lossesordamagesto nonmar$et goodsand services0 therefore,

    most disasterlossdata probably underestimate the full economic impact ofdisasters()ileti

    !---0)itchell and Thomas2!".

    #notherdifficulty with disasterdata isthat many high&magnitude eventsare complex,

    with multiple interrelated perils(Cron et al. 2!2". 4orinstance, strong hurricanesbring

    with them high winds, torrential rain, and storm surge0 these could furthertrigger

    landslides. 'evere stormscouldinclude damage from wind, hail, flooding, lightening,

    and tornadoes 1arthqua$escan trigger tsunamis or fires This ma$es classifying

  • 7/27/2019 An Overview of the Issues

    9/11

    disastersforcomparison acrosseventsdifficult.

    4inally, some countrieshave much betterrecord&$eeping than others. 'ome countries

    may not have institutionsthat are tas$ed with damage estimation, and in some placespost&

    disasterassessmentsmay be difficult. 4urther, developing countriesmay have an incentive to

    exaggerate damagesto gain international aid and, regardless, obtaining good damage

    estimatesin developing countriescan be a challenge because insurance penetration islow,

    boo$ $eeping isoften poor, and much economic activity occursin informal sectors(Toya and

    '$idmore 2D". ery little is$nown empirically about disasterimpactson informal sectorsof

    the economy.

    Thus, all disasternumbersshould be interpreted with some degree ofcaution. %n

    addition, the nature ofthe database can influence the conclusionsdrawn about disasterlosses,

    asnoted by Gall et al. (2-". 4orinstance, different databasesinclude different itemsin the

    estimate ofdamages(e.g., /ust direct damages, orboth direct and indirect", which can causedifferencesin ran$ingsofevents. %n theircomparison ofdisasterdamage estimatesin the

    Hnited 'tatesacrossthree different databases, Gall et al. (2-"find that, although all three

    databasesagree that hurricanesand tropical stormsare the most damaging haard in the

    Hnited 'tates, they differon which haard isran$ed second>earthqua$es, severe weather, or

    floods.

    #t an international scale, three primary data setsare available forcross&country,

    multiple&haard analysis. These are the entre forResearch on the 1pidemiology of*isasters@

    (R1*@s"1mergency 1vents*atabase (1)&*#T", 'wissRe@s'igma, and )unich Re@s+atat'1R%1. 1)&*#T hasa humanitarian focus, and the reinsurance databases('igma

    and +atat'1R%1", not surprisingly, focuson insured and material losses. The databases

    have different thresholdsforthe inclusion ofevents(Cron et al. 2!2". 1)&*#T includes

    eventswith more than ! fatalities, more than ! people affected, a declaration ofa state of

    emergency, ora call forinternational assistance. 1ventsare included in 'igma ifoverall losses

    exceed H'< :5.5 million, insured lossesexceed H'< 63.3 million (both in 2! dollars", or

    there are 2 ormore fatalitiesormissing persons.+atat'1R%1 includesany event in

    which harm to people orproperty damage occurs. #ll ofthese databasesacquire information

    from a variety ofsources.

    1)&*#T ispublicly accessible, whereasthe reinsurance databasesare not, althoughstatistical

    analysesare published by the firms. Thismeansthat foralmost every cross&country,

    multihaard paper, the 1)&*#T data are used. Because ofthis, a fewthingsshould be $ept in

    mind about these data. 4irst, given the threshold forinclusion mentioned above, small events

    are not included, even though frequent lower&impact eventscould still cause substantial

    economic costs. 'econd, 1)&*#T isfocused on aiding humanitarian response. #ssuch, it has

    no threshold fordamages, so eventsin more developed countrieswith a high level ofdamage

    but lowlossoflife and no call forinternational aid may fail to be included. 4inally, 1)&*#T

    data are compiled from multiple sources and are only as good as those sources 'ources

  • 7/27/2019 An Overview of the Issues

    10/11

    include the Hnited+ations, governmental and nongovernmental organiations, insurance

    companies, research institutes, and the press. The sourcesare ran$ed according to their

    trustworthinessin providing accurate and complete data. ollecting disasterdata isa difficult

    process, and R1*should be commended on the wor$ done to create and maintain this

    database. %t isthe best source forconsistent, multicountry natural disasterdata available. That

    said, we would be more confident in ourestimatesofthe economic impactsofnatural

    disastersifmultiple data sourcesall found the same results. With so much ofthe literature

    relying on thisone data source, any problemswith the data will propagate through all

    analyses.

  • 7/27/2019 An Overview of the Issues

    11/11