Presentation on theme: "OECD 2nd International Cat Risks Conference Bangkok, Sept 25th 2009"— Presentation transcript:
1 OECD 2nd International Cat Risks Conference Bangkok, Sept 25th 2009 'How installing and running a high quality hazard instrumentation network can facilitate the creation of parametric triggers' Robert Muir-WoodOECD 2nd International Cat Risks ConferenceBangkok, Sept 25th 2009
2 Agenda 1st and 2nd generation parametric risk transfer structures Instrumentation requirementsRole of governments/multilateral agencies to enable parametric risk transfer
3 Investor/Issuer Mismatch - Developed World IssuersInvestorsPrefer indemnityDislike basis riskSettling for industry loss or modeled lossPrefer parametricDislike indemnityBias for simplicity“When they were designing the Cat bond market, they forgot to invite the investors”
4 Investor/Issuer Mismatch - Developing World IssuersInvestorsNo insurance institutions to process claimsNo reliable historic claimsProportional recovery better than nothingOnly parametric is tenableFor the developing world – it has to be parametric
5 First & Second Generation Parametric 1st generation parametric is based on the parameters of the eventie. earthquake magnitude and location (in a geographic ‘box’)Hurricane intensity and track (and distance from the site)2nd generation parametric is based on measurements of the hazard at multiple locationsEg. peak ground acceleration measurements or peak gust windspeed recordingsWeighted by the exposure values and vulnerability in the vicinity of each recorder, a formula is developed and tuned to create an index that closely matches lossParametric structures require pre-existing Cat model to design correlation with expected loss and risk analysis for the BondHigher basis risk (mismatch with actual losses) likely for 1st generation than second generation structures
6 Caribbean Catastrophe Risk Insurance Facility – 1st Generation parametric trigger Launched Feb 2007 to provide participating governments with immediate liquidity if hit by a hurricane or earthquake. For hurricane - trigger based on ‘calculated’ windspeeds at key exposure concentrationsDerived using predefined Rmax of a hurricaneHowever real radius can vary by factor of 10For earthquake based on magnitude in a box not ground motion on an islandFor a small island – there is the potential for no payout when significant loss as well as payout when there is no loss2nd generation parametric not used in the Caribbean (in 2007) because of lack of suitable pre-existing windspeed and ground motion instrumentation networks.
7 Caribbean Insurance Pool Not Affected by Hurricane Dean INSURANCE JOURNALBy David McFadden August 24, 2007Hurricane Dean will not trigger an insurance pool set up this year by Caribbean countries and the World Bank that provides emergency cash to islands in case of natural disaster, officials said this week.The hurricane, which killed at least 20 people across the region and did extensive damage to bananas and other crops, failed to surpass wind speeds and other thresholds to prompt payments from the disaster pool established in February, according to fund supervisor Simon Young.Jamaica, which sustained the brunt of Dean's destructive path in the Caribbean, came close to being hit hard enough to receive payouts from the Caribbean Catastrophe Risk Insurance Facility."Had the storm been 30 miles to the north that would have triggered immediate payment in Jamaica,'' Young said from the Washington office of Caribbean Risk Managers Ltd., which oversees the insurance program.
8 Requirements for hazard instrumentation networks System must have multiple stationsTo ensure regional coverage (average station spacing 20-30km?)To provide redundancy if a station failsInstruments must be robust and have their own power suppliesDemonstrably able to withstand impacts and/or loss of power from a catastrophic eventMaintenanceMust be maintained by an agency that can demonstrate consistent standards of battery replacement and regular testingIndependence/IntegrityMust be run by an agency seen to be completely independent of potential interference from parties to a financial transaction.(Can be a problem for some government organisations.)
9 2nd generation Parametric Windstorm Index for Europe (based on c 300 windspeed measurements) Lower Basis Risk:Industry loss estimates can be customised by Line of Business and Province/CityIncreased Granularity:RMS Industry Exposure Database (IED) is at postal code resolution, enhancing the accuracy of industry loss estimates.Better Execution:Increased transparency, simplicity and liquidity for investors translates into better execution.
10 Eurowind Index Look-up Table The relationship between insured industry losses and peak-gust wind speeds are probabilistically estimated by application of the Europe Windstorm Model to the Industry Exposure DatabaseThe Index Look-up Table relates an estimated industry loss per 2 figure postcode to a given wind speed in that postcode for each line of business; in effect representing a set of weighted average vulnerability curves.
11 Eurowind Index - Historic Performance DateWindstorm NameRegion ImpactedParadex Europe Windstorm 8.0 Index ValueAnnual Occurrence Exceedance Probability of Index Value25th Jan 1990DariaSouthern England, Netherlands, Belgium and Germany€12,208bn1.82%15th Oct 19871987JSoutheast England and southeast Norway€5,180bn6.37%25th Dec 1999LotharFrance, Germany, Switzerland€6,732bn4.52%25th Feb 1990VivianNorthern Scotland, Scandinavia, Belgium€5,164bn6.40%18th Jan 2007KyrillNorthern Europe€3,298bn10.86%Daria was the most severe of a series of events that moved into Europe in early The storm caused a trail of destruction across southern England (where the highest peak gust wind speeds approached 44 m/s, or 100 mph), the Netherlands, Belgium, and Germany and caused significant insured losses in Denmark, France and LuxembourgThe index value of this event equates to an annual attachment probability around a “1 in 50 year” eventThere have been no examples in recent years of a “1 in 100 year” or “1 in 250 year” event in the European Region
12 But existing stations fail during hurricanes Remember to mention accounting issuesHurricane Charley, 2004Notice the period 8/14 to midday 8/15 – station was inactivated, either by power outage or a planned shutdownVery common problem – of the 44 stations in the Charley footprint, only 4 appeared to record for the duration of the storm.The lack of continuous, high frequency, moving average wind recordings is also an inhibitor to a USHU parametric indexExisting stations are often deactivated during hurricanesThere is “survival bias” – stations that survive are the ones that experienced weaker winds
13 Weatherflow Hurricane Network Ten-meter concrete polesOverall station rated to 140mphAnemometers rated to 220mphCellular link with 1Gb flash backupSolar powered battery backupSecure data transfer and storageTalk about Weatherflow and their involvement in the site choice / station design and construction plans
14 Designing a privately financed windspeed recording network for Florida
16 Second Generation Paradex earthquake Advanced National Seismic System& partner networksShakeMapzip centroid SA valueszipSA0.3sec, SA1.0sec, SA3.0sec(interpolate)Paradex-EQ Event BulletinzipInitial set-up / updateIndustry Loss ELT by zip & LoB?ResC&IResidentialComml & Indf( 0.3sec, 1.0sec, 3.0sec )f( 0.3sec, 1.0sec, 3.0sec )parametricindustry loss index valuesindex valueindex valuezonepre-compiled index value lookup tables(zip < zone < state)
17 New instrumentationHow did they know I dropped my phone in the bathtub?How can we know that the flood was 12 feet deep?Discolored activated indicators provide depth readingCapTechnology in use for hydrological applications (and cellphones)Require routine servicingLow cost (<$500/unit)
18 Low tech solutions have also been employed for 2nd generation parametric triggers Flood heights can leave a mark on buildingsAt pre-surveyed locationsHeights measured within a few days after ‘event’Audited by a highly reputable survey organisation
19 What can Governments and MLF Agencies do to foster parametric risk transfer? 2nd generation parametric triggers provide viable low basis risk alternatives to insurance - in particular for developing countriesRequires installation of networks of hazard recorders around concentrations of exposureHazard recorders need to be highly robust and able to record without external powerSupported by an agency that can ensure the maintenance of the equipment and guarantee the objectivity of the observationsAll organisations/institutions linked with the risk transfer mechanism need to have demonstrable long term integrityUnderpinned by models to structure and define index risk analysisHowever smart instrumentation can promote risk transfer !