Atelier A1 Changement des fréquences en évènements climatiques et impact sur la réassurance 2 février 9h-10h30 Atelier A1 Intervenants Modérateur Alix ROUMAGNAC Président Robert MUIR Chief Research Officer Ernst RAUCH Head Corporate Climate Centre Michel JOSSET Responsable assurances et immobilier Œuvrer pour la diminution des conséquences des événements climatiques Alix Roumagnac : Predict services L’impact des catastrophes : une évolution marquée Source Swiss-re Les inondations, un phénomène planétaire Les risques Naturels : mieux les comprendre pour mieux les gérer Siège : Montpellier 28 ingénieurs spécialistes de la gestion des risques en astreinte 24h24 et 7jours/7 Spécialiste en assistance à la gestion des risques hydrométéorologiques Les Clefs de la gestion du risque SE PREPARER, organiser la réponse opérationnelle Analyser la vulnérabilité Concevoir une organisation opérationnelle de gestion des risques ANALYSER, transmettre une expertise Température Mer Nov 2011 Impacts de foudre ,donnée radar pluie Masses nuageuses - donnée satellite nov 2011 Hauteur/débit via les stations de mesure Intensité des précipitations au km² chaque 5 minutes Transmettre l’information pour la prise de décision et l’action 1/ Information et aide à la décision depuis le vigie PREDICT services 2/ Prise de décision 3/ Action @ EN 2017 Prés de 30 000 Communes Françaises Font face aux risques avec les Services de Predict et de Météo France Le service en 2016 -Une vigie activée en H24 pendant 111 jours pour le suivi de phénomènes à risque avec un record de 24 jours consécutifs pour les inondations du bassin Parisien en juin. -La quasi-totalité des communes ont bénéficié du service en 2016 1988 messages ont été envoyés pour conseiller les communes à activer leur plan communal de sauvegarde . -Des Retours d’expérience systématiquement réalisés, plus de 95% de satisfaction. fréquence gravité Présentation générale du schéma d’alerte Prédict Services •Le footprint industriel de Faurecia est surveillé constament par les spécialistes de Predict Services •Les risques suivis sont : inondation par débordement de rivière, pluies torrentielles, sumersion côtière, tornades, chutes de neige •Une alerte est également générée en cas de séisme important • Les alertes sont générées manuellement par Predict Services quand un évènement menace un ou plusieurs sites de Faurecia •L’alerte est graduée en fonction de l’intensité de la menace et de son développement Présentation générale du schéma d’alerte Prédict Services •Les alertes sont communiquées par mail et SMS pour chaque site concerné :aux personnes suivantes : Directeur d’usine, HSE site, Resp. maintenance, Directeur des Opérations, Resp HSE de la division •La personnée alertée est invitée à se connecter sur la plate forme de Predict services pour consulter des informations sur l’évènement en cours et les recommandations associées •Les superviseurs (assurance Groupe, Assureurs, HSE Groupe) sont informés à partir du stade “Take action” •L’historique des évènements est disponible •Le service est opérationnel depuis juillet 2016 Présentation générale du schéma d’alerte Prédict Services 17 Page d’acueil plateforme Predict Services Alerte de niveau max en cours Sites Faurecia Cyclone Historique des alertes Pluies Message par région 18 Recommendations en cas d’alerte Recommandations en cas d’inondation, séisme, tempêtes… 19 Rapports d’évènements extêmes Rapport d’évènement •Le footprint sur lequel travaille Predict Services footprint et la liste des contacts sera raffarîchie quotidiennement à partir de la base de données centrale risques et immobillier de Faurecia 20 Back office et mise à jour du footprint RIMS real estate database Predict services footprint Daily web service : footprint + contacts Very Severe Cyclonic Storm VARDAH in INDIA December, 6 th 2016 No risk 03h30 UTC (09h00 loc) Position of the storm’s eye Very Severe Cyclonic Storm VARDAH in INDIA December, 09 th 2016 BE AWARE 19h45 UTC (01h15 loc) BE AWARE: Marginal effects of TS VARDAH may be observed in the next 48h. Keep informed for strong winds and stormy rains. Forecast track by the time Very Severe Cyclonic Storm VARDAH in INDIA December, 10 th 2016 BE PREPARED 21h15 UTC (02h45 loc) BE PREPARED: Cyclone VARDAH approaching your area. Preventive measures advised for strong winds and stormy rain. BE AWARE: Marginal effects of Cyclone VARDAH may be observed in the next 48h. Keep informed for strong winds and stormy rains. Very Severe Cyclonic Storm VARDAH in INDIA December, 11 th 2016 TAKE ACTION 13h30 UTC (19h00 loc) TAKE ACTION: Tropical storm VARDAH expected to cross your area. Protective mobilization required against damaging winds and stormy rain. BE PREPARED: Marginal effects of Cyclone VARDAH. Preventive measures advised for strong winds and stormy rain. Very Severe Cyclonic Storm VARDAH in INDIA December, 11 th 2016 TAKE ACTION 13h30 UTC (19h00 loc) Very Severe Cyclonic Storm VARDAH in INDIA December, 12 th 2016 REINFORCE YOUR ACTIONS 08h15 UTC (13h45 loc) REINFORCE YOUR ACTIONS: Extreme weather affecting your area due to VARDAH. Severe risk of gusts and floods ongoing. Reinforce your crisis management. Very Severe Cyclonic Storm VARDAH in INDIA December, 12 th 2016 REINFORCE YOUR ACTIONS 08h15 UTC (13h45 loc) Natural Catastrophes: Trends and Impact on Insurance Ernst Rauch, Munich Re (CorporateClimateCentre@munichre.com ) Role / business model of the insurance industry Risk identification Risk measurement “price tag” Risk transfer Know-How / data transfer Risk consulting Capital substituent  Investment support (Re-)Insurance Clients / Industry / Politics / NGOs © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 30 The five costliest natural catastrophes of the year 2016 – Overall and insured losses 2016: mind the gap… 2016: US$ 175 bn economic losses; only US$ 50 bn were insured Meteorological events (Storm: tropical, extratropical, convective, local) Hydrological events (Flood, mass movement) Geophysical events (Earthquake, tsunami, volcanic activity) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research, NatCatSERVICE – As at January 2017 31 … and the future of insurability of natural catastrophes? 32 Number of loss events Relevant loss events in Europe 1980-2015 Accounted events have caused at least one fatality and/or produced normalized losses ≥ US$ 100k, 300k, 1m, or 3m (depending on the assigned World Bank income group of the affected country). Meteorological events (tropical storm, extratropical storm, convective storm, local storm) Hydrological events (Flood, mass movement) Climatological events (Extreme temperature, drought, forest fire) Geophysical events (Earthquake, tsunami, volcanic activity) Overall losses (in 2015 values) Insured losses (in 2015 values) Overall and insured losses (US$ bn) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research, NatCatSERVICE – As at March 2016 Inflation adjusted via country-specific consumer price index and consideration of exchange rate fluctuations between local currency and US$. 33 Relevant loss events worldwide 1980-2015 Number of loss events Overall and insured losses (US$ bn) Accounted events have caused at least one fatality and/or produced normalized losses ≥ US$ 100k, 300k, 1m, or 3m (depending on the assigned World Bank income group of the affected country). Meteorological events (tropical storm, extratropical storm, convective storm, local storm) Hydrological events (Flood, mass movement) Climatological events (Extreme temperature, drought, forest fire) Geophysical events (Earthquake, tsunami, volcanic activity) Overall losses (in 2015 values) Insured losses (in 2015 values) Inflation adjusted via country-specific consumer price index and consideration of exchange rate fluctuations between local currency and US$. © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research, NatCatSERVICE – As at March 2016 34 Risk information for single risks / locations Munich Re NATHAN Risk Suite (www.munichre.com) Natural hazard exposure analysis tool Combining client risk data with Munich Re natural hazard zoning system Creating risk transparency / information New York, USA © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 35 Risk information for risk portfolios Munich Re NATHAN Risk Suite (www.munichre.com) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 36 Digitalization: resolution of hazard / vulnerability data Climate Change: risk of change (changes of frequency and / or intensity of extreme weather events) Main drivers changing natural hazard risk assessment © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 37 ZÜRS – zoning system for flood (Germany) From global / regional hazard zoning schemes to high resolution hazard information Source: GDV, ZÜRS-Geo 2013 GK 1 GK 4 GK 2 GK 3 River Flood risk: address distribution to risk category / hazard classes (GK) GK1 GK2 GK3 GK4 GK 1, extremely low: flooding less than once in 200 years GK 2, low: once in 50-200 years (dyke protected area) GK 3, medium: at least once in 10-50 years GK 4, high: at least once in 10 years (statistical mean) 38 Increasing number of relevant flood events worldwide 1980– 2015 Accounted events have caused at least one fatality and/or produced normalized losses ≥ US$ 100k, 300k, 1m, or 3m (depending on the assigned World Bank income group of the affected country). Hydrological events (General flood/river flood, flash flood, landslide, storm surge, subsidence) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research, NatCatSERVICE – As at March 2016 39 Globally, there is currently (as per 2016) (re-)insurance capacity available to cover also highly exposed risks Regionally / locally insurance capacity is tight in regions with significant adverse loss experience and / or loss trends Impact on insurance situation (1) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 40 Potential deterioration of losses from natural catastrophes will likely impact supply of risk transfer solutions Proper selection of risk locations and vulnerability reduction measures (including supply chain management) contribute to long-term stability of insurability Impact on insurance situation (2) © 2016 Münchener Rückversicherungs-Gesellschaft, Geo Risks Research 41 ©2013 Risk Management Solutions, Inc. Confidential OUR HISTORY OF INNOVATION Create a universe of events that characterize long- term probabilities Framework for Catastrophe Risk Modeling Define Events Assess Ground Motion Calculate Damage Quantify Risk Calculate the hazard parameter for all sites due to each stochastic event Calculate the average damage and associated uncertainty for the specific exposure. Calculate the financial impact for all perspectives Precipitation •Ant. Conditions •Spatial correlation •Pluvial risk Rainfall runoff Major river Minor river Riverine defenses Drainage systems Inundation Time: 50,000 yr, 1 day resolution, ~850,000 events Space: 13 countries, 40 m resolution, >1 bn cells Stochastic and hazard module Seasonality & clustering of events  MORE REALISTIC TAIL LOSS Events have daily time-stamp  APPLICATION OF HOURS CLAUSE Copyright © 2015 Risk Management Solutions, Inc. . •Bespoke flood defenses •Third party data •Defenses fail stochastically FLOOD DEFENSES Stochastic SoP Design SoP cdf •Hazard values used to set internal underwriting guidelines for screening purposes •Postcode Annualised Loss Ratio’s (ALR’s) = Expected loss per $ insured for building/contents. •Combining ALR’s with building/contents values produces a technical rate. Hazard component: Loss component: Postcode River Catchment 50 year 100 year 250 year 12345 Basin X 0.2 0.8 1.2 12346 Basin X 0.6 1.5 2.1 Postcode Hazard – Flood Depth (m) Postcode Unknown Residential Commercial 12345 0.000724 0.000842 0.000799 12346 0.000470 0.000510 0.000524 Postcode Contents ALR DeclineRefer Addtnl Conditions Write Building Value: €250,000 Contents Value: €15,000 Residential Postcode: 12345 Technical rate = (Building value * Building ALR) + (Contents value * Contents ALR) = €174.86 PRDs, together with the return period hazard maps enable clients to understand risk at location level and define risk adequate premium by relating severity of loss to level of financial loss. PRD Differentiation: Coverage type – buildings/contents Line of business – residential/commercial Peril rating database across whole of Europe SPATIAL CORRELATION OF INLAND FLOOD Dusseldorf Low High Spatial correlation amplifies tail losses Germany and Czech Republic  200yr loss +20% when accounting for correlation London Low High 48 48 Copyright © 2016 Risk Management Solutions, Inc. All Rights Reserved. February 6, 2017 Is climate change altering the occurrence of extremes? Attribution studies 20,000 years of climate simulation Without post 1950 increase in GHGs With post-1950 increase in GHGs The fractional attribution rating scorecard River Flood Cold Wave Wildfire ETC Storm Intense Rainfall Severe Convective Storm Tropical Cyclone Heatwave The fractional attribution rating scorecard River Flood Cold Wave Wildfire ETC Storm Intense Rainfall Severe Convective Storm Tropical Cyclone Heatwave Intensity? Numbers? May/June 2016 Flood Event MAY/JUNE 2016 FRENCH FLOOD CLIMATE CHANGE ATTRIBUTION? ‘For the Seine River Basin an event like this is now expected to occur roughly 80 percent more often due to climate change than it was in the past’ ‘For the Loire River Basin 3-day precipitation extremes in April-June, the team found that an event like this is now expected roughly 90 percent more often due to climate change than it was in the past. ‘In both cases, the increases are at least 40 percent.’ https://wwa.climatecentral.org/analyses/european-rainstorms-may-2016/ 54 54 Copyright © 2016 Risk Management Solutions, Inc. All Rights Reserved. February 6, 2017 Catastrophe models and peril rating databases for Corporate risk management ©2013 Risk Management Solutions, Inc. Confidential •Tools widely used by insurers and reinsurers are now being applied by corporates themselves for how they manage their risks •Across all significant perils (natural and man-made) •Identification of key risk hot spots – as well as actions that can be taken to reduce the risk •Key supply chain risks both for internal and external suppliers •Identify bottlenecks, critical suppliers, critical ports, and model a wide range of potential interrupts •Explore greater redundancy and alternative suppliers (ensuring they cannot be hit by the same catastrophes) RISK MANAGEMENT TOOLS FOR CORPORATES ©2013 Risk Management Solutions, Inc. Confidential DRIVING BETTER FLOOD RISK MANAGEMENT •1) Relocate all ‘mission critical’ equipment above the floodable space •2) Modify heights of defences to withstand a 500 year flood Measure benefits in reduced long term risk costs. EP 1% Combined: Building, Contents & BI 1) 2) 0.2% $ loss EACH ELEMENT OF SUPPLY CAN BE TRACKED USING THE ‘SUPPLIER’S DATA SCHEMA’ •High Res location of supplier •Buildings, Ages, Construction, etc •Lifelines of the supplier •Details (Code) for what is being supplied •Flow of supplies (per day/week) •Route of supply •Key ports/airports •Destination INDIVIDUAL LINKS IN THE SUPPLY CHAIN Tier 1 Tier 2 Tier 3 Tier 4 THE SUPPLY CHAIN RISK MANAGEMENT SYSTEM IS USED TO EXPLORE REDUNDANCY AND MANAGE SHOCKS What if a key supply is interrupted for a week? What if a key supply is interrupted for three months? Flows are viewed passing through the supply chain Onsite Inventory Offsite Warehousing Optional Supplier & ramp-up time Will also require details of the location, building type etc of the warehousing. Current Supplier PROBABILISTIC AND SCENARIO MODELING CAPABILITIES Tier 1 Tier 2 Tier 3 Tier 4 X X X X X Suppliers knocked out by a catastrophe X INTEGRATED USE OF RMS EXPOSURE MANAGER™ •Capability to explore all potential accumulation concentrations •Full insurance and reinsurance structure capability •Complete flexibility about potential event footprints •Peril-independent correlation modelling •Potential to apply full range of potential non-modelled loss footprints BLENDING ACCUMULATION MANAGEMENT WITH RESILIENT CATASTROPHE MODELLING BEST PRACTICE IN CORPORATE RISK MANAGEMENT Shape File Loading Simple Model Building 20 % 10 % 5% 7% 4% 1% 25 % 6% 13% Property Logic Merci Les slides seront en ligne dès la semaine prochaine sur www.amrae.fr