A paradigm shift in natural catastrophe risk management: Why insurance isn t the (only) solution 2017 Global Risk Engineering Conference

Transcription

1 A paradigm shift in natural catastrophe risk management: Why insurance isn t the (only) solution 2017 Global Risk Engineering Conference 18 th May 2017 Amar Rahman, Principal Risk Engineer Natural Hazards Risk Engineering Commercial Insurance

2 Agenda 1. Definition of risk 2. The integrated approach to resilience: Account-level and site-level views 3. Account-level: Catastrophe modeling 4. Site-Level: Assessment of natural hazard risk resilience 5. Summary and closing statements Proposition: - Discuss gaps in risk management approach - To increase group-level natural hazards resilience through optimization of insurance limits AND risk mitigation of critical locations, i.e. implement and utilize two levels of risk management tools: Insurance (post-event, ex post facto) Risk mitigation solutions (pre-event, ex-ante) 2

3 Overview of risk assessment process Understand value chain Identify accumulations Identify high risk regions & individual locations Develop regional/group level loss scenarios Group level analysis Regional/single location analysis Assess quality of infrastructure Identify and evaluate public defense systems Review site-level resilience (organizational and physical) Prioritization based on risk appetite and previous steps Define resilience measures Change in exposures (total insured values) Updated cat models Revisions in structural design codes Change in risk strategy 3

4 Gaps in Risk Management Approach General observations: Group-level analysis, i.e. catastrophe modelling: Poor quality of portfolio data (location, occupancy, construction class) Insurance structure (limits, premiums, etc.) doesn t reflect the risk High-risk sites not identified Value chain is unknown (suppliers) and/or no redundancies in infrastructure Site-level resilience (reduction of material damage & business interruption) rarely considered as a risk mitigation option (usual solution: high risk sites excluded from policy) New projects/acquisitions: Lack of nathaz resilience considerations at due diligence or (conceptual) engineering design stages High costs for subsequent implementation of resilience measures Perception that insurance is the only natural hazard solution option engineering measures largely ignored ( will be covered by the building code ) 4

5 A definition of risk Risk is a function of exposure, event, and controls Event severity (loss value) is not only a function of hazard level, but of all three factors High level of uncertainty in each of the factors related to risk Protection mechanisms are perildependent and far more complex to assess compared to, e.g. fire The factors related to nathaz risks are intangible in contrast to operational risks (fire) Event (Hazard & Trigger) Controls (Vulnerabilities) RISK Exposure (Values at risk OR process that is impacted) Construction risks are associated with a dynamic (time-dependent) component Natural Hazards 5

6 To summarize: High risk is not only a function of high hazard level!! Example: Snow damage (Syria) December 2010 USD 3.4m in damages triggered by 1-day snowfall Event is usually a trigger for an inherent flaw (material and/or design)! Poor quality structure Vulnerability (Controls) Low probability event Hazard RISK Exposure High value, sensitive storage HIGH RISK! Photos used with permission 6

7 Risk definition: Portfolio perspective Output: Define policy structure Identify high exposures Controls (Vulnerabilities) How effective are the protection mechanisms? Primary / secondary characteristics (occupancy, age, construction class/no. of stories) RISK Where is the site? Event (Hazard & Trigger) Exposure (Values at Risk) What are site values/processes? Address Occupancy / Insured values 7

8 Risk definition: Site-level perspective Output: Improve data quality for cat modelling Identify & improve resilience at critical sites Controls (Vulnerabilities) How effective are the protection mechanisms? Compliance with engineering state-of-the-art & best practice RISK Where is the site? Event (Hazard & Trigger) Exposure (Values at Risk) What are site values/processes? Geocode and site layout Site processes, values and value chain 8

9 damage ratio insured value Policy 1 Policy 2 Basics of catastrophe modelling Account Data Stochastic Module Hazard Module Vulnerability Module Financial Module Over Limit Loss source: Gross P2 P2 Co-insurance Importance of data quality cannot P2 be Underlying Coverage 2 Gross P1 Co-insurance P1 overemphasized!!! Underlying Coverage 1 source: Deductible hazard intensity participation Define events Define local hazard parameters Calculate damage Quantify loss Based on historical data Large catalogue of simulated events Local intensity of hazard parameter for each cat event based on geological & topographical features Converts hazard parameter to loss Applies deductibles and limits Calculate losses for different financial perspectives 9

10 Ground-Up Loss (m USD) Catastrophe modeling Output: Probability of loss triggered by a specific event, AAL and OEP, etc. No Coverage Exhaustion RP = 700 Y Policy 200 m USD Activation RP = 125 Captive 200 m USD Account Level Loss Potential (for specific region & peril, e.g. California EQ) Insurance Program Return Period (Year) Check adequacy of insurance limits/coverage, etc. Natural Hazards 10

11 Overview of the resilience approach Step 1: Global (portfolio) analysis Design policy & plan site resilience measures Medium probability event Basements? Construction material? Value distributions? Vulnerability (Controls) Event (hazard/ trigger) RISK Exposure Large number of locations within same watershed Using cat modelling, global hazard identification tools, visual evaluation, etc. identify: 1. Single high RISK locations. 2. Regions with accumulations HIGH RISK! Source: Swiss Re CatNet,

12 Overview of the resilience approach Step 2: Site-level risk assessment Improve resilience Source: ZRE, 2014 Evaluate protection mechanisms (physical & organizational) Vulnerability RISK Use local resources (e.g. structural design codes) and visual evaluation Event (hazard/ trigger) Exposure Assess distribution of values / critical equipment / processes 12

13 Flood hazard is not uniform across the site Hazard Identification: Consider geocodes of individual buildings AND site processes Exposure Address Used with permission Source: Zonierungssystem für Überschwemmungsrisiko und Eisnchätzung von Umweltrisiken (ZÜRS) ( Highest values in assembly hall directly adjacent to the stream Stream level rose from a few centimeters to 1.65 m within 3 hours Flood hazard AND exposure should be correctly identified Site processes understood Controls RISK Event Exposures 13

14 Most critical building has the highest flood hazard Critical: Building with highest contribution to value chain 1 Exposure 1 2 Address 2 2 Used with permission 14

15 15 HAZARD IDENTIFICATION GLOBAL OR LOCAL MAPS?

16 However, there is always uncertainty in hazard maps For example, EQ following events occurred in areas where seismic hazard was designed low according to local maps : 2008 Wenchuan 2010 Haiti 2011 Tohoku 2011 Christchurch 2012 Emilio Romano 2016 Kaikoura (NZ) Situation is even more dramatic for flood (and to some extent, wind), as hazard level is influenced by human activity WHY? Models are only models (cannot reflect the complicated nature of earthquake occurrence) Historical records are too short Hazard maps are not kept up-to-date (long time to collect data, resource intensive, etc.) 16

17 Changes in hazard maps driven by lessons learned from actual events Source: Cubrinovski, M., & McCahon, I. (2011), Foundations on Deep Alluvial Soils, Canterbury Earthquakes Royal Commission Hearings, LINK Liquefaction Source: Macdonald and Mote, 2011 (used with permission) Natural Hazards 17

18 Assessment of protection mechanism quality With respect to physical protection mechanisms, three levels of protection (adapted from UK Environment Agency) Avoidance: Build out of/above the zone Resistance: Building design (strength, material selection, architectural layout, etc.) such that structure not affected by the hazard. Resilience/Repair: Site is impacted but quick recovery due to high resilience, e.g. easy repair, effective emergency response and contingency plans. Greenfield (New Sites) Existing Sites Resistance Resilience All photos: ZRE, with permission Avoidance / Resistance 18

19 Natural Hazards Protection Mechanisms Classification of natural hazards protection mechanisms: Perils Structural (Force-Resisting) Non-Structural (no contribution to force resistance but impact on damage & disruption) EQ Wind Columns, walls, beams, foundations, floor slabs Flood Construction material Drainage systems (building roof, site, road, 3rd party) Topography (surrounding site, elevation of buildings, distribution of contents in the building) Flood protection (mobile, fixed, organizational, architectural) Partitions, piping, equipment, etc. Yard storage, 3 rd party facilities, cladding, windows, doors Type of occupancy Emergency response / business continuity plans Hazard monitoring and response plan Maintenance plan Infrastructure Utilities, roads, bridges Non-structural measures include phyiscal & organizational Infrastructure & public defenses (flood/eq) outside responsibilities of site management difficult to assess Natural Hazards 19

20 There will always be a residual risk Damage: Failure in shear wall Secondary damage (liquefaction) BUT, performed according to design objectives: Structural codes define minimum requirements Codes (especially older generation ones) focus on life safety Example: Grand Chancellor Hotel, 2011 Christchurch Earthquake Source: Macdonald and Mote, 2011 (with permission) Natural Hazards 20

21 Increasing natural hazard resilience Identification of production-critical site processes Below-ground control rooms and critical equipment Source: ZRE (Used with permission) 21

22 Increasing natural hazard resilience Some solutions Wind: Increasing anchorage at building corners EQ: Seismic design of nonstructural elements, e.g. storage racks Snow: Planning for safe snow removal before and during high snow events. Photos: ZRE (with permission) Photos (with permission) M. Bruneau & GC Clifton, (Source: Bruneau, M., Clifton, C., MacRae, G. Leon, R., & Fussell, A. (2011), Steel Building Damage from the Chch EQ of Feb 22, LINK to SOURCE 22

23 Reliability of protection system: Maintenance Issues Good engineering but poor maintenance: Overgrown with vegetation Trees are too close (<10m) to levee Roads over the levee Good maintenance Photos: ZRE, (with permission) 23

24 Reliability of public protection systems: Example Tokyo gas supply network Public utilities in Japan are well-prepared. For example, SUPREME gas supply protection system for Tokyo. Real-time earthquake disaster prevention system. Detects shaking levels that could potentially damage piping and structures and interrupts supply. Seismometers installed at natural gas pump stations (1 sensor per 1 km 2 approx) During Great East Japan (Tohoku) EQ data collection in 5 mins and response. By around 10 minutes after the start of the earthquake, district pressure regulators are remotely operated to halt the supply of gas to areas where large-scale damage is expected. ( State of shaking at the Tokyo metropolitan area detected by seismometer (SI sensors) (including some information by a part of SI sensors of the peripheral gas company) at the Great East Japan EQ. [Source: Tokyo Gas, Used with permission Copyright Tokyo Gas Co. 24

25 Effectiveness of public flood protection systems Impact of Climate Change Hazard Changes in intensity and frequency of extreme events impact of climate change on accuracy of hazard maps Not only climate change but manmade factors Protection mechanisms Reduced design return period (lower effectiveness against low probability events). Changing design boundary conditions, e.g. subsidence, urbanization Aging of infrastructure/poor maintenance/organizational changes Exposure Higher impact due to increased urbanization and development (increased insured values) Huangpu River Flood Defenses: Original design year return period based on data from 1912 to New analysis (data from 1912 to 2002) reveals that original data corresponds to 200 year return period 1. Suzhou River Flood Defense, Shanghai Hanghui, Z., et al. (2016), Model Study on Potential Contributions of the Proposed Huangpu Gate to Flood Control in Taihu Lake Basin, Hyd. Earth Sys. Sci. Disc. Photos: A Rahman, ZRE,

26 Reliability of public flood protection systems Impact of climate change on protection mechanism effectiveness Increase in peak discharges, especially of frequent smaller floods (river). For 100 year return period, water volume will increase by 15% resulting in damage increase by factor of 2 due to 0.60 m of water depth (residential occupancies, existing measures not considered). 100 year defence will be adequate for 20 year return period event. Consider 300 year return period flood for future 100 year event. Event (Hazard/ Trigger) Controls RISK Exposure 26

27 Integrated Approach to Natural Hazard Risk Management Summary Evaluation of nathaz RISK to include not only hazard level, but also quality of protection mechanisms and values at risk/importance to value chain Two-step approach to natural hazards risk management; Step 1: Group level analysis output; Check adequacy of insurance coverage, identify sites with high RISK Step 2: Site level assessments: Increase nathaz resilience (can be done with judicious allocation of existing resources, but requires planning). Impact of natural hazards not only on single locations: Damage to infrastructure (site access, power, gas, etc.), Business interruption can result from damage to the contents, e.g. machinery and equipment, which occurs at much lower hazard levels Data quality has a major impact on insurance policy conditions and risk map. 27

28 Thank you 28

29 The information contained in this document is intended as a general description of certain types of services and insurance covers available to qualified customers. Zurich Insurance Company Ltd or any of its subsidiaries and its employees do not assume any liability of any kind whatsoever, resulting from the use, or reliance upon any of the information contained herein. It does not replace or complement your individual insurance policy, which is the only source for terms and conditions of your respective insurance cover. This is intended as a general description of certain types of services and insurance covers available to qualified customers through subsidiaries within the Zurich Insurance Group Ltd. including, in the United States, Zurich American Insurance Company, 1299 Zurich Way, Schaumburg, Illinois 60196, and, in Canada, Zurich Insurance Company Ltd, 100 King Street West, Toronto ON M5X 1C9, and, outside the US and Canada, Zurich Insurance Plc, Ballsbridge Park, Dublin 4, Ireland (and its EU branches), Zurich Insurance Company Ltd, Mythenquai 2, 8002 Zurich, Zurich Australian Insurance Limited, 5 Blue St., North Sydney, NSW 2060 and further entities, as required by local jurisdiction. 29