AGENDA RISK MANAGEMENT CONSIDERATIONS REINSURANCE IMPLICATIONS CATASTROPHE MODELING OVERVIEW GUY CARPENTER

Transcription

1 AGENDA! CATASTROPHE MODELING OVERVIEW RISK MANAGEMENT CONSIDERATIONS REINSURANCE IMPLICATIONS

2 CATASTROPHE MODELING OVERVIEW 2

3 What is Catastrophe or Cat Modeling? 3

4 What is Catastrophe or Cat Modeling? What? Cat Models are tools that quantify risk How? They examine insured values that are exposed to catastrophic perils such as hurricanes, earthquakes and terrorism and estimate losses Why? Models aid management decision-making on Pricing and underwriting Reinsurance buying Rating Agency/ Regulatory requirements Portfolio management 4

5 Catastrophe Models Help Quantify Plausible But Unobserved Events Significant Global Insured Losses: Though widely used to manage risk, catastrophe models should be viewed with a healthy dose of skepticism $49.7b $30.5b 5

6 What Perils Are of Interest? Natural Man-Made Non-Modeled Hurricane & Storm Surge Earthquake Fire Following Earthquake Sprinkler Leakage Severe Convective Storm Tornado Hail Wind Wildfire Inland Flood Infectious Disease Terrorism Cyber Elements of flood Hurricane induced precipitation Tropical Storm Tsunami Non-California Wildfire Many of these are in development 6

7 What Lines of Business Do Catastrophe Models Cover? Property Personal Farm Commercial / Industrial Builders Risk Lines of Business Covered Marine Yacht Warehousing Docks Utilities Auto Physical Damage Workers compensation Life / Accident and Health Time Element (Business Interruption / Loss of Use) 7

8 Model Approaches Deterministic Versus Probabilistic Deterministic Modeling using a single discrete event The event is assumed to happen without regard to probability Commonly seen as recreations of historic events or single, hypothetical analysis Examples: 2008 Hurricane Ike 2011 Joplin Tornado Probabilistic Model Uses a series of simulated events and accounts for the probability of those events over time 1,000,000+ simulated events in some models Purpose smooth representation of hazard frequency and severity Not achievable through sparse historical events alone 8

9 What are the Advantages/Disadvantages of Catastrophe Models? Advantages Common baseline used in insurance/ reinsurance/rating discussions Minimizes reliance on historical data One additional event in a small database can produce broad fluctuations in results Historical results are not representative of future events in many areas Exposures change over time and onleveling losses is challenging Uses probabilistic distributions to properly address: Low frequency but high severity events (e.g. Hurricane) Geographical distributions of events (fill in the holes) Probabilistic distributions provide for more robust tail, compensating for lack of data Process large volumes of data and complex calculations quickly Disadvantages Model output heavily influenced by three critical areas Quality of the source data (availability, completeness, accuracy) Model methodology (difficult to quantify, different amongst vendors) Model application (added complexity in decision process) Can tempt over-reliance and false sense of precision Sometimes expensive and timeconsuming to develop and operate 9

10 Catastrophe Models Typical Results and Definitions Probability of Non-Exceedance Return Period (Years) OEP Estimated Loss ($000) 99.9% ,884 87, % ,014 61, % ,213 39, % ,731 17, % 50 7,643 8, % 20 1,332 1,554 Average Annual Loss ($000) 712 AEP Occurrence Exceeding Probability (OEP): Results indicate this portfolio has a 1% chance of a single occurrence causing a loss greater than or equal to $17.7 million Said another way, there is 99% confidence of not exceeding a loss of $17.7 million Aggregate Exceeding Probability (AEP): Estimates likelihood that one or more occurrences will combine in a year to exceed the modeled loss threshold Results indicate a 0.4% chance that this portfolio will see a single year with aggregate losses (from one or more events) that meet or exceed $39.7 million 10

11 Catastrophe Models Typical Results and Definitions Average Annual Loss (AAL): The long term average loss (or pure premium) expected in any one year for the book of business for the peril being modeled. Because AAL is an additive statistic, it can be used to subdivide the losses for a portfolio by categories of interest such as: Geography (state, county, etc.) Line of Business Region/Territory Peril Producing Agent Virginia County Average Annual Loss ($000) % AAL of Total Accomack % Virginia Beach City % Middlesex % Northumberland % Norfolk City % Isle of Wight % Suffolk City % Lancaster % Gloucester % Mathews % Subtotal of Top % Subtotal All Other % Grand Total % 11

12 What We Think We Know and Don t Know Sources of Uncertainty in Cat Models Data Inputs Variable construction type Replacement cost values by location Coverage conditions Exposure data completeness and accuracy Scientific and Engineering Interpretation Calibration Datasets HURDAT/Reanalysis Atlantic Multi-Decadal Oscillation (AMO) Damage functions Sparse engineering literature Financial/Statistical Modeling Correlation of portfolio risks and treatment in modeling Treatment of Secondary Uncertainty Applying Insurance Financial Terms Industry Exposure and Claims Data Assumptions Claims settlement practices often differ between insurers Detailed industry data carries uncertainty Code/Construction/Inspection and regional variability How much bias is introduced from each factor and how these, and others, ultimately interact together in the model vendor calibrations or in reality for an event is largely unknown 12

13 ! RISK MANAGEMENT CONSIDERATIONS 13

14 Common Uses of Cat Risk Management Portfolio Management Growth Optimization Allocation Accumulation identification Portfolio Management Example Claims Early warning IBNR detection Adjuster deployment Underwriting Rate development Individual risk Reinsurance Program Structure Cost benefit analysis / pricing Regulatory Regulatory (State / Federal) Rating Agency (AM Best, etc.) Reinsurance Structure Example $100M x $25M 14

15 Current Thinking on Model Usage Is Evolving Model Uncertainty Understood and accepted Leads to a focus on accumulation (not probabilistic modeling) for several perils such as tornado/hail and wildfire Many insurers adopt a management view of risk Commercially available models are just a starting point Challenging the status quo Model customization, blending, governance Affect of model change on reinsurance pricing is dwindling, which can trickle down into primary rates 15

16 Evaluating Models Not All Models Are Created Equal What Does Model Evaluation Mean? Independent approach to model testing and evaluation to synthesize the complex array of multiple vendors, versions, and perils Objective evaluation of cat models Evaluate against scientifically defensible, vendor-neutral datasets Establish confidence where warranted Identify points of caution/correction With model vendor cooperation Scientifically review model documentation Verify our testing methodology Discuss our results with model vendors Expert support from the academic sector Why Does It Matter? We don t need more model answers, but a comprehensive approach to model testing and evaluation Clients can make informed risk decisions that reflect their risk appetite and risk tolerance Property risk decisions often informed by catastrophe modeling Rating agencies rely in part on what a catastrophe model depicts as insurer s risk profile Enables clients to make more informed use of model results, with greater clarity and consistency to form their own view of risk 16

17 Industry Issues USGS Seismic Hazard Maps Update Includes new methods, models, and data obtained since last update in 2008 Modifications to seismic source and ground motion inputs indicate changes of plus or minus 20% over wide areas Generalizations difficult to make and local areas may see larger changes California Insights Discovery of new faults in San Jose, Vallejo, San Diego lead to increased hazard Hazards in parts of San Francisco Bay and Los Angeles increased, while decrease was seen in other parts of these areas. Researchers decreased threat estimates for Irvine, Santa Barbara and Oakland due to new research Source: USGS Open File Report 17

18 Industry Issues Flood Flood Locations First vendor inland flood model recently released 18

19 Industry Issues Wildfire Wildfire has become a serious issue for insurers for both California risks and non- California risks Currently Available Analyses Annual Burn Probability analysis to show key concentrations in high risk areas Probabilistic modeling for California with a countrywide wildfire model on the horizon Experience rating Deterministic scenarios Soberanes Fire 43,400 acres burned 57 structures destroyed Sand Fire 41,430 acres burned 20 structures destroyed 19

20 REINSURANCE IMPLICATIONS 20

21 Vendor Models Influence Management s View of Risk Separate each component of a catastrophe model Evaluate each against up-to-date scientific research and clients exposure and claims data Calibrate the models to reflect a customized view of risk, where warranted 21

22 Model Mechanics Other Factors Influencing Losses Demand Surge The short term inflation in the prices of labor and materials following a catastrophe Inclusion is now standard Loss Amplification For RMS model only Also includes: Claims inflation Super-cat Non-modeled Sources of Loss LAE Inland Flood Mold Fraud Civil Unrest Vandalism/Looting/Arson Pollution Cleanup Off Premises Power Mandatory Evacuations Coverage changes post event Trees Life-line disruptions Long-term health effects The size of the event, industry loss, line of business and geography are all factors that can impact losses. 22

23 Catastrophe Models Are Not Easily Validated Models difficult to validate ASOP 38 State certification / approval (some) User scrutiny License multiple models GC performs testing and reasonability checks Markets and governmental bodies Notable examples of when models have missed : Hurricane Katrina The levy failure was not modeled Hurricane Ike A category 2 hurricane which had a larger footprint and more energy than expected Tohoku Earthquake It was believed that the maximum earthquake magnitude on that fault was about 8 Superstorm Sandy The level of storm surge was not expected 23

24 The Flow of Information in the Catastrophe Modeling Process Determine project scope (what model, what perils, etc.) Evaluate, verify and format data Data quality checks Control totals and Data assumptions Edit the data for model import Review the output Portfolio Management Gather data Raw data from clients is aggregated and scrubbed Import, Geocode, and Run the model Deliver results PML, AAL Exhibits, Maps Reinsurer files SQL Databases (RMS, AIR) Limits Profiles (with Premium) 24

25 Summary of Pricing and Structure at January 1, 2016 Risk adjusted pricing typically declined by 5% to 8% Typical Risk Adjusted Pricing Changes Structure changes were more varied at this renewal as compared to last year; with over half of the renewal programs making changes to limit purchased Approximately 50% of companies placed supplemental coverage around the traditional cat program, being RPP, Underlying Cat, or Aggregate programs Changes to Structures 25

26 Catastrophe Bond Issuance and Capital Outstanding With just under $6 billion from 25 transactions, catastrophe bond issuances fell from their all-time record in 2014.

27 January 1, 2016 Property Renewal Overview PRICING & CAPACITY Rates on line fell in nearly all regions and business segments Global ROL Index fell 9% US ROL index fell 7% Risk adjusted rates decreased 5% - 8% on average Capacity was abundant in all regions Individual renewals experienced a wide range of outcomes dependent on loss activity, past renewal experience and other company-specific features COVERAGE Modifications in terms and conditions were less of a focus for many verses a year ago, while price remains a significant focus Where coverage expansion occurred, hours clauses were a primary focus Multi-year placement activity continued to rise with an increased percentage of placements including some component of multi-year MARKET DRIVERS Stabilizing supply Dedicated reinsurance capital remained flat at roughly USD400 billion Capital from rated carriers fell slightly while convergence capital increased slightly Light global loss activity Approximately USD30.5 billion in 2015, slightly lower than 2014 and well below the 10 yr average of USD50 billion Market flexibility and innovation are providing more bespoke solutions. This approach, along with attractive price points and excess capacity increased global cat limit purchased through 2015 and into the January 1, 2016 renewals. 27

28 Firm Order Behavior Relative to Reinsurer Quotes Range of average quoting behavior was up 11% to down 11%, broadening slightly from last year s range of up 8% to down 8% Relationship of firm order terms to quotes by program firmed slightly as compared to 2015 renewals 6% below the average quote (2015: 8% discount) 1% above the minimum quote received (2015: slightly below lowest quote) Firm Order to Quote comparison is based on quotes entered and available within GC MarketPlace. 28

29 7% drop from January 2015 to January 2016 US Property Catastrophe Rate on Line Index (1990 Jan 1, 2016) *2015 and 2016 represent index points at January 1, all others are full year. 29

30 Conclusion Weak Signals Data Management / Data Capture Product Development and Differentiation Responsiveness to Client Servicing Evaluation and Management of Risk Exposure 30

31 Disclaimer Guy Carpenter & Company, LLC provides this report for general information only. The information contained herein is based on sources we believe reliable, but we do not guarantee its accuracy, and it should be understood to be general insurance/reinsurance information only. Guy Carpenter & Company, LLC makes no representations or warranties, express or implied. The information is not intended to be taken as advice with respect to any individual situation and cannot be relied upon as such. Please consult your insurance/ reinsurance advisors with respect to individual coverage issues. Statements concerning tax, accounting, legal or regulatory matters should be understood to be general observations based solely on our experience as reinsurance brokers and risk consultants, and may not be relied upon as tax, accounting, legal or regulatory advice, which we are not authorized to provide. All such matters should be reviewed with your own qualified advisors in these areas. Readers are cautioned not to place undue reliance on any historical, current or forward-looking statements. Guy Carpenter & Company, LLC undertakes no obligation to update or revise publicly any historical, current or forward-looking statements, whether as a result of new information, research, future events or otherwise. This document or any portion of the information it contains may not be copied or reproduced in any form without the permission of Guy Carpenter & Company, LLC, except that clients of Guy Carpenter & Company, LLC need not obtain such permission when using this report for their internal purposes. The trademarks and service marks contained herein are the property of their respective owners. 31

32