1. MODELING CATASTROPHE RISK Hemant Shah Co-Founder and CEO, Risk Management Solutions Philippe Stephan, CTO2013 Risk Management Solutions, Inc.Confidential

2. PERKINS COIE PALO ALTO, CALIFORNIA3150 Porter Drive Palo Alto, CASoil Class Soil: Soft Rock to Stiff 4 3 2Soil (2.5) Liquefaction: Very Low Landslide: Very Low DTC 4.2 Miles SaAN100%Loss (% structure value)RMS 3 (Reinforced Concrete) 2 stories, built 1966 Approx. 35,000 ft210%1% time-dependent time-independent 0%101001,0001-in-X Probability1 2013 Risk Management Solutions, Inc.Earthquake Risk @ Perkins CoieConfidential10,000 3. Hemant and Weimin with Version 1.0, 1989Photo from RLI Annual Report, 1990 4. OUR HISTORY2013 Risk Management Solutions, Inc.Confidential 5. Growing Sophistication of Models 1,000,000RiskLink 13 500,000+ diskettes100,00010,000 1,000 100IRAS v1.0 17 diskettes10 1I9892014 6. WORLDWIDE CATASTROPHE RISK MANAGEMENT RMS MODEL COVERAGE EarthquakeTropical CycloneWindstorm Severe Convective Storm Winter StormFloodTerrorismPandemicLongevity2013 Risk Management Solutions, Inc.Confidential 7. RISK MANAGEMENT SOLUTIONSWorlds Leading Catastrophe Modeling Firm 1,300 EmployeesHundreds of Clients in the Global Risk Market Subscription Revenue Business Model Individual Client Revenues from $100K - $10MM/Year Models Used to Price, Structure and Underwrite Risk; Assess and Manage Capital; and Define Mitigation Strategies2013 Risk Management Solutions, Inc.Confidential 8. Stochastic Event ModuleHazard ModuleExposure ModuleVulnerability ModuleFinancial Analysis Module 9. SIMULATING PANEUROPE FLOOD HIGH-RES OFF AND ON FLOODPLAINCORRELATION ACROSS 13 COUNTRIES 10. Terrorist Attack ScenarioA Risk Map 11. Exceedance Probability4.0%3.0%2.0%1.0%0.0% $200M$400M$600M Loss (USD)$800M$1B 12. KEY APPLICATIONS PORTFOLIO MANAGEMENTUNDERWRITINGRISK TRANSFEREstablish guidelinesDetermine risk driversDetermine reinsurance needsDifferentiate risksEvaluate capital adequacyStructure risk transferAnalyze policy structuresAllocate capitalCounterparty communicationDevelop pricingEstimate losses2013 Risk Management Solutions, Inc.Confidential 13. Portfolio Management 14. DYNAMIC PORTFOLIO MANAGEMENTNew Quoted Total New Quoted ForecastCapacityBound Expected Renewal In-Force Time 2013 Risk Management Solutions, Inc.Confidential 15. Drill down into your bookView a multitude of metrics all in one place 16. Interact with multiple EP curvesInvestigate the drivers of change 17. Risk Transfer | Cat Bonds 18. 2013 Risk Management Solutions, Inc.Confidential 19. 1.7% ATTACHMENT PROBABILITY8ft AT THE BATTERY. AS SIMPLE AS THAT. 20. CASE STUDY: METROCAT RE 2013-12013 Risk Management Solutions, Inc.Confidential 21. The innovative non-traditional structure allowed MTA to close its storm surge insurance gap Non-traditional deal of the year Bond Buyer magazine 22. METRO CAT BOND IN THE NEWS2013 Risk Management Solutions, Inc.Confidential 23. Supply Chain Risk 24. Tohoku Earthquake 2011 caused supply disruption Major damage: Renesas (40% market share of MCU)GLOBAL SUPPLY CHAINExplosion in Germany 2012 caused supply disruption Evonik damaged 50% market share of Cyclododecatriene(CDT)Thailand Flood 2011 caused supply disruption Major production hub is damaged (25% of computer hard drives in the world) Material Suppliers2013 Risk Management Solutions, Inc.Hurricane Sandy 2012 caused disruption of distribution centersConfidentialParts SuppliersFacilityDistributions 25. EXAMPLE SUPPLY-CHAIN NETWORK IN AUTO INDUSTRYHigh Tech PartsSevere damage in Tohoku areaDomestic DistributionGears Engine AssemblingGlobal DistributionMetal ForgingTransmission Suppliers (Parts) 2013 Risk Management Solutions, Inc.ConfidentialSuppliers (Parts)FacilityDistributions 26. Network Topology and Conceptual ModelEXAMPLE SUPPLY-CHAIN NETWORKAnalytical ModelLoss Model CBI Simulation Engine2013 Risk Management Solutions, Inc.Confidential 27. THE TECHNOLOGY SIDEHive talk February 5th, 2014 Philippe Stephan, CTO2013 Risk Management Solutions, Inc.Confidential 28. Key questions users ask How much for this risk? How is my portfolio? What if something changed? 29. How we get to answers Event D,$ Damage LocationT&CEvent IntensityContract 30. Events are compile-time objects 31. Scale In : 1 portfolio 1Gb of client data Out: 1 model run 5T (* 8bytes) = 40Tb 50K events 100K locations 1K damage samples => Big Re. co: 5K portfolio 200Pb 32. Complexity Non additivity of risk Multiple what-ifsRegulatory framework (keep, encrypt, audit) 33. 100%CPU versus memory access90%% time in MEX80%70%60%50%40%30%0:00:000:07:120:14:240:21:360:28:48MEX/BI Workflow Duration (mins)0:36:000:43:120:50:24 34. 0.700 only realizable in the cloud0.6500.600Exceedance Probability0.550 0.500 0.450 0.400 0.350 0.300 0.250 0.200 0.150 0.100 0.050 -20,00025,00030,00035,00040,000 Max Cores45,00050,00055,00060,000 35. Our stack 36. Whats next for RMS(one) A reference database of subjects at risk An extensible exposure mgt system An ecosystem of models A generic risk exploration system A communication platform 37. Hi from the RMS Sr. Management Team37