new modelling developments - actuaries institute · • a framework for independent catastrophe...
TRANSCRIPT
NEW MODELLING DEVELOPMENTS
JEREMY G T WAITE Head of Analytics Australia and New Zealand – JLT Towers Re
Agenda • The Current Models • The Change in Modelling • The Results – Managing Cat risk… • Analysis/Synthesis/Judgement • Conclusion
The Models
The Models • The basics
– Hazard, Vulnerability, Exposure & Financials and uncertainty • Vendor Models
– AIR V13, EQECAT, RMS & Locally Risk Frontiers • Multimodel Platform
– AIR/RMS
• Open source development – OASIS/GEM/OpenRisk etc.
• Open platform architecture – RMS(One)
The History of Catastrophe Models • Measurement
– Hazard events began in the 1800’s (windspeed, ground motion)
• The Old days – Pre the 1960’s tacks in a wall map to measure concentrations
• Major Vendor Models began – AIR (1987), RMS (1988) , EQECAT (1994)
• In 1989 two large scale disasters – Hurricane Hugo ($4bn), Loma Prieta Earthquake ($6bn) – Encouraged more widespread use of models
• In 1992 Hurricane Andrew ($15.5bn, models predicted $13bn) – Nine insurers insolvent
• Companies turned to modellers for decision support – Modelling firms grew, cat models grew in terms of number, availability and capability
The Change in modelling - Cat Risk Management
8
The New Platforms
“In terms of the three vendors you have got three different levels and it kind of corresponds to their market share. You’ve got EQE who is bringing out something that is hot, you have got AIR who is bringing out something you might call extra spicy and then you’ve got RMS who is proposing something that might potentially be a market disrupter. All of this is pretty exciting because it has come all at once….” – Bruce Dilke, SVP at Argo Group Int’l Holdings “There are three things insurers want: consistency, transparency, flexibility. And they want these features in a more efficient platform. All these new initiatives are trying to address those issues. So I think it is a big deal. This is a very exciting time in cat risk management. I am trying to train people to stop saying cat risk modelling and say cat risk management because it is really about understanding the risk and not just about models.” – Karen Clark, President & CEO of Karen Clark & Company
9
The New Platforms – RMS(one) Supercomputing Power – operating in the cloud requires no costly internal
hardware or infrastructure investments Open environment gives clients control of their assumptions Freedom and flexibility to use their own models or other’s models Speed that will support decision making in what seems like real-time Increases user base from the current technical analysts to underwriters,
portfolio managers, senior and executive-level management Powerful information and analytics
“[RMS(One)] absolutely crushes the latency in the underwriting and portfolio management practices of our clients from months to minutes – that is move the modelling that has been very useful in the back office of insurers and reinsurers absolutely into the heart of the business process so that business practices can be executed to support decision making in real time.” – Hemant Shah, CEO and co-founder of RMS
10
The New Platforms - AIR’s TouchstoneTM Combines functionality of CATRADER, CATStation & CLASIC2 into one
integrated application Deployment options – internally, in a private cloud or in a 3rd-party cloud No pay per use New platform enables:
Much faster imports and runtimes New, improved and enhanced analytics
Geospatial – view exposure & loss data on a map Data Quality Analytics Hazard Analytics – hazard info, risk scores & loss metrics based
on lat/long Ability to review and understand exposure accumulations quickly and
easily
“Touchstone is the product of many years of research and development, but it is also the product of working closely with our clients to identify their evolving needs… As the models have become more sophisticated, so too have model users. Today, companies want greater insight into their risk; therefore the tools which advise them must evolve. More than ever, companies are demanding tools that provide transparency, flexibility, and choice.” - Bill Churney, COO, AIR Worldwide
11
The New Platforms – EQECAT’s RQE Focuses on
Models Covers 90% of the Global GDP
Platform Speed through grid-based computing/modeling Evaluating cloud options New financial module Redesigned databases Enhanced integration and automation features
Analytics and Outputs 300,000 year simulation library allows for better treatment of low
frequency/high severity events 3G correlation allows for better modeling of secondary uncertainty
More features to come in future releases
"We are thrilled that RQE will provide significant and increased value to the global re/insurance market. With so much that differentiates us competitively, we look forward to satisfying the pent up demand for our analytics. Having collaborated closely with leaders from virtually every segment and geography in the global re/insurance business, we are confident that RQE will disrupt the status quo of catastrophe risk modeling.“ - Bill Keogh, former President of EQECAT. “We’ve been big proponents of companies owning their view of risk.” – Paul Little, President, EQECAT
Open Model development • OASIS • A framework for independent catastrophe modelling • An alternative to the 3 main catastrophe modellers (AIR, EQECAT & RMS) • Owned by 22 insurers, reinsurers and brokers across the globe • Provides transparency • Offers greater flexiblity • Utilizes latest technology and operates on a wide range of platforms from laptops to the cloud
• There are others such as Open Risk and GEM
Open Model development (an example) Oasis provides “plug and play” functionality for main components of modeling risks, including: Exposure Data Capture Exposures Policy/Line Data Capture Policy/Line Data Vulnerability Model Damage Functions Hazard Model Frequency/Severity External Data Reference Tables Results Viewing Module Policy/Line Data
14
Technology: Cloud Computing Outsourcing computing power reduces/eliminates:
Hardware purchases A place to store the hardware Service requirements & maintenance
High-Powered Computing Jobs from queue routed to run on capacity that was not available in “silo”-ed architecture
“Compute capacity upon request” Elastic and scalable Simpler process Transparent to user
“Pay-as-you-go” models Infrastructure as a Service (IaaS) Software as a Service (SaaS)
Standardized interfaces Mobility – access the software from any computer
15
Advantages Uses a cloud specifically optimized for catastrophe modeling Cat modeling firm supports both the infrastructure and software Leverages technologies that enable next generation modeling, big data analytics and
business intelligence capabilities
Disadvantages Data security and privacy concerns Potential stability and availability issues/business interruptions Still require IT Support – configuration and support Bandwidth – getting data in and out
AIR, EQECAT & RMS will all be offering an option to deploy in a cloud
Technology: Cloud Computing
Key Comparison • Current Vendor models
– Each model has its own front and back end for dealing with input and output – Main criticism is lack of transparency and lack of efficiency, hence new models…
• Multimodel platform – Allows users to run cat models developed by the vendor & others (e.g. RMS one) – Same interface, same data formats (partners Risk Frontiers, COMBUS etc..)
• Open source – The computer code is freely available – E.g. The Oasis platform will have opens source financial module
• Open Platform – User build their own views of risk, starts with reference models – Full transparency, can test impacts of changes in event freq or severity (e.g. adjust
hazard events), or validate damage functions
The Results
What can go wrong with Models
• Bad Physics • Bad Assumptions • Bad Data • Bad Luck • Too many Black swans
Source: Stein, Geller, Liu 2012
Plate tectonics was only accepted in the early 1960’s by the geoscientific community
Earthquake Epicentres in Australia 1841 to 2000
Source: McCue, K.F, 2001 – Earthquake Epicentres in Australia 1841 – 2000 and recent fault scarps, AGSO – Geoscience Australia, Canberra
• The Australian continent is completely within the Australian plate and consequently there are no major through going active faults like the Alpine fault in NZ or the San Andreas fault in Western USA.
Vendor Model Availability in Australia
Vendor Cyclone Earthquake Bushfire Flood Storm Hail AIR Yes Yes Yes No No No
EQECAT Yes Yes No No No No Risk Frontiers Yes Yes Yes Yes No Yes
RMS Yes Yes No No Combined - Sydney only
The Results • Model uncertainty • Model blending approaches
– Understand the difference! • lower return period losses
– Bushfire, Flood, Storm and Hail – ICRC HR & VR
• Other factors – Understanding what’s included / excluded – Business decisions
• What to do with model output – Approaches for allowing for components not
included – Decisions, Decisions and Decisions – Realistic Disaster Scenario Analysis
Model blending approaches • In Theory there is no difference between theory and practice,
in practice there is.. (Yogi Berra)
• Since all models are wrong the scientist cannot obtain a "correct" one by excessive elaboration. On the contrary following William of Occam he should seek an economical description of natural phenomena. (George Box)
• Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful. (George Box)
• Understand the difference between model results and how they apply to your portfolio, where does the uncertainty lie?
• Where are concentrations
• What’s possible • Track the Cyclones
live.. • What’s the potential
coverage? • Prepare… • Manage the claims
Cyclone ITA, live track from www.bom.gov.au into JLT Towers Re CATography software
Model Output: Realistic Disaster Scenario Analysis • 1 In 200 PML not an event
– What events of this scale are plausible
Hence RDS (used at Lloyds since June 1994)
• Easy to Understand
– Board and investor-friendly
• Qualitative and Quantitative Testing
– Reinsurance arrangements
– Capital impact
• Satisfies APRA stress and scenario test
– Vertical Requirement
– Horizontal Requirement
Cyclone
Earthquake
Bushfire
Flood
Storm
Hail
25
Analysis, Synthesis and Judgement
A New Modelling Hope - positives
• Realisation of the over Reliance of models has lead to improved models demanded by insurers/Reinsurers…
• Increased consistency, transparency and flexibility • More views of risk • Understanding risk, not models
A New Modelling Hope - negatives
• More models could add to confusion… • Move away from modelling firms core competence
into “IT” • Still required interpretation of data and results and
use this to make informed decisions
Uses of results • More ability to sense test results (faster runs more
analytics drilling ability)…give more insight • Understanding uncertainty can lead to strategic
decisions – Avoid, mitigate reduce transfer • Regulatory rules a minimum applied across the
whole industry, in reality one size does not fit all, make risk decision in the context of each company, antifragile decisions around uncertainty…
CONCLUSION • A fool with a tool is still a fool… • The opportunity is to be more able to focus on
model validation in order to understand true risk • This is still just the beginning of the Catastrophe risk
modelling revolution with increased emphasis on Catastrophe risk management…..