RenaissanceRe Services Ltd

1
RenaissanceRe Services Ltd

• Robin Lang
• Puerto Rico
• June 2008

2
Thoughtful Risk Management in a changing
environment

• Robin Lang
• Puerto Rico
• June 2008

3
Outline

• What does good risk management look like ?
• Appropriate risk analysis and modelling is a
  cultural issue
• The power and danger of Risk Models
• The relentless pursuit of excellence in a
  changing environment

4
What does good risk management look like?ERM -
Whats all the fuss ?

• Thoughtful risk management is just good business.
  ERM is a buzz phrase that should be a cultural
  fundamental of all risk taking enterprises.
• Appropriate risk management can provide a
  significant competitive advantage
• Avoiding surprises allows you to take better
  advantage of the opportunities created by
  dislocated markets
• Communicating accurate expectations maintains
  your credibility with your key stakeholders
• Disciplined measurement and aggregation of risk
  decreases the mortality rate for your franchise
• Powerful risk tools allow you to pursue the best
  opportunities.

5
Appropriate risk analysis and modelling is a
cultural issueRisk management is at the core of
RenRe Culture

• Company founded on principle of building better
  exposure management for catastrophe risk
• Risk Management is the responsibility of
  everyone, not just a delegated few
• Significant money, resources and management time
  invested in processes, systems and technology to
  better manage risk
• Management recognized from the beginning that the
  right culture and people were more important than
  the right tools
• As company has matured and grown, we have
  remained committed to maintaining excellent risk
  management in all aspects of our business

6
RenRe Risk Management Principles

• Foster the right culture culture counts as much
  as the models
• Recognize the limits of quantitative modeling
• Seek to understand all the risks being assumed
• Guess dont ignore
• Combine tools and expert users to evaluate
  risk-reward decisions
• Use multiple vectors to evaluate risk
• Crude proxies for risk instead of ignoring risk
• Measure and monitor the aggregate portfolio using
  robust systems and frameworks
• Define tolerances for acceptable levels of risk
• Processes to avoid or eliminate certain types of
  risks
• Actual vs expected feedback loop
• Learn from mistakes
• Undertake a constant process to enhance and
  improve

7
Appropriate Risk Modelling is a Cultural
IssueFundamental principles applicable to all
sources of risk

• Appropriate use of risk modelling can be
  extremely powerful, but can also be extremely
  dangerous
• Understand that all risk models are wrong
• An actual outcome will never match the expected
  case
• But it should be within your distribution!
• If it looks too good to be true it probably is
• Be comfortable with uncertainty
• There is no single answer, merely a range of
  potential outcomes
• Multiple views of risk is critical
• Apply reasonability
• Reality checks are critical
• Dont ignore historical data
• Consider actual vs. expected
• Learn from mistakes and confront them openly
• Questions and feedback loops
• Are you lucky or good?

8
All Models are Wrong- 5 Key Drivers of Model
Error
An attempt at examining the main common sources
of misunderstanding or error (by no means
exhaustive)

• Fundamental difficulties
• Unmodeled risks and hazards
• Modelling malpractice
• Being Careful
• Portfolio error

9
All models are wrong
Model    noun. 4 a simplified mathematical
description of a system or process, used to
assist calculations and predictions.
Source Oxford English Dictionary

• In all sectors of the insurance industry and the
  capital markets there is a tendency to exaggerate
  the predictive power of risk models.
• Over reliance on all forms of risk models is
  common place.
• Manage risk and capital to single modelled points
  
• Blindly accept model change
• We tend to confuse precision and accuracy and
  assume that the more complex a model the better
  it must be

9
10
Fundamental Difficulties

• Building a mathematical model to represent any
  form of natural phenomenon or or human behaviour
  is just hard
• Hazard
• Vulnerability
• How will the legal environment change in the
  future ?
• How will emotion drive market responses
• Models will always be wrong to some degree in
  actual individual events every event provides
  learning
• Unknown EQ Faults?
• Appropriate historical period to estimate
  frequency and severity?
• Wind from the wrong direction?
• Hopefully an appropriate stochastic risk model
  should see actual losses appear somewhere in the
  modelled distribution
• We just may have got the frequency of loss size
  wrong!
• Important to use multiple models and views on the
  same problem
• More Physics Less Statistics

11
Fundamental Difficulties Importance of Multiple
Models

• There is no right answer so use all credible
  inputs
• Single model users invariably Optimise into the
  model
• By design or accident will migrate towards the
  most optimistic representation of risk over time
• All models have biases, some serious. By
  offering different biases, alternative model
  views can mitigate single bias error.
• Dealing with model change
• Given the uncertainty surrounding all underlying
  components of risk models, they may be subject to
  significant change from version to version.
• Hazard
• Vulnerability
• Financial

11
12
Unmodeled Perils

• Significant losses historically from unmodeled
  perils, some known and unmodelled and some
  unknown. The table gives some background on
  estimated property losses sustained in recent
  years this is by no means an exhaustive list
• Memory from region to region seems to be poor
• Did we learn from the 1999 Forrestry losses in
  France and apply that learning to Skandinavia ?

13
Model malpractice

• Poor data
• Incomplete data
• Underestimated values
• Do the characteristics and quantum of data
  provided change dramatically year on year?
• Inappropriate (optimistic?) values and risk
  characteristic coding
• Examples in Katrina well-known
• We should not be surprised if similar problems
  highlighted following a large Japan Typhoon?
• Confusing precision with accuracy
• Is the model fit for purpose?
• More data and higher resolution does not
  necessarily equal more accurate loss estimates
• Hazard Does our ability to model the hazard
  support increasing precision as a deliverer of
  increasing accuracy ?
• Detailed vs. aggregate modelling Cross-check
  detailed model runs against higher-level
  analyses, explain the differences

13
14
Modelling malpractice - What are you covering ?

• How many users actually review the contents of
  large exposure files ?
• This is an excerpt from an actual EDM

15
Modelling Malpractice Inappropriate Risk
Characterisation

• Example of a Mobile Home writer insisting that
  Double Wides would react like a Wood Frame
  house.
• Actual loss experience would indicate that
  modelling as MH is more appropriate

The data used is aggregated and changed from
original but preserves the characteristics
16
Modelling Malpractice Financial Markets Ignoring
the Data ?
Certain sectors of the markets continue to model
the economic impact from these events as very
low probability tail events yet they happen on a
regular basis.  Rather than change their
distributions, many are using these as
deterministics to stress their portfolios but
continue to use VaR as a primary risk measure.
VaR is primary driven by thin tail distributions.

• 1995 Mexico
• 1997 Asia
• 1998 Asia, Russia, LTCM
• 1999 Nasdaq
• 2000 Nasdaq
• 2001 9/11, Enron
• 2002 High Yield, Argentina, Worldcom, Arthur
  Anderson
• 2006 Oil prices
• 2007 US Subprime, Bank Loans, Structured Credit

• 1971 Collapse of gold standard
• 1973 OPEC crisis
• 1979 Iran, Afghanistan, OPEC
• 1982 Penn Square, Mexico, Volcker
• 1987 Black Monday
• 1989 SLs, High Yield
• 1990 Kuwait War
• 1992 ERM

17
Modelling Malpractice Inappropriate Risk Model
Building
The financial markets are seemingly perpetually
surprised by events. Is this due to a
fundamentally inappropriate view of risk and of
risk model building ? Much of the financial
market risk models are based on Normal or Log
Normal distributions, for example VaR These
distributions are more appropriate for data that
has lower volatility between the mean and worst
case scenarios. We were seeing things that
were 25-standard deviation moves, several days in
a row, said David Viniar, Goldmans chief
financial officer to explain the losses of 30
plus for two Goldman Sachs hedge funds. Using
a Log Normal distribution a 25 standard deviation
(Sigma) is about a 600k year return period.
Using a Normal distribution a 7 standard
deviation (Sigma) is about a 13 billion year
return period. It is difficult to believe that
reasonability was being applied to the modelling
here.
18
Appropriate model use - Client historical tornado
hail losses

• What does the historical data tell you?
• Something about annual freq/severity
• Requirement for agg not occ modelling

19
Appropriate model use - Ignore historical data
at your peril
20
Being Careful

• Absence of any judgment or reasonability testing
  on model outputs
• Over reliance on modelled results
• Apply learning from one region to another
• Think you have learned something about NA
  hurricane vulnerability ?
• Where else might an application of this learning
  provide insight or change your view ?
• Japan Australia
• Tail reasonability
• Black Swans

21
Portfolio ErrorInter and Intra Portfolio
Correlation

• Errors from the previous slides compound in a non
  linear fashion, then if you over lay an
  underestimation of intra and inter portfolio
  correlation things can get even worse.
• Intra-portfolio correlation
• Multi locations under blanket policy
• Inter-portfolio correlation
• Correlation in the tails often underestimated
• Treaty/Fac/Energy/Assets/Liability etc.
• Deals/classes where no cat risk was assumed
  covered
• Communications between different risk entities in
  the same company
• The roll up of all of these errors has
  contributed to the wide gap between modelled and
  actual losses in recent large industry events
• (WTC, Katrina, Enron, Sub Prime)

22
How should correlation be interpreted ?

• Linear Correlation is often poorly understood.
• 10 Linear correlation represents only 1
  explanatory relationship between two variables.

23
Constant RefreshWe live and operate in a dynamic
market

• Constantly challenge what you currently do and
  how can you change it do not get lazy
• Our market place is changing constantly,
  (Re)insurers have an obligation to provide
  consistency to their clients whilst maintaining
  dynamic capital models.
• Market Cycles
• Fierce competition for attractive returns
• Capital markets / CatBonds
• Sidecars
• Hedge funds
• Shortened Investment horizons
• New Partners/Markets
• Increased liquidity of capital
• Trading of risk
• It is the same technology but different

24
Constant RefreshWe live and operate in a dynamic
market

• Good technology is key to maintaining an edge in
  existing markets and identifying new
  opportunities for understanding risk. Technology
  must be cultural and decision paths should be
  clear.
• Keep pushing existing technology envelopes
• Forecasting
• What lead time is useful?
• intra/inter Seasonal
• Climate regimes (Heightened Near Term risk?)
• What do you plan to do with the forecast when you
  have it?
• New trading partners/market
• Weather
• Natural Gas

25
Summary

• Probabilistic modelling of risk (Cat and other)
  can be extremely powerful
• Needs to be used with care from the data to the
  results
• Understand limitations of the model
• Augment complex modeling with simple checks
• Analyses must be in the context of the deal
• Appropriateness of Model
• Deal structure
• Curve setting
• Get comfortable with uncertainty
• Constantly challenge

26
Further Information
RenaissanceRe Renaissance House 8 20 East
Broadway Pembroke, HM 19 Bermuda Tel (441)
295-4513 www.renre.com
27
Disclaimer

• These slides represent the work and opinions of
  the author and do not constitute the official
  position of RenaissanceRe Holdings Ltd. or any of
  its directors, officers, employees or
  subsidiaries (RenRe). RenRe does not assume
  liability for the relevance, accuracy or
  completeness of the information provided.

27