Report on CAS Research Working Parties

1
Report on CAS Research Working Parties

• Moderator Donald Mango, FCAS, MAAA
• CAS Vice President of Research and Development
• Director of Research and Development, GE ERC
• 2004 ERM Symposium
• April 27, 2004

2
Agenda

• Working Parties Overview
• Correlations and Dependencies WP
• Executive Decision Making Using DFA WP
• Elicitation and Elucidation of Risk Preferences
  WP
• QA throughout

3
What is a working party?

• Essentially a task force focused on research of a
  specific topic or solution of a specific problem
• Group effort with a single work product
• Modeled on GIRO platform where it has been used
  successfully

4
What can WPs deliver?

• Original Research
• Survey papers
• Overview of research published to date on a
  particular topic
• Could be used as syllabus material
• Compendiums of papers by different authors
• Correlation of Risk working party will use this
  format
• Similar approach taken by Reserve Variability WP
• Studies of industry experience based on special
  data calls

5
What are the current working parties?

• Correlations and Dependencies Among All Risk
  Sources
• Executive Level Decision Making Using Dynamic
  Financial Analysis
• Elicitation and Elucidation of Risk Preferences
• Quantifying Variability in Reserves Estimates
• Implication of Fair Value on Asset Allocation (on
  hold)

6
Report of the Correlation Working Party

• Glenn Meyers
• Insurance Services Office, Inc.

7
Charge of the Working Party

• ERM requires the quantification of the total risk
  of an enterprise. One must consider correlation
  to properly combine the individual risk
  components.
• Considerations
• Theoretical
• Empirical
• Computational

8
Theoretical Considerations

• Conclusion No overriding theory of
  correlation.
• We will provide examples of multivariate models
  that exhibit correlation.
• Experts prefer the term dependencies rather
  than correlation.
• I find myself reverting the common usage so
  nonexperts will know what I am talking about.

9
Empirical Considerations

• Historical problem lack of data
• One observation per year
• If correlation matters, we should be able to find
  data that exhibits that correlation.
• One approach
• Create a model that depends on a driver for
  correlation.
• Use data from several insurers to parameterize
  the driver.
• Example to follow

10
Computational Considerations

• ERM demands the aggregation of segments.
• Simulation
• Iman Conover and Copulas
• Fourier transforms
• Faster than simulations, but less flexible and
  require more setup time.

11
Chapters Written by Individual Authors

• Common Shock Models Glenn Meyers
• The Iman-Conover Method Stephen Mildenhall
• Correlation over time Hans Waszink
• Aggregating Bivariate Distributions David Homer
• Dependency in Market Risk Younju Lee
• Modeling Time Series with Non-Constant
  Correlations Dan Heyer
• Correlations in a General Stochastic Setting
  Lijia Guo
• 4 CAS Members and 3 non members

12
From Meyers ChapterThe Negative Binomial
Distribution

• Select a at random from a gamma distribution with
  mean 1 and variance c.
• Select the claim count K at random from a Poisson
  distribution with mean a?l
• K has a negative binomial distribution with

13
Multiple Line Parameter Uncertainty

• Select b from a distribution with Eb 1 and
  Varb b.
• For each line h, multiply each loss by b.
• Can calculate r if desired.

14
Multiple Line Parameter UncertaintyA simple,
but nontrivial example
Eb 1 and Varb b
15
Low Volatility b 0.01 r 0.50
16
Low Volatility b 0.03 r 0.75
17
High Volatility b 0.01 r 0.25
18
High Volatility b 0.03 r 0.45
19
About Correlation

• There is no direct connection between r and b.
• For the same value of b
• Small insurers have large process risk and hence
  smaller correlation
• Large insurers have smaller process risk and
  hence larger correlations.
• Pay attention to the process that generates
  correlations.

20
Estimating b From Data
0
21
Reliable estimates of b are possible with lots of
data.

• Real estimates provided by Meyers, Klinker and
  Lalonde
• http//www.casact.org/pubs/forum/03sforum/03sf015.
  pdf

22
Sample CalculationsCommon Shocks to Frequency
and Severity

• Multiply expected claim count by a random shock.
• Negative binomial count distributions
• VarShock called covariance generator
• Multiply scale of claim severity by a random
  shock.
• Lognormal severity distributions
• VarShock called the mixing parameter
• Look at spreadsheet

23
Scroll Down
24
(No Transcript)
25
Scroll Down
26
(No Transcript)
27
Parting Message

• Build models of underlying processes.
• Common shock model illustrated here
• Other chapters build other models
• Quantify parameters of models
• Use data! (If data will never exist, why worry?)
• Express parameters in a form that has intuitive
  meaning.
• Correlation is a consequence of the models.

28
Working PartyPresenting DFA Results to Decision
Makers

• Nathan Babcock
• Conning Asset Management

29
Statement of Purpose

• Executive-Level Decision Making Using Dynamic
  Financial Analysis
• Facilitate effective DFA presentations to senior
  management
• Survey existing presentations
• Create tools and documentation

30
Summary of Events

• Enroll volunteers
• Survey past presentations
• Develop graphical template and manual
• Create sample dynamic financial analyses and
  presentations
• Write summary report
• Review presentations, report, template

31
End Products

• Summary report
• PowerPoint template for graphs
• Paper describing concepts behind template
• Three sample presentations (applying template
  graphs)
• Document of presentation guidelines

32
Lessons of the WorkingParty Process

• Chairpersons - management responsibilities
• Communication with oversight committee
• Ability to direct a group of volunteers
• Multiple rewrites of the project timeline

33
Working Party Discoveries

• WebEx - online presentations- paired with
  simultaneous teleconference
• Hyperlinks- useful in documents, not just
  Internet- one reading electronically can page
  within set of manual, template, report
• Graphics

34
Sample Graph Profit Distribution
Stop Loss
35
Sample Graph Liability cash flow
Practically no chance of paying this much
36
Sample Graph Hedging
Hedge is inexpensive,
but useful
37
Working PartyElicitation and Elucidation of
Risk Preferences

• David Ruhm
• The Hartford

38
Risk PreferencesWorking Party

• Motivation
• Most DFA and Risk Management exercises assume
  management has a clear, consistent,
  well-understood, transparent set of risk
  preferences
• Such a set is a prerequisite for many risk
  analysis efforts, including asset portfolio
  composition and reinsurance purchasing.
• The WP members and others do not believe this to
  be the case in general

39
Risk PreferencesWorking Party

• Goals
• Describe or develop techniques and exercises that
  can be used to assist company senior management
  in developing (elicitation) and refining
  (elucidation) a consistent set of risk
  preferences.
• The Working Party may draw from research in
  decision theory, prospect theory, and behavioral
  finance.

40
Risk PreferencesWorking Party

• Framework
• Company level assessment (i.e., not
  riskiness/leverage of individual products sold)
  supporting planning decisions (product lines).
• Provide decision support for the selection of the
  most desirable prospective portfolio mix.
• But moving beyond Markowitz by considering many
  measures and timeframes.
• Consider one-year, two-year, and three-year time
  horizons
• Net Income, both absolute and scaled on PHS
• Several likelihood thresholds (1 in 5, 10, 25,
  50, 100, 250 years)

41
Risk PreferencesWorking Party

• Two Sub-Teams
• Senior Management Silent Auction
• Survey Research of Psychology Studies of
  Preferences
• Silent Auction
• Attempts to elicit consistent preferences while
  correcting for framing effects
• Give to Sr. Mgmt members, then plot their answers
  on a single chart

42
(No Transcript)
43
Risk PreferencesWorking Party

• Behavioral Finance
• Non-Bayesian Forecasting must have a prior
  before you can apply Bayes' Theorem. Also tend to
  overweight recent experience.
• Herding need content providers, expressing
  actual opinions, rather than relativistic
  participants (what do my peers do?). If all the
  peers are asking about the other peers, who is
  putting content in?
• Over-reaction beware of this tendency, which is
  particularly troubling in this buy-and-hold,
  non-exchange market. What should the threshold
  for reaction be? How many bad quarters or years
  in a row?

44
Risk PreferencesWorking Party

• Behavioral Finance
• Myopic Loss Aversion risk premium is a function
  of time horizon. If you look at returns and
  results more frequently (shorter time horizon),
  you will demand a higher risk premium.
• Earnings Management owner's often want
  unrealistic earnings stability, often at odds
  with the desired return.
• Growth/Decline/Reorganization Expansion occurs
  more readily than redeployment and destruction.
  Function in part of myopic focus on sunk costs
  and difficulty of exit.

45
Risk PreferencesWorking Party

• The Road Ahead
• We need a chairperson!
• Finalize research summary
• Finish Senior Mgmt Silent Auction
• Report out later this year

46
Thank you for your attention

• Questions?