Economic Capital

1. Economic Capital Appointed Actuary Seminar September 2004 Doug Brooks

2. Topics • Sun Life’s Risk Framework • Economic Capital in Life Companies • Issues around Economic Capital • Aggregation • Regulatory vs. Economic Capital • Issues – Model Risk

3. RISK COMMUNICATIONS RISK RISK TOOLS PHILOSOPHY CULTURE RESOURCES OBJECTIVES RISK RISK TOLERANCES POLICIES ACCOUNTABILITIES RISK PROCESSES Risk Management Framework

4. Objectives of Risk Management • Avoid risks that could materially affect the value of the company • Contribute to sustainable earnings • Take risks that the company can manage in order to increase returns • Provide transparency of the company’s risks through internal and external reporting

5. Risk Philosophy • Our business is accepting risks for appropriate returns • Risk Filter: • Understand the risk • Capability to manage the risk • Monitoring/reporting capabilities • Actionable mitigation strategies • Appropriate return for the risk taken on • Acceptable impact on total company risk profile • Embedded into the business management practices of every Business Group leader

6. Organizational Attributes Everyone is a Risk Manager • Acting with Integrity • Understanding Impact on Customers • Embedded Risk Management – Discipline • Full and Transparent Communication • Collaboration • Alignment

7. Risk Categorization • Categories • Sub-categories • Source • Exposure Triggers • Direct Consequences

8. MARKET CREDIT RISK RISK OPERATIONAL RISK INSURANCE RISK Risk Categorization

9. Rationale for Common Measurement • Help Ensure Solvency and Viability • Consistent Understanding of Risk & Return • Understand the Impact of Diversification • Understand Contributions of Businesses to Value • Appropriate Arbitrage of Regulatory Capital • Influence Regulators & Anticipate Regulatory Direction • Consistent Disclosure

10. Economic Capital – A Common Risk Measure • Comprehensive coverage of risk types (financial, insurance, credit, operational) • All risks measured on a consistent basis. • Time horizon harmonized across analysis. • Confidence interval linked to strength. • Forward-looking, not historical volatility. • Additive across the businesses

11. Illustration of Economic Capital

12. Economic Capital - Steps • Identify and Model Risk Distributions • Determine Economic Capital by Risk • Aggregate Individual Risks Using Correlations • Allocate Economic Capital to Businesses • Benefits of Diversification? • Determine Risk-Adjusted Return

13. Economic Capital - Issues • Modelling Different Risks - Methodology • Time Horizon and Level of Capital • CTE vs. Percentile • Reflecting “pass through” • Quantifying Operational Risk • Aggregation - Diversification

14. Economic Capital Issues - Methodology • Purpose is to determine a model of the distribution of the risk based on company data • Different models may be appropriate for different risks, or based on materiality or data availability • Stochastic modelling is necessary where there is significant tail risk • Modelling using other distributions may be appropriate for risks with less tail risk, more symmetry, diversifiable risks • Reliance on regulatory factors • Benchmarking where information not available • Approximate means are appropriate where risk is not material

15. Economic Capital Issues – Time Horizon and Level • Time horizon must be able to be modelled • Time horizon must be consistent across risks • Whatever horizon is used must still reflect lifetime liabilities • Level chosen will reflect the time horizon (a greater level will be chosen for a lesser horizon and vice versa – e.g. 99.95% for 1 year, 99.5% for 10 years) • Level is chosen to provide a level of security, and based on desired ratings and commitments to stakeholders

16. Economic Capital Issues – Measure: Percentile vs. CTE • The measure used must reflect risk appropriately • The measure used must be consistent for all risks • CTE is appropriate for measuring tail risks • If there are substantial portion of the company’s risks that have significant tail risk, then CTE may be more appropriate • Level may reflect use of CTE or percentile (e.g. a higher level is appropriate if percentile is chosen, both numerically and to reflect potential for uncaptured risk)

17. Economic Capital Issues – Reflecting Pass-Through • Many insurance products have some degree of pass-through to policyholders • Degree of pass-through varies • Ability to apply pass-through in practice must be assessed to determine appropriate offset to capital • Contractual constraints (e.g. minimum guarantees) • Actual practice of adjusting - expectations • Competitive constraints

18. Economic Capital Issues – Areas with Little Experience • New products: • Policyholder behaviour and policyholder options • Health benefits • Current products in new economic conditions • New markets • Late-duration experience • Concentration risks Conservatism in tail assumptions

19. Economic Capital Issues – Quantifying Operational Risk • Operational risk is an extremely important risk • Most failures have operational roots • What experience is appropriate to use for modelling operational risk? • Must reflect long-term nature of liabilities • Systems and processes and management will change over the lifetime of the liability • Is it worth a lot of investment to model? • Benchmarking approach • Incentive to management • Modifiers based on assessments

20. Economic Capital Issues – Aggregation - Diversification • Risks must be aggregated – requires modelling • Copulas suggested by the IAA Working Group • Should reflect correlation BUT • Must be very careful with assumptions about persistence of observed correlations in the tails

21. Economic Capital Issues – Aggregation • Correlation matrix approach has weaknesses, particularly if non-normally distributed risks • Copulas are a technique to aggregate via a multivariate uniform distribution • T-Copula approach captures both marginal distribution of risks and joint probability of extreme values (tail dependencies of a T-copula are controlled by parameter) • Data issues/calibration - a major factor

22. Example of Aggregation

23. 50 45 40 35 30 25 20 15 10 5 0 99.3 99.4 99.5 99.6 99.7 99.8 99.9 Aggregation - Parameters Gaussian_Copula Correlation T_Copula(V=1) T_Copula(V=4) T_Copula(V=10) T_Copula(V=20) T_Copula(V=100) 50 45 40 35 Correlation Benefit (%) 30 25 20 15 10 5 0 90.0 91.0 92.0 93.0 94.0 95.0 96.0 97.0 98.0 99.0 99.1 99.2 CTE

24. Regulatory vs. Economic View • Creates basis for arbitrage • Appropriate reporting basis • Changes in regulatory capital requirements • Ability to influence regulatory developments

25. Regulatory vs. Economic:Risk Distribution Economic Regulatory

26. Definition of Model Risk: Model risk is a general term referring to the possibility of loss or error resulting from the use of models. This risk has a number of components: • Model misspecification • Assumption misspecification • Inappropriate use or application • Inadequate testing, validation, and documentation • Lack of knowledge or understanding, user and/or management • Inadequate systems structure and change management controls • Error and negligence

27. CAS Stochastic Reserve GMIB Example

28. “Full Run” Relative Error Plot