Enterprise Risk Management and Stochastic Embedded Value Modeling

Transcription

1 Insurance and Actuarial Advisory Services Enterprise Risk Management and Stochastic Embedded Value Modeling ALM Joint Regional Seminar, June 27, 2005 July 4, 2005 Jonathan Zhao, FSA, FCIA, MAAA, MCA

2 Agenda Need for Enterprise Risk Management and Modeling Embedded Value Concept European Embedded Value and CFO Forum Principles Stochastic EV Modeling Case Study Practitioners Guide & Closing Remarks 1

3 Throughout history the industry has misjudged the behavior of its environment, distribution channels and customers. The unbundling of life insurance with the introduction of universal life policies Vanishing premium concept and no lapse guarantees Single premium capital guaranteed products backed by inappropriate assets Embedded guarantees (e.g., GMDB, GMAB, GMWB) in variable and unit-link products Adverse customer behavior has cost the industry dearly in terms of lost profits. Enterprise risk management and modeling is now necessary in order to run the business. 2

4 Enterprise Risk Management Risk is an event or condition that impairs the organization s ability to meet its goals Risk Management is a process of balancing the consequences of risk with the cost of avoiding it or hedging it Enterprise Risk Management is risk management that considers all the enterprise s various business in aggregate and by considering the correlation of the risks Enterprise Risk Modeling is the part of the enterprise risk management that you can model and quantify, and use as a benchmark to take appropriate actions 3

5 Insurance and Actuarial Advisory Services Enterprise Risk Modeling

6 Enterprise Risk Modeling Follows a Logical Path Catalog & Identify Qualify and Prioritize Response Exploit Exploit Optimize Risk Other competitive advantages Catalog & Identify Enterprise concerns & exposures Universe of risks Qualify and Prioritize Use actuarial tools Management judgment Response Avoid Minimize Mitigate risks 5

7 Catalog the Universe of Risks Financial Interest rates, equity market volatility, credit, liquidity Business Mortality, morbidity, economic activity, management failure Operational Process failure, fraud, litigation Event Regulatory change, war, natural disaster Assess the organization s risk tolerance realistically. What risk levels can we take on comfortably? 6

8 Quantify and prioritize Determine tolerance level Create accountability Who s on first? Evaluate RBC, rating agency perspectives To see ourselves as others see us Use available resources first, such as cash flow testing models Do better, if possible The internal and external audit functions may be valuable in this process 7

9 Respond - defensively first Avoid Diversify Reinsure Offset Hedge Strengthen capital Monitor continuously 8

10 Exploit! Risk management is a tool for competitive advantage: PRICING PRECISION SELF HEDGING PRODUCT RATIONING NICHE EXPANSION PRICING ARBITRAGE OTHER 9

11 Enterprise Risk Modeling needs a Global Financial Reporting and Measurement Framework Actuarially robust Reflects local reserving, capital requirements and taxation Acts as an early warning system Reconciles with product pricing Allows for enterprise consolidation Embedded Value reporting is that framework 10

12 Insurance and Actuarial Advisory Services Embedded Value

13 Embedded Value measures management s added value. Embedded Value Use of Embedded Value Basics, Deterministic vs. Stochastic EV European Embedded Value (EEV) Principles Embedded Value at Risk (EV@Risk) Financial statements that help optimize risk 12

14 Uses of Embedded Value Published financials (China, UK, Europe) Incentive compensation (China, US) Information for rating agencies, analysts (Canada) Pricing and planning validation (Global) Enterprise risk modeling (New) 13

15 Basic Components of Embedded Value Actuarial Appraisal Value Value of Future Business Value of Inforce Business Adjusted Net Worth Embedded Value 14

16 Value of Inforce Present Value (PV) of Distributable Earnings = PV after-tax book profits PV Cost of Capital + Target Surplus (TS) at the valuation date Distributable Earnings (year t) = After-tax Book Profit (year t) + TS Released (year t) + After-tax earnings on TS (year t) Allocate assets to business lines, project cash flows using best estimate assumptions, include statutory reserves, discount distributable earnings to present value at a hurdle rate reflecting risk 15

17 PV Cost of Capital Timing difference between TS now and future TS releases net of aftertax investment earnings on assets backing TS = Target Surplus PV Future Target Surplus Released PV Future after-tax earnings on assets backing TS 16

18 Adjusted Net Worth = Statutory capital and surplus + Allocations of surplus, such as AVR + Non-admitted assets with realizable value + Reduced for the value of any obligations not considered in the value of inforce + Adjustment for difference between market value and book value of assets backing adjusted net worth Target surplus is usually included in adjusted net worth for presentation purposes 17

19 Embedded Value Change = Value Added or Achieved Profits Achieved Profit Amount Adjusted Net Worth Starting Embedded Value Value of Inforce S/H Dividend Adjusted Net Worth Value from New Sales Value of Surviving Inforce Ending Embedded Value 18

20 Deterministic EV Best estimate assumptions Reflects overall risks in the discount rate higher discount rate for riskier products Easy to implement and understand Can often use the existing actuarial model However, Focused primarily on interest-rate risk Does not reflect tail exposure Unable to measure the interaction of risks 19

21 Stochastic EV Enables companies to capture the interaction of risks Quantifies risks (total enterprise basis & by line of business) Helps management to determine a comfortable level of risk and to optimize the risk/reward relationships An approach that distinguishes a company from its competitors in terms of enterprise risk measurement and management 20

22 What is Stochastic EV? Examples of simple application Various interest rate scenarios used in asset adequacy testing Sensitivity testing (e.g., increasing lapse rate, lower mortality rate) Formal stochastic approach Identify risk elements (e.g., interest, mortality, and default) Use a stochastic process to define a range of selected risk elements (e.g., Monte Carlo, Economic Scenario Generator) Run EV model over a range for selected risk elements Start with a deterministic model Stochastic Assumption = deterministic assumption x stochastically generated factor 21

23 Insurance and Actuarial Advisory Services European Embedded Value Principles

24 European Embedded Value Principles The CFO Forum launched European Embedded Value (EEV) Principles in 2003 EEV principles are aimed to provide transparent and consistent financial information to the investors across major European insurance companies 23

25 European Embedded Value Principles - continues 24

26 Embedded Value at Risk Concept TM : Difference between the mean EV value and the fifth percentile EV for each risk element (other levels of EV@Risk TM could also be used) Shows variance in EV over a range of economic and non-economic scenarios Quantifies impact to EV for each individual risk elements and in aggregate Demonstrates the correlation effect between different risk elements (i.e., sum of individual risk components is greater than when all the risks are run together) Allows management to determine a comfortable level of risk Requires stochastic EV modeling in order to determine different levels of EV@Risk TM 25

27 Embedded Value at Risk is similar to VAR Difference between mean and 5 th percentile Universal Life - Comparison of Stochastic Runs Embedded Value at 12/31/01 (m) $808 $806 $804 $802 $800 $798 5 th th Percentile Mean $ Stoch Lapse Only Stoch Mortality Only Stoch Lapse & Mortality Base Run 26

28 Insurance and Actuarial Advisory Services Stochastic EV Case Study For a Sample Universal Life Product

29 Ernst & Young Stochastic EV Case Study for ACLI Universal Life (UL) product from the Ernst & Young deterministic EV case study developed for the American Council of Life Insurer (ACLI) CFO Roundtable UL risk elements selected Interest Mortality Asset Default Used 100 iterations for each of the risk elements Too many = long run time Too few = reduce accuracy Experience and testing will help identify optimal number of runs 28

30 Generating UL Stochastic Risk Assumptions The stochastic assumptions are generated for each risk element as follows: Stochastic Assumption (other than stochastic interest) = Deterministic Assumption x Stochastic Generated Factor Distribution Assumption = Normal Distribution Stochastic Interest Economic Interest Rate scenarios were generated using Ernst & Young Economic Scenario Engine (ESE) Stochastic Mortality Mean =1, Standard Deviation = 5%, Max = 2, Min = 0.5 Stochastic Asset Default Mean =1, Standard Deviation = 100%, Max = 4, Min = 0 29

31 Results of Ernst & Young Universal Life Model Deterministic Financial Results GAAP Earnings for 2004 $ 179,232 GAAP Earnings $ 778,388* Embedded Value 12/31/2004 $ 1,324,758* Embedded Value without Target Surplus $ 561,872* * 9.00% 30

32 Results of Ernst & Young Universal Life Model Stochastic Mortality Results Universal Life Results - Stochastic Mortality Percentile Deterministic Mean EaRisk 5th 25th 50th 75th 95th GAAP Eanings 2004 $ 179,232 $ 178,764 $ 4,112 $ 174,652 $ 177,198 $ 178,679 $ 180,672 $ 182,711 GAAP Earnings $ 778,388 $ 777,142 $ 8,756 $ 768,387 $ 773,511 $ 777,888 $ 780,208 $ 785,330 Embedded Value $ 1,324,758 $ 1,321,195 $ 56,637 $ 1,264,557 $ 1,298,890 $ 1,319,582 $ 1,344,527 $ 1,377,207 Embedded Value w/o TS $ 561,872 $ 558,309 $ 56,637 $ 501,672 $ 536,004 $ 556,696 $ 581,641 $ 614,321 Observations: 2004 GAAP earnings at risk is about 2.3% versus the deterministic results while EV@Risk TM is about 4.3% EV@Risk TM is 10.1% when calculated without target surplus 31

33 Results of Ernst & Young Universal Life Model Stochastic Mortality Results Value of Existing Business Universal Life - Stochastic Mortality 1,600,000 1,400,000 1,200,000 1,000, , , , , (200,000) Stochastic Mortality Deterministic Statistics Result % of Mean Mean 558, % Median 556, % Minimum 484, % Maximum 653, % Deterministic 561, % Percentile Result % of Mean 5th 501, % 25th 536, % 50th 556, % 75th 581, % 95th 614, % 32

34 Results of Ernst & Young Universal Life Model Stochastic Interest Results Universal Life Results - Stochastic Interest Percentile Deterministic Mean EaRisk 5th 25th 50th 75th 95th GAAP Eanings 2004 $ 179,232 $ 179,014 $ 56 $ 178,958 $ 178,996 $ 179,018 $ 179,034 $ 179,077 GAAP Earnings $ 778,388 $ 777,616 $ 1,362 $ 776,254 $ 777,206 $ 777,605 $ 778,256 $ 779,014 Embedded Value $ 1,324,758 $ 1,324,485 $ 398,295 $ 926,190 $ 1,226,844 $ 1,384,108 $ 1,463,875 $ 1,538,830 Embedded Value w/o TS $ 561,872 $ 561,599 $ 398,295 $ 163,304 $ 463,958 $ 621,222 $ 700,990 $ 775,944 Observations: 2004 GAAP earnings at risk is about 0.0% versus the mean results while EV@Risk TM is about 30.1%. EV@Risk TM is 70.9% when calculated without target surplus 33

35 Results of Ernst & Young Universal Life Model Stochastic Interest Results Value of Existing Business Universal Life - Stochastic Interest 1,600,000 1,400,000 1,200,000 1,000, , , , , (200,000) Stochastic Interest Deterministic Statistics Result % of Mean Mean 561, % Median 621, % Minimum (142,731) % Maximum 879, % Deterministic 561, % Percentile Result % of Mean 5th 163, % 25th 463, % 50th 621, % 75th 700, % 95th 775, % 34

36 Results of Ernst & Young Universal Life Model Stochastic Default Results Universal Life Results - Stochastic Default Percentile Deterministic Mean EaRisk 5th 25th 50th 75th 95th GAAP Eanings 2004 $ 179,232 $ 178,931 $ 2,349 $ 176,582 $ 177,572 $ 178,645 $ 180,024 $ 182,448 GAAP Earnings $ 778,388 $ 777,722 $ 9,758 $ 767,964 $ 774,227 $ 777,732 $ 782,375 $ 786,094 Embedded Value $ 1,324,758 $ 1,320,977 $ 213,532 $ 1,107,445 $ 1,234,911 $ 1,314,282 $ 1,443,076 $ 1,554,656 Embedded Value w/o TS $ 561,872 $ 558,091 $ 213,532 $ 344,560 $ 472,025 $ 551,397 $ 680,191 $ 791,770 Observations: 2004 GAAP earnings at risk is about 1.3% versus the mean results while EV@Risk TM is about 16.2%. EV@Risk TM is 38.3% when calculated without target surplus 35

37 Results of Ernst & Young Universal Life Model Stochastic Default Results Value of Existing Business Universal Life - Stochastic Default 1,600,000 1,400,000 1,200,000 1,000, , , , , (200,000) Stochastic Default Deterministic Statistics Result % of Mean Mean 558, % Median 551, % Minimum 244, % Maximum 828, % Deterministic 561, % Percentile Result % of Mean 5th 344, % 25th 472, % 50th 551, % 75th 680, % 95th 791, % 36

38 Results of Ernst & Young Universal Life Model Stochastic All Results Universal Life Results - Stochastic All Percentile Deterministic Mean EaRisk 5th 25th 50th 75th 95th GAAP Eanings 2004 $ 179,232 $ 179,116 $ 5,254 $ 173,862 $ 177,220 $ 179,126 $ 180,976 $ 184,316 GAAP Earnings $ 778,388 $ 778,128 $ 14,723 $ 763,404 $ 772,176 $ 778,936 $ 785,524 $ 788,847 Embedded Value $ 1,324,758 $ 1,323,899 $ 369,460 $ 954,438 $ 1,137,281 $ 1,337,677 $ 1,514,501 $ 1,692,022 Embedded Value w/o TS $ 561,872 $ 561,013 $ 369,460 $ 191,553 $ 374,396 $ 574,791 $ 751,616 $ 929,136 Observations: 2004 GAAP earnings at risk is about 2.9% versus the mean results while EV@Risk TM is about 27.9%. EV@Risk TM is 65.9% when calculated without target surplus 37

39 Results of Ernst & Young Universal Life Model Stochastic All Results Value of Existing Business Universal Life - Stochastic All 1,600,000 1,400,000 1,200,000 1,000, , , , , (200,000) Statistics Result % of Mean Mean 561, % Median 574, % Minimum (115,201) % Maximum 1,070, % Deterministic 561, % Percentile Result % of Mean 5th 191, % 25th 374, % 50th 574, % 75th 751, % 95th 929, % Stochastic Interest/Mortality/Defaults Deterministic 38

40 Results of Ernst & Young Universal Life Model Summary of Stochastic Results 2004 Operating Earnings - Universal Life 2004 Embedded Value - Universal Life Percentile All Interest Mortality Defaults 5th 173, , , ,582 25th 177, , , ,572 50th 179, , , ,645 75th 180, , , ,024 95th 184, , , ,448 Mean 179, , , ,931 EaRisk 5, ,112 2,349 Correlation (1,264) Deterministic 179,232 Percentile All Interest Mortality Defaults 5th 954, ,190 1,264,557 1,107,445 25th 1,137,281 1,226,844 1,298,890 1,234,911 50th 1,337,677 1,384,108 1,319,582 1,314,282 75th 1,514,501 1,463,875 1,344,527 1,443,076 95th 1,692,022 1,538,830 1,377,207 1,554,656 Mean 1,323,899 1,324,485 1,321,195 1,320,977 EVaRisk 369, ,295 56, ,532 Correlation (299,004) Deterministic 1,324,758 39

41 Recap of Key Learning Can be used to create an effective decision support framework Competitive advantage Risk optimization Risk measurement and management across risks & product lines is doable Profitability and productivity metrics for distribution system Facilitate capital allocation Segue to IAS accounting and Risk Adjusted Return on Capital (RAROC) 40

42 Closing Remarks Stochastic valuation has arrived Stochastic EV (EEV), ERM Risk) RBC C3 Phase II, Economic Capital, Statutory Reserves, SOP Reserves Actuaries need to elevate their game to be ready Finance theory and conceptual frameworks Modeling & valuation tools Technology Quality assurance Analysis and presentation Communications 41

43 Questions! 42