Professional Documents
Culture Documents
Economic Capital - Overview and Recent Trends: ERM Symposium April 24, 2006
Economic Capital - Overview and Recent Trends: ERM Symposium April 24, 2006
Overview
Aggregation
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
ERM Governance and Economic Capital Governance policies, procedures, tools, resources ERM Framework decision criteria Economic Capital metrics Analytics tools
1. 2. 3.
An appropriate governance structure should be in place to articulate the policies and procedures for managing the enterprise, supported by appropriate tools and resources An ERM framework should specify the firms decision criteria for risk taking and capital utilization The analytics should generate the appropriate financial measures required for making business decisions
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
Basel I requires holding appropriate levels of capital for different financial risks S&P AAA rating requires banks to hold capital at the 99.90% level Basel II expands the concept to operational risks Quarterly reporting of operational risk exposure Insurance companies have picked up the EC concept only in the last several years Rating agencies are starting to give credit for internal models Regulatory changes are accelerating pace of change Larger companies are setting up proprietary stochastic models The European CRO Forum is in the process of recommending standards for the acceptance of internal models for compliance with the new Solvency II capital standards
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
When calculating EC, three alternative approaches are commonly used to measure risk: Risk of ruin, VAR and TVAR
Financial results from a series of Stochastic Scenarios
+
1) TVAR captures the full extent of losses in the event of ruin
2) VAR quantifies the capital required to withstand losses at a particular probability level
Required Economic Capital (REC) = sufficient surplus capital to cover potential losses, at a given risk tolerance level
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
When calculating EC, financial services companies use a range of different confidence levels
Most insurers use expected shortfall risk (CTE or TVAR) or risk of ruin Variable annuity RBC is based on CTE90 level Most banks use VAR Choice of confidence level and implied rating: Most European insurers are using one-year confidence levels ranging from 99.5% to 99.99% European regulators appear to be converging towards a one-year 99.5% confidence level for Solvency II Confidence levels are typically linked to a target risk appetite and financial strength rating Where longer time horizons are used, a lower multi-year confidence level can be justified (e.g., AA over five years vs. AA over one year) Moodys and S&P have suggested using 99th percentile for AA financial strength rating Fitch is using a 98.2 CTE level for AA rating
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
Definitions of EC: Best estimate liability approach Economic Capital is The level of assets, in addition to the Best Estimate Liability, required to pay future policyholder benefits at the chosen Security Factor Economic Capital covers the volatility in: The runoff of existing business The future business (pricing risk) Best Estimate Liability is The best estimate projection of non-investment cash flows Discounted at the asset returns under the best estimate economic scenario Security Factor is Based on a risk of ruin level that is consistent with the companys financial strength rating Commonly used in North America
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
Statutory reserve
Economic Capital is Measured as the difference in market-consistent net assets between normal conditions and stressed conditions The stress tests applied are each calibrated to a probability level over a one-year time horizon, consistent with the companys Normal conditions Stressed conditions financial strength rating Separate stresses are applied to cover a variety of market, credit and insurance risks that might occur over the projected time horizon Results are aggregated using a correlation matrix approach
Net assets MV Assets MCV Liabs MV Assets Net assets MCV Liabs
Net assets
Net assets
Economic Capital
Normal
Stressed
Risk Aggregation
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
One year
Statutory
Risk of ruin
Market Credit
Additive
n years
GAAP
VAR Insurance
Run-off of portfolio
Economic
TVAR or CTE
Operational Liquidity
Six key decisions need to be made, and the approach taken should reflect the nature of the company and managements objectives
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
P/C diversification effect Aggregated P/C Business Cross Sector diversification effect Aggregated Total
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
10
Basel II Solvency II/European CRO Forum OSFI regulation for segregated funds (CAN) C-3 Phase II: RBC for variable annuities (VA) U.S. Proposed stochastic reserves for VAs and UL products U.S. GAAP SOP 03-1: explicit reserves for guarantees U.S. General need to develop risk profiles and perform hedging analysis Measuring Economic Value/Market Consistent Embedded Value (MCEV) Measuring exposure to catastrophic events Demands and increasing scrutiny by rating agencies/regulators Calculating EC is becoming an important tool for insurers in guiding risk-based decision making
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
11
The European CRO Forum is establishing guidelines for admissibility of internal EC models for Solvency I
European Chief Risk Officer Forum is establishing guidelines for calculation of EC and diversification of risk Principles for Regulatory Admissibility of Internal Models (June 2005) Solvency Capital should be set to ensure a standardized likelihood of economic loss to policyholders Internal models should be based on adverse movement in Economic Value of (Assets Liabilities), calibrated to a target annual level of 99.5% probability of solvency All material risks affecting the balance sheet should be modeled Internal risk models should be fully implemented inside the company, and reviewed (at least) annually The CRO Forum advocates the admissibility of diversification benefits In December 2005, the CRO Forum published a subsequent report discussing suggested solutions to major issues in Solvency II EC is to be based on a market-consistent approach
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
12
All major US rating agencies are developing or enhancing their capital adequacy models Standard & Poors (S&P) is currently enhancing its capital adequacy model to reflect changes brought about by the proposed CTE 90 capital for VA RBC Required capital will only be based on the stochastic model (not Standard Scenario) CTE level will be calibrated to rating category Partial credit for hedging, no phase-in EC will be incorporated into S&Ps overall ERM framework by the fall of 2006 Fitch is also revising their VA capital adequacy model to incorporate CTE methodology E.g., AA Fitch rating will require capital based on CTE(98.22) level Partial credit for hedging will be provided AM Best and Moodys are also in the process of enhancing their capital adequacy models
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
13
Leading-edge companies maximize value by relating decisions on the risks they take to decisions on the capital they use
Value Creation
Return on Risk
Value Management
Capital Costs
Economic Capital
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
14
Overview
Aggregation
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
15
AGGREGATION
Aggregation approaches can significantly reduce capital requirements, but must be demonstrably robust Approach 1. Approximate analytical approach using standalone economic capital and correlations 2. Use marginal distribution functions, correlations and aggregate using Monte Carlo simulation 3. Build/approximate joint distribution function and solve analytically/by numerical integration Advantages Quick to execute High degree of accuracy Can attribute diversification easily Easy to understand More accurate Disadvantages Accuracy of the approximation is poorer the more varied the risk distributions Difficult to build Time consuming to run many simulations Difficult to disaggregate Most accurate Almost impossible to build joint distribution in practice Cannot attribute diversification benefits back to risks and businesses
Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
16
AGGREGATION
One aggregation approach is to combine risks based on statistical measures of their variances and covariances
Asset Class 1
Product Class 1
Asset Class 2
Covariance Matrix
Product Class 2
Asset Class 3
Product Class 3
Fundamental construct Each risk is reduced to a statistical distribution of financial outcomes Linkages between risks described via a covariance matrix Aggregate result obtained either analytically or via Monte Carlo simulation
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
17
AGGREGATION
One fundamental approach is to combine risks based on statistical measures of their variances and covariances
The key to this approach is the specification of a correlation matrix Correlations based on available empirical evidence, informed judgment, conservatism Covariances usually assumed to be constant across distribution, facilitating simplified analytic solutions
Correlation Matrix
18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
0 0.2 0.4 0.6 0.8 1
In its most simplistic implementation: Each risk is assumed to be normally distributed Correlation between risks is assumed to be linear Then the Economic Capital for each risk can be combined by the formula Aggregate Capital = i j ij Ci C j
1 2 3 4 5 6 7 8
-0.2
9 10 11 12 13 14 15 16 17 18
Where Ci is the economic capital required for the ith risk at the required confidence level and ij is the correlation coefficient between ith and jth risk While convenient analytic solutions are desirable, they may require simplifying assumptions that are too approximate In these situations, statistical aggregation can still be accomplished, but may require Monte Carlo simulation
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
18
AGGREGATION
An alternative approach is to develop a structural risk model, and then link all asset and product behavior to it
Asset Class 1
Product Class 1
Asset Class 2
Product Class 2
Asset Class 3
Product Class 3
Fundamental approach Each systemic risk is described by a stochastic equation in the ESG; linkages between systemic risks are also built into the equations ESG generates scenarios, representing plausible sets of future conditions All assets and products are subjected to common set of scenarios, with their behavior linked to systemic variables Aggregate results can be obtained additively by scenario
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
19
AGGREGATION
Typically at the product level and up, financial behavior of the product and associated assets are linked to ESG variables Economic Scenario Generator
Inflation Interest Rates Credit Costs Currency Exchange GDP
Company Strategy
Investments Products Capital/Structure Reinsurance/Hedging Operating
Probability
20
AGGREGATION
Both approaches require correlation assumptions; differences are in what is correlated and where in the analysis
Method
Statistical Correlation Approach Structural Scenario Approach
Risk Driver
Correlation assumptions applied within BU model Related by SDE,* including correlations
Risk
Statistics describe risk
BU Capital
Calculated on BU tail statistics
Aggregation
Apply aggregation equation with correlation adjustment to BU capital Sum BU results to get aggregated tail scenario value
Complexity of interactions between risks can influence choice of method: Correlation matrix captures relationship between two risks, but that may not be sufficient for more complex risk dynamics Example: multi-trigger products with payoffs that are functions of event risk, financial risk and policyholder behavior Statistical correlation approach requires correlation factors to be used as inputs in two places: Risk drivers within the BU economic capital calculation BUs at the aggregate economic capital calculation Correlations between risk drivers (systemic variables) represent only one factor in SDE; correlations between BUs emerge from the scenarios and can be evaluated against experience
*Stochastic differential equations.
2006 Towers Perrin Proprietary and Confidential Not for use or disclosure outside Towers Perrin and its clients
21
View of Insurer-developed Economic Capital Models Fitchs Economic Capital Modeling Decisions How to assess Economic Capital Models
Fitch Ratings
Management Review
Strategic vision Appetite for risk Credibility of track record for meeting expectations Controls and risk management capabilities Depth, breadth and succession plans Key executives accomplishments
Operational Review
Distribution mix Market position Franchise value Expense efficiency Underwriting and pricing experience Product mix Administrative / technology capabilities
Financial Review
Level of orderly competition Industry competitive advantage Barriers to entry / exit Regulatory, legal and accounting environment
Operating performance Capital adequacy Investment quality Asset / liability and liquidity management Financial flexibility
FITCHS VIEW OF ECONOMIC CAPITAL MODELS The Long and Winding Road
Principal-based reserving
Solvency II
C3-II ECR
C3-I
Fitch Ratings
Progress
U.S.
Methodology
Factor-based
Transparency
Calibration Organization Assumptions
Resources
Tools
Fitch Ratings April 11, 2006 4
Correlation
Risks
DATA
Fitch Ratings April 11, 2006 5
Fitch Ratings
Simpler
Less than vertical is unresponsive > Arbitrary adjustments > Incomplete scope
Complexity
Past vertical is inefficient > Too much data > Too much time > Lack of comparability
April 11, 2006 7
Fitch Ratings
Need consistency
Between products Between sectors Between countries
Fitch Ratings
Role of Capital
Define the purpose of capital when establishing a capital standard Time period over which a company will assume additional risk producing positions Sources of risk producing exposure
Limit the risk that assets are insufficient to pay claims over the lifetime of policyholder obligations. Include run-off of current obligations and one-year of new business on relevant products. Includes ALM (Market, Interest), Credit, Reserve, Underwriting and Natural Catastrophe Items generating risk exposure will be allowed to fluctuate over the run-off of liabilities - annual time step with maximum of 30 years. The model will focus on Conditional Tail Expectation (CTE) but also produce additional risk measures such as Value of Risk (VaR), Comparables. Calibrated primarily using bond default data along with rule of thumb measures.
April 11, 2006 9
Time period over which risk exposure sources will vary and generate volatility
Risk Measures
Statistical measure to determine relative risk from risk distributions generated by allowing risk exposure sources to vary over the risk exposure horizon Methodology for tuning results of the capital model to meaningful capital indications
Model Calibration
Fitch Ratings
Credit Risk
Incorporates defaults, migration and spread volatility Use common market indices to establish parameters for asset type and quality Over 50 asset buckets. Stochastically model reinsurer default risk
Reserve Risk
Use Mack Method Utilize several checks to ensure data integrity If checks fail, use Industry assumptions leveraged by Underwriting Risk parameters. Incorporates reserve adequacy analysis
Underwriting Risk
Use a collective risk model of frequency and severity of losses. Relies on ELR, Attachments, Limits Factors one year of new business.
Catastrophe Risk
P/C: Use AIR (Catrader) software. Life: Use Lee-Carter methodology trend, volatility, shocks
Aggregator
Each Company will potentially have unique risk curves
Consistent economic scenario set Similar Cat event set Correlated random numbers
Fitch Ratings
10
R E S E R VE R IS K
BAD
BAD
US - Al l Ot h e r US - Au t o P D US - Sp e c i a l P r o p US - P r o d Li a b - C M US - P r od Li a b - Oc c US - R e i ns B US - R e i ns A&C US - In t e r n a t i on a l US - Ot h Li a b - C M US - Ot h Li a b - Oc c US - Sp e c Li a b US - M e d M a l - C M US - M e dM a l - Oc c US - C M P US - WC
GOOD
US - C AL US - P P A Li a b US - H/ O &F/ O
A s s e t R e t u rn
0% 2 5% 50 % 75% 10 0 %
ESTIMATED RISK CHARGES Risk Driver Exposure Asset 22,528,280 Underwriting 14,463,360 Net Reserve 13,695,190 Reinsurance Credit 4,321,680 Catastrophe 3,452,285 A&E 4,134,638 UNDIVERSIFIED REQUIRED CAPITAL SAMPLE Value-at-Risk 99.90 99.00 97.50 12,535,283 4,540,239 2,762,578
CATASTROPHE RISK
Est. Req Cap 2,010,000 1,460,000 3,100,000 40,000 610,000 1,160,000 8,380,000 Asset Risk
10 0 %
75%
50 %
Diversification Liability Risk Benefits Asset v. Liability Total Benefit 1,156,680 7,223,320 Req Cap:
2 5%
Fitch Ratings
11
BBB-
BBB BBB+
A-
A+
AA-
AA
AA+
AAA
Avoid arbitrariness
BB
BBB
Fitch Ratings
13
Fitch Ratings
14
Capital Adequacy
Regulatory Requirements
Creates Consistent Principles
Fitch Ratings
15
Basic Precepts
Sophistication should match the complexity of the enterprise Platform needs to be fully vetted
Inputs Models Outputs
INPUT DESIGN IMPLEMENTATION
MODELING METHODS
Transparency
A Standardized Platform breaks down a complicated process, involving many resources over several years Garbage in, Garbage out
OUTPUT CRITERIA
Fitch Ratings
16
Model Methodology
Discount rate (portfolio rate, risk free?) Source of Generated Economic Scenarios and Parameterization Available Capital Definition
Available capital
Depends on accounting regime - important not to double count some elements included as part of required capital
Aggregation of results
How to sum results between entities and across regions
Calculation of correlations
What experience was used? Where is the expert opinion?
Fitch Ratings April 11, 2006 18
Fitch Ratings
19
10
Summary
Infancy stage of Economic Capital Models There is general consensus but no consistency Substantial challenges
Resources Understanding Pay back
Necessity
Insurance is the risk-assuming business Statistics: The only science that enables different experts using the same figures to draw different conclusions. Evan Esar, American Humorist (1899 1995)
Fitch Ratings
20
11
Framework Cycle
Risk Focused Examinations Off-Site Risk-Focused Financial Analysis Internal/External Changes Priority System (CARRMEL) Supervisory Plan Insurer Profile Summary
CARRMEL
Priority System Based on Ratios and Analysis to Measure:
Capital Adequacy Asset Quality Reinsurance Reserves Management Earnings Liquidity
Risk Assessment Results Financial Analysis Handbook Process Ratio Analysis (IRIS, FAST, Internal Ratios) Actuarial Analysis Update with internal/external changes
Risk-Based Capital
1. 2. 3. 4. 5. 1. 2. 3. 4. 5. 6. Asset Risk-Affiliates Asset Risk-Other Insurance Risk Interest Rate Risk and Health Credit Risk Business Risk > 250% > 200% to <250% > 150% to < 200% > 100% to < 150% > 70% to < 100% < 70% No action level Trend test level Company action level Regulatory action level Authorized control level Mandatory control level
RBC Checklist
Determine whether concerns exist regarding the insurers risk-based capital position. Review the Five-Year Historical Data Schedule, Risk-Based Capital Analysis.
Is the ratio of Total Adjusted Capital divided by Authorized Control Level Risk-Based Capital less than or equal to 250%?
Yes
No
Has there been a significant change (XX%) in the ratio of Total Adjusted Capital divided by Authorized Control Level Risk-Based Capital from the prior year?
No No
Has the Authorized Control Level Risk-Based Capital increased by more than YY% from the prior year?
If required by statute, obtain a copy of the insurers RBC Plan and review. Review all items that generate significant RBC charges to ensure that these factors have been considered in the overall analysis of the insurer.
Line (23) was added to LR043 Exemption Test: Cash Flow Testing for C-3. Cash Flow Testing for C-3 RBC
LR043 Exemption Test: a company may be required to perform cash flow testing to determine its RBC requirement. A practical method of measuring the degree asset/liability mismatch exists
https://www.actuary.org/life/phase2.asp
Principle-based Approach
Captures all of the material risks, benefits and guarantees in the contract using basic risk analysis and risk management techniques
Provides an appropriate level of conservatism that is consistent with the objectives of statutory reporting Only requires a modeling and or stochastic approach be used when necessary to properly capture the risks of the contract For some products, a deterministic, single scenario approach is adequate to capture the risks of the contract
For products with, material tail risk of a high level of uncertainty in cash flows arising from policy owner options in the contract, a stochastic approach may be necessary A stochastic approach does not require that all assumptions be stochastically modeled
LTC (cont.)
A ratio of premiums to claims is calculated for the current year and prior year. For the loss ratios to be used, the current year and prior year premiums must be greater than zero. In addition, the current and prior year claims incurred must be greater than or equal to zero. Otherwise, the current year incurred claims are used instead of the two-year loss ratio multiplied by the current year premiums.
LTC (cont.)
The result of the claims ratio or the current years incurred claims is multiplied by tiered factors as follows: If Current Year LTC Premium Positive: Negative or Zero: Long-Term Pre-Tax Pre-Tax Care Claims Claim Factor Claim Factor First 35 Million Over 35 Million 0.385 0.569 0.123 0.185
The LTC premium charge and claim charge would be combined and then reduced by 35% on the tax adjustment page.
The non-cancelable long-term care business should be excluded from the new Long-Term Care page that was added for 2005.
Non-cancelable long-term care premium receives a pre-tax RBC charge of 15.4%. The after-tax factor is approximately 10%.
Annual Statement
Five-Year Historical Data (publicly available) Current and four previous years Risk-Based Capital Analysis
Authorized control level risk-based capital Total adjusted capital Trend of ratios indicates stability, growth, or decline
Independence
Should be defined in regulation or qualification standard SEC and NAIC definitions may differ