2009 Seminar for the Appointed Actuary - PfAD Colloque pour l’actuaire désigné PED 2009 Seminar for the Appointed Actuary - PfAD Colloque pour l’actuaire désigné PED Canadian Institute of Actuaries Canadian Institute of Actuaries L’Institut canadien des actuaires L’Institut canadien des actuaires

Outline Changes in Standards of Practice New Educational Note 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 1

Changes in Standards of Practice Three changes in Exposure Draft Increase claims development’s upper limit to 20% Decrease investment return risk’s lower limit to 25 basis points Recognise of stochastic techniques to be within accepted actuarial practice in Canada Timing Comments due October 31 Effective December Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 2

Process Notice of Intent (NOI) released June 5 Comments from 13 individual CIA members 1 consulting firm with input from 9 members P&C RMCR Very detailed memo from Designated Group (DG) to membership (expected release September 2009) with all comments summarized Comments were important input for drafting of PfAD Ed Note 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 3

Proposed Changes Investment return risk – reduction to 25 bp –Focus shift away from current economic environment –Instead consider P&C insurers invested in high quality government bonds and fully immunized from mismatch risk Claims development – increase to 20% reflects feedback and situations like Alberta auto Stochastic techniques – accepted, but no proscribed methods or measurements (different from life SOP) 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 4

Educational Note - Purpose Provide guidance to actuaries in selection of MfAD for P&C insurers Also valuable to actuaries working with self-insurers and captives Flexible enough to allow for future developments in –IFRS –Use of stochastic techniques in valuation of policy liabilities 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 5

MfAD Refresher MfAD reflects degree of uncertainty of best estimate assumption SOP ( ): deviations of actual from expected experience may result from one or more of the following: Error of estimation – past experience data may be insufficient or unreliable – future conditions may differ from conditions that generated past experience Deterioration or improvement of expected experience as result of influences which actuary does not anticipate Statistical fluctuation MfAD not expected to be so high that probability of unfavourable development is < 1% or 5% (i.e., scenarios under DCAT), purpose of capital 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 6

Cost of Capital Methods Focus of Ed Note is MfAD derived from deterministic or stochastic analyses Cost of capital approach – another alternative approach to determining MfAD PCFRC believes that cost of capital methods are important area of future development Many unresolved issues around the use of such methods What capital: economic capital, regulatory required capital, rating agency capital, capital used for pricing, other How to determine cost, how frequently is cost updated, should it vary by contract or claim type or vary by duration of contract or claim See April 15, 2009 report Measurement of Liabilities for Insurance Contracts: Current Estimate and Risk Margins, prepared by Risk Margin Working Group (RMWG) of IAA for further details (IAA Risk Margin Report) 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 7

Organization of Educational Note 1.Introduction 2.Terminology 3.Desirable risk margin characteristics 4.Three categories of margins for adverse deviations 5.Explicit assumptions - margins for adverse deviations using a deterministic analysis 6.Relevant statistical concepts 7.Stochastic techniques 8.Three P&C product examples 9.Quantile approaches 10.Comparison of risk margin methods 11.Documentation and reporting 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 8

Information Sources IAA Risk Margin Report Feedback from CIA members received in response to MfAD NOI Checklists for MfAD from Appointed Actuary reports 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 9

Desirable Risk Margin Characteristics - IAIS Addressed in Second Liabilities Paper (2006, paragraph 57) Does not prescribe any one method “ … methodology for calculating the risk margin should share certain characteristics.” The less that is known about current estimate and its trend, the higher the margin Risks with low frequency and high severity will have higher risk margins than risks with high frequency and low severity For similar risks, contracts that persist over longer timeframe will have higher risk margins than those of shorter duration Risks with wide probability distribution will have higher risk margins than risks with narrower distribution To extent that emerging experience reduces uncertainty, risk margins will decrease, and vice versa IASB Identified same properties as desirable 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 10

IAA Risk Margin Requirements 1.Apply consistent methodology for entire lifetime of contract 2.Use assumptions consistent with those used in determination of corresponding current estimates 3.Be determined in manner consistent with sound insurance pricing practices 4.Vary by product (class of business) based on risk differences between products 5.Be easy to calculate 6.Be consistently determined between reporting periods for each entity that is, risk margin varies from period to period only to extent that there are real changes in risk 7.Be consistently determined between entities at each reporting date, that is, two entities with similar business should produce similar risk margins using the methodology 8.Facilitate disclosure of information useful to stakeholders 9.Provide information that is useful to users of financial statements 10.Be consistent with regulatory solvency and other objectives 11.Be consistent with IASB objectives 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 11

CIA Risk Margin Characteristics Characteristics cited by both IAIS and IAA are consistent with characteristics noted in CIA SOP Section and : A larger margin for adverse deviations (compared to the best estimate assumption) is appropriate if the actuary has less confidence in the best estimate assumption, an approximation with less precision is being used, the event assumed is farther in the future, the potential consequence of the event assumed is more severe, or the occurrence of the event assumed is more subject to statistical fluctuation. A smaller margin for adverse deviations is appropriate if the opposite is true Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 12

CIA SOP - MfAD Three categories of MfAD Claims development – % of claim liabilities excluding PfAD Recovery from reinsurance ceded – % of amount deducted on account of reinsurance ceded excluding PfAD Investment return rates – deduction from expected investment return rate per year SOP note that according to how considerations so vary, selected margins should vary between: Premium liabilities and claim liabilities Lines of business Accident years, policy years, or underwriting years 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 13

Explicit Assumptions MFAD Using a Deterministic Analysis 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 New concept – not excessive May be circumstances for MfAD above high margin SOP: “for unusually high uncertainty or if the resulting provision for adverse deviations is unreasonably low because the margin for adverse deviations is expressed as a percentage and the best estimate is unusually low.” (See attachments to presentation material – new SOP and consideration tables) Page 14

Considerations – Deterministic Approach (Explicit Assumptions) Ed Note contains numerous tables with considerations for explicit assumptions approach Not exhaustive list, but representative of key issues to consider In some situations, considerations may not be relevant or applicable May be additional unique considerations specific to organization For each consideration, spectrum between situation necessitating low margin or high margin For many insurers, particular circumstances for any one consideration may dictate selection of margin between low and high values set out in SOP When faced with situation in which some considerations indicate low margin and others indicate high margin, actuary would use professional judgment to determine priority of considerations and resulting final margin 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 15

MfAD for Investment Return Rates Addresses several different types of risk: Mismatch risk between payment of claims and availability of liquid assets Error in estimating the payment pattern of future claims Asset risk including credit/default risk and liquidity risk In economic environment of low interest rates, mismatch risk and credit/default risk still remain Following CIA SOP, could derive discount rate adjusted by MfAD less than 0% In practice, actuaries may limit discount rate to 0% in such situations Two examples of alternative formula-based approaches presented in Ed Note 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 16

Relevant Statistical Concepts Not intent to present detailed description of statistics Expected that actuaries using stochastic methods for determination of MfAD have expertise in fundamentals of statistical modeling Key concepts: Risk distribution Normal distribution Standard deviation Coefficient of variation (CV) Skewness 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 17

Stochastic Techniques Stochastic simulations can be powerful techniques for quantifying risk exposures underlying P&C insurance policies Users require full understanding of stochastic risk modelling for successful implementation and rational interpretation Expect actuaries who use stochastic approaches to determine MfAD would generally use stochastic methods to determine policy liabilities Beyond scope of Ed Note to address stochastic modeling techniques except as applicable to determination of PfAD 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 18

Stochastic Techniques (cont’d) Refer reader to CIA’s August 2001 Research Paper Use of Stochastic Techniques to Value Actuarial Liabilities under Canadian GAAP prepared by Working Group on the use of Stochastic Techniques (Working Group) of CLIFR Specifically, recommend review of: Section 3. When to Use Stochastic Simulation Methods for Actuarial Liability Valuation Section 4. General Overview of the Stochastic Valuation Method for Actuarial Liability Valuation Section 6.5. Correlation Section 7. Practical Issues 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 19

Stochastic Techniques (cont’d) Reference to Mack and bootsrapping – no details – refer to literature Actuaries using stochastic techniques for developing MfAD would also take into account considerations presented in Section 5 of Ed Note 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 20

Stochastic Techniques (cont’d) Actuaries considering change from deterministic approach to stochastic approach would engage in discussion with insurer’s management and auditors to determine if such change represented potential change in accounting policy  important consideration would be materiality of any resulting change Since stochastic techniques may be more subject to variability from valuation date to valuation date, ongoing communication between actuary and insurer’s management and auditors may be required 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 21

Stochastic Techniques (cont’d) When using stochastic models, important to recognize that MfADs do not cover inherent or statistical volatility arising from particular model  expected that large and small insurers would generate similar MfAD when using same model MfADs do, however, cover uncertainty in whether actuary has the “right” model or “right” parameters Thus, actuary working with large volumes of data or more years of experience will likely have more confidence that the selected model is “correct” and resulting margins will likely be lower for larger volume or more established data than for smaller volume or less reliable data 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 22

Mandating Assumptions for Stochastic Techniques In the CIA 2001 Research Paper, the Working Group struggled with issue Impractical, inappropriate, and unmanageable for following reasons: –Would entail significant and time-consuming testing and review of experience data from across the industry –Would potentially require large number of possible assumptions or variations in assumptions to be covered –Would be difficult to anticipate all unique company circumstances that can legitimately affect valuation results and therefore cause prescription to be inappropriate –Ranges would need periodic updating to reflect emerging experience –Would undermine integrity and responsibility of Appointed Actuary Considerations equally applicable to P&C insurers today Actuary reminded of his/her responsibilities 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 23

Sample Products – Discussion Stochastic modelling will typically be beneficial for products with skewed cost distributions with low frequency of occurrence but high severity and/or material variability in the cost distribution Examples: Stop loss reinsurance Catastrophic P&C insurance risks Credit, warranty, and mortgage guarantee insurance Long-tail lines of business such as professional liability 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 24

Three P&C Product Examples 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 25

Quantile Approaches Establishing which statistical measurement is most appropriate for determination of MfADs based on stochastic techniques is important decision Difference between selected measurement and mean result establishes dollar PfAD Following approaches reviewed: Multiples of standard deviation Percentile or confidence levels, Value at Risk or VaR CTE, Tail Value at Risk or TVaR Currently, no generally accepted method (from regulatory, accounting, or actuarial perspective) for determining appropriate quantile for purpose of determining risk margins 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 26

Quantile Approaches (cont’d) Multiples of Standard Deviation Simplicity Practicality Percentile or Confidence Levels Most commonly used Australia, Singapore, South Africa – 75% confidence level requirement (Australia, at least equal to ½ std deviation) 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 27

Quantile Approaches (cont’d) CTE The 2001 Research Paper states: Setting the liabilities for life insurers in excess of CTE(80%) would not normally be an acceptable practice as the resulting coverage would be excessive and inconsistent with GAAP. Provision for more catastrophic, implausible or unknown events is done through required capital, which would normally be established at a much higher CTE percentage Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 28

Quantile Approaches (cont’d) CTE (cont’d) Unlike life insurance, no specific statistical measurement or percentile mandated by CIA SOP for P&C insurance Examples prepared by RMWG of IAA, indicate that range of CTE(60%) to CTE(80%) is likely too high for many traditional P&C lines of insurance 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 29

Three P&C Product Examples 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 30

Evaluation of Quantile Methods In evaluating various methods for developing risk margins, IAA Risk Margin Report suggests that two aspects of insurance liabilities be considered to measure risk margin: Time – the rate at which risk is released over time (i.e., settlement pattern) Shape – the risk distribution of possible outcomes around the mean value, at the reporting date, over a specified time horizon 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 31

Comparison of Risk Margin Methods Summary of observations regarding the various methods Comparison of methods from quantitative perspective Comparison of methods from qualitative perspective Qualitative review includes comparison of each method to desirable characteristics of risk margins identified by IAIS and IAA 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 32

Summary Observations In quantile family of methods, CTE approaches theoretically more sound than confidence level approaches Differences can be significant for products with more skewed risk distributions Regulatory oversight or actuarial practice could apply higher confidence levels for products whose risk distributions are more highly skewed Explicit assumptions best considered as useful approximations for implementing a quantile method Consistency among insurance products and between insurance and other industries is challenging using purely explicit assumption approach 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 33

Quantitative Comparison 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 34

Qualitative Comparison – Compliance with 5 IAIS Characteristics The less that is known about current estimate and its trend, the higher the margin Risks with low frequency and high severity will have higher risk margins than risks with high frequency and low severity For similar risks, contracts that persist over longer timeframe will have higher risk margins than those of shorter duration Risks with wide probability distribution will have higher risk margins than risks with narrower distribution To extent that emerging experience reduces uncertainty, risk margins will decrease, and vice versa 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 35

Explicit Assumptions Although could be constructed in manner to address characteristics, do not necessarily satisfy any As an implementation approach, explicit assumptions, selected by product, could be made to approximate percentile method If approximation sufficiently close, explicit assumption approach would satisfy characteristics to same extent as method it approximates 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 36

Quantile Methods All quantile methods fail characteristic (c-2) Confidence level method does not necessarily satisfy characteristics (a), (b), (d) or (e) Highly skewed distributions (e.g., Product C) can result in negative risk margins Examples show that as distributions become more dispersed and more skewed, risk margins implied by a fixed confidence level include fewer standard deviations Violates spirit of characteristics (a), (b), (d), and (e) throughout and letter of those in the extreme CTE and methods based on multiples of standard deviation generally satisfy characteristics (a), (b), (d), and (e) better than do confidence level method 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 37

Table 10.2 Comparison of Risk Margins with IAA Desirable Characteristics 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 38

Documentation and Reporting CIA SOP: “Documentation is an integral part of work that affects the application of nearly all standards.... Appropriate documentation describes the course of the work and the actuary’s compliance with accepted actuarial practice.” Actuary would document his/her process for determining MfADs Important regardless of whether actuary uses explicit assumptions or stochastic techniques Documentation for both explicit assumptions and stochastic techniques would include support for key decisions 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 39

Documentation and Reporting (cont’d) With respect to reporting, normally in user’s interest to be aware of MfADs selected by actuary Accordingly, reasonable that actuary would usually at least consider some disclosure regarding MfADs within actuarial work product for both internal user and external user reports Must also take into account complexity of concept, potential importance of concept to user, as well as sophistication of user who will be receiving work product Also important for actuary to communicate with insurer’s auditors, particularly regarding any significant change either in value of MfADs or process for determining such values 2009 Seminar for the Appointed Actuary Colloque pour l’actuaire désigné Seminar for the Appointed Actuary Colloque pour l’actuaire désigné 2009 Page 40