Discussion of Unpaid Claim Estimate Standard

1. Discussion of Unpaid Claim Estimate Standard Raji Bhagavatula Mary Frances Miller Jason Russ November 13, 2006 CAS Annual Meeting San Francisco, CA

2. Background • Guidance in estimating unpaid claims, applicable to most work, not just opinion work • Exposure draft released in March • Comment period through June • 32 comments received • In process of reviewing comments and amending proposed standard • Have not finalized revisions to standard – thoughts expressed today are still being considered

3. Main Points of Discussion • Title of standard • Scope of standard • Consideration of purpose and use • Handling of constraints • Identification of scope of work • Understanding of nature of claims • Methods and models • Assumptions • Uncertainty • Communications and Disclosures

4. Title of Standard • “Unpaid Claim Estimates,” not “Reserves” • Comments received argued both ways • “Reserves” is a line item on a US financial statement, not the estimate of future payments itself, but improper usage common in US • “Unpaid claim estimates” is more accurate, more generic and better accepted globally

5. Scope of Standard • Includes evaluation of self-insureds, not just insurance entities • Excludes ratemaking purposes • Some comments expressed disagreement, but deemed necessary to limit scope • Excludes oral communication • Some comments expressed disagreement, but practical necessity • Excludes any actions taken after production of unpaid claim estimate

6. Consideration of Purpose and Use • Should identify in communication • Has an impact on many items, such as • Selection of intended measure • Choice of method or model • Whether uncertainty needs to be measured • Presentation of results • Documentation and disclosures • Once the intended measure is decided, the purpose and use should not bias the estimate relative to that intended measure

7. Handling of Constraints • Sometimes constraints exist in the performance of an actuarial analysis, such as those due to limited data, staff or time. Where the actuary believes that such constraints create a significant risk that a more in-depth analysis would produce a materially different result, the actuary should notify the principal of that risk and attempt to discuss the constraints on the analysis with the principal.

8. Scope of Work • Must identify what you are estimating: • Intended measure • Gross or net • Collectibility risk • Unpaid claim adjustment expenses • Cohorts of claims • Other items needed to sufficiently describe scope

9. Intended Measure • Want it to be clear what the actuary is estimating • “Best Estimate” or “Actuarial Estimate” not sufficient • Best estimate of what? Mean? Mode? Low or High? Risk margin included? Etc. • Original exposure draft introduced term “Actuarial Central Estimate” to use as a default • Many comments concerning whether a default should exist and if so what it should be • Concerns that ACE as default advocates something biased lower than desired • Discussing option of removing default – require disclosure of intended measure in all cases

10. Actuarial Central Estimate • Defined in exposure draft – “An estimate that represents a mean excluding remote or speculative outcomes that, in the actuary’s professional judgment, is neither optimistic nor pessimistic. An actuarial central estimate may or may not be the result of the use of a probability distribution or a statistical analysis. This definition is intended to clarify the concept rather than assign a precise statistical measure, as commonly used actuarial methods typically do not result in a statistical mean.” • Intention was for definition to cover estimates produced by “typical” actuarial methods such as loss development • Never intention to imply this is estimate actuary should produce • Considering retaining term with reworded definition, but not as default – gives actuary option to use term rather than re-define in own communication.

11. Understanding Nature of Claims • Should have an understanding appropriate to the analysis • Will not include a long list of considerations as existed in original Principles • Such educational material is better suited for practice notes than for standards or principles

12. Methods and Models • Consider what is appropriate, even if same method as used before • Original exposure draft only mentioned “methods”, adding “models” to language in response to comments • Significant discussion on language concerning use of multiple methods • Want to encourage use of more than one method • Under what circumstances can the actuary use just one method?

13. Assumptions • Consider what is reasonable and appropriate given intended measure • Assumptions should “have no known significant bias to underestimation or overestimation of the identified intended measure” • Bias with regard to an expected value estimate would not necessarily be bias with regard to a measure intended to be higher or lower than an expected value estimate • Consider sensitivity of result to alternative assumptions • May rely on assumptions provided by principal, with appropriate disclosure

14. Uncertainty • Some comments received stating actuaries should be encouraged or required to measure uncertainty • This may or may not be appropriate given the situation • Language will require consideration of uncertainty, but not necessarily measurement

15. Communications and Disclosures • General guidance in other ASOPs • ASOP 9, ASOP 23, ASOP 41 • Additional communication and disclosure requirements specific to this standard • This does “raise the bar”, particularly for company actuaries

16. Communications and Disclosures • Clearly convey the intended purpose or use • More than one intended use? • Any compromises in order to produce a single work product for multiple intended uses? • Any resource constraint issues? • Clearly define the scope of the estimate • Include discussion of the uncertainties in the estimated claim liability

17. Communications and Disclosures - Dates • Actuarial communication should include • Accounting date • Valuation date • Review date may have to be included in certain circumstances As an example, “This unpaid claim estimate as of December 31, 2005 was based on data evaluated as of November 30, 2005 and additional information provided to me through January 17, 2006”

18. Communications and Disclosures – Significant Events and Assumptions • Include explicit discussion of any significant assumptions or events underlying the estimate that may not be obvious to the intended audience, including significant assumptions regarding the accounting basis or application of an accounting rule • Where the final work product reflects a material assumption or methodology that differs from what the actuary believes to be reasonable • Disclose the dependency of the final result and • Disclose the source

19. Communications and Disclosures – Ranges • What is the range intended to convey? • For example: • a range of estimates of the actuarial central estimate; • a range representing a confidence interval within the range of outcomes produced by a particular model or models; • a range representing a confidence interval reflecting both process and parameter risk; • some other clearly defined range

20. Communications and Disclosures – Material Changes • If the analysis is an update, disclose any material changes in assumptions to the extent known by the actuary • Disclose the reasons for the change • Not required to quantify the effect