risk transfer – actuarial perspective

Click here to load reader

Download Risk Transfer – Actuarial Perspective

Post on 08-Feb-2016

83 views

Category:

Documents

0 download

Embed Size (px)

DESCRIPTION

Risk Transfer – Actuarial Perspective. Casualty Actuarial Society – Washington, D.C. September 18-19, 2008 Ian Sterling, FCAS, MAAA. Agenda. FAS 113/SSAP 62 Methods of Testing Metrics Next Steps. Reinsurance Accounting Guidance. GAAP - PowerPoint PPT Presentation

TRANSCRIPT

  • Risk Transfer Actuarial Perspective Casualty Actuarial Society Washington, D.C.September 18-19, 2008Ian Sterling, FCAS, MAAA

    Presentation titlePage *

    AgendaFAS 113/SSAP 62Methods of TestingMetricsNext Steps

    Presentation titlePage *

    Reinsurance Accounting Guidance GAAP FASB Statement No. 113, Accounting and Reporting for Reinsurance of Short-Duration and Long-Duration ContractsStatutorySSAP No. 62, Property and Casualty ReinsuranceSimilar to FAS 113

    Presentation titlePage *

    Short-Duration Risk Transfer FAS 113Risk Transfer Conditions:Paragraph 11 Test:The reinsurer assumes substantially all of the insurance risk relating to the reinsured portion of the underlying insurance contracts, orParagraph 9 Test: (a) The reinsurer assumes significant insurance risk under the reinsured portions of the underlying insurance policies.Transfer of insurance risk refers to:Ultimate amount of net cash flows between parties, andTiming of the receipt of cash(b) It is reasonably possible that the reinsurer may realize a significant loss from the transaction. Risk factors do not include recognition of reinsurance costs, investment risk, taxes, or credit risk

    Presentation titlePage *

    SSAP 62Risk Transfer Conditions:Indemnification of the entity company against loss or liability relating to insurance risk in reinsurance requires both of the following:a. The reinsurer assumes significant risk under the reinsured portions of the underlying insurance agreements; andb. It is reasonably possible that the reinsurer may realize a significant loss from the transaction

    Presentation titlePage *

    Reinsurance Attestation Supplement 20-1Risk Transfer Testing Practice NoteAmerican Academy of Actuaries Committee on Property and Liability Financial Reporting November 2005

    Presentation titlePage *

    Methods of Testing(1) Reasonably Self-EvidentPurposeNeed to DocumentConsiderationsSubstance of the arrangementExistence, impact and role of risk-limiting factorsUse of professional judgmentContract Terms to make this less likely?

    Presentation titlePage *

    Methods of Testing(1) Reasonably Self-EvidentExamples of Safe Harbors:A straight QS with no risk-limiting features other than a loss ratio cap with negligible effect on the economics of the transactionSingle year property cat and casualty clash contracts with little or no risk limiting features apart from a reinstatement premium common to these types of contractsMost facultative and treaty per risk excess of loss arrangements with rates on line well below the present value of the limit of coverage, or without aggregate limites, sub-limits, or contingent features

    Presentation titlePage *

    Methods of Testing(1) Reasonably Self-EvidentExamples of contracts not reasonably self-evident:Aggregate excess of loss contractsContracts with experience accounts, experience rating refunds, or similar provisions, if such provisions have a significant impact on the contracts economicsMultiple year contractsQS contracts with risk limiting features

    Presentation titlePage *

    Methods of Testing

    Presentation titlePage *

    Methods of Testing(2) Scenario TestingHistorical results by yearComparison of All Underwriting Downside ScenariosComparison of Cedent and Reinsurer Expected Underwriting Deficits

    Presentation titlePage *

    Methods of TestingHistorical Results by Year

    Presentation titlePage *

    Methods of TestingComparison of All U/W Downside Scenarios

    Presentation titlePage *

    Methods of TestingComparison of Cedent and Reinsurer Expected Underwriting Deficits (EUD)

    Presentation titlePage *

    Methods of Testing(3) Simulation Testing

    Types of ModelsAggregate Loss ModelsFrequency-Severity ModelsCombination ModelsConsiderations

    Presentation titlePage *

    Methods of TestingModeling ConsiderationsNeed to model:Contract lossesContingent PremiumCommissionsOther Contract FeaturesReinsurance Underwriting ExpensesPotentially not considered:Tax impacts

    Presentation titlePage *

    Metrics(1) 10-10 Rule (Value at Risk)Initial rule of thumb, and still somewhat used todayDefinitionShortcomingsUnintended Consequences

    Presentation titlePage *

    Metrics(2) Tail Value at Risk (TVaR)DefinitionCriteria similar to 10-10 = TVaR > 10% at 10th percentileAdvantagesDisadvantagesDoes this solve the shortcomings of VaR?

    Presentation titlePage *

    MetricsVar and TVaR Example

    Presentation titlePage *

    Metrics(3) Expected Reinsurer Deficit (ERD)DefinitionHow relates to previous methodsCriteria of ERD > 1% similar to 10-10 (1% = 10% x 10%)Avg Loss Severity = TVaR at the economic breakeven LRAdvantagesDoes this solve Shortcoming 2?Similar to Financial?

    Presentation titlePage *

    Metrics(4) Other MethodsOther methods:1. Right Tail Deviation (RTD) - Wang2. Mean Square Adverse Deviation3. Conditional Expected Downside4. Some combination of (2) and TVaRAdvantageDisadvantage

    Presentation titlePage *

    SummaryCan a bright-line test be used?AdvantageDisadvantageGuides10-10, TVaR, ERD, RTD, etc.Methods of Testing Risk TransferReasonably Self-Evident, Scenario Testing, SimulationNext Steps

    Holly will discuss these in more detail, under the accounting perspective, which is important b/c risk transfer ultimately is an accounting decision. Actuaries cannot provide formulas that reduce analysis of intent of the reins contract. Failure to meet a quantitative analysis does not imply reins should be denied. Formulas can be manipulated through fraud. However, although ultimately an accounting decision, we need to continue to work with the accountants to provide our expertise.substantially all and reasonably possible is very vague, which has lead to much discussion on this topic and various methods to address.Paragraph 11 if fail either 9-a and 9-b. However, look at other way around is easier to test b/c if passes 11 dont have to worry about 99-a: test a met if the reinsurer has risk of variation in both timing and amount of payments, and payments must be timely to meet this criterion;9-b: test b requires an examination of possible outcomes. To meet this test, at least some of the outcomes have to produce a loss for the reinsurer, where a loss is determined using present value of all cash flows.11: portion of polices has to be fully ceded, where portion is restricted to percentage or section and the only risk the cedent can retain on this portion must be trivial, having no importanceSimilar to FAS 113Again, Holly will discuss in more detailFAS 113 is accounting guidelines, this workpaper represents some actuarial views.Intended to provide advisory, non-binding guidance to P/C actuaries regarding testing for risk transferWritten by actuaries, for actuaries and not attended to be professional accounting guidanceSome contracts are self-evident that there is risk transfer. It is too onerous to provide analysis for all contracts, especially where not needed and professional judgment can come into play.This is a principles-based standardPurpose: To eliminate and/or avoid the time and expense associated with unnecessary analysisDocument: Amongst other things business purpose and signoff (from management and external auditors/other team members) that risk transfer was believed to be reasonably self-evidentContract Terms to make this less likely:- Experience accounts- Profit commissions- Additional premiumsHistorical results by year: Use judgment or other tests which will be discussed later to determine- need to consider parameter risk (such as rate level history not accurate, claim trend assumption wrong, trend incorrect)Comparison of All Underwriting Downside Scenarios: if cedents margin >= reins margin for cedents breakeven LR implies U/W loss and reins assumed substantially all of downside risk- compare over a range of LR on the original unreinsured portfolio- downside risk ins risk relating to the underlying contracts- easier to understand but not always conclusiveComparison of Cedent and Reinsurer Expected U/W Deficits: if EUD faced by reinsurer >= EUD of cedent, then substantially all test is met- more complicated but can always be applied2 out of 10 last years, resulted in a loss to the reinsurer- Risks: LDFs variable (especially in immature years), Rate adequacy (Were premiums adequate), Investment Risk- How to measure? Will be discussed laterSince above Cedents breakeven ratio of 80% the Reins Margin is close to Cedent Margin implies reinsurer has assumed substantially all of the insurance risk of the reinsured policiesCed Margin = 1 - LR - Ced Exp RatioReins Margin = 1 - LR - Reins CommIf cedents margin equal or exceeds the reinsurers margin for the loss ratios that imply an underwriting loss, then the reinsurer has assumed substantially all of the insurers downsideSince reinsurers expected underwriting deficit > cedents expected underwriting deficit implies reinsurer has assumed substantially all of the ceding companys downside riskEUD = pure prem of an aggregate XOL attaching at breakeven loss ratio; or= frequency x severity of U/W LossIf EUD of reinsurer >= to EUD of cedent, then substantially all test is metAgg Loss Models: log-normal loss ratio modelFreq-Sev: For each year (simulation) simulate frequency (# of claims) and loss distribution (size of each loss per claim)- Poisson for frequency (can assume Negative Binomial but makes tail fatter)- Weibull for Severity

    Combination: Agg model for Normal Losses and Freq-Sev model for Shock/Cat Losses

    Considerations: For Freq-Sev: dividing losses into three segments (small, medium, and large claims)- Be conscious of uncertainty in model parameter and reflect this in the structure of the model- less relevant historical experience implies more parameter uncertaintyTax impacts: Further discussion by KevinArose as way to test f

View more