Session 55 Panel Discussion When Is Your Own Data Not Enough? · Robert E. Winawer, FSA ....

Session 55 PD, When is Your Own Data Not Enough?

Moderator:

Robert E. Winawer, FSA

Presenters: Leonard Mangini, FSA, MAAA Timothy S. Paris, FSA, MAAA

SOA Antitrust Disclaimer SOA Presentation Disclaimer

https://www.soa.org/legal/antitrust-disclaimer/https://www.soa.org/legal/presentation-disclaimer/

2018 SOA Life & Annuity SymposiumSession 55- When Is Your Own Data Not Enough?PBR Developments that Might Surprise You!Tue, May 8, 2018Leonard Mangini, FSA, FRM, FALU, CLU, MAAAPresident and Managing Member, Mangini Actuarial and Risk Advisory LLC

Presenter BiographyLeonard Mangini, FSA, FRM, FALU, CLU, MAAAPresident and Managing Member, Mangini Actuarial and Risk Advisory LLCMr. Mangini brings clients over 28 years of industry expertise, holding senior Product, Reinsurance,Financial, and Risk Management-related industry roles at Manulife, ACE, AXA, and USLIFE and assistingclients with Product Development, Financial Reporting, Underwriting, Reinsurance, Risk Management,M&A, and Litigation issues as a consultant with E&Y, Milliman, and now his own advisory firm.

In his last direct company role, Leonard was Deputy Global Corporate Chief Actuary supervisingprinciples-based assumption and margin “unlocking” for over 100 products sold in 19 business unitsacross the US, Canada, and Asia and also served on the Global Product Risk and Global ALM Committees.

In prior reinsurance roles, Mr. Mangini served as an internal Board member, President, Chief Actuary,Chief Pricing Officer, and Chief Risk Officer, and co-founded a US life reinsurer. He’s one of the fewactuaries credentialed by exam as a medical underwriter, and has priced mortality, morbidity, longevity,and policy behavior for products issued in fully underwritten and alternative distribution channels.

Leonard serves on both the SOA Product Development and Joint Risk Management Section Councils,previously Chaired the Financial Reporting Section Council, and was a member of the Marketing andDistribution and Reinsurance Section Councils. He currently Chairs he Academy’s PBR Life Reserve WorkGroup (LRWG), is a Member of the Academy Life Practice Council (LPC) and Life Valuation Committee(LVC) and serves on Academy’s ASOP 11- Credit for Reinsurance- Update Committee, and is part of theLeadership team of the SOA Assumption Discussion Group. Mr. Mangini is a Fellow of the Society ofActuaries (FSA), a Certified Financial Risk Manager (FRM), Fellow of the Academy of Life Underwriting(FALU), a Member of the Academy of Actuaries (MAAA), and earned a MS in Computational Finance.

2018 SOA LAS: Session 55 When is Your Own Data Not Enough? May 8, 2018

2

Liability Disclaimer, Copyright, Use of Slides

Although I’ve attempted to faithfully capture the letter and spirit of legal, regulatory and ActuarialStandard of Practice (ASOP) constraints, you have a personal professional duty to familiarize yourselfwith the original source material and apply professional judgment as to its specific application to yourown work and those working under your direction as you perform covered Actuarial Services. Thenature of your work, and other professional designations you hold, may require you to be bound byadditional professional requirements from other organizations as well.

This material has been prepared for general informational purposes only and is not intended to berelied upon as accounting, legal, tax, or other professional advice, nor is it an Actuarial Opinion byLeonard Mangini, Tim Cardinal, Arnold Dicke or their respective firms Mangini Actuarial and RiskAdvisory LLC, Actuarial Compass LLC, or AADicke LLC. Please refer to your advisors for specificprofessional advice. The views expressed by the presenter are not necessarily those of ManginiActuarial and Risk Advisory LLC, Actuarial Compass LLC, AADicke LLC, or the Academy of Actuaries.

Much of the original source material on VM-20/PBR and Professionalism is copyrighted material of theAmerican Academy of Actuaries, Society of Actuaries, or National Association of InsuranceCommissioners. This presentation paraphrases these for educational purposes to capture the intent ofexisting and proposed regulations and standards of practice or may paraphrase the results of SOAresearch, and every attempt has been made to identify and cite original sources, where applicable

These slides may NOT be copied, redistributed, or otherwise furnished to any party without priorwritten consent of Mangini Actuarial and Risk Advisory LLC, other than as required to comply with anaudit of the meeting attendee’s annual CPE compliance.

3

2018 LAS: Session 55 ©2016-18 Mangini Actuarial and Risk Advisory LLC May 8, 2018

Society of Actuaries’ Presentation Disclaimer

Presentations are intended for educational purposes only and do not replace independent professionaljudgment. Statements of fact and opinions expressed are those of the participants individually and, unlessexpressly stated to the contrary, are not the opinion or position of the Society of Actuaries, its co-sponsors orits committees. The Society of Actuaries does not endorse or approve, and assumes no responsibility for, thecontent, accuracy or completeness of the information presented. Attendees should note that the sessionsare audio-recorded and may be published in various media, including print, audio and video formats withoutfurther notice.

4

SOCIETY OF ACTUARIESAntitrust Compliance Guidelines

Active participation in the Society of Actuaries is an important aspect of membership. While the positive contributions of professional societies and associations are well-recognized and encouraged, association activities are vulnerable to close antitrust scrutiny. By their very nature, associations bring together industry competitors and other market participants.

The United States antitrust laws aim to protect consumers by preserving the free economy and prohibiting anti-competitive business practices; they promote competition. There are both state and federal antitrust laws, although state antitrust laws closely follow federal law. The Sherman Act, is the primary U.S. antitrust law pertaining to association activities. The Sherman Act prohibits every contract, combination or conspiracy that places an unreasonable restraint on trade. There are, however, some activities that are illegal under all circumstances, such as price fixing, market allocation and collusive bidding.

There is no safe harbor under the antitrust law for professional association activities. Therefore, association meeting participants should refrain from discussing any activity that could potentially be construed as having an anti-competitive effect. Discussions relating to product or service pricing, market allocations, membership

restrictions, product standardization or other conditions on trade could arguably be perceived as a restraint on trade and may expose the SOA and its members to antitrust enforcement procedures.

While participating in all SOA in person meetings, webinars, teleconferences or side discussions, you should avoid discussing competitively sensitive information with competitors and follow these guidelines:

• Do not discuss prices for services or products or anything else that might affect prices• Do not discuss what you or other entities plan to do in a particular geographic or product markets or with particular customers.• Do not speak on behalf of the SOA or any of its committees unless specifically authorized to do so.• Do leave a meeting where any anticompetitive pricing or market allocation discussion occurs.• Do alert SOA staff and/or legal counsel to any concerning discussions• Do consult with legal counsel before raising any matter or making a statement that may involve competitively sensitive information.

Adherence to these guidelines involves not only avoidance of antitrust violations, but avoidance of behavior which might be so construed. These guidelines only

provide an overview of prohibited activities. SOA legal counsel reviews meeting agenda and materials as deemed appropriate and any discussion that departs from the formal agenda should be scrutinized carefully. Antitrust compliance is everyone’s responsibility; however, please seek legal counsel if you have any questions or concerns.

5

AGENDA: When is Your Own Data Not Enough?A PBR Mortality Assumption Perspective That Might Surprise You!

A Quick Primer on VM-20 Mortality• Prudent Estimate Mortality• Margins: Credibility and Other AdjustmentsWhat Happens if You Sub-Optimize Aggregation?• Creating PBR reserves which are bigger then XXXSituations that Can Cause Trouble if You’re Careless• New Underwriting Tests, Accelerated Underwriting, New ChannelsVM-20 and ASOPs- What Do They Already Say About AggregationYou May Have Enough Data!...Recently Exposed APF• 2018-17- Aggregation of Mortality Segments for Credibility- comments until May 10Optimized Aggregate Mortality and Segment-Level ImpactsOther Sources- What Might You Do If Your Own Data Really Isn’t Enough

6

2018 LAS: Session 55 ©2018 Mangini Actuarial and Risk Advisory LLC May 8, 2018

A Quick Primer on VM-20 Mortality

7

2018 LAS: Session 55 ©2016-18 AADicke LLC and Mangini Actuarial and Risk Advisory LLC May 8, 2018

VM-20 Mortality Credibility and Modeled Reserves

The Operative VM for 2018:• Has NO “Deterministic Exclusion Test” for Term• NO current exposed APFs to alter this in time for VM applicable in 2019

àMUST calculate NPR and “Modeled Reserve” to see which one “wins”àImpacts 2018 issues, almost certainly 2019 issues, probably longer…àSomething YOU have to deal with in Pricing since reserves impact pricing


8

VM-20 Definitions- Prudent Estimate (Section 9.A)

“Prudent Estimate”:• Used for ANY risk factor NOT prescribed or stochastically modeled• Must comply with Section 12 of SVL (Included as VM-05)

“Anticipated Experience”+ Margin • Adverse Deviation/Estimation Error

Anticipated Experience: Company OWN experience if relevant, credible• Mortality shall combine relevant company experience with industry

experience data in deriving anticipated experience consistent with statisticalcredibility theory and accepted actuarial practice

à ASOP 1: “shall” means that NOT doing so is a deviation!


9

VM-20 Definitions- Margins (Section 9.C.5)

Separate Margins for Company and Industry Experience rates

A percentage Increase applied to each rate

• Company margin varies by attained age and credibility- table lookup based on whether Company chooses Bühlmann or Limited Fluctuation Method

• Once Company chooses one of these two methods “locked in” unless Company requests and receives approval from Commissioner to switch

• Industry margin varies ONLY by attained age- i.e. assumed fully credibleSize of Margin Increased for:

• Imprecise experience studies or “staleness”• Underwriting or risk selection criteria have changed since experience study• Lack of homogeneity• Unfavorable environmental or health developments- e.g. a pandemic• Changes in marketing or administration that create anti-selection risk• Ineffectiveness of underwriting compared to expectations


10

VM-20 Definitions- Sufficient Data Period (Section 9.C.6)

Use Company Experience in policy durations with “Sufficient Data”

“Sufficient Data Period” End: Last policy duration with 50 or more claims

Determines when/how to grade between Company and Industry rates


11

VM-20 Mortality Grading

Splits Policy Period into 3 “Eras”• Company Period: 100% Company-Based Assumptions• Grading Period: Blends Company and Industry Experience• Industry Period: 100% Industry-Based Assumptions

If Credibility < 20%• NO Company Period• NO Grading Period

à Applicable Industry Data


12

Inputs Into Calculations Are Projects in of Themselves!

VM-20 Section 9.C.1.a• Establishing Mortality SegmentsDetermine Applicable Industry Table• Using Relative Risk Tool to map into Relative Risk Tables (RR80, RR100 etc)

Creating/Choosing an Internal Company Table• Collecting relevant and credible data BY AMOUNT• Performing actual/expected experience studies• Taking account of “considerations” for underlying data and margins


13

VM-20 Mortality SegmentsVM-20 §9.C.1.a:“Company shall determine mortality segments for purpose of determining separate prudentmortality assumptions for groups of policies that the Company expects will have differentexperience than other groups of polices”

Clear that this means different underwriting classes:• Male vs. Female• Non-Tobacco vs. Tobacco User• Preferred vs. Standard vs. Table Rated

BUT many companies issue a variety of policies using the same or similar• marketing PROCESS and• underwriting PROCESS

14


Company and Industry Credibility

VM-20 Section 9.C. requires the use of Look-Up Tables to Find Loads

For Company Experience

• Attained-Age Ranges as Rows• Level of Credibility as Columns

For Industry Experience

• Attained-Age Ranges as Rows• Level of Credibility is NOT a factor- since industry is assumed 100% Credible

à“Output” is a Percentage Load to ADD to the “Raw” Mortality


15

Determination of Company Experience (Section 9.C.2)EACH mortality segment:• Company Mortality derived from Company experience data (9.C.2.a)• Company experience NOT available/limited, Company can choose an applicable

industry table (9.C.2.a)

Company experience shall be based on (ASOP 1 meaning of shall):• Actual books of business within mortality segment (9.C.2.b.i)• Other books of business within Company with similar underwriting (9.C.2.b.ii)• Other sources, if available and appropriate, such as actual experience data from

one or more mortality pools in which the policies participate under the terms of a reinsurance agreement (9.C.2.b.iii)

• Other source “appropriate” if underwriting and expected mortality experience characteristics similar to policies in mortality segment (9.C.2.b.iii)

Shall not be lower than mortality Company expects to emerge, and can justify and disclose in VM-31 PBR Actuarial Report (9.C.2.c)


16

What Happens If YouSub-Optimize Aggregation

17


Suppose Company Experience from “Granular” Segment

Suppose set segments at Underwriting Class level for Particular Product

Super-Preferred Risk Class:• Expect best mortality amongst all underwriting classes• BUT, few might qualifyà so low count and amount in best class• Could even have < 20% credibility à forced to use Industry Table• Even if > 20% credibility, prudent estimate + high marginà valuation mortality for

Super-Preferred > valuation mortality for “next lower” Preferred Class!• Few Females? à Could have Female Qx > Male Qx for same rating class• Could have

Other Potential Undesirable OutcomesDiversity of interpretation between Companies as to conditions when usingaggregate mortality is acceptable despite having similar underlying risk profiles.

à Different mortality margins, SDPs, and grading from Company to Industrymortality for two Companies otherwise similarly-situated

Diversity of interpretation between jurisdictions’ conditions regarding whenusing aggregate mortality is acceptable despite similar underlying risk profileà Similarly-situated Companies but different domiciles with differing mortalitycredibility margins, SDPs, grading and thus different reserves patterns

Threatens Intent of Valuation Manual:• Harmonize valuation across jurisdictions• Similarly situated companies have similar levels of reserves• Similar products with similar risk and experience have similar level of reserves


19

Risk of PBR Reserves with Low Credibility > pre-PBR XXX !A Company ignoring legitimate CHOICES available to determine mortality experienceat an optimal level of aggregation may even produce reserves > pre-PBR “XXX”…!

A pair of articles looking at smaller insurers* and larger insurers** illustratesgraphically how the sum of the Deterministic Reserve (DR) and the Deferred PremiumAsset (DPA), i.e. the DR+DPA, varies for different levels of Buhlmann Credibility.

* “A VM-20 Mortality and Credibility Factor Observation”by Tim Cardinal, Principal, Actuarial Compass LLC

September 2017, SOA “Small Talk” newsletter, pp.14-15

https://www.soa.org/Library/Newsletters/Small-Talk/2017/september/stn-2017-iss48.pdf

** “A VM-20 Mortality and Credibility Factor Observation”by Tim Cardinal, Principal, Actuarial Compass LLC

December 2017, SOA “Financial Reporter” newsletter, pp.20-22

https://www.soa.org/Library/Newsletters/Financial-Reporter/2017/december/fr-2017-iss111.pdf

20


https://www.soa.org/Library/Newsletters/Small-Talk/2017/september/stn-2017-iss48.pdfhttps://www.soa.org/Library/Newsletters/Financial-Reporter/2017/december/fr-2017-iss111.pdf

Situations that Can Cause TroubleIf You’re Careless

21


Situations Potentially Leading to Over-Reserving

Performing Experience Studies at an “overly-granular” level when:• Other products in the Company have similar underwriting/expected mortality• Accelerated Underwriting doesn’t alter the broader business acquisition and underwriting

process that underlies the risk classification system being used• Medical advances (e.g. Hga1c diabetes tests) upgrades a portion of an existing risk-

classification system but doesn’t fundamentally alter the overall process• New underwriting data sources (e.g. Rx databases) collect similar risk classification

information as existing sub-processes of an overall classification system• Technology (Predictive Analytics) replaces portions of the risk classification process• New distribution (e.g. FinTech) produce similar mortality as traditional channels

You MIGHT be able to demonstrate similar expected experience to other business within the Company or perform experience studies at a more aggregate level with greater credibility


22

VM-20 and ASOPs:What Do They Already Say About

Aggregation?

23


Company Discretion in the Current VM- CredibilityVM-20 (Section 9.C.4.b) already currently states that: “Credibility may be determined at either the mortality segment level or at a moreaggregate level if the mortality for the sub-classes (mortality segments) wasdetermined using an aggregate level of mortality experience.A single level of credibility shall be determined over the entire exposure period, ratherthan at each duration, within the exposure period.This overall level of credibility will be used to:i. Determine the prescribed margin for company experience mortality rates.

ii. Determine the grading period (based on the credibility percentage shown in column (1) in the applicable table in Section 9.C.6.b.iii) for grading company experience mortality rates into the applicable industry basic table.”

24


Company Discretion in the Current VM: SDP and Grading

VM-20 Section 9.C.6.b.ii. already currently states that:“In determining the sufficient data period (SDP) the company shall first identify thelast policy duration at which sufficient company experience data exists (using all thesources defined in Section 9.C.2.b).The sufficient data period then ends at the last policy duration that has 50 or moreclaims (i.e., no duration beyond this point has 50 claims or more) subject to the limitsin Column 2 of the applicable table in Section 9.C.6.b.iii.b.The sufficient data period (SDP) may be determined at a more aggregate level thanthe mortality segment if the company based its mortality on aggregate experienceand then used a methodology to subdivide the aggregate class into various sub-classes or mortality segments.”

25


Aggregate Experience Studies that Are Subdivided

The common link between Sections 9.C.4.b and 9.C.6.b.ii is the concept ofperforming experience studies at an aggregate level and using a conservationof deaths method to subdivide that aggregate experience into sub-classeswhere these sub-classes:

• Respect Section 9.C.1.a- i.e. separate mortality segments reflect groups of policiesthat are expected to have different mortality

• Respect Section 9.C.2.c- namely, the mortality for each segment shall be no lowerthan the Company expects to emerge and which the Company justifies anddiscloses in VM-31

• Comply with ASOP 12 on Risk Classification

26


Actuarial Standard of Practice 12Risk Classification for All Practice Areas:Guidance when designing, reviewing, changing risk classification when classifying into groups intended to reflect relative likelihood of expected outcomesà Obviously applies for PBR Mortality Segmentation

Scope- Section 1.2• Setting of rates, contributions, reserves, benefits, dividends, experience refunds• Analysis or projection of quantitative or qualitative experience or results• Actuaries performing activities likely to have material effect, in actuary’s judgment on

intended purpose

Source: ASB Board Website-• http://www.actuarialstandardsboard.org/wp-content/uploads/2014/07/asop012_101.pdf

27


http://www.actuarialstandardsboard.org/wp-content/uploads/2014/07/asop012_101.pdf

Actuarial Standard of Practice 12§ 3.2/3.3: Actuary SHOULD consider the following in selecting Risk Characteristics: • Expected outcomes- demonstrate variation in actual/reasonable expected anticipated experience

correlates with risk drivers à Causation NOT required • Demonstration: may use relevant information from any reliable source, including statistical or

mathematical analysis, may use clinical experience or expert opinions • Equity- Rates considered equitable if differences reflect material differences in expected cost • Interdependence- Should consider interdependence to extent expect to have material impact • Inferences without Demonstration- sometimes appropriate to infer without demonstration• Objectivity- should select risk characteristics capable of objective determination (measurable)

based on readily observable facts• Law and Business Practices: Should consider constraints of applicable law, industry practices• Homogeneity (§3.32): variation in outcomes within a risk class too greatà subdivide. If too

granularà consider combining proposed risk classes to balance predictability and homogeneity

Source: ASB Board Website-• http://www.actuarialstandardsboard.org/wp-content/uploads/2014/07/asop012_101.pdf

28


http://www.actuarialstandardsboard.org/wp-content/uploads/2014/07/asop012_101.pdf

Amendment Proposal (APF) 2018-17Aggregation of Mortality Segments for the Purpose of

Determining Credibility

29


Amendment Proposal Form 2018-17:Aggregation of Mortality Segments for the Purpose of Determining Credibility

• Proposed by Academy Life Reserve Work Group (LRWG) to clarify and provide guidancesurrounding sub-dividing aggregate mortality segments ALREADY permitted by 9.C.4.b

• Argues Company may issue a variety of policies using the same or a similar marketingprocesses and the same or a similar underwriting processes which may lead them to

examine mortality experience and credibility at an aggregate level AND sub-divide such

aggregate experience class into sub-class mortality segments under 9.C.1.a with different

expected mortality but the SAME credibility produced at the aggregate experience studylevel and that such practice is both SOUND and in the spirit of the current VM wording

• Proposes a “geography change” of part of VM-20 Section 9.C.4.b to a new 9.C.4.c intendedmerely to separate existing language on how credibility is USED from how it’s determined

• Proposes changing remaining wording of Section 9.C.4.b AND adding a Guidance Note to9.C.4.b. The intent of both of is to clarify and provide guidance on how credibility is

determined in the aggregate by clarifying the level of aggregation permitted by the VM

Source: L. Mangini, Chair, LRWG, Academy of Actuaries’ Presentation to LATF 3/22/2018

30

2018 LAS: Session 55: When Is Your Own Data Not Enough? May 8, 2018

Proposed New VM-20 Section 9.C.4.b- Opening Part

The revised language for Section 9.C.4.b starts with the following language:

“b. Credibility may be determined at either the mortality segment level or at a more aggregate level. Experience for different mortality segments may be aggregated if the following three conditions are met: i. The company based its mortality on the aggregate experience and then used amethodology to subdivide the aggregate class into various mortality segments and adescription of the methodology for sub-dividing is provided in the VM-31 PBR Actuarial Report;ii. All mortality segments to be aggregated were subject to the same or similar underwriting PROCESESES; and iii. The mortality segments were sold by similar distribution systems and to similar market segments.”

The first romanatte rephrases the first paragraph of the current 9.C.4.b using language from9.C.6.b.ii, while the “AND language” requires that ALL three romanette conditions be metbefore the Company can consider aggregation.


31


Proposed New VM-20 Section 9.C.4.b- Middle Part

New VM-20 Section 9.C.4.b:

“Underwriting processes that utilize new methods, but which are expected to produce similarmortality, are considered to be similar to previously established underwriting processes ifthese expectations regarding mortality are supported in the VM-31 actuarial report bymedical, clinical, actuarial, or industry studies; predictive analytics or statistical analyses; ormodeled demonstrations.If the distribution system or target market for a mortality segment differs from that of theother mortality segments within the aggregate grouping, the mortality experience cannot beaggregated for credibility purposes unless the company expects and can demonstrate thatthe mortality experience for the segments is similar to that of the other mortality segment,and support is provided in the VM-31 Actuarial Report .”

We’ll see on the next slide that this “mortality-outcome-expectations” criteria for similarityand what is considered acceptable evidence to support these expectations both appear tobe aligned with relevant sections of acceptable actuarial practice in ASOP 12 on RiskClassification. The distribution system language above also addresses the underwritingdynamism reality discussed above, namely that properly designed direct-to-consumermarketing may in actual practice produce mortality similar to that of “agent-sold business”.

Source: L. Mangini, Chair, LRWG, Academy of Actuaries’ Presentation to LATF 3/22/2018 and APF 2018-17

32


Alignment of APF with ASOP 12 and Clarification of Evidence ASOP 12 Section 3.2.1:“The actuary should select risk characteristics that are related to expected outcomes. A relationshipbetween a risk characteristic and an expected outcome, such as cost, is demonstrated if it can beshown that the variation in actual or reasonably anticipated experience correlates tothe risk characteristic. In demonstrating a relationship, the actuary may use relevant informationfrom any reliable source, including statistical or other mathematical analysis of available data. Theactuary may also use clinical experience and expert opinion.”

The LRWG argued that similar expected mortality IS an objective measure for demonstrating that agroup of policies should be aggregated for credibility purposes, and actuarial techniques such asconservation of deaths could be used to divide develop experience assumptions for the mortalitysegments. They also argued that ASOP 12 language is clearly consistent with using medical, clinical,actuarial, or industry studies, predictive analytics, statistical analyses, and/or modeleddemonstrations.This is important since VM-20 Section 9.C.2.f.1 governs adjusting Prudent Estimate Assumptions:“supported by published medical or clinical studies or other published studies that correlate a specific riskselection criteria to mortality or longevity experience (for example, criterion and correlations determined throughpredictive analytics)”.

and many Company experience studies or analyses are internal and proprietary so NOT published


33


Proposed New VM-20 Section 9.C.4.b- Closing Part The last portion of the new VM-20 Section 9.C.4.b language states:

“If the company determines mortality and credibility at an aggregate level, then the futuremortality experience of each of the mortality segments within the aggregate shall be studiedseparately and the emerging results for each of these segments shall be presented withinthe VM-31 actuarial report”

This language reflects the prudential caution in the LRWG presentation that if theCompany has developed company experience mortality rates from aggregate experienceand then subdivided the aggregate class into mortality segments, that actual mortalityexperience that emerges from the mortality segments might diverge over time, so ongoingmonitoring and disclosure is essential to demonstrate that the Company’s initialexpectations regarding the appropriateness of the initial aggregation continue to be well-founded at subsequent valuation periods.


34


Proposed New VM-20 Section 9.C.4.b- Guidance Note The APF also proposes including the following guidance note for 9.C.4.b:

“Guidance Note: The intent of this section is to allow aggregation of different types of lifeinsurance products (such as term, whole life, universal life (UL), etc.) and different underwritingand risk classes within these products for purposes of determining credibility when theunderlying underwriting processes are similar. In the evaluation of similarity, considerationshould be given towards differences in mortality based on distribution and target marketcharacteristics. The intent is not to allow broad aggregation of disparate underwriting methodssuch as guaranteed issue or simplified issue with fully underwritten products. Mortalityexperience of products such as corporate- and bank-owned life insurance (COLI and BOLI)products, can also be aggregated for credibility if the company expects mortality experiencewill be similar to that of fully underwritten products.”

Source: Publicly Exposed APF 2018-17

35


Key Takeaways: If APF 2018-17 is ApprovedConsider performing experience studies at the highest level of aggregation you can

justify in VM-31 in compliance with ASOPs 12, 23 and 41, and use conservation of

deaths to subdivide into mortality segments aligned with Section 9.C.1.a, factoring in:

• Products targeting similar market demographic segments and underwriting processes• Groups of policies with similar mortality expectations• Whether a new distribution system produce similar mortality as an established one• Whether a medical innovation is just a new process to measure data already being used

in risk selection and correlates to an existing process with a lengthy credible study

• Whether a new underwriting data source is just replacing a sub-process within a broader underwriting and marketing or distribution process that otherwise hasn’t changed

materially over the length of an already credible experience study

• Whether Accelerated Underwriting is merely a sub-process of a larger system that shares the same or similar underwriting and marketing process and produces similar mortality

• Whether current similar expected mortality is expected to diverge over time• Whether investment segments drive policyholder behavior that alters expected mortality

36

2018 LAS: Session 55 ©2018 AA Dicke LLC and Mangini Actuarial and Risk Advisory LLC May 8, 2018

Alternative Source of Data:What to Do When Your Own Data REALLY Isn’t Enough

37


What if Your Data Really Isn’t EnoughDrawbacks of direct carrier using traditional reinsurance to “boost credibility”• High Quota Share percentages cede away profits just to optimize reserves• Reluctance to share proprietary data- reveals profit margins• Concentrating counter-party risk in one reinsurer

M&A can boost exposure and claims volume impacting credibility and SDP• Expensive and time-consuming- combining cultures, admin systems, strategy• Requires regulatory approvals

Direct carrier can act as a reinsurer• Flow-reinsurance takes time to build scale• Block acquisitions can be as complicated as M&A

38

2018 LAS: Session 55 ©2018 AA Dicke LLC and Mangini Actuarial and Risk Advisory LLC May 8, 2018

Contact Information

39

Leonard Mangini, FSA, FRM, CLU, FALU, MAAAPresident , Mangini Actuarial and Risk Advisory LLC

E-mail: [email protected]: www.manginiactuarial.comMobile: (516) 418-2549

2018 Life & Annuity Symposium

Session 55 – When Is Your Own Data Not Enough?May 8, 201810:00-11:15am

Timothy Paris, FSA, MAAARuark Consulting LLC

DisclosuresThis material is not intended for general circulation or publication, nor is it to be reproduced, quoted or distributed for any purpose without the prior written permission of Ruark. Ruark does not accept any liability to any third party. Information furnished by others, upon which all or portions of this material are based, is believed to be reliable but has not been independently verified, unless otherwise expressly indicated. Public information and industry and statistical data are from sources we deem to be reliable; however, we make no representation as to the accuracy or completeness of such information. The findings contained in this material may contain models, assumptions, or predictions based on current data and historical trends. Any such predictions are subject to inherent risks and uncertainties, as future experience may vary from historical experience. The reader should consider the applicability of these models, assumptions, and predictions for the future, and whether additional margins for conservatism should be included. Ruark accepts no responsibility for actual results or future events. The opinions expressed in this material are valid only for the purpose stated herein and as of the date indicated. No obligation is assumed to revise this material to reflect changes, events or conditions, which occur subsequent to the date hereof. All decisions in connection with the implementation or use of advice or recommendations contained in this material are the sole responsibility of the reader. This material does not represent investment advice nor does it provide an opinion regarding the fairness of any transaction to any and all parties.

2

Source: LIMRA

VariableAnnuities

Fixed IndexedAnnuities

Gross Sales (p.a.) ~$150 billion ~$100 billion

Net Sales (p.a.) ~$0 billion ?

% Qualified 65% 55%

% GuaranteedLiving Benefit 77% 68%

3

Overview of VA Industry Experience

VA Industry Data

25 participating companies

2008 to present

77 million contract years of exposure

$905 billion account value

5

Surrenders vary by living benefit type

0%

30%

7 ormore

6 5 4 3 2 1 0 -1 -2 -3 ormore

Surr

ende

r Rat

e

Years Remaining in Surrender Charge Period6

GMIBGLWB

NoneGMWB

Experience varies by company, but why?

50%

100%

150%

7

GLWB, Normalized by Years Remaining in Surrender Charge Period

Average

Most GLWBs are actuarially out-of-the-money

8

7%

89%

36%

7%

54%

4%3% 0%

Nominal Actuarial

ITM 5-50%

ATM

OTM

ITM 50+%

GLWB moneyness basis matters

0%

25%

OTM 25%+ OTM 5 - 25% ATM ITM 5 - 25% ITM 25 - 50% ITM 50 -100%

ITM 100%+

Surr

ende

r Rat

e

Spike - nominal

Spike - actuarial

9

GLWB income utilization affects surrenders

0%

25%

7 ormore

6 5 4 3 2 1 0 -1 -2 -3 ormore

Surr

ende

r Rat

e

Years Remaining in Surrender Charge Period

None

Excess

Full or Less Than

10

Income utilization varies by age and tax status

11

0%

100%

Income commencement is the key question

12

GLWB Partial Withdrawal Frequency

Commencement

Continuation

0%

100%

50-59 60-64 65-69 70-79 80+Attained Age

Nonqualified

Qualified

Nonqualified

Qualified

GMIB annuitizations are low

13

0%

20%

ITM 100+% ITM 50 -100%

ITM 25 - 50% ITM 5 - 25% ATM OTM 5 - 25% OTM 25+%

Actuarial basis

70-7960-64

65-69

80+

Guarantees can affect mortality too

14

0%

150%

1 2 3 4 5 6 7 8 9 10

% o

f Rua

rk M

orta

lity

Tabl

e

Duration

GMDB only

GLWB

Mortality effects are amplified by policy size

15

0%

150%

Overview of FIA Industry Experience

FIA Industry Data

16 participating companies

2007 to present

17 million contract years of exposure

$215 billion account value

17

FIA surrenders vary based on interest credited

18

0%

35%

10 ormore

9 8 7 6 5 4 3 2 1 0 -1 -2 -3 ormore

Surr

ende

r Rat

e

Years Remaining in Surrender Charge Period

0-2%

All others

Behavioral Analytics Framework

20

IndustryData

Statistical Techniques

Expert Judgment

Traditional Analysis

20

Model Development

Start with maximum data set (industry)Extract relevant subset for a companyDevelop a model on this basisDo likewise using only company’s dataCustomize model to reflect both, so that most important factors are included, with stable coefficients, balancing goodness-of-fit and predictive power

You can go far with Generalized Linear Models (GLM)

21

22

Logistic Regression Model

ln𝜇𝜇

1 − 𝜇𝜇= 𝛽𝛽0 + �𝛽𝛽𝑖𝑖𝑥𝑥𝑖𝑖

“Log of odds” is a linear function of key factorsBinary values, such as surrenders or deaths

23

Goodness of Fit

Predictive Power

23

Bayesian Information Criterion

Rewards goodness-of-fit to historical data, but penalizes for additional factors used in your model

One of many metrics to help guide your model selection process

24

Actual-to-Expected Ratios

“Predictive Power” in the new vernacular

Develop E using train data, compare to A from test data

Out-of-sample, out-of-time, and k-fold cross-validations

Examine in aggregate, by cohorts, and over time

Look at range of outcomes and tails

25

Expert Judgment is Vital

Business context, sensibility, materiality, parsimony

Let the data speak

More data usually beats more complex models

Build simple models for complex data, and complex models for simple data

26

Sample Model

VA GLWB / GMIB Income Utilization

Attained Age Tax StatusHistorical Income

UtilizationContract Size Interactions

28

Using industry data

For each factor coefficient, standard error terms 𝜎𝜎𝜇𝜇

are typically very small ~ 1/300 to 1/100.

Then testing predictive power using 5-fold cross-validation, average A/E errors are also very small ~ 1/700.

29

Using company-only data

In some cases, company-only data is insufficient to even identify the key factors observed in the industry data, or it demonstrates factor coefficient estimates that are not sensible.

Even if they do, the coefficient standard error terms 𝜎𝜎𝜇𝜇

can be 20x larger.

Similarly, the average cross-validation A/E errors can be 10x larger.30

Combining industry and company-only data

A customized combination of industry and company-only data can produce a vastly superior model with much better fit and predictive power.

Such a model should identify and quantify the effects of each additional factor in the presence of the others, and the interactions between them.

Confidence increases with additional data.

31

The power of more data

For GWLB / GMIB income utilization, need to address complexities of frequency and severity relative to guarantee amounts.

Customized model using industry data can reduce error by half where it matters most, for Full income utilization.

32

Discussion

Cover pageManginiParis

Session 55 Panel Discussion When Is Your Own Data Not Enough? · Robert E. Winawer, FSA ....

Documents

Transcript of Session 55 Panel Discussion When Is Your Own Data Not Enough? · Robert E. Winawer, FSA ....