Individual Life Insurance and Annuity Product Development ......health, annuity, long-term...

W hat might the product world looklike in 2010? A number of forcesare developing that will heavily

influence the products that characterize thelife insurance and annuity market in the mid-term future. Some are already taking effect,while others will evolve over several years.

Risk Management

Asset-liability management was in the spot-light in the early 1980s when interest ratesspiked. It returned to the spotlight in recentyears because of the risk concentrationcreated by many insurers’ heavy reliance onvariable annuities and the sale of derivative-based benefits added a new dimension ofconcentrated risk. The impending arrival ofC3-Phase II risk-based capital requirementsfor variable products, the scrutiny of ratingagencies and the scarcity of reinsurancealready are causing a retreat in equity put-based benefits and will probably lead tomore severe reductions.

The availability of guaranteed living bene-fits and guaranteed minimum death benefitswill be limited by an insurer’s ability tohedge them through internal product balanc-ing and diversification. Dynamic hedgingand reinsurance will remain in the back-ground. The result will be significantly lessprominence for these currently popular guar-anteed benefits.

Equity-indexed benefits, which gainedpopularity as a product designed for theconservative equity investor, will gain newpopularity as a product for the prudentinsurer to deliver guarantees. Their ability tobe hedged with call options will be valued.

continued on page 3

Comments from the ChairThe Shape of Products in 2010by Noel J. Abkemeier

Product Matters!The newsletter of the Individual Life Insurance and Annuity Product Development Section

Published in Schaumburg, IL by the Society of Actuaries July 2003 • Issue No. 56

Published by the Individual Life Insurance

and Annuity Product Development Section

of the Society of Actuaries

475 N. Martingale Road, Suite 800

Schaumburg, IL 60173-2226

Phone: 847-706-3500 •Fax: 847-706-3599

Web: www.soa.org

This newsletter is free to Section members. A

subscription is $20.00 for nonmembers. Current-

year issues are available from the Communications

Department. Back issues of Section newsletters

have been placed in the SOA library and on the

SOA Web site: (www.soa.org). Photocopies of back

issues may be requested for a nominal fee.

2002-2003 SECTION LEADERSHIP

Noel J. Abkemeier, Chairperson

Kevin J. Howard, Vice-Chairperson

Anne M. Katcher, Secretary

Nancy M. Kenneally, Treasurer

Keith Dall, Council Member

Abraham S. Gootzeit, Council Member

Paul A. Haley, Council Member

Susan Kimball, Council Member

Kelly A. Levy, Council Member

Douglas C. Doll, Newsletter Editor

Tillinghast-Towers Perrin

3500 Lenox Road, Suite 900

Atlanta, GA 30326-1119

PHONE: (404) 365-1628

FAX: (404) 365-1663

E-MAIL: [email protected]

Joe Adduci, Senior DTP Coordinator

• NAPP Member

PHONE: (847) 706-3548

FAX: (847) 273-8548


Clay Baznik, Publications Director


Lois Chinnock, Staff Liaison


Facts and opinions contained herein are the sole

responsibility of the persons expressing them and

should not be attributed to the Society of Actuaries,

its committees, the Individual Life Insurance and

Annuity Product Development Section or the

employers of the authors. We will promptly correct

errors brought to our attention.

Copyright © 2003 Society of Actuaries.

All rights reserved.

Printed in the United States of America.

2 Product Matters! • July 2003

1 Comments from the Chair:The Shape of Products in 2010Noel J. Abkemeier

4 Mortality Anti-selection—Different Versions of Dukes/MacDonaldDouglas C. Doll

6 Estimating Worksite Mortality—A Structured ApproachAdrian R. Pask

10 Is Your Reinsurance Creating EnoughValue?James W. Dallas

13 AAA—Illustrations Work GroupMichael S. Taht

14 Shadow Account Products: Beyond PricingJeremy A. Bill

16 VUL Secondary Guarantees: Catalyst for Sales ReboundThomas E. Norton and Robert Grotyohann

21 U.S. Population Mortality ImprovementDouglas C. Doll

22 Variable Annuity Risk and Seeking “The Perfect Hedge”Douglas L. Robbins

23 Risk-Relevant ResourcesSOA Risk Management Task Force

24 Pricing for Internet DistributionW. Howell Pugh

26 2001 CSO Table—The Road to 2009William Carroll

30 Upcoming PD Sessions at the Annual Meeting in Orlando

32 SOA Report on Mortality Improvement SurveyDouglas C. Doll

COVER

CONTENTS

FEATURES

Product Design Flexibility

Periodically over the last 17 years there hasbeen a push to make broad revisions to nonforfeiture requirements, andone of the quinquennial efforts is now occur-ring. Whether change occurs now or five yearsfrom now, change is inevitable. The StandardNonforfeiture Laws for Life Insurance (1942)and Deferred Annuities (1976) both have theirroots in eras that predated today’s computercapabilities, liberalized financial regulationstructures, available financial products, andconstantly evolving consumer needs and pref-erences. The formulaic requirements of thecurrent laws will be replaced with more flexi-ble approaches that will allow new productdesigns.

A single product may be capable of accom-modating multiple risks equally, e.g., life,health, annuity, long-term disability, home-owners, auto, etc., rather than accommodatethem only as small ancillary benefits on aprimary product. Life cycle products thatstart as life insurance and evolve into anannuity and then long-term care may beavailable. Products may have personallydesigned balances among death benefit,premium and cash value, including no cashvalue despite sizable premiums.

Increased Disclosure

Historically, sales disclosure has beenpushed by regulators for the purpose ofconsumer protection. The ever-growingculture of litigiousness and the wideningcircle of class action law suits has intersectedwith increasing product complexity and willcause insurers to take a leading, if notcontrolling, interest in providing full disclo-sure. The product complexity will becharacterized by personalized product designand/or personally adaptable sales illustra-tions that demand disclosure.

Genetic Testing

Genetic testing will have moved from thelaboratory to being used by individuals. Acommon-sense business approach towardunderwriting will evolve despite some

legislative pressures to the contrary, and itwill allow genetic test results to be treatedlike other components of medical history,namely something that must be disclosed onan application and is available for formingan underwriting decision.

Immediate Annuity Creativity

Limited attention has been paid by insurersto immediate annuities because the insur-ance industry never achieved enough sales toform a critical mass. The aging of the babyboomers, the increasing awareness thatmanaging one’s own investments is not aseasy as it once looked, and emerging educa-tion about the risk of unpredictably livingtoo long will lead to the long-awaited emer-gence of the immediate annuity market. Thiswill lead to much broader and more competi-tive offerings of underwritten annuities,including those that are attractive topurchasers with impaired health. The degreeof creativity that has marked the deferredvariable annuity market for the last 10 yearswill expand into immediate annuity prod-ucts; however, insurers will be more cautiousabout the new types of product design risksthey take on.

Will these changes all occur? Let’s checkback in seven years (when I have purchasedan attractive and fully disclosed immediateannuity that contains multiple benefits, noneof which imposes great risk on the insurer).

* * * * *

Before we get to 2010, there are enoughcurrent risks that insurers face. The ProductDevelopment Section has initiated aresearch project called the Analysis ofProduct Guarantees that will analyze thevarious guarantees provided in fixed andvariable life insurance and annuity products.The study is intended to identify the guaran-tees and their associated risks, describepricing methods and measures and quantifythe impacts on policyholder behavior. Theresults should be a valuable resource forproduct development actuaries. We will keepyou apprised of the progress of the study.�

July 2003 • Product Matters! 3

Chairperson’s Corner • from page 1

Noel J. Abkemeier, FSA,

MAAA, is a consulting

actuary at Milliman USA in

Williamsburg, Va, and is

chairperson of the Product

Development Section. He

can be reached at noel.

[email protected].

A common method for projecting themortality associated with highlapse rates is to use the so-called

Dukes/MacDonald approach. I have foundmore than one version of Dukes/MacDonaldbeing used in practice. It is important thatwe are aware of which version is beingused, so we understand how much extramortality we are projecting. The purpose ofthis article is to provide some backgroundon anti-selection formulae (for those who,

unlike me, are not old enough to have beenaround when they were developed), and todescribe the different forms of“Dukes/MacDonald” that I have seen.

Anti-selection Formulas

This topic came to the forefront during the“term wars” of the early 1980s, when ever-decreasing term rates caused very high lapse

rates on existing term products. The develop-ment of select and ultimate rate scales forterm insurance was expected to lock in highlapse rates, as healthy lives had significantincentive to lapse and start over on a newselect scale. Finally, term products withexplicit re-entry provisions required theactuary to estimate the mortality of the nonre-entered group as well as the re-entries.

Three major methods to calculate themortality of the persisters were published inthe 1980s. They are similar in their underly-ing theory, but somewhat different inmechanics and results.

Shapiro/Snyder Method1

The mortality of the persisters is expressed asratios to standard mortality. Each duration anew ratio is calculated equal to the prioryear’s ratio, plus an increment to the ratiocalculated assuming that the extra lapsersare fully select. Refinements to the modelinclude an assumption that lapsers are notfully select (by introducing an “effectiveness”percentage), and by grading off over time theincrements to the mortality ratio.

Dukes/MacDonald Method2

This method uses the concept of conservationof total deaths. The excess lapsers areassumed to be fully select at the time oflapse, but their mortality grades to ultimatein normal fashion.

Mortality Anti-selection —Different Versions of Dukes/MacDonaldby Douglas C. Doll


Features

1) “Mortality Expectations Under Renewable Term InsuranceProducts”, Proceedings of the Conference of Actuaries inPublic Practice, Vol. XXX.

2) “Pricing a Select and Ultimate Annual Renewable TermProduct”, Transactions of the Society of Actuaries, Vol.XXXII.

The mortality of the persisters is assumedto be the difference between total aggregatemortality and the mortality of the excesslapsers. Note that the effect of one year’sexcess lapse goes away after 15 years, if a 15-year select mortality table is being used. Thefocus of Dukes/MacDonald’s method was onexcess lapse due to re-entries to term prod-ucts, and the method assumed ananti-selection effectiveness of 100 percent.

Becker/Kitsos Method3

This method starts with the Dukes/MacDonald method and refines it by addingan effectiveness factor, similar in concept toShapiro/Snyder effectiveness. In theBecker/Kitsos method, excess lapsers areassumed to have mortality equal to fullyselect, plus an extra mortality equal to aportion of the initial difference between theselect and the persisting group. This extramortality is graded off over a 15-year period.

The Different Forms ofDukes/MacDonald

The typical formula used today is a modifica-tion of Dukes/MacDonald, whereby aneffectiveness percentage less than 100percent is assumed.

The different versions that I have seenused differ based on which group of “persis-ters” the excess mortality is spread over. Thethree methods are as follows:

� Method 1: Persisters are those who continue their policy in-force.

� Method 2: Persisters are those who continue in-force, plus the nonselect excess lapsers.

� Method 3: Persisters are those who continue in-force, plus the nonselect excess lapsers, plus the base rate lapsers.

To illustrate the impact of the three meth-ods, consider the following example:

� Base lapse rate is 10 percent� Total lapse rate is 85 percent

� Effectiveness is 80 percent

� Select and point-in-scale mortality rates are .01 and .03, respectively

Assuming 100 lives, I now calculate themortality ratios for the in-force business forthe three methods:

� Base lapses = 10

� Excess lapses = 85 – 10 = 75

� Select excess lapses = .80 * 75 = 60

� Nonselect excess lapses = 75 – 60 = 15

� Extra mortality on persisters = 60* (.03 - .01) = 1.20

� Method 1 mortality ratio = (.03 + 1.20 /15) /.03 = 367%

� Method 2 mortality ratio = (.03 +1.20 /30) /.03 = 233%

� Method 3 mortality ratio = (.03 + 1.20 /40) .03 = 200%.

The differences among the three methodsare significant and demonstrate that it isimportant that you know exactly howmortality deterioration is calculated in yourpricing models.�


Douglas C. Doll, FSA,

MAAA, works at

Tillinghast-Towers Perrin in

Atlanta, GA. He is editor of

Product Matters and can

be reached at doug.doll@

tillinghast.com.

3) “Pricing for Profitability in ART”, Best’s Review,September 1984, and “Mortality and Lapse Assumptions inRenewable Term Insurance”, Reinsurance Reporter, August1984.

F or the pricing actuary, estimatingworksite mortality is a challengingtask. This kind of coverage has char-

acteristics of both individual and groupinsurance because it is an individual policythat is sold at an employee’s place ofemployment. Additionally, life insurancesold in the workplace is a voluntary benefit,meaning participants elect to purchase thecoverage and choose the coverage amount.The voluntary nature of this insurancecreates the opportunity for antiselection

because individuals in poor health will electcoverage more often than healthy individu-als. On the other hand, insurance companyunderwriters accept or reject the best risksbased upon the information contained inthe application. The result is two competingforces driving worksite mortality: antiselec-tion by the applicant population andprotective selection by the underwriters.

The problem for the pricing actuary is thatthe two competing forces are compounded bytwo elements of the product design process:the number of questions on the applicationused by the underwriters to select the bestrisks and the minimum required participa-tion level. Agents want to streamline thesales process by removing as many questionsfrom the application as possible. Removing

questions from the application reduces theprobability that underwriters can identify asubstandard individual. Additionally, theworksite group may have differing levels ofparticipation. Participation is the percentageof the employees at the worksite who applyfor insurance. The lower the participationrate, the greater the intensity of anti-selec-tion, because a limited number ofsubstandard individuals will be a largerpercentage of the insured population. Thepricing actuary is often required to answerquestions such as:• “What is the impact on mortality if we

streamline the application by removing a question about prescription medication?”

• “If we lower our required participation from 40 percent to 30 percent, how muchwill our mortality and resulting premiums increase?”

One solution is to divide the actively atwork population into risk classes and vieweach selection decision as a screen that elim-inates individuals in each risk class. For thiscase study the population is divided intothree risk classes: standard risk with 100percent of standard mortality, substandardrisk with 300 percent of standard mortality,and HIV positive risk with 2000 percent ofstandard mortality. An HIV-positive risk isused in this example because this risk repre-sents a “mortality time bomb” that isidentifiable through a question or blood test.The first screen in the process is theemployee electing to purchase coverage. It isreasonable to assume that substandard indi-viduals will elect coverage more frequentlythan standard individuals. This screen willskew the applicant population toward thesubstandard risks. The second screen is theunderwriting process that further reducesthe applicant pool. In opposition to the appli-cation screen, the underwriting screenselects individuals who are standard andweeds out substandard individuals, skewingthe insured population back toward the stan-dard risks.

Estimating Worksite Mortality—A Structured Approachby Adrian R. Pask


Features

The process involves six steps:1. Estimate actively at work mortality

2. Estimate the actively-at-work risk class distribution

3. Calculate a standard risk mortality table

4. Estimate the participation rate

5. Estimate the underwriting acceptance rates

6. Calculate the insured mortality table

This process has been used successfully toprice worksite products. The following exam-ple should be considered illustrative anduses assumptions that do not reflect anyparticular product or pricing situation. It iscritical to develop the assumptions in eachstage of the process using professional judg-ment while considering the impact of thetarget market and product characteristics.

Step 1: Estimate Actively-at-WorkMortality

One of the key mortality advantages of lifeinsurance sold in the worksite is that all theapplicants are actively at work. Employeespass through a powerful screen by showing upfor work regularly. Major mortality risk factorssuch as terminal cancer or serious drug abuseare reduced because it is difficult for theseindividuals to remain full-time employees.

The problem facing the actuary is selectingan appropriate actively-at-work mortalitytable. The requirements for this mortalitytable are that it:• Reflects current experience

• Is sex and smoker distinct and

• Reflects the actively-at-work selection criteria.

Both the actuary and underwriter mustevaluate the target market, evaluate therisks of the target industry and adjust theactively-at-work mortality tables accordingly.This method can be applied equally to insur-ance for miners or insurance for officeworkers if the actively-at-work mortalitytable is adjusted correctly.

Step 2: Estimate Distribution forActively-at-Work Ratings

The actively-at-work population contains bothstandard and substandard risks. Substandard

risks are more likely to purchase insurancethan standard risks in the worksite as theyface stricter underwriting standards andhigher rates if they decide to purchase insur-ance as an individual. For this reason is itcritical to estimate the number of lives in eachrisk class.

Input from the underwriting departmentis critical for this step. Underwriters can useinformation from fully underwritten applica-tions and industry statistics to estimate thedistribution of the actively at work risk classdistribution. Table 1 shows the assumeddistribution for this case study and how thelives are distributed into the risk classes.

Step 3: Calculate the StandardMortality

The third step is to calculate the mortality fora standard, 100 percent rated, individual. Theconservation of deaths principle says that theactively-at-work population can be brokeninto risk classes and the result must sum tothe population mortality, creating a uniquemortality rate each year. The net effect is tohave one unknown, the standard mortalityrate, per year. Subsequent mortality rates canbe estimated by a bootstrapping method.Table 2 illustrates the method.

An important point is that the standardmortality rates calculated in this step arenot the mortality rates used for pricing theworksite product. The “standard mortalityrate” represents the mortality table for afully medically underwritten product. Thefollowing steps adjust the standard mortalityrate for both underwriting selection andparticipant anti-selection to arrive at aninsured mortality table that can be used inproduct pricing.

Step 4: Estimate the ParticipationRate

The fourth step is to estimate how manyindividuals in each risk category will applyfor insurance. For this case study Table 3shows three scenarios: 20 percent, 40percent, and 60 percent participation rates.Twenty percent participation means that 20percent of the standard risk class elects to

Table 1: Actively-At-Work Risk Class Distribution

Risk Class Numerical Rating Actively-at -Work Lives

Standard 100% 950

Substandard 300% 45

HIV Positive 2000% 5


continued on page 8

apply for insurance. The substandard riskshave a higher participation rate as theyknow that they are receiving value. However,it is unlikely that every member of a riskclass, even severely impaired classes, willapply for insurance. There are some individ-uals that will not elect to purchase insuranceregardless of the economic value. The appli-cation rate for impaired risks should bebounded between the standard participationrate and 100 percent.

It is important to note that the 20 percentparticipation scenario will generate anaggregate participation rate greater than 20percent because the substandard risks electto purchase insurance more frequently thanthe basic 20 percent. If an aggregate partici-pation rate of 20 percent is required, you cansolve for the standard rates that will yield a20 percent aggregate participation rate.

Step 5: Estimate the UnderwritingAcceptance Rates

The underwriting acceptance rate is thepercentage of applicants who are acceptedfor life insurance. All the standard individu-als should be accepted by the underwritingscreen. For substandard individuals thereare three ways that they can pass throughthe underwriting screen:

• The underwriting questions do not iden-tify them as substandard

• The individual knowingly commits fraud

• The individual is ignorant of his/her health conditions

The first situation gives the actuary aninteresting basis for dialog with the under-writer. The conversation could go somethinglike this:

Actuary: You say we have a 50 percentchance of identifying and rejecting asubstandard individual. What can we changeto move that percentage to 70 percent?

Underwriter: Adding an additional ques-tion about prescription medications willmove that percentage to 70 percent.

Actuary: Our producers are asking for ashorter application. What is the impact ofeliminating the question about medicaltreatment in the last five years?

Underwriter: Given the protective value ofthe other questions, that change will resultin identifying and rejecting 30 percent fewersubstandard risks.

Table 4 shows a sample underwriting accept-ance rate.


Worksite Mortality • from page 7

Table 2: Actively At Work Standard Mortality Calculation

Number of Lives

Year qx At Standard Substandard HIV+ Qx

At Work Work (100%) (300%) (2000%) (100%)

0 1,000.00 950 45 5

1 0.001185 998.82 950 * (1-q1) 45 * (1-3 * q1) 5 * (1-20 * q1) 0.001

=949.05 =44.87 =4.90

Table 3: Participation Rate

Standard Substandard HIV+ Aggregate

20% Participation 20% 70% 70% 22.5%

40% Participation 40% 80% 80% 42.0%

3 0.003538 992.93 944.31 44.19 4.42 0.003

2 0.002366 996.45 947.15 44.60 4.70 0.002

4 0.004695 988.27 940.53 43.66 4.07 0.004

5 0.005833 982.83 935.83 43.01 3.66 0.005

60% Participation 60% 90% 90% 61.5%

Step 6: Calculate the InsuredMortality Table

Calculating insured mortality involvesprojecting the number of lives in each riskclass and using the total number of lives toconstruct an insured mortality table, asshown in Table 5.

Expressing the resulting mortality as apercent of the actively-at-work mortalityillustrates the combined effect of anti-selec-tion and underwriting selection. Apercentage less than 100 percent meansthat the underwriting selection predomi-nates. A percentage greater than 100percent means that applicant anti-selectionpredominates.

Perform What-If Analysis

When pricing worksite products the actuaryoften has to quantify the answers to “What-if ” questions. What if we change theminimum participation requirement? Whatif we change the level of underwriting? Table6 illustrates how this process allows the

actuary to quantify the impact on mortalityof different proposed product designs.

Conclusion

There are many factors that are critical tothe success of this process.• Input from the underwriting department

is critical for both estimating the risk distribution of the actively-at-work mortality and the underwriting screen.

• It is critical to recognize that the stan-dard mortality table created in step three is not the table used for pricing the coverage. The table must be adjusted for participation and anti-selection to produce insured mortality.

• Cooperation and understanding between the underwriters and actuaries is critical.

• At the end of the process, the insured mortality must be reasonable.�


Worksite Mortality

Table 5: Insured Mortality

Year Standard lx Substandard lx HIV+lx qx Insured qx I/qxg

0 950*40%*100% 45*80%*50% 5*80%100% - -

=380 =18 =4

2 378.86 17.84 3.76 0.002550 107.76%

3 377.72 17.68 3.54 0.003803 107.50%

1 380*(1-0.001) 18*(1-0.003) 4 * (1-0.02) 0.001279 107.90%

=379.62 =17.95 3.92

Table 4: Underwriting Acceptance Rate

Standard Substandard HIV+

Guarantee Issue 100% 100% 100%

Simplified Issue without HIV Testing 100% 50% 100%

Simplified Issue with HIV 100% 50% 0%

20% 40% 60%

Participation Participation Participation

4 376.21 17.47 3.25 0.005028 107.11%

Table 6: First Year Mortality Ratios

Guarantee Issue 132.96% 114.12% 107.23%

Simplified Issue without HIV Testing 123.91% 107.90% 102.27%

Simplified Issue with HIV Testing 97.31% 92.02% 90.18%

Adrian R. Pask, ASA,

MAAA, is an associate

actuary at Milliman USA in

Windor, Conn. He can be

reached at adrian.pask@

milliman.com.

5 374.33 17.20 2.93 0.006219 106.62%

“F irst-dollar” reinsurance agree-ments have become common-place. In fact, figures indicate

that approximately 60 percent of new lifeinsurance sales are reinsured, driven by firstdollar reinsurance programs. In addition,many of these programs are being imple-mented on a coinsurance basis for level termbusiness—as opposed to yearly renewableterm (YRT) programs.

A direct writer of level term business cedessuch business on a coinsurance basis to areinsurer for two primary reasons.• Reinsurers are typically on the forefront

with mortality trends, and will be more aggressive on their assessment of mortality levels. This aggressiveness then gets passed on in their pricing.

• Direct writers like shifting the burden of the onerous Regulation XXX reserves on to the reinsurers.

Combining the above two reasons oftenleads to a leverage of returns for the directwriter. A product with sub-par profitabilitywithout reinsurance suddenly becomes a prof-itable product through the use of coinsurance.

There are recent signs that the use ofcoinsurance and the inherent shift of theburden of the reserves to the reinsurers areputting stress on the reinsurers’ capacitylimitations. Reinsurers are able to shoulderonly so much of the burden of the reservestrain caused by Regulation XXX. Whatreinsurers do best, and know best, though,is mortality risk.

If your company is seeing signs fromyour reinsurers that this is the case, thisarticle will present some thoughts toconsider as to why sales of level term prod-ucts may still be acceptable, even withoutthe support of coinsurance programs.Placing less reliance on coinsurance willtake some pressure off reinsurers to findthe capacity for reserves, while maintainingthe ability of direct writers to leverage off of

life reinsurers’ ability and willingness to beon the forefront regarding mortalityassumption levels.

Value Creation Analysis

Much of the discussion below will make useof embedded value (EV) analysis. EV is anexcellent means to better understand howmuch value is being created for a company.EV is the present value of distributable earn-ings, where distributable earnings aredefined as after-tax book profits, less the costof holding required surplus. EV has becomepopular in Europe, as well as in Canada. It isalso gaining steam as an accepted and moreinsightful form of reporting in the U.S., aswell (when compared to U.S. GAAP).

The discount rate used to calculate thepresent value of the stream of distributableearnings varies from company to company inperforming an EV calculation. In today’senvironment, the discount rate typicallyused to discount the distributable earnings isin the 7 percent to 9 percent range. The rateused to calculate EV is called the RiskDiscount Rate (RDR).

For a product priced with a double digitinternal rate of return (IRR) on distributableearnings, discounting the same stream ofdistributable earnings at a 7 percent to 9percent rate will generate a value greaterthan zero.

Table 1 on the next page provides a simpleexample of the development of IRR and EVfor a given flow of anticipated distributableearnings for a hypothetical book of life insur-ance business.

Assuming that EV is the appropriatemeasure of value, the pricing departmentshould develop products that maximize EV.Maximum EV is not necessarily tied to prod-ucts with the highest IRR. Maximizing IRRcertainly helps, but maximizing IRR does notalways maximize EV. For example, one way tomaximize IRR is to minimize the investment,but a large IRR on a tiny investment couldequate to a small value for EV.

Placing lessreliance on coinsurance will take somepressures off reinsurers to findthe capacity forreserves...


Is Your Reinsurance Creating Enough Value?by James W. Dallas

Features

EV is also a function of gross sales. Thehigher the IRR, presumably the lower thesales. When you multiply the sales timesthe EV per unit of sales for a given level ofIRR, your optimal EV does not necessarilyoccur at the largest IRR. For a simple

example, see Table 2 above. Optimization ofEV does not occur at the point that IRR perunit is at its maximum.


TABLE 1: IRR AND EV FOR A STREAM OF DISTRIBUTABLE EARNINGS

TABLE 2: OPTIMIZATION OF EV

Pricing IRR per Anticipated Units EV per EV

Scenario Unit of Sales Unit Created

Year After Tax Book Effect of Required DistributableProfits Surplus Earnings

(+) (-) (=)

1 - $105 $5 - $110

2 $35 $1 $34

3 $35 -$2 $37

4 $35 -$1 $36

5 $35 -$3 $38

IRR: 11.8%

EV at 7%: $11.65

EV at 9%: $6.47

1 9% 110 10 1,100

2 12% 100 12 1,200

3 15% 60 14 840

TABLE 4:KEY FIGURES PER UNIT ISSUED 80% COINSURANCE

Pricing Embedded Embedded

Scenario Value at Value at

7% RDR 8% RDR

No Reinsurance 1.90 -0.13

9% IRR After

Reinsurance 0.81 0.38

11% IRR After


10% IRR After


12% IRR After


TABLE 3: KEY FIGURES PER UNIT ISSUED

Pricing First Initial Maximum Embedded

Scenario Year Statutory Statutory Value at 7%

Loss Reserve Reserve Over RDR

Pricing Horizon

No reinsurance: 33.16 32.28 33.52 1.90

continued on page 12

EV and Coinsurance of Level Term Products

To see how EV analysis could apply to thepricing of level term business and the asso-ciated use of reinsurance, I developed anillustrative term product to guide usthrough value creation under various rein-surance programs. To begin, I develop EVwithout use of reinsurance and with theuse of coinsurance. Below is a briefsummary of the assumptions:

No Reinsurance• 20 years of level premiums, fully

guaranteed

• No reinsurance

• Profit margin after tax and after use of required capital of 5 percent, which resulted in an IRR of just under 8 percent

Table 3 on page 11 summarizes some of thekey figures for products without reinsurance.Ithen solved for various coinsurance allowancestructures that accomplished varying levels ofleverage for my baseline product. I did this byfirst assuming 80 percent first-dollar quotashare coinsurance in all instances and a 100percent of premium first year allowance. Ithen solved for the renewal allowance struc-ture that would leverage the IRRs to 9percent, 10 percent, 11 percent and 12percent.

Table 4 on the bottom of page 11 summa-rizes the results. As you can see, the

embedded value at a 7 percent discount doesnot increase over and above the base caseuntil sufficient allowances are received fromthe reinsurers to increase the IRR above 11percent. So, certain coinsurance programsmay actually decrease the value beingcreated, even though they may leverageIRR.

Note that the analysis above is depend-ent on the level of the RDR. Using an 8percent RDR demonstrates that any coin-surance program that increases the IRRabove the baseline IRR of 8 percent willincrease value for the company over and theno reinsurance scenario.

EV and the Use of YRT

What happens, though, if I use YRT reinsur-ance, rather than coinsurance, for myillustrative product? In this analysis, I usedthe same base product, but reinsured 80percent of the mortality risk on a YRT basis,rather than on a coinsurance basis.

To analyze the effect of various YRTscenarios, I varied the level of YRT rates thatI might receive from my pool of reinsurers. Ilooked at the results under a variety of rein-surance rates ranging from 5 percent belowmy mortality assumption built into my baseproduct, to 5 percent higher than my base-line mortality assumption.

Table 5 below summarizes summarizessome of the key figures per unit.

By comparing Table 5 EV results to theTable 4 EV results under coinsurance, you cansee that using an RDR rate of 7 percent, the


Pricing XXX Reserves... • from page 11

TABLE 5:KEY FIGURES PER UNIT ISSUE YRT REINSURANCE

Pricing Embedded Embedded

Scenario Value at 7 % Value at 8%

RDR RDR

No Reinsurance 1.90 -0.13

YRT Rates 5%

Below Pricing 2.25 0.25

YRT Rates 2%

Above Pricing 1.75 -0.20

YRT Rates 2%

Below Pricing 2.04 0.05

YRT Rates 5%

Above Pricing 1.54 -0.39

EV for mortality rates as high as 2 percentmore than the pricing mortality assumption ishigher than coinsurance allowances that lever-age the return up to more than 11 percent. IfI’m able to secure reinsurance rates at orbelow my pricing mortality assumption, theEV is better than coinsurance allowances thatleverage as high as 12 percent.

This example illustrates that by using EVanalysis, consideration of both coinsuranceand YRT reinsurance programs could proveinsightful. However, you will notice, bycomparing the 8 percent RDR columns, adifferent RDR rate will lead to differentconclusions. Your company’s final results willdepend on your company’s RDR rate.

It is also useful to analyze the effects ofvariations in results from that expected. Forexample, the effect of mortality fluctuationswill be different for the coinsurance and YRT

scenarios. You’ll see that the YRT scenariocreates more leverage on the EV results thandoes the coinsurance program, with EVbeing increased more when mortality is lessthan expected, and the opposite effect whenmortality is greater than expected.

Conclusions

Despite the fact that coinsurance programsremove much of the strain and reserves, ananalysis using embedded value could shedsome light on the sensitivity of results undervarious combinations of YRT and coinsur-ance programs. Using YRT rather thancoinsurance, may give you the ability tomaximize your embedded value, while at thesame time giving you the opportunity toleverage off your life reinsurers’ expertiseregarding mortality expectations.�


Pricing XXX Reserves...

AAA—Illustrations Work Groupby Michael S. Taht

R ecently, the Life ProductsCommittee of the AmericanAcademy of Actuaries “the

Academy” formed a new work group. TheIllustration Work Group “IWG” has beencreated to research and review issues withrespect to life insurance illustrations. Basedon the particular issue, the IWG will providefeedback to constituents such as theAcademy’s Life Products Committee, NAIC,LHATF and the Academy’s general member-ship. The first issue that the IWG willexplore is whether mortality improvement isimplicitly being utilized in certain late dura-tion mortality assumptions supporting someillustrations.

The explicit use of mortality improvementin the preparation of Illustration ActuarialCertifications is prohibited by the ModelIllustration Regulation. However, thereappears to be great divergence in the level oflate duration cost of insurance rates onuniversal life and variable universal lifeproducts. While this may be the result ofsignificantly different opinions on late dura-tion mortality, the IWG is researchingwhether this is also the result of mortalityimprovement being implicitly assumed by

some actuaries in setting mortality assump-tions. The existence of significant differencesin late duration mortality assumptions couldalso be due to the lack of late durationmortality experience on business issued intoday’s underwriting and mortality environ-ment. No matter what the cause for thedifference in late duration mortality is, if lateduration mortality turns out to be signifi-cantly higher than assumed, there is aconcern expressed by some actuaries that lifeinsurance illustrations will not be support-able in the future.

The Life Insurance Illustrations ModelRegulation’s goal was, in part, to ensure thatlife insurance illustrations do not misleadpurchasers of life insurance. Consistent withthis goal, the IWG is researching and identi-fying situations where divergent opinionsexist regarding late duration mortality andthen providing further guidance (above andbeyond the current ASOP and practice notes)to actuaries, in setting mortality assump-tions at late durations.

For more information regarding the IWG,please contact its chairperson, TraceyPolsgrove at [email protected].�

James W. Dallas, FSA,

MAAA, is a senior consult-

ant at Tillinghast-Towers

Perrin in St. Louis, MO.

He can be reached at jim.

[email protected].

Michael S. Taht, FSA,


actuary at Towers Perrin in

Atlanta, GA. He can be

reached at mike.taht@

tillinghast.com.

O ne of the most successful productsin the current market is the death-benefit-focused UL. This type of

product offers guaranteed lifetime protection(using a secondary guarantee and a maturityextension) at a very affordable price. Manycompanies have utilized a shadow accountapproach to make the guaranteed coverageeven more affordable.

The shadow account is a value that iscalculated similar to a UL cash value.However, the shadow account has its own setof charges. As long as the net shadowaccount value is positive, the policy will stayin force regardless of whether the cash valueis sufficient to cover the insurance charges.

Pricing a product with a shadow accountcan be very difficult because of the need toreflect the reserves required under GuidelineAXXX. However, once the product is success-fully priced, there is still the difficult task ofgetting the product to market. This task canbe especially difficult for companies withlegacy administrative systems.

Two Account Values

The biggest challenge to using a shadowaccount is calculating two separate accountvalues: one for the shadow account and onefor the actual cash value. When premiumsare paid, they must be applied to both theshadow account and the cash value.

One approach is to use a single plan(policy record) that contains the charges forboth the shadow account and the actual cashvalue. When premiums are paid, the plancalculates both the shadow account valueand the actual cash value (See Figure 1).

This approach can be difficult because ofthe need to store two sets of charges. Thiscan be especially difficult if the charges arevastly different. For example, actual COIcharges may be on an attained age basis, butshadow charges could be select and ultimate.

Another option for keeping track of twoaccount values is to use two plans. One plancontains the charges for the shadow account

and one contains the actual charges used incalculating the cash value. In this case, thepaid premiums are sent to both plans. Thefirst plan calculates the shadow value andthe second one the cash value (See Figure 2).

The biggest challenge under this approachis making sure both plans contain the sameinformation. The same premium must besent to both plans. Moreover, when a changeis made to the policy, it must be made to bothplans. Also, there must be a link between thetwo plans to indicate when the shadowaccount is no longer positive and the guaran-tee is no longer in effect.

Overall, there are difficulties with bothapproaches to calculating two separateaccount values.

Communication

Communicating the concept of a shadowaccount is another challenging endeavor.Several terms are used in the actuarialcommunity that may not be appropriatewhen marketing to potential policyholders.For instance, the term “shadow” maysuggest something secretive or deceptive,which is probably not the image that isdesired. Moreover, referring to an “accountvalue” may give the impression that theshadow account has a surrender value,which is not the case. Terms such as“balance” or “threshold” may be moreappropriate.

Once the terminology is in place, theshadow account can be described to policy-holders and agents. The shadow accountcalculations can be explained in great detailor the discussion can be limited to the guar-antee provided by the shadow account. Ineither case, it is important for agents to befamiliar with the unique features of ashadow account guarantee.

Policy Form

Since the shadow account provides a guaran-teed death benefit, the policy form for the

The biggest challenge to using a shadowaccount is calculating twoseparate accountvalues....


Shadow Account Products: Beyond Pricingby Jeremy A. Bill

Features

shadow account must contain enough infor-mation so that the policyholder canreproduce the shadow account calculations.At least two approaches can be used toaccomplish this goal:1) Use the description of the cash value

found in the base contract and describe any differences. This approach produces a shorter form, but can be confusing if there are a lot of differences between the shadow account calculations and the cash value calculations.

2) Describe all of the calculations from scratch, but apply them to the shadow account values. This approach takes up more space but is effective in separating the shadow aaccount from the cash value.

When describing the charges that will beused in calculating the shadow account, it’s

again important to consider the terminologythat will be used. Also, since these chargesare guaranteed, they will need to be printedsomewhere in the policy form. A separateschedule of charges can be included thatapplies only to the shadow account. This canhelp to distinguish between the guaranteedcharges that apply to the actual Cash Valuesand the charges that are guaranteed for theshadow account calculations.

Conclusion

All in all, the process of administering andfiling a shadow account product can be evenmore difficult than pricing the product.However, with hard work and creativity, it ispossible to effectively market and administera very competitive, complex product.�


Figure 1 – Using One Plan to Calculate Shadow Account Value and Cash Value

Figure 2 – Using Two Plans to Calculate Shadow Account Value and Cash Value

Premiums

One Plan

ShadowCharges

ShadowValue

CashValue

ActualCharges

Premiums L ink

Two Plans

ShadowCharges

ShadowValue

CashValue

ActualCharges

Jeremy A. Bill, ASA,

MAAA, is an actuarial

assistant at Midland

National Life Insurance in

Sioux Falls, SD. He can

be reached at jbill@mnlife.

com.

T he extended bear equity market hasdriven down sales of VUL andcaused potential buyers to desire

downside protection. The UL market with itslifetime guarantees based on very aggressivepremium levels has set the standard for thisdownside protection. Distributors, and to alesser extent buyers, focus on UL requiredpremium levels more than the requiredpremium level of other VUL products withlifetime guarantees.

In addition to the aggressive UL requiredpremium levels, developers of VUL second-ary guarantees must overcome conservativereserving requirements, the absence of viablereinsurance options and the negative percep-tion of senior management due to the lossesexperienced by many carriers on variableannuity guarantees. The incidence of riskunder VUL death benefit guarantees is muchdifferent and smaller than guaranteescontained in variable annuities. These chal-lenges may cause the optimum path to beone of including a “UL Lifetime Guarantee”within a VUL Product.

Reserves:

Reserves for this guarantee provide for deathbenefits that exceed those which exist in theabsence of the secondary guarantee. Thereserve methodology is contained in theVariable Life Model Regulation and is clari-fied in Actuarial Guideline XXXVII (AG 37).AG 37 enables consistent reserve treatmentof VUL guarantees even though the earlierversion of the model regulation (which is theonly version approved in many states) didnot anticipate current types of long-termVUL guarantees.

The reserve is the greater of (1) the one-year term reserve (OYT) and (2) the

attained age level reserve (AAL). The OYTreserve is designed to cover extremecircumstances that could occur prior to thenext annual statement. The AAL reserveprovides for the full risk period but isdesigned to build and decrease slowlythrough periods of poor and favorableinvestment experience, respectively.

The OYT reserve essentially is equal tothe total term cost for up to one year that isnot covered by a “reduced” account value.The “reduced” account value is equal to thevaluation date account value after the sepa-rate account portion is reduced by one third.

The AAL reserve equals the “residue” ofthe prior year’s AAL reserve increased by adefined payment (which can be positive ornegative). The “residue” is the prior year’sAAL reserve adjusted with persistency andinterest and reduced by tabular claims dueto the guarantee. The defined paymentequals (A) the present value of projectedfuture guaranteed minimum death benefitsless (B) present value of projected futuredeath benefits in the absence of the guaran-tee less (C) the “residue” divided by a levelannuity factor for the guaranteed period.

AG 37 clarified the following calculationdetails:• Secondary guarantee must be assumed

to remain in force if contractually possible.

• Projections and discounting are based onvaluation mortality and interest andignore product loads.

• XXX select factors do not apply.• In the calculation of AAL, at all points

during the guarantee period the quantity (A)-(B) is floored at zero.


VUL Secondary Guarantees:Catalyst for Sales Reboundby Thomas E. Norton and Robert Grotyohann

Features

Competitive Overview:

The death benefit guarantees currently offeredunder VUL products can most logically beseparated by the length of the guaranteeperiod. Most VUL contracts offer a short-termbenefit and may also offer an intermediateterm or a lifetime guarantee or both. Short-term benefits provide a three- to 10-yearguarantee. They essentially enable a lowerminimum premium by deferring the need for apositive surrender value. Intermediate termbenefits usually provide a guarantee to “retire-ment” (i.e. until the later of age 65 or 70 and20 years). The “best” intermediate termdesigns usually require payment of thecommissionable premium. Lifetime guaran-tees are currently gaining most attention, buthave very high required premiums.

The death benefit guarantees can also beseparated by the methodology for measuring“required premiums.” The CumulativePremiums Method requires that the sum ofpremiums paid less partial withdrawals lessoutstanding debt exceeds the sum ofrequired monthly premiums since issue.With the Interest-Adjusted PremiumsMethod, premiums and withdrawals areadjusted with an interest factor from point ofpayment before comparison to the requiredmonthly premium adjusted with interestfrom the assumed due dates. The third andfar less common method requires a positiveshadow account value based on actual premi-ums, withdrawals, debt and assumed returnand charges. All three of these methods maybe utilized with or without a catch-up provi-sion. Without catch-up, the premiumrequirement (or a positive shadow accountvalue) must be met on each processing date(or sometimes annually).

If it is deficient, a lapse notice is sent indi-cating that the guaranteed benefit will be lostif the required premiums are not paid. Withcatch-up, the owner is only required to havepaid sufficient premiums at the point thesurrender value is negative and the contractwould lapse without the secondary guarantee.A grace period is allowed for paying sufficientpremium to cause a positive surrender valueor to meet required premiums.

Recently two companies have addedsecondary guarantees maintained solely or

primarily based on “premiums” allocated tothe fixed account. This essentially creates aUL with a secondary guarantee and a tax-advantaged side fund. The requiredpremiums for the first of these two productsare competitive with UL products (we havenot yet seen required premiums for thesecond product). At a competitive ULpremium level (30 percent to 40 percent of aguideline annual premium for superpreferred), significant additional premiumscan be paid and allocated to generate retire-ment income or to increase corridor deathbenefits.

Figure 1 provides a sampling of requiredpremiums for lifetime VUL death benefitguarantees. The majority of these benefitshave required premiums greater than 85percent of a guideline annual premium. Oneof the products has required premiums inthe range of 65 percent to 70 percent of theguideline annual premium. Even at thislower level, it is significantly higher thansome UL guarantees.

Benefit Cost Analysis:

With some stochastic cost analysis, therequired premium levels shown in Figure 1can be logically explained. For our modeling,we captured the following key benefit cashflows on a stand-alone basis.• PV of the change in year by year

secondary guarantee reserves• PV of foregone monthly deductions due

to investment return scenarios whichtotally dissipated account values

• PV of benefit specific charges ($0.01 permonth per $1,000 of a specified amountin our case)

We discounted all cash flows at 12 percentinterest and expressed all PVs as an amountper $1,000 of an initial specified amount.Our analysis only reflects standard contractlapse and does not attempt to add benefitlapse or adjust contract lapse dynamicallybased on investment returns in each stochas-tic scenario.

For our modeling we created a genericdeath-benefit-oriented VUL for a male


With catch-up, theowner is only

required to havepaid sufficient

premiums at the point the

surrender value is negative...


preferred non smoker with issue age 55. Fora $500,000 specified amount, Death BenefitOption 1 and a gross return of 9.00 percent,our current assumption level premium tocarry to age 100 is $5,855.

In our model, investment returns wererandomly generated based on the independentlognormal distribution. The baseline analysisassumes a mean return of 10 percent and a 15percent standard deviation. Initially 1000scenarios were run. The results were thenordered (lowest cost to highest). Then everyfourth scenario was extracted for use on allother runs (250 scenarios).

Figure 2 displays our cost and revenuecomponents at the following three differentpremium levels.• $16,813 (95 percent of the guideline

annual premium)• $15,044 (85 percent of the guideline

annual premium) • $13,274 (75 percent of the guideline

annual premium)

These mean results clearly show that thereserve cash flows dominate the cost andgreatly increase as the required premiumdrops below 85 percent of the guidelineannual premium. At this mean and variancewe did not generate any scenarios where theaccount value was fully liquidated. Figures 3and 4 display the distribution of net costs forboth 85 percent of GAP and 75 percent ofGAP respectively.

Next we looked at sensitivity testing of themean and standard deviation of the invest-ment return by utilizing the following twocombinations and 85 percent of GAP.• Mean 10 percent/Standard deviation

20 percent• Mean 6.5 percent/Standard deviation

15 percent

The results from these distributions areshown in Figure 5. The reserve costs increasesomewhat, but the incidence of totally dissi-pating the account value stays at or close tozero. These limited results support arequired premium of roughly 85 percent of

GAP if the full statutory reserve effect isreflected. If nonstatutory measures are usedto set the profit goals or if some mechanismis employed to utilize reserves closer to theeconomic cost, then lower required premiumscan be justified.

Effect of 2001 CSO:

Since the XXX select factors do not apply tothis benefit, the lower mortality ratescontained in the 2001 CSO Table helpconsiderably. This can be viewed in Figure 6,which compares the reserve costs at severalrequired premium levels.

Other Considerations:

Financial reinsurance with a methodologysimilar to that utilized for UL secondaryguarantees would lower costs considerably.Reinsurance on this basis is essentially notavailable. Reinsurers are unwilling to spendlimited capital on an agreement that yieldsminimal incremental revenue. In addition,losses on annuity guarantees are causingmany reinsurers to avoid the equity guaran-tee market. Companies may have more luckpiggybacking their UL deals with a “UL life-time guarantee” within a VUL.

Another possible cloud on the horizon isthe proposed C-3 changes which wouldrequire risk capital based on stochasticmodeling at the modified CTE 90 level. Themodified CTE 90 level is the arithmetic aver-age of the worst 10 percent of all scenarios,with no scenario being calculated as a posi-tive value of accumulated surplus. Currentreserves for a lifetime benefit with arequired premium at 85 percent of GAPprobably exceed this level.

Conclusions:

The “UL lifetime guarantee” within a VULmay be an easier path for financial reinsur-ance and more in sync with buyers’ risktolerances at the tail of a bear market.Alternatively, early use of the 2001 CSOTable for death-benefit-oriented VUL shouldgenerate positive results.�


VUL Secondary Guaranteess... • from page 17

Thomas E. Norton, FSA,

MAAA, is president of

Norton Consulting Group

LLC in Emerson, NJ. He

can be reached at tom@

thenortonconsulting

group.com.

Robert Grotyohann, FSA,

MAAA, is vice president of

Norton Consulting Group

LLC in Emerson, NJ. He

can be reached at rob@

thenortonconsultin

group.com.


VUL Secondary Guaranteess...

Sample Required Premiums (Lifetime Benefit)Face Amount = $500,000 – Best Class – Male NS – DB Option 1

Company/Product Issue Age 45 Issue Age 55 Issue Age 65

Comp. A/Accum. $10,129 $16,192 $25,765

BENEFIT COST ANALYSIS (1980 CSO)Required Premium Level – Mean Costs – Issue Age 55

Premium PV of PV of Foregone PV of Net

Level Reserve Changes Revenue PV

95% of GAP $0.05 $0.00 $0.61 -$0.56

85% of GAP $3.56 $0.00 $0.61 $2.95

85% of GAP $13.92 $0.00 $0.61 $13.31

Comp. A/DB $9,215 $15,069 $24,659

Comp. B $9,985 $16,635 $29,495

Comp. C/Accum. $10,432 $16,896 $28,421

Comp. C/DB $10,020 $16,325 $27,750

Comp. E $8,790 $14,600 $24,365

Comp. D/DB $6,324 $10,160 $17,486

Benefit Cost AnalysisRequired Premium Level = 85% of GAP

0

10

20

30

40

50

60

70

-0.5 0.1 0.7 1.3 1.9 2.5 3.1 3.7 4.3 4.9 5.5 6.1 6.7 7.3 7.9 8.5 9.1 9.7 10.3 10.9 11.5 12.1 12.7 13.3 13.9 14.5 15.1 15.7 16.3

PV Reserve + PV Waived Chgs - PV of GDB Revenue per 1,000 of Face Amount

Figure 1

Figure 2

Figure 3


VUL Secondary Guaranteess... • from page 19

Benefit Cost AnalysisRequired Premuim Level = 75% of GAP

0

5

10

15

20

25

30

35

40

45

0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 13.5 15.0 16.5 18.0 19.5 21.0 22.5 24.0 25.5 27.0 28.5 30.0 31.5 33.0 34.5 36.0 37.5 39.0 40.5 42.0

PV Reserve + PV Waived Chgs - PV of GDB Revenue per 1,000 of Face Amount

BENEFIT COST ANALYSIS (1980 CSO)Investment Return Effect – Mean Costs – Issue Age 55

Mean/Standard PV of PV of Foregone PV of Net

Deviation Reserve Charges Revenue PV

10%/15% $3.56 $0.00 $0.61 $2.95

6.5%/15% $5.25 $0.01 $0.61 $4.61

10%/20% $4.56 $0.00 $0.61 $3.94

BENEFIT COST ANALYSIS (2001 CSO)Cost Comparisons with 1980 CSO – Mean Costs – Issue Age 55

Premium Level 1980 CSO 2001 CSO

PV of Reserve PV of Reserve

85% of GAP Exceeds 2001

(1980 CSO) $3.56 CSO GAP

65% of GAP

(1980 CSO) $28.04 $0.79

75% of GAP Exceeds 2001

(1980 CSO) $13.92 CSO GAP

60% of GAP

(1980 CSO) $36.23 $4.71

Figure 4

Figure 5

Figure 6

Y ou may have seen in the newspaperslast March a press release announc-ing that the U.S. had reached

another record for life expectancy. TheNational Center for Health Statistics(NCHS) had just released their report on2001 U.S. population mortality. (This reportis preliminary—the final report, whichincludes additional detail, likely will comeout this fall.)

The report is available on the internet atwww.cdc.gov/nchs/releases/03news/lifeex.htm. I used the report to update a table onU.S. population mortality improvementrates. Some comments on the table results,focusing on the four years ending 2001, areas follows:

• There has been robust improvement (2 percent+) for males ages 55-74. This is a key age group for life insurers when you weight by expected mortality.

• Males ages 45-54 did not fare as well in1997-2001, although they had done quite well the prior four years.

• You cannot tell much from the male ages 25-44 statistics. Looking at the entire history from 1987, you see deterioration

through 1993 on account of AIDS, and then substantial improvement as the AIDS deaths reduced.

• Female mortality at older ages continues to have smaller improvements than male mortality. In addition, female mortality at ages 35-44 has deteriorated in recent years.

• You might be wondering how much impact the September 11 event had on 2001 mortality population rates. Theanswer is “not much.” Consider that eachdecennial age group from 25-54 has about 40 million lives. An additional 1000 deaths in each decennial group would raise the death rate by only 2-3 per 100,000.

The table shows, for comparison, ProjectionScale G, which frequently is used to projectannuitant mortality. It looks not unreasonablefor males, although high at younger ages. Forfemales, Projection Scale G appears high. (Infact, when the Annuity 2000 table was devel-oped, it was developed by taking the priortable and projecting to 2000 using ProjectionScale G for males, but only half of ProjectionScale G for females.) �


U.S. Population Mortality Improvementby Douglas C. Doll

Annual Mortality Improvement – U.S. Population

Death Rates per 100,000 Annual Rates of Improvement

Age 1987 1990 1993 1997 2001 Males 1987-90 1990-93 1993-97 1997-01 1987-01 Scale G

25-34 189 204 212 163 143 -2.7% -1.3% 5.3% 3.0% 1.6% 0.5%

35-44 290 310 329 275 259 -2.4% -2.1% 3.9% 1.4% 0.7% 2.0%

45-54 638 610 603 548 544 1.4% 0.4% 2.2% 0.2% 1.0% 1.8%

55-64 1,626 1,553 1,480 1,343 1,192 1.5% 1.5% 2.2% 2.7% 1.7% 1.5%

65-74 3,636 3,492 3,411 3,170 2,914 1.3% 0.8% 1.7% 2.0% 1.3% 1.4%

75-84 8,206 7,889 7,700 7,055 6,842 1.3% 0.8% 2.0% 0.7% 1.1% 1.2%

Females

35-44 135 138 145 135 148 -0.7% -1.7% 1.6% -2.4 -0.7% 2.2%

25-34 74 74 74 68 66 0.0% 0.0% 2.1% 0.6% 0.7% 1.0%

45-54 367 343 334 310 316 2.1% 0.9% 1.8% -0.5% 0.9% 2.0%

55-64 910 879 868 806 754 1.1% 0.4% 1.7% 1.6% 1.1% 1.8%

65-74 2,070 1,991 2,010 1,937 1,892 1.3% -0.3% 0.9% 0.6% 0.6% 1.8%

75-84 5,102 4,883 4,824 4,832 4,764 1.4% 0.4% 0.0% 0.3% 0.5% 1.5%

O f the many concerns on the mind ofvariable annuity (VA) pricing actu-aries, one of the highest today is

investment risk. First, how do we price forit? Better yet, is there a way we can dispensewith it outright?

After all, we would like nothing betterthan a consistent predictable profit stream,risk-free. For VAs, our profit stream comeslargely from fees charged as a percent ofseparate account values, so fluctuations inthose values are counter to our mainpurpose. The addition of guarantees makesthe situation worse. And the proposed RBCrequirements for variable products has addi-tional pricing implications regarding thoseguarantees.

As risk managers we seek to acquiresomething that is negatively correlated tothe profit stream we now possess. The closerthat “something” is to 100 percent negativelycorrelated with VA profits, the closer we aremathematically to being able to create afixed profit stream. The negatively correlatedinstrument is well known to risk managersas a “hedge.” We might seek to hedge ourposition regarding either guarantee risk (inorder to minimize required capital) or totalprofitability.

Obtaining the perfect hedge is easier saidthan done. The first complicating issue is cost.The fact that we are seeking to “acquire”

something generally implies that we have topay something for it. If the price exceeds thevalue we are trying to protect, there is obvi-ously no future in the transaction.

If we attempt to build a strategy on theasset side, via options and/or futurescontracts, we may be able to hedge very wellindeed. But the more perfect we attempt tomake the hedge, the more trading we willhave to do to keep the assets and liabilitiesin balance as conditions change, increasingtrading costs. (Uncertainty about futuredecrement rates makes this especially toughfor some benefits.) More importantly, marketvolatility could turn out to be higher thananticipated when a hedging strategy isdevised. If so, our hedging cost will turn outto be higher than budgeted as well, all otherthings being equal.

Hedging our risks via reinsurance mightseem attractive, and a few reinsurance solu-tions for guarantee risk still can be found attimes. One factor making reinsurance poten-tially appealing is that the cost is oftencharged as a percent of asset fees. This leavesus to simply enjoy the remainder, if any.Another nice feature is that, if the reinsur-ance completely covers the risk, we do in facthave 100 percent negative correlation. Theguarantee-based reinsurance cash flows (netof premiums) exactly mirror our direct flows.

The problem with reinsurance recently isthat many programs do not completely covertail risk. Such “partial tail coverage”programs often assign losses over a certainlevel back to the direct writer. This meansthat the 100 percent negative correlationonly extends to the cap. Scenario testing ofsuch a program might determine thatrequired capital levels are barely reduced bythe intended hedge. Thus the hedge fails toachieve the intended pricing impact (reduc-ing RBC). Lastly, one more risk to consider isreinsurer insolvency.

One relatively new rider to variable annu-ities was initially billed as a good hedge for

Variable Annuity Risk and Seeking “The Perfect Hedge”by Douglas L. Robbins


Features

total VA profitability, at essentially no netcost to the writer. That was the enhancedearnings death benefit (EEDB). For an addi-tional asset fee, this rider would add 40percent or so of contract gains to the annuitydeath benefit, ostensibly to cover incometaxes on gains. The gains considered in theadditional benefit were capped at somemultiple of premiums paid, but because thisdeath benefit paid off in an up market(rather than a down market, like most otherguarantees), it was considered to be nega-tively correlated with other VA features andriders.

In fact, stochastic pricing of a VA with anEEDB added in produces quite similar tailresults to those of the VA alone. At the uppertail, the cap on the EEDB limits its potentialcost to the insurer, although its asset feesgrow just like those of the base VA (andother riders), in very good scenarios.

More importantly, at the lower tail, wherethe EEDB costs nothing, the additionalEEDB fees generated (since they are asset-based) are also at their lowest levels. And atthat point, the worse things get for the VAand GMDB, the worse they get for the EEDBtoo.

In fact, like the situation discussed abovewith “partial tail coverage” reinsurance, thenegative correlation between the EEDB andVA/GMDB is only over a band of fund valueranges. In this case, it is a narrow band,between premiums paid and the benefit cap.More importantly, the correlation outsidethat band is not merely zero. It is positive,due to EEDB charges being based on thesame fund value as other asset charges.

It is apparent that many things mightappear to be hedges, but fail to do the job insome cases. We have been in a market overthe past three or so years that would havebeen perceived as unlikely at its outset. Thisscenario would have defeated some of thestrategies we have discussed here, but not allpotential ones.

Product development actuaries are contin-uing to seek the “golden bullet” for managingVA risk. They will continue to look on boththe asset and liability side, and may alsoconsider hedges between different products(e.g., VA vs. EIA, a topic for another day). Themost diligent actuaries will carefully testany potential new strategy over a wide rangeof potential future scenarios.�


Risk-Relevant Resourcesby The SOA Risk Management Task Force

L ooking for timely, thought provokinginformation on risks affecting yourline of business? Why not visit the

SOA Risk Management Task Force Web siteat http://www.soa.org/sections/rmtf/rmtf.html

Created back in 2002, task forcesubgroups have been researching and writ-ing about all facets of risk that affect theindustry. Not only will you benefit from theresearch and documentation available on thesite, you’ll find useful links to other risk-oriented resources, network opportunitiesand events. Subgroups include:

• Economic Capital Calculation and Allocation

• Enterprise Risk Management

• Equity Modeling

• Extreme Value Models

• Health Risk Management

• Policyholder Behavior in the Tail

• Pricing for Risk

• Risk Based Capital Covariance

• Risk Management Metrics

Please take this opportunity to visit thesite, add it to your list of favorites forfrequent review and send your comments,questions and considerations to RMTFcontacts.

The RMTF welcomes and needs yourparticipation too! If you would like to learnmore about the Risk Management TaskForce in general or any of its subgroups,contact Dave Ingram or Valentina Isakina [email protected] or [email protected].�

Douglas L. Robbins, FSA,


actuary at Tillinghast-

Towers Perrin in Atlanta,

GA. He can be reached at

doug.robbins@tillinghast.

com.

W hat does Internet distributionlook like three years into thedownturn? There is some indica-

tion of stability in the marketing groupsafter they suffered sharp drops in sales. Arecent survey that I performed found 175sites that sold life insurance on the Internet.Over 100 of those were multi-company sitesthat sold at least 10 companies’ products.

The most prominent sites probablyaccount for the majority of sales, although Ido not know of any measurements of thetotal market. Two of the main sites arepublicly quoted companies and thus havepublished their own results.

QuoteSmith had its second best sales yearin 2002. Their reported term sales (numberof policies) are showing stability over lastyear’s sales and overall appear to parallelthe industry pattern. (See table on top ofpage 23).

InsWeb reports the fee income it gets forproviding term leads to companies. Theirdrop in sales occurred in 2002. The secondhalf of the year is down by one-third fromthe previous year (See table on bottom ofpage 23).

Term still remains the most commonproduct sold on the Internet. The belief isthat it is less complicated and can be morereadily self-bought by the consumer. Thereare several attempts to sell annuity productson several sites. Almost no one is attemptingto sell universal life.

Underwriting Techniques

The biggest area of innovation is on theunderwriting front. Attempts to bring newunderwriting tools to replace full underwrit-ing are coming from new directions on afrequent basis. In general, the idea is tosupplement simplified issue questions withnonconventional underwriting tools such ascredit reports and prescription databases.The goal is to have the trade-off between

price and convenience at an acceptable levelto minimize anti-selection. The Internet alsoallows easier access to some affinity groups,which will help mitigate anti-selection.

Pricing Assumptions

Distribution costs have a very differentprofile for Internet sites than for other director agent distribution. Many of the sites haveup-front fees to handle setup of your policyand rates on their system. They also willcharge an annual fee to carry your producton the site. These costs must be spreadacross the expected sales. You can compareyour product to those on the site and esti-mate your expected traffic.

In addition to the bulk charges for theWeb site, there are often per-lead or applica-tion fees. They typically range from $80 to$125. Thus, a key pricing assumption is thepull-through rate—the percentage of applica-tions that turn into policies. Sometimescompanies have requested a guaranteedminimum on the pull-through rate or evenhave negotiated to not pay for leads abovetwice the number who close. This is a criticalitem to manage the cost, in order to achieveoverall profitability.

Internet distribution is more likely tohave a flat commission per policy or lead.Such an arrangement is not typical foragent distribution. This means that there isa need for good information on average sizeor age. However, the Web site may be able tofilter out applications for small amounts oryoung ages and reduce the less profitablecases to a very small percent of your prod-uct’s sales.

Internet distribution is also commonusing company-owned Web sites or affinitysites. The sites are accessed by a clickthrough button on a general company homepage. Sometimes a banner ad is used tobring customers into the Web site. The mainadvantages that the sole company Web site

Pricing for Internet Distributionby W. Howell Pugh

Features


W. Howell Pugh, FSA,

MAAA, is president of

Howell Pugh Consulting in

Indianapolis, Ind. He can

be reached at howellp@

comcast.net.


brings are that the prospects are moreuniform and costs can be better controlled.

Underwriting and issue costs are similar todirect marketing costs. The use of a call centerto manage the application process is recom-mended. This can also be a key weapon tomanage the pull-through rate. Exam require-ments generally are similar to agent-soldproducts. Some companies are attempting touse technology to control these costs.

Lapse experience for some of the earliercompanies is about half the rate of agentsold business. The buyers are in control.They feel good about their choice of productand the shopping process to obtain it. Thereis not a need to repeat the shopping experi-ence. There is no agent to push for newapplication.

Mortality for fully underwritten Internetdistribution is the same as for agent distri-bution. However, because of the Internet, if

you have an “underwriting hole,” word canspread quickly. Since Web sites now haveflexibility to handle substandard conditions,it is important to monitor your own under-writing rules and how the Web sites haveconstructed their rule set to make sure thatyou have not created an outlier.

Reinsurance is used for Internet term tomitigate the XXX reserve strain. Sometimesthe reinsurance allowances can help mitigatethe risk on the direct side. Using a normalallowance structure would mean that per-policy distribution costs would bemismatched by percent of premiumallowances. See if the reinsurer will sharesome of the policy issue risk.

Internet term is a product for a newdistribution. Some of the pricing assump-tions are unique, while other assumptionsare familiar to those who price agent-soldterm.�

QuotesSmith 2002 Reported Term Sales (number of policies)

Year Quantity Sold

2002 17,622

2000 33,491

1998 10,920

1996 6,649

1999 17,039

1997 8,755

2001 16,915

InsWeb Quarterly Reports (fee income $1000s)

2002 2001 2000

4th Quarter $733 $1,202 $642

3rd Quarter 938 1,411 544

2nd Quarter 1,100 1,142 422

Total $3,814 $4,677 $2,119

1st Quarter 1,000 970 511

Introduction

O nce in a generation a new mortalitytable becomes the legal standard forstatutory reserve liabilities and

nonforfeiture benefits for ordinary life insur-ance in the United States. In December2002, the National Association of InsuranceCommissioners (NAIC) adopted a modelregulation formally known as RECOGNI-TION OF THE 2001 CSO MORTALITYTABLE FOR USE IN DETERMININGMINIMUM RESERVE LIABILITIES ANDNONFORFEITURE BENEFITS MODELREGULATION. This action marked the endof an extensive development process and thebeginning of a transition period that mayextend until 2009.

State laws and regulations specify stan-dard mortality tables to measure lifeinsurance reserve liabilities required to beheld in insurance company financial state-ments filed with state insurance authorities.Mortality tables are also used under statelaw as standards for minimum cash surren-der values and paid-up insurance benefitsprovided in life insurance policies. Underfederal income tax law, these tables affect lifeinsurance company reserve deductions andthe definition of life insurance for federalincome purposes.

The 2001 CSO Table

The 2001 CSO Mortality Table was devel-oped by the American Academy of Actuariesworking with the NAIC Life and HealthActuarial Task Force (LHATF). The tablewas based on the 2001 Valuation Basic Tabledeveloped by the Society of Actuaries for thatpurpose. This marks the first time that amajor valuation table was developed as ajoint project of the AAA and the SOA withthe Society acting in its basic research roleand the Academy serving in its public inter-face role. There was extensive liaisonbetween the organizations at every step ofthe process. Hopefully this will serve as amodel for future joint projects.

Like the 1980 CSO Table that it replaces,the 2001 CSO Table is a family of tables.There are nonsmoker/smoker versions, as

well as composite versions that do not distin-guish by smoking habits. There aregender-blended versions for nonforfeiturepurposes that combine male and femalemortality rates. There are select and ulti-mate versions as well as ultimate versions.The select/ultimate tables are presented inthe classical format, rather than in the formof select factors used with the 1980 CSO. Thelength of the select period is 25 years, whichis significantly longer than the 10 years usedwith the older table. There is no newCommissioners’ Extended Term (CET) Tableto replace the 1980 CET Table; therefore, thenew minimum basis for the computation ofvalues related to extended-term insurancebenefits will be the 2001 CSO MortalityTable. The volume of insurance continuedunder extended-term insurance provisionshas declined, and the SOA task force did notfind sufficient differences in experienceunder that provision to justify recommenda-tion of a new CET Table.

A complete discussion of the developmentand extensive analysis of the 2001 CSOMortality Table, including all the tables inExcel spreadsheet format, may be found onthe SOA Web site (www.soa.org) in the Reportof the American Academy of Actuaries’ CSOTask Force. The Society of Actuaries Report onthe Valuation Basic Table is also included asan appendix to that report.

The NAIC Model Regulation

The NAIC model regulation sets forth therules for use of the 2001 CSO MortalityTable.

Authority to adopt the new commissioners’mortality tables by regulation is found in theStandard Valuation Law and the StandardNonforfeiture Law for Life Insurance. Animportant exception to this is the state ofFlorida where legislative action is necessary toauthorize adoption of the regulation.

Unlike the 1980 CSO Table, where thesmoker/nonsmoker and gender-blendedversions were adopted at different timessubsequent to the composite table, the 2001CSO Mortality Table is a complete package.Consequently, the entire enabling regulationis contained in one NAIC model regulation.

2001 CSO Table—The Road to 2009by William Carroll

This marks thefirst time that amajor valuationtable was devel-oped as a jointproject of the AAAand the SOA...


That regulation was developed by LHATFafter extensive discussion among regulatorsand industry. Regulators were nearly unani-mous and all industry concerns wereadequately addressed. Consequently, itshould eventually be adopted by all states.

Timetable for Use of Table

Under the NAIC model, companies arepermitted to elect to use the 2001 CSOMortality Table as the minimum standard ona plan-by-plan basis for policies issued on orafter January 1 of the year next following orcoincident with the effective date of stateadoption of the regulation.

Companies must use the 2001 CSOMortality Table as the minimum standard forpolicies issued on or after January 1, 2009.

Key Provisions

The flexibility that developed over the past20 years is continued under the model.

There was never any thought given byLHATF to restricting the flexible use of thesmoker/nonsmoker or composite tables.Companies may continue to use either thecomposite table or the smoker/nonsmokertables for all valuation and nonforfeiturepurposes. Or, companies may use thesmoker/nonsmoker tables to determine thevaluation net premiums and additional mini-mum reserves, if any, under Section 8(“Deficiency Reserves”) of the StandardValuation Law (SVL) and the composite tablesfor basic reserves and nonforfeiture benefits.This remains optional on a plan-by-plan basis.

Flexibility was not so easily arrived atwhen it came to flexible use of the ultimateor select/ultimate tables. One leading stateactuary believed that a company should notbe able to use one form of the table for basicreserves and the other form for deficiencyreserves. This issue first arose with theintroduction of the 10-year select factors foruse with the 1980 CSO table.

The SVL merely says that the table foruse in determining the minimum standard isthe 1980 CSO with or without 10-year selectfactors. Soon after adoption of the 1980amendments to the SVL, the practice ofusing the ultimate form for basic reservesand the select/ultimate form (the 10-yearfactors) for “deficiency reserve” testingemerged. When the NAIC adopted the initialversion of Valuation of Life Insurance ModelRegulation (Regulation XXX) in 1995, it was

made clear that the choice of table for basicreserves and deficiency reserves was inde-pendent. After much discussion by LHATF,this flexibility was continued.

The 2001 CSO Mortality Tables also applyto Regulation XXX. Actuaries who enjoydetails will relish this section of the model.Others will be satisfied to learn that the2001 CSO replaces the 1980 CSO and every-thing else remains about the same. An entiresection of this model regulation is devoted tothis issue. The NAIC staff has painstakinglylocated every reference in the existingRegulation XXX to a mortality table or rateand has indicated which part of the 2001CSO Table applies. To oversimplify withoutmuch loss of accuracy, parts of the 2001 CSOreplace corresponding parts of the 1980 CSO.This is not quite precise because the twotables do not have exactly correspondingparts. (For example, where Regulation XXXpermitted 1980 CSO with or without 10-yearselect factor, but did not permit the 20 yearselect factors developed for Regulation XXX,the 2001 CSO model permits only the 2001CSO ultimate table.) Importantly, the use ofX-factors with the select 2001 CSO Table iscontinued for deficiency reserve subject togenerally the same conditions.

There was brief, but never serious consid-eration, of extending the use of X-factors tobasic reserves. One new aspect is that theactuarial demonstration required when X-factors are used may not combine the resultsof tests that utilize the 1980 CSO MortalityTable with tests that utilize the 2001 CSOMortality Table unless the combination isexplicitly required by regulation or is neces-sary to be in compliance with relevantActuarial Standards of Practice.



Companies may use the gender-blendedversions of the 2001 CSO Mortality Tablesfor nonforfeiture purposes. Those with akeen interest in this issue will notice thatsome of the minor limitations that applied tothe use of the 1980 CSO Gender-BlendedTables have been removed. Use of thesetables is also optional on a plan-by-planbasis. Protection against violation of theunfair trade practice statute is continued forthose that take advantage of the plan-by-plan feature. In response to a request forclarification, new wording was added to clar-ify that these tables are not permitted bythis regulation as the minimum valuationstandard. Only a handful of states explicitlyprovide that gender-blended tables may beused as a valuation standard.

Finally, under an important new provi-sion, use of the 2001 CSO Mortality Tablestriggers a requirement that the actuarialopinion filed by companies with the annualstatement be based on asset adequacy analy-sis. This will affect companies that file thealternate form of the actuarial opinionincluded in the older version of the NAICActuarial Opinion and MemorandumRegulation that is still found in most states.

Unresolved Issues

The American Men (5) Mortality Table neverreplaced the American Experience Table asthe valuation standard in all of the states.Critics of that table pointed out that it wasbased on the average experience of thecontributing companies and, consequently,not adequate for all of the industry. Thesame issue arose during the LHATF discus-sions of the 2001 CSO Table. New Yorkadvocated that companies be required by theNAIC model regulation to annually filemortality experience data with the states.This would facilitate state review of companymortality experience and deal with thedecline in the number of companies partici-pating in the SOA experience studies.LHATF rejected this, choosing instead to relyon asset adequacy analysis to protect againstinappropriate use of the new table.

Discussions of this issue have continuedin New York. An announcement from theNew York Superintendent of Insurance tothe New York domiciled companies on thismatter is expected shortly. Indications are

that the letter will strongly encouragecompanies to participate in the SOA-LIMRAMortality Study and indicate that New Yorkwill consider a regulatory solution if volun-tary participation is not satisfactory. TheAmerican Council of Life Insurers (ACLI)also will write an announcement asking itsmembers to consider voluntary participationin the study to avoid a regulatory mandate.

Federal Income Tax Reserves

When the 2001 CSO Mortality Tablebecomes permitted in computing reservesunder the insurance laws of at least 26states, it will become the “prevailing commis-sioners’ standard table” for reserves undersection 807(d) of the Internal Revenue Code.The law provides a full three-year transitionperiod during which the previous prevailingtable may be used. For example, if the 26thstate adopts the 2001 CSO Model Regulationduring 2005 with a permissive date ofJanuary 1, 2006, then the 2001 CSO Tablewould become prevailing on January 1, 2006.Tax reserves for policies issued during 2006-2008 could be based on either the 1980 CSOor the 2001 CSO. For policies issued on orafter January 1, 2009, the 2001 CSO Tablewould apply. It is not coincidental that thishypothetical date coincides with the manda-tory use date in the NAIC model. Of course,the state regulators could act more quicklyand the 2001 CSO Table could becomerequired for tax reserves before the 2009state mandatory date.

The 2001 CSO Table, like the 1980 CSOTable, is a family of tables composed of morethan one table with options under thosetables. Consequently, more than one tablemeets the criteria for prevailing table. Inthese cases, section 807(d) calls for use of thetable that “generally yields the lowestreserves.” The 2001 Valuation Basic Tableand the 2001 CSO Mortality Table weredeveloped with this issue in mind. TheReport of the American Academy ofActuaries’ CSO Task Force says essentiallythe following about the relative magnitude ofreserves under the available options:

• Reserves on the Ultimate Table are generally less than Reserves on the Select/Ultimate Basis.

• Reserves on the Smoker/Nonsmoker

New York advocated thatcompanies berequired by theNAIC model regulation toannually filemortality experi-ence data with the states.


CSO Mortality... • from page 25

Basis are generally the same as reserves on the Composite Basis.

This is the same fact pattern that waspresented for the 1980 CSO Table.Consequently, it is reasonable to expect thattax reserves would continue to be based onthe ultimate table and that either smokerdistinct or composite tables could be used fortax reserves.

Definition of Life Insurance

Section 7702 of the Internal Revenue Codeprovides that in order to be a life insurancecontract under the Internal Revenue Code,the contract must be a life insurancecontract under applicable law and must alsomeet one of two alternative computationaltests. Both of these tests include an elementfor “reasonable mortality charges.” The codefurther provides that reasonable mortalitycharges shall not (except as provided in regu-lations) exceed the mortality chargesspecified in the prevailing commissioners’standard tables. The Department ofTreasury Notice 88-128 set forth “interimrules” for the determination of the reason-able mortality charge requirement.Specifically, the Notice provided a safeharbor for the use of the 1980 CSO Tablethat taxpayers continue to rely on.

Unlike section 807, section 7702 has noexplicit transition provisions that wouldprovide for an orderly transition in theprevailing table. The ACLI concluded thatthe most immediate need for guidancerelates to transition rules and the effect ofthe new tables on existing contracts. InOctober 2002, the ACLI requested thatTreasury provide the following transitionguidance:

1) The provisions of Notice 88-128 with certain clarifications shall continue to apply to all life insurance contracts issued before January 1, 2009 that do not utilize the 2001 CSO Tables in the underlying computations.

2) For contracts issued using the 2001 CSO Tables, there is a safe harbor of 2001 CSO Tables for all reasonable mortality charge computations to the same extent as Notice 88-128 provided a safe harbor

for use of 1980 CSO, except that the reasonable mortality charges for a contract cannot exceed the amount of mortality charges guaranteed under the contract.

3) After December 31, 2008, the only safe harbor will be the 2001 CSO Tables for newly issued contracts. Existing contracts can continue to use 1980 CSO to the same extent as provided under Notice 88-128.

4) Contracts using 1980 CSO Tables that undergo material changes or adjustment events can continue to use their existing mortality tables. In the case of an exchange of a contract for a newly issued contract, however, the date of issue of the new contract will determine the appro-priate mortality table. This exception should apply only in the case of an actual exchange and not in situations where an insurance contract merely undergoes a modification, such as the addition or removal of a rider.

The ACLI met with representatives of theDepartment of Treasury in late 2002 todiscuss this request for guidance. While noformal action was taken regarding transitioncourses, ACLI remains optimistic that workon the requested guidance will proceed thisyear.

Current State Activity

As this article is being written, only a fewmonths have passed since the NAIC adoptedthe model in December 2002, so it should notbe surprising that only two states, NewMexico and Texas, have adopted the regula-tion and only three others, Oklahoma,Pennsylvania and Utah, have issued propos-als. Bureaucratic machinery moves slowly.An apparent slow start should not be takenas an indicator of future developments.When the number of states adopting the rulereaches the 26 required to make the newtables prevailing for FIT purposes, pressureto complete the remaining states will natu-rally arise. Sooner than we realize, 2009 willbe upon us and the implementation effortthat lies ahead will be history.�


CSO Mortality

William Carroll, FSA,

MAAA, is an actuary for

the American Council-Life

Insurers in Washington,

D.C. He can be reached at

[email protected].

DO YOU KNOW HOW MUCH YOU’RE

SPENDING?–THE HIDDEN COSTS OF PRODUCT

COMPLEXITY

This session looks at the behind the scenescosts that are often missed when designingnewer, more complex products. They can runthe gamut from increased systems and admin-istrative expenses to the need for more robustmarketing materials and, in the extreme,market conduct costs. Participants:• Develop a better understanding of new

product design• Learn methodologies for estimating the

additional cost of new product features• Learn how to measure, account for and

mitigate market conduct risk

RISK MANAGEMENT THROUGH PRODUCT

BALANCING

Most pricing risk management tools are usedfor one specific product. However, risk in oneproduct may be offset by expanding acompany’s product portfolio. This sessionexplores the possibilities of reducing risksusing a combination of products. Some of therisks that are discussed include asset andliability durations, mortality risks and multipleguarantees within a product design.

Several case studies are presented that allowattendees to gain increased understanding ofrisk management through product balancing.

EVERYTHING OLD IS NEW AGAIN—THE REVIVAL

OF TRADITIONAL LIFE PRODUCTS (WITH A

TWIST)This session examines the resurgence of wholelife, term and universal life products, both froma market and product perspective.

At the conclusion of this session, participantshave a better understanding of:• Renewed growth in life insurance market

share of these products• Customer demographics and economic

trends driving the resurgence• New product features that are modernizing

traditional life products• New reserve and tax regulations to consider

EQUITY-INDEXED ANNUITIES OR VARIABLE

ANNUITIES GUARANTEED LIVING BENEFITS?WHICH IS BETTER?Annuity purchasers are attracted by theprospect of equity-based returns while retaininga floor of protection. Equity-indexed annuitiesand guaranteed living benefits on variable annu-ities are two different ways to address thisdesire, but the returns and risks for bothpurchasers and insurers are very different.What are the important considerations for aninsurer in this market? Panelists debate themerits and risks of these two products. Thissession addresses:• Differences in benefits that can be offered• Insurer investment risks under each product• Impact on distributors• Influence of reinsurance and available

hedging

PRODUCT DEVELOPMENT SECTION HOT

BREAKFAST

The Product Development Section invites sectionmembers to a hot breakfast buffet, a short busi-ness meeting and a speaker. This is a great wayto meet and socialize with other section membersand learn about section activities.

The breakfast is open to Product DevelopmentSection members only. There is a nonrefund-able charge of $10. Please include theadditional fee with your registration.

LESS UUNDERWRITING: MORE PROFITS OR

MORE PROBLEMS? Many insurers are moving toward less under-writing on their life products. Panelists discusswhat measures they are taking to make theirunderwriting requirements less stringent andhow the market is reacting to these measures.Specifically, table-shaving programs andsimplified issue underwriting are addressed.

The panelists answer the following questions:• How are insurers handling the competitive

pressure if others are paring down their underwriting requirements?

• How are reinsurers handling this reduced under-writing in their reinsurance quotes?

Upcoming PD-Sponsored Sessions at theAnnual Meeting in Orlando


• What are the cost/benefit trade-offs of these changes?

Attendees learn the underwriting changesbeing considered and the impact of this onmortality and pricing.

“ONE SHOE DOESN’T FIT ALL:” USING THE

APPROPRIATE PRICING MEASURE

There are many measures of profitability in thelife insurance and annuity product develop-ment processes, including internal rate ofreturn, return on equity, return on assets, profitmargin and break-even year. Which are themost appropriate, and what are they tellingyou? How does the answer differ by productand why? How might profit targets differ underchanging circumstances?

Attendees develop a better understanding ofpricing measures and how they can be mostappropriately applied to varying products andenvironments.

WHICH DISTRIBUTION CHANNEL WILL WIN?Insurers struggle to distribute products thatare customer-driven and profitable. Variouscompeting distribution channels attempt toachieve the most prominent position. Theyinclude national and regional marketing organ-izations, agency forces of various kinds, groupand worksite specialists, Internet marketersand others.

Representatives from three of these channelsdiscuss:• Their value proposition for customers and

insurers• Their relative strengths and weaknesses• The reasons they will increase their position

with insurers and customers in the future

Attendees increase their understanding of thedynamic development of various distributionchannels.

RISK-BASED CAPITAL REQUIREMENT ON

VARIABLE ANNUITIES WITH GUARANTEES

New risk-based capital requirements for vari-able annuities with guaranteed benefits suchas GMDBs and GMIBs have been developed bythe AAA and are being considered by the NAIC.The proposed requirements may significantlyincrease the level of capital required on suchproducts from today’s typical levels. As well, the

approach to determining the required capitalinvolves stochastic modeling of guaranteedbenefits.Industry experts discuss:• Overview of the new capital requirements• Implementation issues such as model

validation, number of scenarios, grouping,interest rates andpolicyholder behavior

• Anticipated impact• Methods to reduce the necessary number of

scenarios

Attendees benefit by learning about the newrequirements and how they affect the bottomline.

UNITED STATES VS. CANADA—REGULATORY

PRODUCT FILING PROCESSES

This session involves a brief discussion on theregulatory filing processes in the United Statesand Canada followed by a lively debate on thecurrent and future processes for each country.Regulators from each country discuss currentfiling processes and other product-relatedissues.

Attendees develop a better understanding ofthe regulatory filing issues from each countryas well as the possible evolution of theseprocesses.

NORTH AND SOUTH OF THE BORDER–PRODUCT

DISTINCTIONS

The Canadian marketplace is one wherepreferred term insurance is relatively new, crit-ical illness insurance products are provensellers and life insurance policy forms are notsubject to provincial approval. In the UnitedStates, current developments include shadowaccounts, speedy issue products and guaran-teed benefits on variable annuities. Whatmakes these neighbors different?

Representatives from both sides of the borderdiscuss:• Market and regulatory differences and simi-larities• The comparison of product offerings in each

country• An overview of current issues in product

development

Attendees gain increased understanding of thedifferences and similarities between the UnitedStates and Canadian marketplaces.�


T he Society of Actuaries released inMarch the report of the MortalityImprovement Survey Subcommittee.

This article highlights just a few of the inter-esting results in the report. The completereport can be accessed on the Internet atw w w. s o a . o r g / r e s e a r c h / m o r t a l i t y _improvement.html.

The report provides results of a survey oflife insurance company practices in thesummer of 2000 regarding the use of mortal-ity improvement assumptions in pricing. 61U.S. companies and six Canadian companiesresponded. All companies were direct writers(i.e., not reinsurers).

The survey asked about “generational”improvement (projecting historical experi-ence up to the current date) and “durational”improvement (projecting current mortality toimprove into the future). I am less interestedin the generational improvement results.Although only 35 percent of respondents saidthey used generational improvement, most ofthose who did not use it said it was becausetheir experience table already reflected

current expectations or were already up-to-date.

With regard to durational (future) mortal-ity improvement, only 25 percent ofrespondents said they use this. Of those thatuse this, in the first 10 policy years, theannual improvement for male age 45 bestnonsmoker preferred class ranged from .50percent to 2.00 percent, with an average of.89 percent. Approximately half of thesecompanies grade the improvement to zeroafter a period of 10 years or so. Althoughonly 25 percent of companies priced usingdurational improvement, 52 percent ofrespondents believe it is appropriate to usedurational improvement.

The reasons given for why durationalmortality improvement is not assumed arelisted in the table below.

Of course, the issue of mortality improve-ment should not be considered in isolation—it should be combined with consideration asto the steepness of the underlying mortalitytable, which is an issue described in anotherarticle in this newsletter (See page 13).�


SOA Report on Mortality Improvement Surveyby Douglas C. Doll

Features

Reasons Durational Improvement was not Used

Reason Percentage of Respondents

Creates problems with Illustration Certification 62%

Creates problems with XXX X-Factors 36%

Other 28%

Company does not believe durational improvements factors are needed 28%

Company does not believe durational improvement factors are appropriate. 56%

Individual Life Insurance and Annuity Product Development ......health, annuity, long-term...

Documents

Transcript of Individual Life Insurance and Annuity Product Development ......health, annuity, long-term...