Longevity risk is often overlooked by plan sponsors, but may have a significant impact on plan valuations and funding in the future.

A recent Watson Wyatt mini survey of 40 major defined benefit (DB) pension plan sponsors in the U.K. revealed that the risk of mortality improvements is of far greater concern to them than the risk of a fall in equity prices or a fall in bond yields. North American plan sponsors may find this result surprising. Until recently, there has been relatively little talk on this continent about longevity risk in the funding and expensing of DB plans.

While Canadian plan sponsors devote a great deal of effort to managing their investment risks and reviewing their asset-liability mismatch, longevity risk largely remains a shadow lurking in the background. For most ongoing plans, the actuarial gains and losses caused by differences between actual and assumed mortality experience are usually far smaller than those created by differences between actual and assumed investment returns in plans whose assets and liabilities are significantly mismatched. This comparison likely explains why Canadian plan sponsors haven’t paid much attention to longevity risk in the past—there are few, if any, Canadian pension plan stories in which the use of a particular mortality table has become a history-changing event.

The U.K. Situation

In the U.K., however, arguments over mortality tables have put the brakes on some major merger and acquisition deals. What’s happening in the U.K. to cause such an emphasis on longevity? A few factors come into play.

• There is a prevalence of closed DB plans in the U.K. An aging plan membership necessitates a greater focus on potential mortality losses, as the other sources of actuarial gain or loss gradually diminish in relative importance over time—especially if the mismatch between assets and liabilities is purposely reduced.

• It is a common U.K. practice to provide inflation protection to pensions in payment, which significantly boosts the actuarial value of such payments in the latter part of a member’s life and makes it more important to accurately predict how long the member will live.

• U.K. legislation requires companies to cover the cost of buyout with an insurer if they want to terminate a pension plan, which forces companies with ongoing pension plans to keep an eye on market buyout pricing terms. Most of the jurisdictions in Canada require full funding on plan windup as well, and we also have solvency funding rules that cause ongoing plans to fund toward windup costs. However, annuity prices in the U.K. tend to be based on a projection of mortality rates that improve continually into the distant future—unlike some sellers of Canadian annuities who, it appears, are willing to price more aggressively. This adds to the pressure on U.K. plan sponsors to fund their plans using a conservative mortality table.

• The U.K. is also experiencing a phenomenon known as the “cohort effect.” Simply put, those born between 1925 and 1945 have experienced a particularly rapid improvement in mortality rates.

One explanation for this phenomenon, especially for those born in the earlier part of this period, is that they are likely to have smoked when young, quit before too much damage accumulated and subsequently lived longer. Japan, the country with the highest life expectancy in the world, also has a pronounced cohort effect; however, Canada does not. It’s not immediately apparent what causes these inter-country cohort differences.

Mortality Tables

What about longevity risk in Canada? While automatic indexing of retirement benefits is far less common in the Canadian private sector, the current low interest rates and potentially maturing pension plans make future mortality improvements an important factor in DB plan valuations.

DB plan valuations are based on mortality tables—statistical tables that show the expected death rates or “mortality rates” of people at each age. Creating a mortality table begins with studying the actual rates of death experienced by a particular population in the recent past. Advances in medical research and technology, gender and the decline in smoking all influence mortality rates. Income also has an impact on longevity, although recent research from the U.K. suggests that marriage may have a more significant effect on longevity than income levels. Additionally, mortality rates for employee groups tend to be lower than mortality rates for the Canadian population as a whole, due to the assumption that a minimum level of health is necessary to participate in the workforce.

Longevity Risk is the risk typically borne by pension funds or annuity providers from an unanticipated reduction in mortality rates. From a plan member’s perspective, it’s the risk that the individual’s retirement assets may not be sufficient to maintain the income required to sustain him or her for life

 

While past mortality rates are a matter of fact, it’s extremely difficult to project what these rates will be in the future. Canadian actuaries have been responsive to signals of improving mortality, moving to new standard tables as they become available and as experience dictates the need. Many are now using the UP94 table with projection to 2015 using the “AA scale” (a methodology for projecting improvements from one year to the next). Improvements differ between males and females. However, fully recognizing potential future mortality improvements would require the use of a generational table. A generational table does not stop reflecting future improvements beyond a pre-set future year such as 2015; instead, for each age, it features rates that gradually reduce year over year into the future. Compared to the more common tables that project to a pre-set future year (usually not much beyond 2015), a generational table can add considerably to the actuarial liabilities. This is especially true for a less mature plan, in which the average age of the plan members means that the vast majority will live many years beyond the pre-set future year used in the more common tables.

Here’s an example to show how the use of different mortality tables affects future liabilities. In the year 2017, a 55-year-old male would have an expected age of 84.5 at death based on the UP94 table with full generational improvements, compared to only 80.9 using the UP94 table projected to 2007. Those extra years of pension payments would have to be accounted for in the liabilities.

For example, the liability for that 55-year-old in 2017 would increase by approximately 7.3% when moving from the “projected to 2007” table to the generational table, if the pension is not indexed. And the increase is even more significant—approximately 10.1%—if the pension is fully protected from inflation, meaning that focusing on longevity risk is even more important with an indexed plan.

Future Projections

Should actuaries project mortality rates using a generational table or simply stop projecting improvements beyond, say, 2015? It’s a question of balance. The latter approach helps contain the plan sponsor’s costs, but reduces the benefit security available to the members. And, of course, actuaries don’t know whether life expectancy will continue to increase year over year forever, or if there is a certain age beyond which it is impossible for humans to live.

Projection refers to assumed future improvements in mortality rates, often expressed as a projection to a particular pre-set future year.

 

The Society of Actuaries recently organized a “Living to 100 and Beyond” seminar, which included a paper by Jacob Siegel entitled “The Great Debate on the Outlook for Human Longevity: Exposition and Evaluation of Two Divergent Views” analyzing the arguments for and against continued mortality improvement. The paper concluded that it is impossible to reconcile the arguments and anticipate future developments with any real certainty.

Similarly, James Vaupel of the Max Planck Institute for Demographic Research and Jim Oeppen of the University of Cambridge co-authored a paper in 2002 showing that, for the last 160 years, life expectancy has steadily increased by one-quarter of a year per year: “In 1840, the record was held by Swedish women, who lived on average a little more than 45 years. Among nations today, the longest expectation of life—almost 85 years—is enjoyed by Japanese women. The four-decade increase in life expectancy in 16 decades is so extraordinarily linear that it may be the most remarkable regularity of mass endeavour ever observed.” In fact, research by the Society of Actuaries has led the Canadian Institute of Actuaries to conclude that “future mortality improvement rates from the AA scale are more than likely to be insufficient” for annuity mortality in Canada.

On the other hand, Professor Jay Olshansky of the University of Illinois at Chicago has argued that extrapolating past trends doesn’t make sense, pointing out that most of the increase in life expectancy during the 20th century occurred as a result of reductions in early age mortality. Long-term reductions in the future would be due to fundamentally different reasons and would also depend on the pace of development of new technologies to slow the aging process and/or greatly reduce the number of deaths from heart disease, cancer and stroke.

Yet plan fiduciaries and actuaries need to determine whether to pre-fund future mortality improvements for a fixed period of years or beyond, and pension accounting numbers in company financial statements must be based on management’s best estimate for future mortality improvements. Management has no choice but to take a position.

Hedging Techniques

While the Watson Wyatt mini survey revealed key findings about U.K. plan sponsors, the increasing volume of European writings on longevity risk suggests that the focus on longevity is continent-wide. For example, there is much discussion in Europe on how to immunize a pension fund against increasing life expectancy. The traditional method has been to purchase life annuities; however, this has meant eliminating future investment gains that could have emerged through investing in equities.

The French bank BNP Paribas tried to help plan sponsors resolve this issue by establishing a “longevity bond” that pays coupons proportional to the survival rate of a given population. Reception for the bond was lukewarm, and it has since been withdrawn. However, a newly established U.K. firm, PensionsFirst, is now launching a similar offering.

Other innovative mortality-focused investment products will undoubtedly emerge in due course. For example, U.S. institutions such as JPMorgan, Credit Suisse First Boston and Goldman Sachs Group have recently created “longevity indexes” to help holders of longevity risk (e.g., pension funds and annuity writers) price their exposures in the capital markets.

It’s also possible to hedge against improving mortality experience using plan design. One large single-employer DB plan in Canada has an “escalation account,” credited/debited with 50% of pensioner mortality gains and losses, which is used to finance additional increases to pensions in payment. With the creation of longevity indexes, single-employer DB plans can more easily be redesigned to scale back benefit accruals or to increase employee contributions if the population’s mortality experience improves faster than anticipated— or vice versa, if mortality experience worsens relative to expectations. Most multi-employer plans already have a built-in mechanism to absorb mortality gains and losses, either into the promised benefits or partially into the employee contributions.

And DB plans are not alone in having to manage longevity risk. The gradual shifting of risk from plan sponsors to plan members as they move from DB to defined contribution (DC) plans or Group RRSPs means that these members must shoulder the risk of increases in life expectancies—either through ever-increasing annuity prices or, for those who self-annuitize by drawing monthly payments from their DC or RRSP accounts, by eventually running out of money as they live longer than anticipated. Longevity hedging techniques other than buying a life annuity may make their way into the DC marketplace over time.

Mortality tables may not be the most engaging topic to discuss at a dinner party; however, plan sponsors should keep this issue in mind when considering pension risk management. Plan sponsors need to protect themselves against the Woody Allens of the world: those who wish to achieve immortality not through their work, but simply, by not dying.

Ian Markham is director of pension innovation for Watson Wyatt Canada ULC. Andrew Fung is a senior actuary at Watson Wyatt Canada ULC. ian.markham@watsonwyatt.com; andrew.fung@watsonwyatt.com

For a PDF version of this article, click here.

© Copyright 2008 Rogers Publishing Ltd. This article first appeared in the February 2008 edition of BENEFITS CANADA magazine.