ISSUES & ANSWERS

A Deeper Look at Longevity Risk

by Richard Brown, Longevity Consultant, Club Vita Canada Inc.

The average Canadian life span has increased dramatically over the past century. In fact, men in B.C. now enjoy a life expectancy in line with that of the longest lived OECD countries. Slowly but surely, the number of years Canadians spend in retirement and the number of years they spend at work have begun to equalize. But a longer retirement comes at a higher cost for defined benefit (DB) plan sponsors – and the heightened risk of individuals in defined contribution plans outliving their retirement savings. This means longevity risk should be understood, measured and managed better by pension plan sponsors.
 
Breaking Down Longevity Risk

Longevity risk tends to get overshadowed by other key financial risks – particularly investment and interest rate risk. This isn’t surprising, considering the impact these risks can have on the potential financial outcomes and costs of a DB pension plan. However, while pension plan sponsors spend considerable time and effort monitoring and managing investment and interest rate risks, most don’t spend any time measuring their longevity risk. Yet, as plans take steps to reduce or more closely manage their other financial risks, a natural extension is to see what can be done to better understand and manage longevity risk.

At its core, longevity risk is fairly straightforward. Part of the reason it isn’t as well understood as other risks is that there are inherent challenges in measuring and managing longevity risk. It’s therefore helpful to break down longevity risk into three key components to better understand these challenges:

 
  1. Measurement risk;
  2. Experience risk; and
  3. Trend risk

Measurement Risk

Pension plan sponsors have a wealth of detailed economic and investment information available at their fingertips. However, the same level of detail hasn’t traditionally been available to measure the longevity of their members. While Statistics Canada publishes an annual look at the current expected lifespan of Canadians, the data used is typically a few years old and examines the general Canadian population. Therefore, the vast majority of pension plans instead measure longevity based on the findings of longevity studies performed by actuarial associations, which focus only on members receiving pensions from registered pension plans. The downside for pension plans is that these pensioner longevity studies are typically performed infrequently – traditionally about every 10 to 15 years. As a result, when DB pension plans adjust their longevity assumptions following the publication and adoption of a new study, they have historically experienced one-time liability increases in the range of 5% to 10%. More frequent re-measurement of longevity would help to reduce the impact of these changes.

Another important factor is the makeup of the pensioner population included in longevity studies, and how the findings are applied to individual plans. The Canadian Institute of Actuaries’ 2014 Final Report on Canadian Pensioner Mortality (CPM study) broke ground as the first longevity study based solely on Canadian pensioner data. Previously, Canadian plans relied on U.S. pensioner longevity data. The shift to Canadian data is a significant improvement, given that
recent OECD data shows Canadians as a whole live longer than our U.S. counterparts. However, Canada’s size and diversity poses challenges when it comes to applying country-wide data to a particular pension plan. For example, Figure 1 shows that life expectancy at age 65 varies by about 2.5 years depending on an individual’s province of residence. The CPM study has prompted plan sponsors (and their actuaries) to more carefully consider how to apply the general longevity findings to individual plans – particularly by industry and pension size. Actuarial practice is moving away from the one-size-fits-all approach of the past.

Figure 1: Life expectancy at age 65 by province and gender (
SOURCE: Statistics Canada, Life Tables, Canada, Provinces and Territories, 2009-2011)



It’s worth noting that diverse (i.e., heterogeneous) pensioner populations have more potential measurement risk than very uniform (i.e., homogeneous) populations. It is, therefore, relatively easier to develop longevity expectations for homogeneous groups. The Ontario Teachers’ Pension Plan is an example of a large plan with a homogenous population – all pensioners had the same profession, were employed in the same province and likely had similar income levels. Many pension plans, in contrast, have much more diverse membership populations. Setting a single longevity view for such plans can miss the true characteristics of the plan.

Experience Risk

The risk that a particular plan’s longevity experience will differ from expectations over time varies greatly with the size of the plan. In general, the larger the plan, the smaller the experience risk, all other things being equal. This is because even the most robust longevity assumptions are based on how long pensioners are expected to live on average, based on factors that plan sponsors can observe. This misses some key factors. For example, some people are genetically inclined to live for many years; of course, this isn’t something plan sponsors have any insights into. These kinds of very individual factors can be significant for a small plan. As plan size increases, these individual differentiators are largely diversified away.

For large plans, the cumulative impact of longevity experience poses a risk. If a plan’s longevity assumptions underestimate its true expected experience (which has tended to be the case historically), experience losses can snowball. This underestimation can be purely due to Canadians just living longer in general, or the result of mismeasurement of the plan’s longevity. Unlike investment returns, which tend to correct, past longevity experience losses have typically tended to not be offset by future gains.
 
Trend Risk

Trend risk involves uncertainty about how long people will live in the future. While measurement risk can be reduced considerably through more robust analysis, and experience risk unfolds over time, trend risk is by far the hardest to mitigate.

History tells us that lifespans have consistently increased over time. In Canada, this upward trend can be tracked back as far as the early 1920s for both males and females. Figure 2 shows life expectancy improvements observed by decade. In the first decade of the 21st century alone, Canadians gained an additional 2 and 1.4 years of life after age 65, for males and females, respectively. The drivers of these past increases have been diverse, and it’s hard to determine the specific impact of each driver on life expectancy. On top of this, the drivers of the past are often unlikely to drive future increases. For example, the reduction in smoking rates has played a key role in increasing average Canadian lifespans over the past 50 years. People continue to smoke, of course. But given the current low levels, future reductions in the prevalence of smoking will not have the same impact on life expectancy.

Figure 2: Improvement in life expectancy at age 65 by decade (
SOURCE: Human Mortality Database, 2011 data)



In some ways, the common investment warning about the harm of focusing on past performance may be even more applicable to longevity. That said, humans are quite efficient at figuring out new ways to live longer, and given the current exponential growth in technological advances, there are no signs of us running out of ideas. An important step in better understanding longevity will be for plans to have better insights into their trend risk over the next 10 or even 20 years. Currently the vast majority of Canadian pension plans have the same view regarding future lifespans and aren’t considering how things may differ for their plan based on its characteristics.

As Canadian pension plans continue to mature and take steps to further mitigate investment and interest rate risks, longevity risk will inevitably receive increased attention. This outcome aligns well with the common anecdote that Canadian pension practices tend to follow the U.K. by about 10 years – a forecast that has proven fairly accurate when it comes to plan closures, DC practices and DB de-risking. By understanding the key components and challenges of longevity risk, plan sponsors can begin to better control their longevity risk and avoid the costly “longevity surprises” of the past.

Richard Brown is a longevity consultant with Club Vita Canada Inc., a wholly owned subsidiary of Eckler Ltd. Richard provides longevity analytics services to the Canadian pension industry and oversees Club Vita Canada’s longevity modelling and research efforts. Richard has more than 10 years of experience providing risk management consulting to pension plan sponsors. He is a fellow of the Canadian Institute of Actuaries and the Society of Actuaries, and is a CFA® charterholder.