Sections

Longevity in the spotlight

Scheme-specific adjustments to standard industry tables can help smaller pension funds take into account the effect of longevity, according to Michael Kelly and Glyn Bradley 

As interest rates have declined, defined benefit pension liabilities have become increasingly sensitive to rising life expectancy. A few years ago we would have said that a one-year increase in life expectancy would increase a scheme’s liabilities by about 3-4%; today, with discount rates even lower (so making pensions more expensive to provide), it can be 5% or more. In recent years, many schemes and sponsors have taken steps to reduce their risk exposure to changes in their asset values, interest rates, inflation and employer covenant. Life expectancy, or longevity, risk is thus growing in significance as anunhedged risk for schemes and their sponsoring employers. However, new actuarial techniques, computing power and financial products are helping pension schemes gain an accurate understanding of their longevity risk, and opening new routes to manage it.

In smaller pension schemes the survival, or not, of a handful of individual members to older ages than expected can make a material difference to the amount of money that needs to be put aside to cover benefits. Even in mid-sized schemes, liabilities are often concentrated among a handful of retired senior managers. Generally speaking, and for any age group, over 10% of a scheme’s members will be expected to live until their late 90s or even past 100, even if their ‘life expectancies’ are much lower. Computer simulations can quickly roll the dice thousands of times to help assess the financial impact of this random timing effect.

The next risk is that the current life expectancy estimates for members may not be correct. The profiling techniques outlined below can help reduce but not eliminate this risk. Non-pensioners in particular remain a significant risk as the membership characteristics of a scheme may change over time. We would look to the life expectancies of pensioners with similar demographic profiles, but there remains the risk that the factors that explain pensioner life expectancies today might not work the same way in a few decades’ time.

For larger and less mature schemes the biggest longevity risk is that life expectancies rise faster than expected. Since mortality rates are less certain the further into the future we look, the average lifespans of active and deferred members will depend on uncertain developments in health spending, cancer drug progress, and regenerative medicine over the next 30 or so years.

“For larger and less mature schemes the biggest longevity risk is that life expectancies rise faster than expected”

The first step in dealing with life expectancy risk is to understand it. Until 10-15 years ago, pension schemes made little distinction between life expectancy assumptions for different groups of workers. There was an over-reliance on broad regional and socio-economic class variations in life expectancy. It is now recognised that we need to factor in multiple characteristics of a scheme’s membership when analysing its life expectancies, including the concentration of liabilities among the most affluent members. Members’ postcodes can be used to split them into groups likely to have similar socioeconomic status (occupation, qualifications and so on) and lifestyle habits such as smoking and obesity, without the expense of underwriting each member. It is then possible to compare members’ details to a database of mortality experience to calculate scheme-specific adjustments to standard industry tables. This approach is suited for small schemes and forms an inexpensive and routine part of assumption setting. Larger schemes, however, should use more sophisticated modelling techniques that take into account the actual death records of their own scheme – not just matching their membership details to other schemes’ experience. 

Looking so far into the future, it is not surprising that most people have struggled to understand the extent to which life expectancies are projected to rise, or why life expectancies might increase even further than estimates have allowed for.

• What are the underlying drivers of life expectancy now and in the future?
• How would the end of smoking or an improvement in people’s lifestyles accelerate life expectancies?
• How much of a drag on life expectancy is rising obesity?

To date, most improvement assumptions have been based on a combination of long-term trends and expert opinion, as well as on the assumptions used by insurers, official projections and other pension schemes in a scheme’s peer group. But, it has been hard to understand what might bring about these changes in life expectancy in the real world.

Risk modellers such as RMS, the industry leader, are now able to relate life-expectancy improvement assumptions directly to real changes in people’s lifestyles, medicine, health care and developing technology through a robust framework. Their scenarios are able to show a view of the future grounded in their modelling, integrating the latest research across a range of fields and allowing the different possibilities to play out against one another. But, perhaps more importantly, they are also able to show how different views on future trends – rising obesity, for example – would lead to different life expectancies. These projections can then be related back to the cost of providing pensions so that trustees and scheme sponsors can understand their funding risks.

It is sometimes suggested that longevity risk emerges slowly, since it is not until scheme members have died that we know their actual lifespan. However, if a membership is surviving for longer than expected then it will be paying out more benefits than expected, and its life expectancy assumptions are likely to be revised upwards. 

Value-at-risk modelling suggests the potential movement is likely to be small relative to financial and investment risks. However, a significant part of calculating life expectancies is driven by assumptions about how mortality rates will improve in future – and views on future improvements can change faster than a few years’ extra mortality data would suggest. For example, if the number of deaths over the next three years turns out close to that expected but, for example, vaping leads to a sharp reduction in tobacco smoking, then life expectancy assumptions are likely to be revised upwards in anticipation of faster future improvements. This means that trustees and sponsors of defined benefit schemes risk being forced to increase their life expectancy assumptions significantly over a few years, either from peer-group pressure, auditors/regulator insistence, or from the revised anticipated cost of passing the longevity risk on.

Some trustees and sponsors will want to offload their longevity risk, either through buy-ins of buyouts, or through swaps. There is a cost to this, but, as with any insurance, it can be good value when it is compared to the plausible possibilities. Other schemes will want to retain their longevity risk, perhaps upping their life expectancy assumptions or adopting an overall higher funding target. Either way, the important thing is to recognise that how long members may plausibly live is a much wider range than often recognised in the usual debates over increasing long-term improvement rates by 0.25% a year or so.

Michael Kelly is a principal and UK leader for longevity and mortality analysis at Mercer. Glyn Bradley is a principal in the retirement innovation, policy and research team at Mercer

Have your say

You must sign in to make a comment

IPE QUEST

Your first step in manager selection...

IPE Quest is a manager search facility that connects institutional investors and asset managers.

  • DS-2382

    Closing date: 2017-12-14.

  • QN-2383

    Asset class: Residential Property.
    Asset region: Ireland.
    Size: EUR 10m.
    Closing date: 2017-12-18.

  • QN-2384

    Asset class: Equities Switzerland (Large Caps).
    Asset region: Switzerland.
    Size: CHF 550 – 600 mn.
    Closing date: 2017-12-15.

  • QN-2385

    Asset class: Liability Driven Investment.
    Asset region: Europe.
    Size: Size: EUR 1 Billion, Liability size: EUR 3 Billion.
    Closing date: 2018-01-08.

  • QN-2386

    Asset class: Fixed income.
    Asset region: Global developed markets.
    Size: CHF 500 -1000m.
    Closing date: 2018-01-15.

  • DS-2392

    Closing date: 2017-12-21.

  • QN-2393

    Asset class: All/Large Cap Equities.
    Asset region: Europe.
    Size: EUR 200m.
    Closing date: 2017-12-21.

  • QN-2394

    Asset class: Real Estate Industrial.
    Asset region: Europe.
    Size: EUR 10m.
    Closing date: 2018-01-04.

Begin Your Search Here