How should schemes use mortality studies

After investment and inflation risk, longevity is the largest risk involved in running a defined benefit scheme.

It is well recognised that when setting life expectancy assumptions schemes should use evidence-based methods that take into account the characteristics of a scheme’s membership.

Surprisingly, surveys have shown that up to a third of pension schemes do not set scheme-specific life expectancy assumptions. The main reasons schemes are reluctant to carry out such studies are the perceived costs, lack of suitable products and complexity of analyses – but are these fears borne out?

In an ideal world, schemes would maximise the information they have on members. Generally, the information available for setting assumptions is limited to that held on the scheme records including gender, pension amount and postcode.

Members’ postcodes can be used to segment the membership into groups likely to have similar characteristics (occupation, socio-economic status, etc) and lifestyle habits such as smoking and obesity.

Often described as ‘postcode models’, this is a cost-effective, well-established and evidence-based assessment of life expectancy that compares members’ observable characteristics with the current mortality rates for pension scheme members in much larger databases who have the same characteristics.

This approach is well suited for small schemes and now forms a comparatively inexpensive and routine part of assumption setting. Schemes with a large pensioner membership and reliable recording of deaths, however, should consider the past experience of their own members.

As the number of scheme members rises, the relevance of the scheme’s own experience in setting the assumption rises. For the biggest schemes it is likely to be the most relevant source in setting an assumption for the scheme’s life expectancy.

Where the number of members is in the thousands, rather than tens of thousands, schemes often consider their own mortality experience alongside the experience of other scheme members with similar characteristics, as described above.

Evidence-based studies ought to provide a solid platform for the appropriate life expectancy assumption for a scheme, as part of funding discussions between trustees and sponsors.

They also play a very important role in setting assumptions used for paying out the right level of benefits to members, such as cash commutation factors or transfer values.

Where a scheme is considering entering into a transaction with an insurer for a buy-in or longevity swap they form a near-essential part of evaluating any pricing a scheme might receive.

Even for the smallest of schemes, the resulting cost of members living a few more months than expected will dwarf the cost of a scheme-specific analysis.

Understating life expectancy can lead to increases in the funding reserve emerging over time, and might make the cost of transferring the risk to an insurer through a buy-out or longevity swap look more expensive than it actually is.

Conversely, overstating life expectancies can lead to a scheme being better funded than it thinks, and with an increasing number of schemes setting funding-level based triggers to reduce risk, this could lead to missed opportunities elsewhere.

Having carried out whatever form of study is most appropriate, trustees and sponsors should recognise that such exercises are only models of the real world and that the actual lifetimes of pension scheme members cannot be known with certainty until the last pensioner dies.

Part of using any study is to ask “what happens if life expectancy turns out to be higher or increases faster than we have assumed, and what is the impact of that on both the sponsor and the pension scheme?”.

An important part of running any pension scheme is making the best assumptions possible, understanding the associated uncertainty and planning accordingly for what happens if things turn out differently than expected.

Michael Kelly is Mercer’s UK Leader of Longevity and Mortality Analysis

Asha Suresh is the co-ordinator of Mercer’s mortality area of focus