The imminent publication of new self-administered pensions schemes (SAPS) mortality base tables - developed using data from UK occupational pension schemes rather than insurance companies - are expected to provide trustees with a more reliable set of data. But experts predict they may still be a less than perfect solution.
Mortality assumptions used by trustees for actuarial valuations of liabilities are at present made up of two components - base tables that look at current mortality rates and projected improvement rates for future mortality.
Steven Robinson, director at HSBC Actuaries & Consultants, suggests 95% plus of pension schemes currently base mortality assumptions on data captured by insurance companies in the early 1990s, called the pensioner annuity (PA) tables, and the PA92 tables in particular.
"There's increasing recognition that these tables are out of date, so what trustees boards and actuaries do at the moment is make adjustments to these tables," says Robinson. "There are many trustee boards using different assumptions and they adjust them in different ways. So there is no level starting point."
The latest PA tables were published with data from the early 2000s (PA00); experts suggest there are problems with basing mortality on insurance data as people purchasing insurance policies are likely to be wealthier and in better health compared to the average occupational scheme member, while the fund could include members who may be in poorer health or who take early retirement.
Insurance policyholders are generally expected to have a longer life expectancy than scheme members so by using the PA92 tables, Robinson points out, occupational pension trustees could be "overstating mortality and driving up costs unnecessarily". To rectify this, the Continuous Mortality Investigation (CMI) branch of the Actuarial Profession is publishing the SAPS tables that are based on data submitted by actuaries from around 350 self-administered pension schemes, each with at least 500 current pensioner members, for the years 2000-2006.
"These are a much more up-to-date studies of mortality, and as well as being much more relevant they are more reliable as the size of the data set is probably around three to five times as big in terms of the number of members covered," continues Robinson.
He predicts, once the final tables are released - the expected launch is the end of October - that SAPS will be quickly adopted as it is possible they will reduce liabilities. However, he warns: "Quantifying the change is difficult as schemes are using different assumptions so the percentage difference will change".
That said, Robinson claims a scheme could cut its liabilities to a 65-year-old pensioner by almost 4% - compared to assumptions using a PA92 table, medium cohort projection with 3% yield and 50% spouses pension - "which is quite material in pension scheme terms", he notes.
Jane Beverley, principal and head of research at Punter Southall, also points out the CMI is producing not just one but three SAPS tables - including a ‘heavy' and ‘light' version - which allows schemes to use a different table for different sectors, so the ‘heavy' table might only apply to low-paid workers, and the ‘light' table to wealthier members.
However, despite the choice of tables, she points out the data is "still a fairly blunt instrument" as the tables do not currently look, for example, at postcodes as an indicator of a person's longevity, which means even if they start with the SAPS tables, trustees may still need to adjust them to make the data applicable to their particular scheme.
"SAPS give you the average rate but you might not fit the average. The tables are only a base and they don't look at future changes, so trustees are still going to have to make assumptions for projections," Beverley adds.
"If trustees are making decisions on mortality, they first need to work out the base tables and then look at future improvement rates," says Robinson. The adoption of these mortality tables is voluntary so, while Robinson thinks SAPS will become the industry norm within a short time, Beverley suggests it will take schemes a while to decide, partly as pensions tend to work on a three-year cycle.
"The point is that it's not possible for trustees to get a table that exactly reflects what's going to happen. SAPS are not perfect, but then no table could be. But they are narrowing in on scheme members, rather than policyholders, and that's a good start," says Beverley.
No comments yet