Longevity turbulence - How are pension schemes responding to higher mortality?

UK longevity improvement rates are attracting attention like never before. John Cridland’s independent review of state pension age and the debate on the funding of care for the elderly show the importance to the political agenda. The debate raging in actuarial circles is whether the higher mortality levels observed in the national population are a blip or a trend back towards lower rates of improvement.

Pension actuaries are looking for positive news to alleviate the effect of lower interest rates. You might imagine that lower longevity improvements would have a profound impact on funding levels of pension funds, but most actuaries are already anticipating a reversion to slower rates of improvement.

At Hymans Robertson we have cautioned against changing course in response to recent data points, and in April we released provisional findings from our own analysis of improvement rates of members of DB pension schemes.

One swallow does not a summer make 

We cautioned pension schemes against sharp changes in response to recent data points for three main reasons: 

1. Longevity improvements before 1990 were a bumpier ride – the 1990s and 2000s were unusual for their smooth upward trajectory; 
2. Actuaries struggle to justify longevity improvements to finance directors (despite the tapering to lower rates of improvement) – increases are far harder to implement than cuts and;
3. Before changing course, understand the drivers of change, and particularly whether the pattern is uniform across the population. 

There have been a few important recent contributions to shed light on what’s going on beneath the surface:

• First, the Continuous Mortality Investigation (CMI) has released the latest vintage of its model. The default version of this model is calibrating against recent trends in national population data rather than DB pension scheme members. The CMI’s report pointed to a sharp slowdown in rates of improvement at the national average level. Its report also includes analysis of a smaller body of data in the Self-Administered Pension Scheme (SAPS), which shows a contrasting picture of higher improvements.
• Secondly, research reports studying the performance of the National Health Service in recent years as our nation’s ageing population has increased the demand for its services. These suggest that higher levels of “excess” winter mortality may be connected to the capacity of the system to cope.
• Finally, Hymans Robertson released provisional findings from our own analysis of improvement rates of members of DB pension schemes. We split up our population into three groups of men and two groups of women, which we call VitaSegments. The methodology that we use for allocating pensioners – based on postcodes and pension amounts – can be found here. Very broadly, the Comfortable VitaSegment relates to those pensioners with reasonable levels of DB pension (>£150 pw), the Hard-Pressed those who live in the more deprived areas and the Making-Do being the in-between group. 

The key results are for men, who dominate the liabilities of most DB schemes.

Source: Club Vita. Table shows annualised improvements (reductions) in mortality over sucessive 5 year intervals. Values in brackets indicate width of a 95% confidence interval. Difference in improvements between comfortable and hard-pressed statistically significant for 2005-2010 and 2010-2015. 

We’re seeing evidence of a sharp decoupling/divergence of socio-economic groups taking place – see in box in bottom right of the above table. During the noughties, we observed slightly higher improvement rates for Making-Do and Hard-Pressed compared to Comfortable. Over the latest 5 years (2010-2015), we’re seeing the opposite. The improvement rates in the Comfortable group have barely changed – an average annual drop of over 2% a year in each of the three successive 5 year periods. This is particularly important for pension schemes as their liabilities are generally concentrated in this Comfortable group.

So, the exceptionally strong tailwind of improvements observed in the 1990s and 2000s decades has abated. But this has not necessarily turned into a headwind either: the death rates of all our groups still fell over the 5 years from 2010-2015. Instead, we are observing a more turbulent period, and one in which the socio-economic longevity gap could widen again.

Our key message to pension schemes is that even with high deaths in recent years, longevity risk has not reduced.  Care must be taken before jumping to the latest CMI model in its core form, and an understanding of the trends by socio-economic group is essential in forming a well considered view.

Insurers face much the same issues – albeit generally with a much better capacity to interpret and make use of the range of sources of data.  For more details of our recent analysis, please see the replay of our webinar on our socio-economic improvements research.

Contributing Authors:

Andrew Gaches

The material and charts included herewith should not be considered a definitive analysis of the subjects covered, nor is it specific to circumstances of any person, scheme or organisation. It is not advice and should not be relied upon. Hymans Robertson LLP accepts no liability for any errors or omissions.