[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Mortality: Cause and effect

Benjamin Franklin wrote, “In this world nothing can be said to be certain, except death and taxes”, but while death might be a certainty, its timing is far from certain. Earlier this year, over 200 delegates attending one-day seminars held in London and Manchester heard the latest thinking and research on mortality and longevity.

Pooled data, shared insights
The Continuous Mortality Investigation (CMI) highlighted the pace of current improvements in mortality at these seminars. Following particularly large falls between 2002 and 2003, the most recent experience (2006) shows actual deaths have been occurring at around 85% of the rates expected under the ‘00’ series for both temporary assurances and pensioners. Although smoker mortality rates continue to be elevated at around 170% of the combined rates, similar falls have been seen for both smokers and non-smokers.

The seminar provided a preview of the new CMI prototype projections model which blends current rates of mortality improvement into long-term rates set by the user. By designing a widely accessible and flexible model, the CMI hopes to focus debate on the most important parameters and encourage a move away from the widespread use of the outdated interim cohort projections. This is especially relevant with the cohort effect being less clearly centred on the 1926 generation in the latest CMI data.

This observation was echoed by Club Vita research that shows a 10-year delay in the cohort in occupational pension schemes compared to the interim cohort projections. Club Vita also highlighted the different longevity improvement patterns being observed in pension schemes for men and women and for manual and non-manual workers (see Figure 1), along with the importance of considering the implications of these differences for projections.

Cause for thought
Most actuaries are familiar with the challenge of trying to explain projections of future life expectancy to decision-makers, be they finance directors, life office boards or pension scheme trustees. It is often met with surprise when we confess our models do not look directly at the trends in underlying causes. Actuaries avoid causal models for a number of reasons including concerns over the reliability of cause of death records and, above all, the complex interactions between different causes of death. This, however, might be set to change.

The Mortality Projections by Cause Research Group has made substantial progress in developing a causal-based model, the broad mechanics of which were outlined by Adrian Pinington in the April 2008 issue of The Actuary. Around 500 projections have now been made by the group, looking at different groups of causes (for instance, cancers) and different sub-causes (for example, lung cancer). Mortality rates are declining for many causes, but the historic trends rarely reveal conclusively how much further mortality can fall for each cause. The group’s projections include flat and steep variants which can be thought of as ‘close to bottoming out’ and ‘some way to go yet’ scenarios.

Improvements in cardiovascular disease have dominated the last 30 years. The group’s projections (Figure 2) indicate that, while it may have some way to go (for older ages), the impact at the aggregate level is diluted owing to a flattening trend in mortality from this cause with diminishing contribution to total mortality. Correlations between causes mean the results of causal models cannot necessarily be combined in a simple additive manner. Given knowledge of the marginal survival functions for each cause, researchers from Cass Business School demonstrated that, by using the theory of copula functions, they can be combined to get an all-cause survival probability. The impact of partial to full elimination of individual causes on expected present values of annuity and life contracts was illustrated using US mortality data and a range of copulas to capture different possible interdependencies. For example, under a two-cause model with positive dependency, elimination of cancer as a cause of death could increase life expectancy from birth by up to 8.6 years.

An alternative approach of using medical knowledge to understand disease processes and how risk factors and treatments interact with these processes was described by Daniel Ryan (Watson Wyatt). Interestingly, both approaches came to similar conclusions, with the results of the modelling broadly equivalent to medium cohort projections with an underpin.

Expert opinion has an important role to play in these models and the speakers highlighted the promising, but costly, areas of gene therapy and nano-technology. Dr Philip Smalley (RGA) provided further insights from the medical profession including how 37% of over 75-year-olds have three or more chronic diseases. An understanding of how the same underlying disease can manifest in different causes of death, of how diseases can lead to further diseases or how the treatment of one disease can cause another is crucial. For example, Alzheimer’s is the seventh leading cause of death but is often manifested in death from pneumonia or other co-morbidities such as stroke and heart attack.

Putting theory into practice
Experience data by cause of death is also being built up within the growing enhanced annuity market. Matt Trott (LV=) described how market segmentation abilities, accurate pricing and high quality service are crucial to success. The need to treat customers fairly has brought product design under scrutiny, and the audience was left with some food for thought as to the appropriateness of offering investment-linked annuities to those in poor health.

Actuarial advice is also coming under greater scrutiny. Huw Evans (Watson Wyatt) described how the separation of corporate and trustee pensions advice, large deficits and increased sensitivity to longevity following investment de-risking are just some of the reasons why mortality advice will be examined more closely. He suggested a number of areas where pensions actuaries might be challenged:
>> How credible is the data on which the advice is given?
>> How much attention has been paid to base rates/improvement rates at higher ages?
>> Has proper consideration been given to trends including the timing, breadth and duration of the cohort effect?

Killing off the risk?
The seminars ended with a lively panel discussion on the prospects for the longevity risk transfer market. If 2008 was the year of the credit crunch, will 2009 be the year of the longevity swap? The panel was optimistic that it would be, with suggestions for several big transactions in the second quarter of 2009 to stimulate the market. They were clear that the development of the market will not be without challenges: for example, the time frames to which pension trustees operate. The capital market representatives posed an interesting challenge to delegates as well: ‘Just what price do you put on insuring longevity risk?’

Grateful thanks are given to Hannah Bolton at the Actuarial Profession for organising the seminars, along with the time given by all of our speakers.

Steven Baxter and Tony Cox were co-chairmen for this year’s mortality and longevity seminars. Steven is a longevity consultant at Hymans Robertson LLP and Tony provides mortality and longevity de-risking advice and broking services at Aon Benfield