[Skip to content]

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

From projections to forecasts?

The driving theory test nowadays has a ‘hazard perception’ section. It involves a driving simulator, with a moving film showing the advancing road, as seen from the driver’s seat in a car. The rear view mirror is included, as hazards can appear there first.
Actuarial projections of mortality have often been likened to driving using the rear view mirror, and as mortality continues to improve, actuaries have become better at recognising signals from the recent past. An example is the cohort effect, where the pace of improving population mortality has been shown to differ markedly depending on birth year. Used first by the Government Actuary’s Department (GAD) in the UK population projections, it became evident some time later in CMI data and was reflected in the CMI’s interim cohort projections for the 1992 series tables.
Projection methodologies provide a strong base for looking at the future, but what about when improvements happen suddenly? What, for example, might be the possible impact of a widespread ban on smoking, or the development of a new drug to cure a particular disease? One approach is to look at the potential future causes of death: medical, social, and behavioural, and to model what might happen if deaths from a particular cause were to reduce or increase substantially. This might then be used to calibrate future projections.
The profession has been active in this area for some considerable time, despite the impression given by some media comments. Recognising its continued importance, ‘Mortality and Morbidity Developments’ is one of the main research themes as the profession adopts a more strategic approach to research in 2007 and 2008. Since many readers will be aware of the CMI’s work in this field, I will focus on some other research initiatives supported by the profession.

Actuaries Panel on Medical Advances
The Actuaries Panel on Medical Advances (APMA) was established in the late 1990s. Chaired by Howard Waters, it sets out to identify medical advances or changes in environment that may affect future levels of mortality or morbidity, to model likely changes and to quantify their financial impact.
APMA is currently running a major project, funded by the Engineering and Physical Sciences Research Council (EPSRC) and the profession: modelling the demographic and financial effects of medical advances. The project, based at Heriot-Watt, aims to develop a stochastic model of disease processes so that changes in outcomes can be assessed. For example, the model could be used to assess the impact of the widespread introduction of simple treatments such as aspirin and statins for those at risk of ischaemic heart disease and stroke. The model’s parameters were set using medical research findings, and calibrated against external data from the US-based Framingham longitudinal research study. Interim results look interesting, and the research team is due to report in 2007. APMA is now looking at further potential projects.

Longevity in the 21st century
A Mortality Study Task Force was set up in 2003 and produced an outstanding paper: ‘Longevity in the 21st Century’ (Willets et al 2004). Starting with an insightful analysis of the reasons behind the trends in UK mortality throughout the 20th century, the paper went on to look at a variety of medical and social developments that have the capacity to influence future mortality. This included positives, such as medical advances, and negatives such as infectious diseases.
The paper discussed the likely pattern of future mortality improvements, and concluded with an assessment of the implications for life assurance and pensions, society in general, and the actuarial profession.
If I were to pick one comment that highlighted the difficulty in projecting future longevity, it would be this:
It seems probable that, in some cases, the mortality assumptions currently used by actuaries do not make sufficient allowance for future improvement mortality rates at some ages are now just 2% of the rates that applied 100 years ago. This magnitude of change could scarcely have been thought possible at the beginning of the last century. Would our actuarial predecessors have imagined such an improvement was possible?
Presented both at the Faculty and the Institute, the paper provides an excellent background to mortality developments, and gives far-reaching comments on the financial and social implications of these developments for the future.

Recent medical advances
Developing further on the issue of cause of death, a multi-disciplinary project has started at University College London. The project focuses on the contribution of medical advancements to mortality improvements from 1970 to 2000. It will aim to test several models to measure the impact of factors such as medical advances, behavioural changes, material changes, and socio-political issues on longevity.
One of the most interesting potential outcomes from this research is expected to be tables of contributions, showing the estimated reduction in population death rates relating to the various explanatory factors. This might allow the researchers to compare the impact of various contributory factors to the decline in mortality across different socio-economic groups, age cohorts, and possibly different countries.

Scoping mortality research
Focusing on mortality by cause of death must of course draw on the understanding and activity of other professional disciplines. This is nothing new for the profession we already benefit from clinicians, gerontologists, statisticians, and demographers in our research projects. But if we are to benefit fully from an understanding of research already done in other disciplines, it will be helpful to take a wide-ranging review: looking at findings to date and how they interrelate with our own research.
The profession has therefore launched a scoping mortality research project to look at research findings on mortality developments from medicine, epidemiology, gerontology, demography, health economics, medical sociology, social policy, and psychology. The idea is to identify gaps in the spectrum of research and point the way to potentially valuable areas of further multidisciplinary research. If we get this right we should, as in the past, be able to call in funding from the major research agencies such as the ESRC (Economic and Social Research Council) and the EPSRC.

Other work in progress
At the time of writing, the Faculty Mortality Research Group is preparing a paper on reasons for mortality improvements and the CMI is looking further at the need for guidance, tools, and projections. I am conscious that there is other research being supported by the profession, and hope that the researchers will be encouraged to submit further articles to The Actuary.

Mortality rate projections
If mortality rates at some ages are now just 2% of the rates that applied 100 years ago, where does that point us for the future? CMI and GAD projections have been criticised at times for being too radical, and at other times for being too tame. Yet both have projected substantial further improvements in mortality.
Regardless of how good we get at analysing the past and projecting the future, we have to accept that uncertainty applies: wars, pandemics, and wonder drugs can produce unpredictable results. Stem cell research, genetic engineering, and the potential impact of global warming add a particularly 21st-century edge to uncertainty. Stochastic modelling can help measure the impact of uncertainty, but we shall need different models for different purposes.
Is it possible to factor in the potential effects of the above-mentioned changes and produce forecasts that can be relied on? I doubt it. Perhaps, referring back to the driving theory test, we should focus on ‘hazard perception’ rather than certainty. We need to accept the uncertainty inherent in any projection of the future. That is understood by those working on projections; it needs to be clear to all those who use projections to support financial decision-making within life insurance, pensions, and elsewhere.
What we can do is work with others to understand better the factors affecting mortality improvement (and deterioration), find ways of allowing for the risks, and contribute to the effective management of those risks. The rear view mirror will remain vital; what we could do with now is a clearer view out of the front windscreen!

07_04_04.pdf