[Skip to content]

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

Signs of ageing

Longevity, health, economic resources, and social and family circumstances are inexorably linked. In the light of such linkages, how much do we really know about the fundamental causes of ill-health, disability, and ultimately mortality outcomes in old age? For that matter, how much do we really know about the fundamental causes of particular economic, financial, or social outcomes in old age? The answer is, unfortunately, not much.
When it comes to data on ageing, the supply of statistics showing correlations is plentiful. Such statistics can be useful in the short term for forecasting purposes but, by confusing symptoms with causes, do not allow policy-makers to target their interventions appropriately. Nor do these statistics allow financial institutions to design products that address the true needs of their potential customers.
Hopefully this is beginning to change. Economists are now building past health behaviours and current health outcomes into models of retirement and are acknowledging the links between health, education, and the accumulation of economic resources over the lifecycle. Epidemiologists are looking at economic and social resources, along with inequalities in such resources, as risk factors for morbidity and mortality. And both groups are taking up the challenge of showing true causal pathways within and across dimensions.

The depth and breadth of good data
To investigate these casual pathways requires good data and existing sources have so far proved inadequate. Representative surveys focusing on one dimension at a time do not allow investigators to consider the correlation between dimensions at the individual level, and hence are not always useful. Those surveys that do cover all dimensions tend to incorporate crude measures that do not facilitate appropriately scientific analysis. Finally, analysis of official data or commercial databases can only be limited in scope, typically focusing on financial dimensions alone with minimal other information available. In addition, such analyses cannot account for family or social effects, nor look at the attitudes or expectations of individuals in various circumstances.
To tackle the limitations of existing data sources, researchers based at University College London, the Institute for Fiscal Studies, and the National Centre for Social Research have been designing and collecting new data with which we can begin to understand the economic, social, psychological, and health elements of the ageing process. The English Longitudinal Study of Ageing (ELSA) is the first study anywhere in the world of its type collecting detailed biomedical, economic, and social data on the same individuals on an ongoing basis. ELSA is jointly funded by a consortium of government departments and the US National Institute on Aging, and it will yield public-use data, while revolutionising empirical work on the ageing process and associated policy questions.

The ELSA sample
The study recruited over 12,000 respondents aged 50 or over to be interviewed every two years and given nurse visits every four years. The data provide a representative sample of the population of England aged over 50 living in households. As study members move into institutions the ELSA sample will become representative of the whole English population aged 50 and over. Crucially, since the same individuals are followed as they age, the series of economic choices and outcomes and their dependence on health and social factors can be analysed at the individual level, providing an unparalleled empirical picture of the evolution of life over 50 in England. The first wave of ELSA data was collected in 2002/3 and the second wave is currently being collected.
At the core of ELSA are extremely detailed biomedical, economic, and social measurements. Blood samples are taken and analysed for known risk factors and respondents undergo physical and cognitive performance tests. In addition, the health questionnaire collects information on self-reported health and functioning, previous health histories and behaviours, and limitations in activities of daily life. On the economic side, full details of all sources of income, employment, asset, and wealth holdings and private pension arrangements are collected, along with measures of consumption and housing. Furthermore, ELSA collects respondent’s National Insurance numbers and requests their consent to link their data to government records on their history of National Insurance contributions and all benefit receipts (including tax credits and state pensions). Around 80% of respondents give permission to link their records, and similar rates are recorded for permission to link to hospital health records. Respondents are currently being approached for their permission to extract and store genetic material from their blood samples.

Future expectations
The potential of such data is huge and the first fruits are already coming to bear. The 2002/3 wave of data provided much-needed information when the report and first tabulations were published in December 2003. Headlines tended to focus on occupational and socio-economic patterns in disability and morbidity. Also important was the first evidence of how holdings of total assets and wealth vary across the population, how pension coverage and particular pension arrangements vary by wealth and income, and how employment probabilities at older ages vary by pension arrangements and health status. Data from ELSA will also be providing key evidence for the interim report of the Pensions Commission this year.
One further dimension of the ELSA data is both important and novel in UK surveys; this is the collection of quantitative information on expectations of the future. Respondents are asked the chances of various events happening to them in the future, on a scale of 0 to 100, where ‘0 means that you think there is absolutely no chance an event will happen, and 100 means that you think the event is absolutely certain to happen’. On this basis expectations are collected for, among other things, the chances of living to various ages, the chances of being in paid work at age 60, the chances of leaving a bequest and receiving an inheritance, and the chances of future financial problems. Such questions have been used in the US, and UK trials showed them to be operating successfully over here, so ELSA became the first non-US study to use such questions.

How long do you expect to live?
Analysis of expectations data has already shown potentially important results on individual longevity expectations on average people approaching retirement are underestimating their chances of living to age 75 (see James Banks, Carl Emmerson, and Zoë Oldfield, 2004). And women tend to underestimate these chances by more than men. On average, 6064-year-old women reported a 65% chance of living to age 75. Even estimates of the ‘true’ probability based on cross-sectional life tables would suggest this chance to be more than 80%, and any projections of cohort increases in longevity would widen the gulf between these two numbers. (See figure 1.)
Such underestimation of longevity would lead to a demand for retirement saving that was ‘too low’ and individuals being unwilling to annuitise wealth on reaching retirement. As such, it suggests that, when thinking about the government’s ‘informed choice’ agenda and arming individuals with greater information on their pension arrangements, we should think of providing more information on the more fundamental issue of longevity as well. It also suggests that we need to know much more about the way in which individuals form their expectations of the future.
Further research on the 2002/3 ELSA data is already under way. Some of it, partly sponsored by the actuarial profession through contributions to the IFS consortium on pensions and retirement saving, will look directly at the links between public and private pension arrangements and other factors (including expected longevity). Research will also investigate the way in which cognitive functioning is associated with financial expectations and holdings of financial instruments. As each wave of data builds up, the returns to such ongoing empirical research will only increase. And in the future, as the academic, policy-making and business communities generate new research questions, we will now have the data already in place with which to investigate solutions. o