The danger of underestimating how long we will live

21.06.2004

People retiring in the next decade or so will live considerably healthier, more active and longer lives than their predecessors. But according to research by James Banks and colleagues, many are drastically underestimating the chances of their retirement lasting at least 10 years – and hence may not be saving ‘enough`.

The first results of Banks et al’s study of people’s expected longevity – which draw on data gathered in the English Longitudinal Study of Ageing and are published in ESRC’s new report Seven Ages of Man and Woman as part of Social Science week – show that:

People currently approaching retirement will receive the most generous state pensions of any generation – and will enjoy considerably healthier and longer lives.

But on average, both men and women aged 50-64 are underestimating their chances of living to 75.

Women report chances only slightly greater to those of men. As a result, women underestimate their longevity chances by considerably more than men.

The ‘true’ probability of a 60-64 year old woman reaching the age of 75 is greater than 80%. But, on average, 60-64 year olds report a chance of just 65%.

Those with more education and income report a higher chance of living to 75 than those with less, although the effects are rather small – less than five percentage points – and less significant for men than for women.

Different dimensions of health are strongly correlated with expectations: those with better self-reported health report higher chances, and those with previous conditions such as heart attack, along with smokers, report lower chances.

The data also include measures of attitudes to age that may capture unobserved vitality and health as well as attitudes, optimism or differences in reporting behaviour. These are positively correlated with longevity expectations over and above the measured health and socio-economic variables – those who think old age ‘begins’ later also report higher chances of living to later ages.

Social engagement (as measured by the frequency with which people see friends) is also positively correlated with subjective life expectancy.

Expected longevity is a key variable for individuals and governments planning their retirement resources. Errors in expectations could have important policy implications. If, as seems likely, people take decisions based on their subjective expectations rather than the true probabilities, unrealistic expectations could well generate adverse consequences in the future.