Nov. 6, 1997
Vol. 17, No. 4

current issue
archive / search

    How long will older Americans live?

    Researchers say free market offers most accurate method of finding out

    By William Harms
    News Office

    Taking a new approach to the study of longevity, researchers at the University have used life insurance prices to assess life spans for older Americans and have found that longevity is expected to increase at the rate of one to two months a year for each age group currently 60 and over. Persons who are now 60, for instance, can expect to live more than a year longer than their older brothers and sisters who are 70 and have similar health characteristics.

    The study, by Tomas Philipson, Associate Professor in Economics, and Charles Mullin, a graduate student in Economics, is the first effort to use data on the prices of life insurance contracts to infer what the market believes about old-age longevity.

    The research is based on the belief that life insurance projections of life spans are more accurate than others because they are made by people whose livelihood depends upon making sound predictions. An error in computing how long people live can result in business failure for insurance companies, Philipson pointed out. Other predictions, based on census data, surveys and health statistics, are inconclusive.

    "Forecasts by different researchers and public institutions vary greatly. This is striking given that all the groups work with the same mortality data," Philipson said.

    Philipson and Mullin published their findings in "The Future of Old-Age Longevity: Competitive Pricing of Mortality Contingent Claims," a paper issued by the National Bureau of Economic Research Inc.

    Estimates of future old-age longevity have been based on difficult and subjective assumptions, such as future medical care, what physicians think is the natural limit on life and what demographers identify as longevity determinants.

    According to the researchers, the incentives of market-based forecasts are useful because there is so much disagreement among experts. Physicians, for instance, are not in agreement about a natural limit on longevity, and demographers often work with incomplete survey information, Philipson said.

    Philipson and Mullin based their research on an examination of the rates charged by life insurance companies for policies drawn by people who turned 60 between the years of 1990 and 1996. Competition in the free market causes prices to reflect the longevity of policy holders. Unlike survey data, which is often incomplete, the records these prices are based on are highly accurate, as relatives have a reason to report deaths in order to collect on the policies.

    Using the life insurance data, Mullin and Philipson found that among the group of female non-smokers who turned 60 in 1990, nearly 73 percent would live until age 80, while among those who turned 60 in 1996, nearly 82 percent would still be alive at age 80.

    Among male non-smokers, 70 percent of the men who turned 60 in 1990 will be alive by age 80, while 73 percent of those who turned 60 in 1996 are expected to live to age 80, according to the same data.

    Philipson is reluctant to project the impact of the findings on the future of social security, because more work needs to be done with a wider range of age groups and other types of data. He does see incentive-based forecasts using life insurance price data as an excellent method for assessing future trends in aging.

    "Looking at the projections made by people with an incentive to be correct is an important means of determining what we should expect in old-age longevity," Philipson said.

    "This is particularly true because people who rely on data and have no special interest in being correct in their estimates have come up with quite different projections. This study suggests that the future value of public annuity spending, such as Social Security, could be priced out by market prices rather than subjective forecasts by academics or public bureaucrats," he said.