Aging and work capacity

Declining health with age can limit individuals’ work capacity, increasing the likelihood of mismatch between their abilities to perform certain tasks and the minimum demands of the jobs available to them. Traditional measures of health status are insufficient for understanding how labor supply outcomes are influenced by the match between individuals’ abilities and job demands. We use unique survey data on individuals’ self-reported ability levels, harmonized with occupational ability requirements from the O*NET database, to develop a new measure of work capacity. We find that average abilities overall and across different domains are high relative to average occupational demands. At the same time, age-related declines in abilities are modest, at least through age 70. Putting these elements together, individuals’ work capacity is relatively stable with age. Finally, we show that our measures of work capacity are predictive of current and expected future labor supply outcomes, with and without controls for standard health variables.

This research was supported by grant number UM19-02 from the U.S. Social Security Administration (SSA) through the Michigan Retirement and Disability Research Center (MRDRC) and by grant number R01AG056239 from the National Institute on Aging. Its contents are solely the responsibility of the authors and do not represent the opinions or policy of SSA or any agency of the federal government. We thank Kate Bent, John Pencavel, seminar participants at the Tinbergen Institute, Tilburg University, Netherlands Bureau for Economic Policy Analysis, Erasmus University, KU Leuven, UQAM, the University of Chile and USC, and participants of the 2019 MRDRC Workshop, 2019 Retirement and Disability Research Consortium Conference, 2020 Stanford Working Longer Conference, 2022 CIPHER conference, 2024 Society of Labor Economists Annual Meeting, 2024 American Society of Health Economists Annual Conference, 2024 London Economics of Longevity and Ageing Conference, and the 2024 Association for Public Policy Analysis & Management Fall Research Conference for helpful comments and suggestions, and Michael Jetsupphasuk, Patrick Rhatigan and David Zingher for excellent research assistance. The data used in this article are available online at https://alpdata.rand.org/.