Retirement and Social Security: A Time Series Approach

Brendan Cushing-Daniels C. Eugene Steuerle

WP#2009-1

Abstract

Traditional analyses of retirement decisions focus on the age, from birth, of the individual making choices about how much to work, consume, and save for old age.  However, remaining life expectancy is arguably a better way of examining these issues. As mortality rates decline, people at a given age now have more remaining years of life expectancy than they did in the past. If participation rates at older ages remain constant (or decline), then average time spent in retirement will increase. Additionally, because health status and mortality are correlated, adults with more expected years of life are generally in better health (and better able to work) than those with fewer years of remaining life.

This paper examines labor force participation rates of older workers considering both chronological age and remaining life expectancy.  Results show that participation by remaining life expectancy declines for men through the early 1990s, leveling off in the next decade.  However, participation by age has been rising for men in their sixties since the mid-1990s.  Whether we specify the empirical model by age or by remaining life expectancy, ages 62 and 65 both have strong negative effects on participation, confirming a major role in retirement decisions for Social Security.  Finally, we find that controlling for other factors – education, marital status, and business cycle effects – magnifies the decline in participation attributable to cohort effects for men born between 1900 and 1960, but reduces the importance of cohort effects for women born in these years.