New Methods for Analyzing Active Life Expectancy

Abstract

The increment-decrement life-table methods used in several recent analyses of active life expectancy depend on parameters representing rates of movement between functional states such as "active" or "disabled." Available data often pose severe problems for the derivation of these parameters. For example, panel-survey data typically fail to record functional status between interviews. The time intervals between interviews also tend to vary across respondents, often substantially. The Longitudinal Study of Aging, used in this research, exhibits these problems. The authors develop a discrete-time Markov chain model of functional status dynamics that accommodates these features of the data and present maximum-likelihood estimates of the model. Also introduced is a new technique for the calculation of active life expectancy: microsimulation of functional status histories. The microsimulation technique permits the derivation of several new indexes of late life-course outcomes.