Five period measures of longevity

This study provides a summary of recently proposed alternatives period measures of "longevity" and assesses whether empirical differences between these measures are consistent with predictions from analytic studies. Particular attention is given to the tempo effect. Three of the five period measures are virtually equal to one another in a simulated population in which mortality follows a Gompertz model with a constant rate of improvement. Similar results are observed among females in Denmark, England and Wales and Sweden in the last quarter century. However, these three measures differ substantially from the conventional period life expectancy when mortality changes over time. These findings are consistent with theoretical analysis by Bongaarts and Feeney (2002, 2003, 2005) which demonstrated that this deviation is caused by a tempo effect whose size varies with the rate of change in mortality.