Modeling Mortality Surfaces: The Influence of Period and Cohort Components on Elderly Mortality

Elisabetta Barbi, Università di Roma "La Sapienza"
Carlo G. Camarda, Max Planck Institute for Demographic Research

In this paper we consider two main questions: To what extent do early-life and current factors determine trends in old-age mortality? Did early life factors determine the decline of adult and old mortality in the second half of the 20th century in the European countries? We present a new explanatory but flexible relational model to analyze mortality surfaces including period and cohort indicators. This two-dimensional model is estimated via a semi-parametric approach by smoothing the general mortality trend over age and time with penalized B-splines. This allows, on one hand, to keep the non-linear structure of mortality dynamics over the long term, and on the other hand to evaluate simultaneously the relative importance of period and cohort effects. We apply the described model to two case studies, Sweden and Italy, which have been chosen because of their different past mortality dynamics.