Analysis on Cohort Effects in view of Differential Geometry and its Applications

Abstract

This paper analyzes birth cohort effects and develops an approach which is based on differential geometry to identify and measure cohort effects in mortality data sets. The measurement is quantitative and provides a potential method to compare cohort effects among different countries or groups. Data sets of four countries (e.g. U.k., U.S., Canada and Japan) are taken as examples to explain our approach and applications of the measurement to a modified Lee-Carter model are analyzed. In fact, this paper is an upgrade version of our paper arXiv:1504.00327. There is a new section which gives applications of our approach based on the Lee-Carter and APC models.