A simple method for age–period–cohort decomposition of firm survival data

摘要

In this paper, the aggregate data on firm survival—classified by firm age and foundation year—are decomposed into age, period, and cohort effects. Since there is a linear relationship such that age = period − cohort, the three effects cannot be uniquely identified. A simple decomposition method is newly proposed and applied to aggregate data on the hazard rates of Japanese manufacturing firms. The performances of the proposed method are compared with those of four alternative methods via Monte Carlo simulations.