The Intrinsic Estimator for Age-Period-Cohort Analysis: A Simulation Study

Kenneth C. Land, Duke University
Sam Schulhofer-Wohl, Princeton University

Age-Period-Cohort (APC) multiple classification/accounting models have long been objects of attention in statistical studies of human populations. A new approach to the statistical estimation of APC accounting models for rectangular age by period tables, called the Intrinsic Estimator (IE), recently has been developed. Some finite sampling properties and asymptotic properties of the IE have been proven. But these mathematical proofs are sufficiently abstract and difficult to understand that further exposition and analyses, especially analyses using straightforward and replicable numerical illustrations, are needed. To address this need, this paper presents results of simulation studies of properties of the IE. Generally, the results show that the IE performs well as a statistical estimator under most conditions. The results of the simulation studies may be used to inform empirical studies that use the IE for statistical estimation of APC accounting models.

  See paper

Presented in Session 135: Age-Cohort Methodological Innovations and Findings: Mortality