<Article> A maximizing covariation approach to age-period-cohort analysis with application to smoking behavior--Comparison of reliability with an intrinsic estimator-- (Special Issue of the Faculty of Business Administration)

IR

Search this article

Description

type:01 紀要論文 / Departmental Bulletin Paper

In an additive age-period-cohort model, the behavior of the dependent variable comprises separable components for the three intrinsic effects and inseparable components capturing their linear dependency. This paper proposes a constraint only on the part of the parameter vector that generates the inseparable component to solve the identification problem. This parsimonious constraint will reduce the bias because the intrinsic effects are not biased. The three effects are identified while maximizing covariation between them; consequently, the inseparable component is divided equally between the birth factor (cohort effect) and the subsequent environmental factor (age+period effect). The reliability of the identified estimator under this model was compared with that of the intrinsic estimator proposed by Fu (2000). The difference in relaibility between the estimators in two models depends on the three effects.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top