When Can Age, Period, and Cohort be Separated?
1978; Oxford University Press; Volume: 57; Issue: 1 Linguagem: Inglês
10.1093/sf/57.1.282
ISSN1534-7605
Autores Tópico(s)Insurance, Mortality, Demography, Risk Management
ResumoAny attempt to separate age, period, and cohort effects must distinguish between three levels of analysis: measuring differences, inferring effects, and imputing causes. The three differences that must be measured are longitudinal, cross-sectional, and time-lag. The inference of effects is based on the fact that each difference is composed of two of the three possible effects: age, period, and cohort. Only under specified conditions and with certain assumptions is it possible to separate and estimate the values of these three effects. The imputation of causes for these effects must be based on evidence from outside this model. This model is compared to other methods and illustrations of its utility are discussed.
Referência(s)