ABSTRACT
In this chapter, the author discusses the pattern of the age-period-cohort (APC) data in the rectangular tables and illustrates the relationship among the age, period, and cohort. He examines that such inherent relationship leads to model complexity and induces a complex parameter identification problem in APC models. The author then explores special patterns in APC data, that is, the so-called Lexis diagram, paying special attention to age, period, and cohort groups. In the Lexis diagram, a simultaneous move-up or move-down in the row and column by one cell will land on the same diagonal of the table. Such a special pattern determines a linear relationship among the rows, columns and diagonals. The APC models with discrete linearly dependent covariates can resolve the identification problem with all covariates involved in the model because of the increased parameter space and the model degrees of freedom through discretization of the factors.
