Rotation of the age-varying parameters in multi-population mortality models

Abstract

In addition to the well-known Lee–Carter model, two versions of a multi-population mortality model, known as augmented common factor models, were fitted to Hungarian data in the present study. The two subpopulations considered in the analysis were men and women. When predicting mortality rates, it is important to not only predict the trend in mortality change (improvement) given that the age-specific coefficient of this time-varying parameter also changes over time. The phenomenon of this time dependence of the age pattern and consideration thereof in mortality projections are known in the literature as rotation. As a result of the present research, possible trajectories for the life expectancy of men and women in Hungary up to the year 2050 were determined by predicting rotated and nonrotated versions of three different mortality models.