Cohort mortality pattern modeling – model application to Swedish cohort data

Petr Mazouch, University of Economics , Prague (VSE)
Klara Hulikova, Charles University in Prague

Mortality modeling is one of the traditional and fundamental demographic issues. The purpose of the mortality modeling is to find relations and hidden regularities and patterns in the mortality development. Knowledge of these patterns could be used among others for the forecasting of the future development of mortality. There exist many various methods of mortality forecasting leading to more and less accurate outlooks. This paper introduces a simple alternative approach which is based on cohort mortality patterns modeling. The assumption of constant change of the force of mortality between two following ages across cohorts is the fundamental base of the proposed model. This assumption was verified on time series of Swedish data which are long enough for this purpose. It was proved that the mentioned changes of mortality between ages were really nearly constant for all the cohorts and ages used for this verification. Moreover, the created time series of mortality changes meet the requirements for further statistical modeling. Finding of the coefficients of mortality increase between two ages and its intrinsic patterns across generations could be used for its very simple application to not yet extinct generations and through that for the estimation of their future mortality development. The verification of this potential usage of the proposed model was realized on the shortened cohort data of the Swedish population. After the application of the methodology described in this paper, the comparison of the estimated and real development was carried out and assessed.

  See extended abstract

Presented in Poster Session 3