A comparison of Poisson or Negative Binomial Regression and Lee-Carter Models of forecasting Norwegian male mortality

Abstract

Mortality rates are the ratio of death counts and estimates of the population exposed to risk of deaths in matched intervals of time and age. Mortality forecasting is the use of historical data of mortality to determine future pattern of trends. Mortality forecasting remains a major challenge for the planning of pension systems and management of annuity businesses. Having high mortality forecasting accuracy requires a model that provides good fit to the historical mortality data for consistent predictive performance. Poisson or Negative binomial regression and Lee-Carter (1992) demographic models are some of the predictive models being consistently used in mortality forecasting. Hence this thesis seeks to compare and evaluate forecasting accuracy of these two models in predicting Norwegian male mortality.