Abstract:
Originally, time and age have been considered by Lee and Carter for mortality rate forecasting since 1992.
The objective of the proposed model is to incorporate two additional factors, gender and region in order
to forecast Thai mortality rate. Multiple regression is used to construct the forecasting equations for five
groups of age, i.e. 1-19, 20-39, 40-59, 60-79 and 80-99 years old. The forecast accuracy in form of Mean
Square Error (MSE) and Variance Accounted For (VAF) are determined and compared with original Lee-
Carter model. The result shows that both MSE and VAF of the proposed model are less than those of
Lee-Carter model for all age groups except age group 20-39 for in-sample data. MSE and VAF of outsample
data are similar to those of in-sample data but Lee-Carter model yields better result for age
group 40-59 only. Generally, the mortality rate forecasted by the proposed model is lower than those of
Lee-Carter model. Finally, Thai mortality rates are forecasted for years 2019-2023 for five age groups of
male/female in seven difference regions across Thailand.
King Mongkut's University of Technology North Bangkok. Central Library