SMOOTHING OF MORTALITY CURVE: THE USAGE OF MOVING AVERAGES AND GOMPERTZ-MAKEHAM FUNCTION IN THE CASE OF THE CZECH REPUBLIC

摘要

Our paper discusses the possibilities of levelling the age-specific death rates of the Czech population using different lengths of moving averages and the Gompertz-Makeham (G-M) function. Moving average is a technique that belongs between the simpler one, G-M function between the most used approaches for levelling of mortality curves at higher ages. These approaches are very useful in the conditions of the Czech Republic. The lengths of moving averages were used according to Czech literature and empirical studies and finally set to 3, 9 and 19 values with different weights. The best results of smoothing by moving averages is achieved with the use of moving averages of 19 values, on the other hand we lose most initial observations. The G-M functions were estimated in STATA software and results are different e.g. in comparison with 19 values length moving averages. Results from this study will be used for future research to implement this technique into extrapolation script that will be able to smooth the values and then extrapolate them to the highest ages (110 years or above).