engleski

The Influence of Seasonal and Other Factors on Population Mortality in Croatia

Analysing monthly series of deaths in Croatia during the period 1996-2003 we tried to identify seasonal and other influences. Two methods for seasonal adjustment were applied: X-12-ARIMA, which is empirically based, and TRAMO/SEATS, which is model-based method, the first one still dominating statistical practice. We compared results obtained by both methods. We identified significant seasonality in population mortality with both methods. Seasonal adjustment quality index was 3.366 for X-12-ARIMA and 1.808 for TRAMO/SEATS, thus giving the advantage for model-based approach. Only one outlier was detected (additive outlier in June 1999). The Easter Effect was significant as well as Trading Day Effects. X-12-ARIMA found six regressors (Monday to Saturday) in trading day effects to be significant, but TRAMO/SEATS detected one regressor (number of working days during the week) being significant. The influence of working days within the week is very interesting signalising problems in heath care during weekend. The highest number of deaths occurred in December and March, the lowest in September. It is due to climate influences. There is a change in seasonal factors which could be attributed to global warming, thus seasonal factors in the winter getting lower towards the end of series. Seasonal factors in September are getting higher due to rise in average temperatures during summer. Analysing trend, it is interesting to point out stability of population mortality within the series span despite the fact that Croatian population is getting older due to fall in live births. The explanation is the rise of life expectancy.

mortality; seasonality; population changes

nije evidentirano