engleski

The analysis of seasonal variations in the presence of intervention

Seasonal analysis of economic time series becomes much more important for the transition country like Croatia. Since she is fast heading towards market-oriented economy, seasonally adjusted series could provide the information for economic policy and decision making people in the government as well as in the companies. The problem for economic time series in recent past is the presence of the intervention, namely war. It does effect the seasonal adjustment procedure, as well as trend and irregular component. In some series (e.g. unemployment) seasonal component suddenly disappears or becomes very disorted. In this case, it is very difficult to make decomposition of time seies and to forecast its short-term future values. We have used several methods to analyse such a series influenced by the intervention (war) such as seasonal ARIMA with the intervention and structural models (Harvey & Durbin). For seasonal adjustment we used X-11-ARIMA method and the newest method X-12-ARIMA. The use of these methods showed substantial problems in the identification process of the appropriate ARIMA model incorporated in seasonal adjustment procedure, especially in the case when seasonal component dissapears during the war, and does not appear completely like before the war.

seasonal variations; intervention

nije evidentirano