Fluctuations caused by abnormal levels of seasonal factors (e.g., extreme weather conditions or unusual circumstances) are visible in the seasonally adjusted data because they exceed or fall below the average seasonal values. Deviations of an occasional nature and unusual changes for which there is an economic explanation (e.g., effects of changes in economic policy, large procurement, natural disasters, etc.) are not eliminated in the season adjustment, i.e. they are part of the seasonally adjusted data. Seasonal adjustment also includes the elimination of calendar effects, as far as they result from the difference in the number of weekdays and weekends, and can be quantified. Seasonally adjusted data are data that are removed periodically repeated within period components of the series - seasonal and calendar variations. So seasonally adjusted data highlights the behavior of the studied variables (changes in the direction of development and changes related to the business cycle). The seasonally adjusted lines contain random component (noise) having different levels in different series and the amount depends on the economic nature of the study, and applying the method of seasonal adjustment . Seasonal adjustment of statistical indicators in the NIS is organized according to the "Guide of the European statistical system for seasonal adjustment". The procedure is performed using the software developed by Eurostat product Demeter, where by the algorithm TRAMO / SEATS. The procedure for obtaining seasonally adjusted data includes correction of the series in terms of calendar effects and abnormal values. Calendar effects are expressed as : - differences in the number of working days in a given period;
- effect of leap years ;
- holidays with variable dates (Easter).
Adjustments for differences in the number of working days are made to obtain seasonally adjusted data, which are independent of the number and structure of days (number of Mondays, Tuesdays etc. / number of weekdays and weekends) in the quarter. Abnormal values (outliers) are unusual changes in time series of a given indicator. Their manifestation can be in several ways, the most significant are the peak values (abnormal values of individual observations in the order), short-term changes (series of abnormal values to change the level of the indicator for a short period of time), changes in the level (a series of abnormal values ??with constant size and time- effect on the level of the indicator). |