Seasonal adjustment, which is a routine activity in statistical offices nowadays, and the connected mathematical background have been a subject of theoretical investigations for several decades. However, methods and tools of seasonal adjustment are still under development and perpetual debates focus on them. Furthermore, there is significant flexibility regarding applied adjustment settings and model selection, which may lead to subjective and ambiguous results. As the number of the series to be adjusted is rapidly increasing and the quality of official seasonally adjusteddata is increasingly important, the need of recommendations and guidelines is indisputable. ESS Guidelines on Seasonal Adjustment (2009) can be regarded as benchmark working material in this topic.
The goal of this module is to discuss important issues on seasonal adjustment, providing a guide on how to deal with them describing practices and giving some references to achieve further information. One of the most essential issue is providing temporal and cross-sectional consistency of time series. Although forcing consistency may hold disadvantages, it may be required to fulfill accounting constraints (as in the quarterly national accounts). Owing to statistical investigations and the available significant computer resources, nowadays, this task is much less demanding than it was a few years ago.
The module also focuses on the choice between indirect or direct approach to seasonally adjust time series derived as aggregation of other component time series. Choosing between these approaches is not obvious, comprehensive analyses have been performed in order to eliminate uncertainty. Practices and guidelines are described in the related subsection. Revision is also a crucial element of the seasonal adjustment: the updating of unadjusted data and the use of bilater filters lead to revise the seasonally adjusted data previously released undermining the credibility of the producer agencies.
The financial crisis seriously undermined the reliability of the results of seasonal adjustment. The seasonal pattern and the behaviour of time series may change significantly. Therefore, it is necessary to take the impacts of the crisis into consideration. Although, this aim is available through outliers and ramp effect, monitoring time series is also essential part of treatment.
Seasonal adjustment and the comparison with raw data are often subject of confusion among users. Consequently, the details of publication and communication policy is essential to prevent misunderstanding.
To read the entire document, please access the pdf file (link under "Related Documents" on the right-hand-side of this page).
Your feedback is appreciated. Please send your remarks, suggestions for improvement, etc. to email@example.com.