Presenter(s): Duncan Elliot & Craig McLaren, ONS
Date: 3 December 2021, 4:00 PM (CET)
It is standard practice in official statistics to seasonally adjust monthly and quarterly time series using the methods recommended in the ESS guidelines on seasonal adjustment. The COVID-19 pandemic accelerated demand for official statistics time series data to be published at a higher frequency than the more usual monthly or quarterly periodicity of indicators. There are well developed software for the seasonal adjustment of monthly and quarterly data, such as JDemetra+ and X-13ARIMA-SEATS, but neither of these are currently designed to handle daily or weekly series.
The aim of this seminar is to introduce participants to some of the experimental methods for seasonally adjusting higher frequency time series that have been explored at the Office for National Statistics. We present examples of published experimental weekly and daily time series and discuss some of the challenges in seasonally adjusting these series.
Webinar learning outcomes
Participants will be introduced to the concept of seasonal adjustment in official statistics, the challenges of seasonally adjusting weekly and daily data, and examples of how this has been achieved in practice.
The webinar discusses:
- What is seasonal adjustment and why do we do it?
- Seasonal adjustment standard practice in official statistics
- Higher frequency time series (weekly and daily data examples)
- Methods and software for seasonally adjusting higher frequency time series
- Experimental high frequency seasonally adjusted outputs at the Office for National Statistics
Prerequisites for the webinar
No formal prerequisites but a basic statistical knowledge is assumed.
Further reading and resources
ESS guidelines on seasonal adjustment https://ec.europa.eu/eurostat/documents/3859598/6830795/KS-GQ-15-001-EN-N.pdf
Handbook on seasonal adjustment https://ec.europa.eu/eurostat/documents/3859598/8939616/KS-GQ-18-001-EN-N.pdf
Recording of the webinar