英國勞動力調查的季節調整－－相平移、移動節日、以及2010年閏周的效果修正

Seasonal Adjustment of the UK Labour Force Survey: Correcting for Effects of the Phase Shift, Moving Holidays and the 2010 LeapWeek

英國國家統計局（ONS）所發佈的重要統計數字大多是經過季節調整後的結果， 以期有助於解讀及詮釋時間序列的變動。季節調整的目的在於去移因季節性所產生的影響（比如聖誕節的高銷售金額）或來自於日曆排定的影響（如復活節可能出現在3 月或4 月）。每一年的12 個月份在英國勞動力調查（LFS） 中均以一個「4 周」或「5 周」長的調查月表示， 故時間序列必須透過regARIMA 模型的「日曆化」， 來改進季節調整估計的品質。這樣的步驟一直被檢測著， 直到2010 年10 月， 英國LFS 受限於一個1 周的調查空窗期， 或稱「閏周」， 以便它們的報告期間與歐盟其它國家一致，也因此出現了一個新奇的問題。本文解釋LFS如何以先驗調整及必要的後續調整，來克服2010年閏周引起的額外複雜性。

In the Office for National Statistics (ONS), most headline statistical outputs are seasonally adjusted to aid interpretation of movements within the time series. Seasonal adjustment aims to remove effects associated with the time of the year (for example, higher sales at Christmas) or the arrangement of the calendar (for example, Easter moving between March and April). Each of the 12 Gregorian calendar months is represented in the UK Labour Force Survey (LFS) by a four- or five-week-long survey month, and the time series must therefore be “calendarised”, through the application of regARIMA modelling, in order to improve the quality of seasonally adjusted estimates. This process is tried and tested but, in October 2010, the UK LFS was subject to a one-week-long survey break, or “leapweek”, in order to align its reporting periods with those of the rest of the European Union – which presented a novel problem. This article explains howLFS prior adjustments are constructed in general, and the adaptations necessary to overcome the additional complications caused by the 2010 leapweek.