1. WHAT IS SEASONAL ADJUSTMENT?
5. QUALITY OF SEASONAL ADJUSTMENT
6. SPECIFIC ISSUES ON SEASONAL ADJUSTMENT
Monthly and quarterly time series are often characterised by considerable seasonal variations, which might complicate their interpretation. Such time series are therefore subjected to a process of seasonal adjustment in order to remove the effects of these seasonal fluctuations. Once data have been adjusted for seasonal effects by X-12-ARIMA or some other seasonal adjustment tool, a clearer picture of the time series emerges.
For more information on seasonal adjustment: metadata on methods: seasonal adjustment
Due to fixed and movable holydays and holidays in July and December the figures of man-weeks worked vary across the year. Also the number of employed persons and unemployed persons vary across the year, especially for the young people. Among other matters this is because of summer substitute and young people searching fore summer jobs.
We seasonally adjust 2 age-divided series (more/less than 24 year) separately for employed persons and unemployed persons, and add the two series afterwards.
For man-weeks worked we seasonally adjust 3 gender- and age-divided series: persons 16-24 years, males 25-74 years and females 25-74 years.
Due to no sample uncertainty, Registered unemployed and registered unemployed + government measures are also seasonally adjusted as a supplement to the LFS. The Register-figures are seasonally adjusted separately for 4 gender- and age divided (more/less than 24 year) series.
Pre-treatment is an adjustment for variations caused by calendar effects and outliers.
Calendar adjustment involves adjusting for the effects of working days/trading days and for moving holidays. Working days/trading days are adjustment for both the number of working days/trading days and for that the composition of days can vary from one month to another.
Comments:
Comments:
LFS is surveying persons attachment to the labour market in whole reference weeks. The reference weeks are not split to follow the exact months of the calendar. Therefore, all the monthly data files include as many Mondays as Tuesdays and so on. The monthly data files include either 4 or 5 reference weeks, and the inflation factors adjust fore that. Due to this the number of working days in a month is only affected by holidays and national days off.
Comments:
The seasonally adjusted time series for employed persons over 24 year and unemployed persons over 24 year are pre-adjusted if Easter is in March.
Due to sensitivity to single holidays or common days off, the man-weeks worked series in addition are pre-adjusted for 2nd Easter day, May 1st, May 17th, Whit Monday and Ascension Day. (We also take account of the effect when two of these days occur on the same date, for instance Ascension Day on May 17th in 2007 and 2012 and on May 1st. in 2008).
In addition we pre-adjust man-weeks worked series for the number of common days off that falls on week days in December, and for the number of normal days in the week between Christmas and New Year's Eve that falls in December/January in the LFS. Also we pre-adjust man-weeks worked series for the number of week days in June the in the LFS falls in July, due to the fact that we don’t divide reference weeks (Monday – Sunday) between months in the LFS, like it is done in the calendar. All regression variables are deseasonalized by subtracting the long term monthly average. Also the regression variables are fine tuned by a week multiplier to take account of that some months represent 4 reference weeks and other 5 reference weeks.
Comments:
We also take account of the fact that LFS don’t divide reference weeks (Monday – Sunday) between months, like in the ordinary calendar. For instance the holiday May 1st may in the LFS be in April. Correspondingly may New Year's Day be in December in the LFS.
Outliers, or extreme values, are abnormal values of the series.
Comments:
Only additive outliers are detected automatically.
Pre-treatment requires choosing an ARIMA model, as well as deciding whether the data should be log-transformed or not.
Comments:
The pickmdl procedure in Version 0.3 of X-12-ARIMA is used with the standard method (=first) in the yearly identification of ARIMA models. (Then these automatically identified models are hard coded in the specification files.)
In the list of ARIMA models that are checked i pickmdl procedure, 3 extra ARIMA models are included in order to take better account of the data acquisition in the LFS where interviewees are interviewed every 3 months for 2 year. The extra ARIMA models are:
(0,1,[1,3])(0,1,1)12
(0,1,3)(0,1,1) 12
(0,1,1)1(0,0,1)3(0,1,1)12
The decomposition scheme specifies how the various components – basically trend-cycle, seasonal and irregular – combine to form the original series. The most frequently used decomposition schemes are the multiplicative, additive or log additive.
In some series, consistency between seasonally adjusted totals and the original series is imposed. For some series there is also a special relationship between the different series, e.g. GDP which equals production minus intermediate consumption.
Comments:
The labour force is not seasonally adjusted. The labour force, seasonally adjusted is defined as the sum of employed persons, seasonally adjusted and unemployed persons, seasonally adjusted.
The equality is imposed by indirect approach.
Direct seasonal adjustment is performed if all time series, including aggregates, are seasonally adjusted on an individual basis. Indirect seasonal adjustment is performed if the seasonally adjusted estimate for a time series is derived by combining the estimates for two or more directly adjusted series.
When performing seasonal adjustment of a time series, it is possible to choose the period to be used in estimating the model and the correction factors. Correction factors are the factors used in the pre-treatment and seasonal adjustment of the series.
Comments:
Due to a major revision of the LFS in 2006, we now let the time series start in 2006.
Seasonally adjusted data may change due to a revision of the unadjusted (raw) data or the addition of new data. Such changes are called revisions, and there are several ways to deal with the problem of revisions when publishing the seasonally adjusted statistics.
Comments:
Revisions second months in the quarter.
The figures from the first months in the quarter are preliminary. They are re-estimated, at the same time as the data from the second months in the quarter are estimated, supplemented with the extra interviewees that we got after the preliminary deadline and with more up to date or better auxiliary register variables.
Quarterly. The figures from the first two months in the quarter are preliminary. They are re-estimated at the end of the quarter supplemented with the extra interviewees that we get after the preliminary deadlines and with better auxiliary register variables.
Yearly. After seasonal adjustment, the levels are corrected to make the annual averages of the seasonally adjusted figures equal to the corresponding unadjusted figures from the LFS. The annual averages of a year are not ready until we publish the 4th quarter figures. At that point we use new factors of level correction. Until the 4th quarter publication is available, preliminary level correction factors from last year have to be used.
● Partial concurrent adjustment
The model, filters and calendar regressors are re-identified once a year and the respective parameters and factors re-estimated every time a new or revised data becomes available.
Comments:
Additive outliers are detected automatically and parameters re-estimated every time new data becomes available.
Due to relatively short time series, we will consider re-identifying the models and filters every quarter in 2012.
Table of quality measurement for this statistics.
For more information on the quality indicator in the table see: metadata on methods: seasonal adjustment
Comments:
As from 2006 the LFS was revised, among other things the lower age limit covered by the survey was lowered from 16 to 15. In order to avoid breaks in the series, we start all the time series in 2006.
Comments:
Due to large random variation (sample uncertainty) and small seasonal variation, we do not seasonally adjust all the main variables of the LFS gender divided. This is the case for unemployment, employment and for Employed persons man-weeks worked for persons 16-24 years. Afterwards, gender divided figures are constructed from seasonally adjusted figures by utilizing monthly gender distributions calculated from trend-cycle figures from additional unofficial seasonal adjustments of the LFS.
Options:
Comments:
In order to reduce the uncertainty, the presented series are three months moving averages of the seasonally adjusted figures. For instance the figure from September is the average of the estimates from August - October.
See previous releases for historical data.
Statistics Norway’s metadata on methods: seasonal adjustment
EUROSTAT: Seasonal Adjustment. Methods and Practices
US census: X-12-ARIMA-manual
Dinh Quang Pham: Time series and basic seasonal adjustments and X-12-ARIMA, Reports 2001/2, Statistics Norway
ESS-Guidelines X-12-ARIMA-manual Seasonal Adjustment. Methods and Practices