All comparisons between successive periods are based on working day adjusted and seasonally adjusted index numbers in constant prices. Comparisons between the current period and the corresponding period of the previous year are based on working day adjusted index numbers in constant prices.
When the new orders index and the delivery index for a new month are published, the index for previous months is also revised. The material is normally revised for five months retroactively, except for the reference months June and July when the material is revised six and seven months respectively. This is mainly due to new and revised information having been received. The seasonally adjusted series and the trend series are always revised from January 2000 and onwards.
When seasonally adjusting the new orders and deliveries in industry, every series is individually seasonally adjusted using the Tramo Seats program. Tramo-Seats was introduced for deliveries and orders in the publication on data for January 2011. Tramo-Seats is recommended by Eurostat for seasonal adjustment of official statistics.
An important part of this method is to identify a model for the actual time series for each industry. These models are used for forecasts of the actual Industrial new orders and deliveries for two years forward and the forecasts are used in the seasonal adjustments. Because the forecasts are completely dependent on the choice of the model, this may cause revisions of seasonally adjusted data to a greater or lesser extent. To reduce the effects of revisions in seasonally adjusted data, Statistics Sweden has chosen to follow Eurostat's recommendation to use fixed models during the period of one year. According to Eurostat recommendations, seasonal adjustments should be reviewed at least once every year, including the selection of models. This recommendation has been applied by Statistics Sweden to for example the National Accounts and Industrial production index.
A new model review was conducted in connection to the publication in April 2012 concerning data for February 2012 due to the correction of the original series published in March 2012.
The next review of the model is planned to take place in connection with the publishing concerning January 2013.