Reasons for revisions to original data:
The main reason why the data become revised is the estimation used for activities 26-27. Some listed companies do not provide their data within the production timetable of the statistics so these data have to be estimated. This has to be done eight times per year. For instance, the data for the April to May period were estimated. At turning points of economic trends, the estimation models react slowly to changes.
Estimation had been increased as of 2010. The Tax Administration's periodic tax return data are used to compile statistics on the output of enterprises with fewer than 50 employees. Enterprises with fewer than 50 employees account for around 13 per cent of industrial output and these data cover nearly 70 per cent of the production of enterprises with fewer than 50 employees. These data are not at Statistics Finland's disposal by the release date of the volume index of industrial output. For this reason the data on these enterprises are estimated for the month the release concerns. The periodic tax return data will be available when data for the month concerned are released for the second time.
The volume index of industrial output is based on a sample. The sample of the direct data collection comprises almost 1,000 establishments or enterprises. Every month, data are missing for an average of just under 20 establishments. This may cause revisions to the data but its impact has been quite minor. The Tax Administration's periodic tax return data are used to calculate data for enterprises with fewer than 50 employees in 19 activities at the 2-digit level. The data cover approximately 15,000 enterprises. The data are not available for the month for which data are being released. The data for the month concerned must be estimated. The Tax Administration's periodic tax return data may become revised several months backwards.
Correcting of data
Data suppliers may correct their data retrospectively. However, as a rule this does not have a major impact at the level of total industries.
Checking of weights
In accordance with the chain index principle, establishment and industry-specific weights are changed annually. This has also had some impact at the level of total industries, although not of the same magnitude as estimation.
Heading-specific weights were revised in the release for June. The change in weights has had some impact on the point figures and change percentages for the early part of the year.
Calculation of annual index
The sample-based monthly volume index has been retrospectively benchmarked to the annual volume index calculated from structural statistics. This has influenced the indices especially at the 3 and 2-digit level activity categories.
Changing of seasonally adjusted figures
Seasonally adjusted index point figures may change even if original index point figures do not. The changes in the seasonally adjusted and trend point figures may also stretch several years back. Especially at turning points of economic trends seasonally adjusted series fail to react sufficiently quickly to sharp changes.
Figures become revised faster than before
Below is an example of how the original annual change percentages for April and May 2011 became revised in a few main activities.
|Revisions to April data|
|Industry||1st release||2nd release||3rd release|
|Total industries BCDE||4.4||3.7||5.7|
|Manufacturing excl. 26-27||10.4||6.0||6.9|
|Electrical and electronics industry 26-27||0.3||0.2||9.6|
|Revisions to May data|
|Industry||1st release||2nd release|
|Total industries BCDE||6.4||9.7|
|Manufacturing excl. 26-27||5.3||11.2|
|Electrical and electronics industry 26-27||2.8||5.3|
Last updated 10.8.2011