New method for estimation implemented in internationel trade in good statistics from statistical year 2023
In international trade in goods statistics improvements of estimation methods has been made and they were implemented from January 2023 statistics. New estimates are calculated retroactively from January 2022 in order to calculate the monthly and yearly percentage changes correctly. The new methods are both for estimation of trade below threshold and for non-response. The impact on total trade is minor but estimates are now included in lower levels of all classifications.
1. Estimation of trade below threshold
In trade between EU countries thresholds are in use to define which enterprises are obligated to declare for Intrastat. Each EU Member State defines the threshold value to be applied by it at the compilation of international trade statistics independently and complying with the coverage limits set for the compilation of data by the statistics regulations on internal trade. In Finland, in year 2023 the threshold is 800 000 euro for both exports and imports. If the trade value of an enterprise is less than the threshold value under a period of a year, it is not obligated to declare for the trade. The trade below threshold is estimated and added to statistics.
The new method of estimation is based on VAT data of Tax Administration. In addition, also VIES data and estimates are used if needed (VIES = Value Added Tax Information Exchange System).
The major change with the implementation of the new methods is that estimates are included in all CN-classes, from CN 8-level to CN 2-level with few exceptions. The estimates are divided among CN-classes according to data from Intrastat in previous months. Under the earlier method, estimates where shown in code CN 9969 9999 “unspecified”.
Estimates for trade under threshold are included also in all other classifications at the lower levels (SITC, CPA, BEC, NACE) as well as in the calculation of volume index for international trade in goods with CPA-classification.
|27||MINERAL FUELS, MINERAL OILS AND PRODUCTS OF THEIR DISTILLATION|
|88||AIRCRAFT, SPACECRAFT, AND PARTS THEREOF|
|89||SHIPS, BOATS AND FLOATING STRUCTURES|
|98||COMPLETE INDUSTRIAL PLANTS|
|3002 4110||(2022--.) Vaccines against SARS-related coronaviruses *SARS-CoV species*, for human medicine|
|7108 2000||Monetary gold|
Estimates for trade below threshold are also included in country specified data, i.e. country of origin, country of consignment and country of destination. Estimations are added only for EU-countries. The country specific estimates are calculated separately every month.
2. Estimation of non-response
Non-response in international trade in goods statistics means late incoming data or missing data. The share of non-response has been minor in Finland, estimated to round 1.5 per cent yearly.
Estimation of non-response is also based on VAT data of Tax Administration. If VAT data is not available, the non-response trade is estimated with a multivariate time series forecasting model. In multivariable time series models not only an enterprises data but the whole industry´s data is used in the estimation model. Estimations are calculated based on previous data. With the new method for non-response, estimates are included in all CN-classes in the same way as for estimates for trade under threshold. In table 1 is shown the CN-classes where estimates are not included. Also estimates of non-response are divided to lower levels of other classifications and are included in calculation of volume index with CPA-classification.
Estimates for non-response trade are also included in country specified data, i.e. country of origin, country of consignment and country of destination. Country of consignment can only be an EU member state but country of origin can also be a non-EU country. For Russia and North Korea no estimates are calculated. In exports, estimates are included in data for country of destination, which can only be an EU member state.
Also the previous model for estimation was based on VAT data of Tax Administration, but only an univariate time series forecasting model was used. The calculation was made with time series from individual enterprises. Non-response estimates where shown in code CN 9979 9999, “unspecified”. Estimates were included in country specific data.
Sofia Lahdensuo, tel. +358 40 301 4039
Christina Telasuo, tel. +358 40 332 1828