Trade Statistics unit carries out a range of quality assurance checks on the administrative data obtained from CHIEF (and the Intrastat survey data), with specific reference to data elements critical for trade statistics. Checks are carried out at both transaction (line) level and at aggregate level for accuracy of trade data.
Individual declaration checks
Validation of all essential fields work as a back-up to CHIEF front-end validity. TSu also identifies any duplicated lines that may occur due to replication or multiple declarations on CHIEF.
Credibility parameters are produced (a form of ‘sense’ check) to assess the risk of the declaration. An indicator called the ‘magnitude of failure’ is calculated based on patterns from historical declarations.
Items that are high value, weight or excessively different from the ‘average’ expected for that product will receive more assurance checks. Also goods that are not normally traded or may involve embargoed items are also investigated.
Items that are identified as potentially incorrect are prioritised by the above system. These trade lines are then queried by TSu data management staff who are responsible for the line level assurance to clarify the accuracy of the declarations. Any invoices and shipping details are requested to further prove declaration accuracy.
The Intrastat Operations team perform a key role by ensuring timely declarations by Intrastat businesses, as well as critical function around education, audit and penalty. This supports the HMRC strategy of ‘Promote, Prevent, Respond’.
A number of key estimates are added to the EU trade data which ensures the accuracy of the trade statistics and minimises the need for revisions. This information can be viewed in the Estimate methodology.
Account Sign off
A number of final checks are actioned to the trade data to ensure that all key declarations at both detailed product and high level aggregates have been assured. For example, a process of checks that track time series values at aggregated trade levels. The trade data is ‘signed off’ following all these checks by the responsible statistician, with an audit log recording the process.