Findings

2.1 Our findings are summarised into five key areas:

• communication with users
• the revisions of trade in services estimates
• trade asymmetries
• quality and methods documents
• resource, business continuity and capability

2.2 We have set out four requirements related to these areas. These requirements supersede the previous requirements from our assessment in 2015. We expect ONS to meet these requirements within six months and to update us on its progress.

ONS’s communication with users

2.3 ONS engages with a wide range of trade statistics users and stakeholders through monthly meetings. Users told us that there is a strong network among users of trade statistics. ONS also engages with a wider audience through its monthly Economic Forum. Users felt that the level and frequency of engagement with the trade team were good. Users told us that there are opportunities to engage with other users of trade statistics and that the team is usually responsive to queries. A key user also felt that the team at ONS is generally transparent about upcoming changes to the statistics and notifies them early if the decision will impact them.

2.4 However, when users’ needs cannot be met, ONS should be transparent about and communicate why (for example, tight resources or a lack of suitable data) and any constraints. Users told us it was not always clear why their views or comments were not taken onboard.

2.5 A key issue that users raised with us is the lack of a comparable back series update following the structural break in data from 2021 as a result of HMRC data collection changes after the UK’s exit from the European Union (EU). ONS published articles analysing the impact of the data collection changes on the statistics. These articles were clear, useful and consolidated key information, and explained why ONS decided against making any adjustments to account for the discontinuity. Many users expressed disappointment that a back series wasn’t released as they would like to be able to make comparisons of trade with the EU before and after the UK’s exit. ONS understands this is a problem for users and discussed it with users at the monthly trade data strategy meeting and at wider meetings with users. ONS explained that the reason the back series could not be adjusted was due to a lack of available data to be able to make any adjustments. As a result, ONS concluded that any adjustment would not have been sufficient quality.

2.6 In the monthly bulletin, ONS provides additional insights around the statistics and has produced ad hoc analysis in response to external events. In our analysis, we found that the release covers information about drivers, similar to comparable economic statistics releases such as the monthly GDP statistics. Additionally, ONS has published a series of articles about specific trade topics, for example on Russia following the war in Ukraine, to help users understand how global events impact UK trading patterns. ONS also publishes an annual trade year in review article, doing a deep dive into particular commodity drivers across the year, taking into account commodities which users have expressed an interest in. Recently, ONS published an article analysing the specifics of the UK’s trade relations with the US and clearly explaining the reasons for the trade asymmetry, in response to media and policy interest in trade with the US in early 2025. We commend ONS’s proactive response to the interest in trade earlier this year. However, some users told us that they would welcome additional analysis in the bulletin. We encourage ONS to work with users to understand any gaps in the bulletin and, if possible, address them.

2.7 Users would benefit from a clear overview of the different datasets and how to navigate known limitations such as trade asymmetries. For example, ONS could publish a user guide that pulls together relevant information for users.

2.8 In the last year, three corrections have been made to the bulletins (covering data from December 2023, May 2024 and August 2024), and ONS uses the appropriate box at the top of the bulletin to notify users of any corrections. Each of the corrections was a very simple error: mixing up “exports” and “imports” within the commentary, slight errors in the numbers, and failing to upload the new dataset for one of the tables. Although small, mistakes such as these could be prevented in future releases by improved quality assurance and sign-off processes, as they appear to be mostly the result of human error. More recently, ONS discovered an issue with the HMRC trade in goods data and an error in the trade in services data in March 2025, which delayed the publication of January 2025 data. ONS has kept users up to date, investigated the errors and provided users with preliminary indicative estimates of the impact of the errors. In the March 2025 publication, they published the revised estimates to account for the errors in line with the National Accounts Revisions Policy.

2.9 A key user raised that they have repeatedly asked ONS to let them know if they find an error or make any changes to the dataset. The user feeds the trade datasets into a reproducible analytical pipeline to produce their own datasets and this pipeline does not work if a minor thing has been changed in the dataset. The trade team informed us that it has introduced a new process to let key users know about errors as soon as they are found. ONS should assess the effectiveness of this process by engaging with the relevant key users.

Revisions to the trade in services estimates

2.10 A key area of concern for users is the revisions to the monthly trade in services estimates. To meet user needs, ONS added trade in services to the monthly publication, whereas before they were published in the quarterly release. Revisions are an expected part of producing statistics, especially when the underlying data are survey-based, but users told us that they are especially notable in the trade in services data.

2.11 It is helpful that throughout the bulletin ONS refers to the trade in services figures as “early estimates” to indicate their volatility, but ONS should consider improving the information available on the uncertainty of the data.

2.12 It is good that ONS already publishes revisions triangles for the total trade balance, imports and exports. Users told us that they consider the trade in goods data to be very reliable. ONS should consider publishing revisions analysis for trade in services separately because users recognise that these are subject to large revisions. Doing so would help reassure users that ONS is aware of the size of the revisions. In our analysis, we found that the revisions made to the trade in services estimates are large, with the mean absolute revision (MAR) for imports being £695 million and £799 million for exports after one month (£536 million for the total trade balance). In terms of the relative mean absolute revision (RMAR), this was 3.4% for imports and 2.4% for exports after one month (4.3% for the total trade balance). Additionally, it appears that there is an upwards bias in the revisions. See Annex 2 for details on our revisions analysis.

2.13 Users told us that the revisions mean that they need to use large caveats with the monthly trade in services data as the revision is likely to be large and could ‘change the story’ the month after if the revision is in the opposite direction. As a result, the monthly estimates are difficult to use and have limited value to users, and some users have had to refer instead to the quarterly releases. Users also feel this issue has worsened and wonder about any potential upwards bias in the revisions and if this could be addressed. ONS should reassure users about these revisions. In addition to conducting analysis of the revisions, more information on how the estimate is modelled and how much of it is based on other existing data would help users understand the reliability of the estimate.

Trade asymmetries

2.14 ONS has done crucial development work around the asymmetries in trade data since our previous assessment. Trade asymmetries are the differences between the published trade data of a given country and the data of its partner countries. Each country collects its own data on imports and exports. In theory, each country’s exports to another country should match the other country’s imports from them, and vice versa. In reality, figures often do not match exactly. These asymmetries are especially pronounced in trade in services data. Previously, users had concerns that the UK had significantly higher asymmetries when compared to other countries.

2.15 ONS has made significant progress to understand and begin quantifying the factors and causes of the UK’s bilateral trade asymmetries. Between 2018 and 2020, it published articles on asymmetries in trade data that cover analysis on trade in services asymmetries and explain some of the potential causes of the asymmetries, showing comparisons and the relative strengths of different estimates. More recently, in 2022, ONS published an article focused on the challenges of measuring trade asymmetries. ONS has also done work internationally to improve its understanding of this topic and proposed the creation of an informal Expert Group on International Trade in Services Asymmetries in the Organisation for Economic Co-operation and Development (OECD) Working Party on International Trade in Goods and Services Statistics. ONS was influential in the group being set up and chaired it for multiple years.

2.16 In February 2021, the Economic Statistics Centre of Excellence published the paper Using a model to understand more about UK trade in services asymmetries. This paper uses a methodology developed by the OECD and the World Trade Organisation to produce a reconciled asymmetry dataset for UK trade in services. The resulting dataset confirmed that the most challenging measurement issue for UK trade statistics lies in recording trade in services imports.

2.17 Therefore, improving the quality of trade in services estimates is an important part of minimising trade asymmetries where possible. Since our previous assessment, ONS has expanded the International Trade in Services (ITIS) sample to include microbusinesses (with 0–9 employees), which is the key data source for trade in services. The survey sample has also been re-designed to enable a better understanding of the service trade activity of small UK businesses. This will help to improve the quality of estimates of trade in services, particularly estimates of imports. The methods changes for the 2020 ITIS were detailed in an ONS article in 2022. Once the impact of the sample improvement can be seen in the data, ONS should consider publishing an up-to-date impact analysis.

2.18 When we spoke to users, none expressed significant concerns about trade asymmetries. Users would like to see a document that brings together all the information and work that ONS has done on asymmetries and how it should be used to interpret the statistics. It would be good to see ONS publish such a document and continue working internationally to increase understanding and minimise the asymmetries.

Updating quality and methods information

2.19 Following our previous assessment, ONS published multiple trade development plans that set out its approach to improving the trade statistics, including an assessment of the quality of the trade statistics, a strategy for user engagement, and documentation of system capability and risk mitigation. The most recent development plan was published in 2022.

2.20 Since 2015, ONS has redeveloped its website and significantly improved its monthly publication, which now provides a useful overview of headlines for both trade in goods and trade in services. The accessibility was also greatly improved, and ONS should continue its work to maximise accessibility for users. The monthly bulletin provides relevant links to related data, bespoke articles and other important information. The publication has engaging interactive graphs that allow users to explore the data.

2.21 As well as improving its methods and quality assurance process, ONS also updated the information about methods, clearly covering the specifics set out in the requirements from our previous assessment, such as presenting the overall framework for the trade statistics, information on strengths and weaknesses, and details about the different data sources. It is good to see that ONS has become more transparent.

2.22 It was positive to hear that users generally understood the methodology used to produce the monthly trade data and felt it is well established. Users also appreciated ONS offering training to their staff to explain their methods. The publication of articles that detail changes to methods, such as the series covering the impact of trade in goods data collection changes on UK trade statistics following the UK’s exit from the EU, is a great example of ONS providing additional information when helpful for users. Some users mentioned that they would like more details on how the trade in goods data from HM Revenue and Customs (HMRC) are transformed to be on a balance of payment basis for the ONS trade publication and the associated quality assurance process.

2.23 However, the two documents covering methods, the Quality and Methodology Information (QMI) report and methodology document, should be updated to reflect the current state of affairs. The QMI was last updated in May 2021 and the methodology document in March 2017. The QMI and methodology document should provide up-to-date information on users and user engagement, updates to ITIS, and more information about how HMRC data are transformed. It is especially crucial that the QMI reflects the changes to the methods, processes and data collection of trade in goods data as a result of the UK’s exit from the EU. Additionally, ONS should ensure that the difference between the two documents is clear and that the two documents reference each other appropriately. Alternatively, ONS should consider whether two separate documents are needed and if it might be clearer for users to only have one document that covers everything related to quality and methods.

Resource, business continuity, and capability in the trade team

2.24 During the review, ONS has told us that the trade team has experienced issues with resourcing and knowledge retention. Since the development work after the UK’s exit from the EU, the trade statistics development project has been de-prioritised and was shut down two years ago. As a result, the trade team has received less support.

2.25 The trade team has also experienced high staff turnover, which has meant a loss of significant subject matter expertise. Subsequently, the slow nature of the recruitment process has meant that the vacancies have remained unfilled for extended periods of time, leaving a smaller team to handle the business as usual of the monthly publication. It is important that the trade team is appropriately resourced and has the necessary expertise to meet the requirements of the Code of Practice.

2.26 Additionally, the trade team told us that the legacy system used to process trade in services data is inefficient, difficult to work with and no longer fit for purpose, this puts strain on the limited resource and restricts capacity for continuous improvement. The team has already taken steps to progress work on moving away from the legacy system and is developing requirements for a new system. As part of a deep dive into their systems, quality assurance checks of the trade in services data highlighted an error in the annual processing system. This error, alongside a separate error in the trade in goods data received from HMRC, meant the publication of the January 2025 data was It was published two weeks later on the same day as the Quarterly National Accounts and the Balance of Payments, where all corrected data could be brought through in line with the National Accounts revisions policy. Together with ensuring appropriate resource, a better processing system is needed to prevent future errors and improve the efficiency of the team.