Dear Scott 

Statistics on International Trade in UK Nations, Regions and Cities

We have completed our compliance check of ONS’s experimental statistics on International trade in UK nations, regions and cities, which present estimated values of exports and imports of goods and services for various sub-UK geographies. It is good that users have been very supportive of ONS’s decision to publish subnational trade estimates. There has been considerable development and extension to these statistics since they were first published.

The statistics team has further development plans, for example converting all results onto a harmonised industry basis between trade in goods and trade in services. We support these continued improvements and that development plans had been introduced following extensive engagement both within ONS and externally with users and stakeholders. Whilst key users are aware of these plans through engagement meetings, we recommend that you look to set out some of the principal elements of future developments transparently in the statistical bulletin as a way of informing a wider range of users and enabling their engagement.

The statistics team told us that, while these estimates contribute insights around the importance of trade to local economies, users have expressed a preference for statistics at even lower geographies such as local authority and Local Enterprise Partnership levels. Estimates at International Territorial Level 3 (ITL3) and city region levels are the lowest geographies currently published.

Uncertainty around the estimates increases as the trade data are apportioned at lower and lower geographical levels. ONS warns in its Quality and Methods Information for these statistics that data at the local authority level as it currently stands would not be sufficiently robust in terms of quality and statistical disclosure. We concur that producing estimates at levels of geography below ITL3 could only be introduced at acceptable quality levels and sufficient disclosure by adopting a more robust apportionment method and examining the impacts of such a method on suppression of low-level data. 

To avoid users placing too much reliance on low-level estimates you have set out comprehensive quality and methods information. To enhance the steps you have already taken, we recommend:

• Improving descriptions of uncertainty: Many users access the statistics directly from the data tables, so it is also important that key messages about potential uncertainty are available there – we recommend that you introduce a solution such as colour coding cells to visually convey the uncertainty.
• Considering whether the current statistical disclosure control policy strikes the correct balance between access and confidentiality: These statistics are broad proxies of actual trade transactions. The disclosure control policy is thus designed to avoid the potential of the proxy data clearly signalling real, potentially confidential, trade data information about a business entering the public domain. Since users are more interested in data about local geographies rather than trade that can’t be allocated to a particular region, we support your preference to suppress data for unknown regions. We also commend your response to users’ preferences by focussing data suppression at the lowest level of geography to preserve disclosure at the higher geographies. We encourage you to continue work to minimise data suppression and to update periodically your examination of the risk of proxy data signalling real trade data.

As part of our discussions with the statistics team we discussed the benefits to users of using alternative data for apportionment and weighting to the current method of using employment data on the Inter-Departmental Business Register. We encourage you to further examine the feasibility of producing estimates of export-supported jobs at local levels through linking your trade and jobs data. There would appear to be user and stakeholder appetite for using estimates of local jobs with relatively high trade-dependency. 

We would like to thank you and the statistics team for engaging with us during the short review. I am copying this letter to Helen Tuck and Chris Goldsworthy, Subnational Trade team, ONS. 

Yours sincerely

Mark Pont
Assessment Programme Lead