Improving the Quality of Regional Economic Indicators in the UK: A Framework for the Production of Supply and Use and Input Output Tables for the Four Nations

With increased devolution of powers, Brexit and Covid-19 changing the economic structure and linkages between different regions, timely regional economic statistics are needed to support regional and national policymaking in the UK. Regional Supply Use Tables (SUTs) provide devolved administrations with a disaggregated insight into the structure of a given region. Regional Input-Output Tables (IOTs) derived from the SUT facilitate estimation of regional impact assessments and economic models which can be used to analyse the effects of regional and national policies on different regions.

Of the four UK nations, only Scotland and Northern Ireland currently produce their own SUTs and IOTs on a regular basis. In this report, we develop a strategic framework for the production of SUTs across the four UK nations. For those unfamiliar with SUTs and IOTs and the difference between them, we begin by introducing these tables and discuss how they are used by economists and statisticians in policy and academia.

We then describe methods for producing regional SUTs and IOTs. We first discuss the bottom-up approach which involves detailed data collection at the regional level. While giving a higher level of accuracy, this approach is also more resource intensive and faces several practical, statistical and conceptual challenges. Many of these challenges arise from the fact that it is more difficult to measure regional activity than national activity. Top-down approaches involve regionalising the national UK SUT or IOT using an indicator variable.

We then focus on the UK data landscape, discussing the UK’s sampling frame, the interdepartmental business register. We also consider the UK’s Regional Accounts which record estimates of regional Gross Value Added (GVA), Gross Fixed Capital Formation and Gross Disposable Household Income produced using top-down methods. We then go on to discuss how SUTs and IOTs are compiled in the UK, Scotland and Northern Ireland. Importantly, Scotland and Northern Ireland both adopt a hybrid approach, using a combination of regionalised UK data and nation-specific data sources. They also constrain their totals to the Regional Accounts, although, for some sectors, Scotland has moved away from full consistency with the Regional Accounts.

We demonstrate how regional IOTs for the four nations can be produced by regionalising the UK IOT using location quotients (LQs). While producing regional IOTs from the published UK IOT is possible, there are some significant differences between our IOT and those produced using the bottom-up approach. The first key issue is the assumption that the GVA to output intensity for each region is the same as the national average leading to an unrealistic amount of imports and exports to the rest of the UK. The second issue is that using the LQ method produces much smaller intermediate sales and purchases than the bottom-up approach.

In our recommendations for compiling regional SUTs for the UK nations we consider two scenarios. The first scenario sets out how four SUTs for Scotland, Northern Ireland, Wales and England could be constructed using a predominately bottom-up approach. While this scenario is ambitious, it is also pragmatic and sets out how a bottom-up approach could be developed using the existing sampling frame, the interdepartmental business register, and existing business surveys administered by the Office for National Statistics (ONS) and devolved administrations. A bottom-up approach would lead to the four nations adopting similar data collection strategies facilitating comparability and compatibility. This would allow users to understand: (i) the production structure of a given UK nation, (ii) differences in production structure across UK nations and (iii) the production structure of the UK as a whole.

The second scenario is more modest and sets out how four SUTs could be constructed using a hybrid approach. This would involve using the Scottish and Northern Irish approaches as a starting point to develop a framework to produce SUTs for the four nations. Ultimately, this approach would allow users to understand the production structure of a given nation but accuracy declines perhaps rendering comparisons across the SUTs of different nations more problematic. Unlike a bottom-up approach, a hybrid approach may not facilitate similar data collection strategies across nations with an imposition of consistency potentially preventing regions from incorporating useful nation-specific data sources. Regional SUTs are also typically constrained to the UK Regional Accounts produced using top-down methods.

To address these two scenarios we have a number of recommendations.

  • First, when collecting data on Scottish, Welsh and English activity the feasibility of asking Great Britain Reporting Units (RUs) to report on the activity of their Scottish, Welsh and English Local Units (LUs) should be investigated further given that this approach has proven successful in Scotland and Wales, for example, when collecting interregional trade data. Taking this one step further, it may be possible to “create” regional RUs whose industrial classification reflects the dominant activity across regional LUs.
  • Second, surveys issued by the ONS such as the Annual Business Survey (ABS) and Annual Purchases Survey should have sample sizes which facilitate the estimation of statistics for the four UK nations as well as the UK as a whole.
  • Third, building on the Whole of Scotland Economic Accounts Project, a fifth SUT could be used to capture foreign production as well as offshore oil and gas extraction preventing the distortion of regional activity.
  • Fourth, we recommend that the Canadian approach to allocating central government and head office output be investigated in relation to the UK again to prevent distortions of regional activity.
  • Fifth, recognising that for some industries a top-down approach to regionalisation will be required, we recommend strengthening existing data sources by: exploring the possibility of developing regional GVA to output intensities using ABS microdata; mapping household consumption to industries; and collecting data on internal trade and regional exports, building on existing data collection by the Scottish Government and Northern Ireland Statistics and Research Agency (NISRA).
  • Sixth, given the issues associated with LQ based top down regionalisation, in particular, the underestimation of interregional exports and imports, we recommend a review of top-down regionalisation methods with respect to the UK IOT. In our unique policy context, it would be beneficial to assess how different top-down methods perform when other regional data (for example, data on interregional trade available from some the devolved administrations) is used to inform the regionalisation process.
  • Seventh, we would recommend that the four nations publish SUTs annually following a common timeline. The UK SUT, however, could be published earlier each year since the regional SUTs may need to utilise proportions derived from the UK SUT. We also recommend that the four nations agree on the minimum number of industries and products to include in their respective published SUTs. The 64 industries and products used by NISRA may act as auseful starting point. Importantly, each nation could still choose to compile a more detailed regional SUT for their own use.
  • Eighth, bottom-up data should, where possible, gradually replace the Regional Accounts produced using top-down methods. Where this is not possible, a reconciliation process should take place between the regional SUTs, UK Regional Accounts and UK SUTs with the devolved administrations identifying where Regional Accounts estimates are inappropriate.
  • Last, we recommend that all four nations also produce industry by industry IOTs annually since these tables are a crucial input for regional economic modelling. In the short-run, this process could be automated through regionalisation of the UK SUT using LQs and strengthened using additional data sources as detailed above.

This ESCoE paper was first published in May 2022


Sharada Nia Davidson

Sharada supports the Institute’s projects which develop new economic indicators and statistics or require time series data analysis. She has undertaken work for the Economic Statistics Centre of Excellence.  Previously, Sharada worked as a Consultant and PhD Trainee for the European Central Bank, modelling the impact of macroprudential policies.

James is a Fellow at the Fraser of Allander Institute. He specialises in economic policy, modelling, trade and climate change. His work includes the production of economic statistics to improve our understanding of the economy, economic modelling and analysis to enhance the use of these statistics for policymaking, data visualisation to communicate results impactfully, and economic policy to understand how data can be used to drive decisions in Government.

Kevin is a Chancellor's Fellow in the Department of Economic with a focus on the use of regional economic models for policy analysis. Areas of interest include; energy and climate change, poverty and tourism.

Mairi is the Director of the Fraser of Allander Institute. Previously, she was the Deputy Chief Executive of the Scottish Fiscal Commission and the Head of National Accounts at the Scottish Government and has over a decade of experience working in different areas of statistics and analysis.