Post by Professor Calvin Jones, Academic Advisor, Knowledge and Analytical Services, Welsh Government

Darllenwch y dudalen hon yn Cymraeg

In 2022 Welsh Government began the process of compiling and publishing a set of Supply and Use (SUTs) and Input-Output Tables (IOTs) for Wales. These were to be the first holistic picture of the Welsh economy produced following Scotland (which has published SUTs and IOTs for decades) and Northern Ireland (which has done so for the post EU exit period).

The publication was intended to improve our understanding – inside and outside Welsh Government – of how the Welsh economy ‘worked’ in terms of production, trade and product use. More importantly, it was to determine whether collating such a picture – consistently, across all products, industries and users – was feasible, whilst guaranteeing a reasonable level of quality and robustness. The project delivered an Input-Output publication for Wales (for base-year 2019), detailing the structures and activities of 55 industries in Wales, along with related product use and supply. The Supply and Use and Input-Output Tables (hereafter called IOTs for simplicity) have been published as official statistics in development, together with an extensive methodology document.

The uses and demands of Input-Output analysis

Input-Output tables (IOTs) form part of Systems of National Accounts (SNA): the framework whereby economic activity that’s undertaken within a country, or by its resident entities, is collated and reconciled so that comparisons can be made across sectors, final demands, inputs and outputs, and importantly across time. SNA codification and guidance developed by intergovernmental agencies means that we can also reliably compare key metrics – like Gross Domestic Product (GDP) – across countries. Separately, IOTs form the basis of various kinds of economic analysis – employment and output multipliers, modelling, supply and value chain studies, productivity analysis etc. – which then inform other non-economic tools like eco-, materials and carbon footprinting, and social accounting.

IOTs are then very useful. Their outputs are extremely difficult to duplicate in other ways. Their compilation however requires the reconciliation of a huge variety of data from different sources – business, trade and consumption surveys, company administrative and customs data, and public sector accounts to name just a few. At subnational scale, including in the UK, this is a stretching task.

Regional Economic Statistics in the UK

Historically – and critically, pre-devolution – economic statistics were largely the purview of the Office for National Statistics, supplemented by local or bespoke surveys (e.g. for specific sectors). The statistical landscape was largely ‘even’ across England and Wales especially, with Northern Ireland and Scotland long having more differentiation (for example, their own Censuses of Population). The undifferentiated nature of economic statistics (arguably) mattered less when economic policy was effectively developed and implemented for the UK as a whole. The dawn of the 21^st Century saw changes to that model. Economic development was devolved to the new Scottish Parliament and Welsh Assembly, and the new English Regional Development Agencies (RDAs).  This required data on regional economic structures and relationships that did not exist in the form required. Whilst the RDAs are no more, Scottish and Welsh Governments have increasingly gained the power to enact policy that is different to that of Westminster  and the new City-Regions, and combined mayoral authorities have their own distinct data needs.

As devolution has increased over the last quarter of a century, the statistical landscape to inform that devolved decision making has not always developed in the same way.

The Wales Input-Output Tables: Data Gaps

There are several areas where currently available data sources do not adequately support the compilation of reliable regional Supply and Use and Input Output Tables for Wales.

Key metrics – IOTs require the ‘balancing’ of supply and demand for industries and products at the relevant spatial scale. The structure of UK business surveys like the Annual Business Survey and Annual Purchases Survey means that aggregation of IOT-important variables to a reliable Wales-total is difficult. This also affects UK administrative company data on, for example, trade. Essentially, we are missing some key details on how the business population in Wales is structured.

Information on products – Intelligence on industries (and on employment) is generally more developed for the UK than that on the supply and use of products. The main product-sales survey for the UK, PRODCOM, does not currently report at sub-UK level, and data on regional service provision, and on regional trade (both with other parts of the UK and internationally) are either unavailable, insufficiently detailed or apportioned using indirect measures. This means we do not have much information on what products and services are actually made in Wales, using what inputs, from where, and for which final users. 

Apportionment – The above issues often arise from the fact that UK data structures – sampling or administrative reporting – were not designed to report robustly at regional scale. Where these data have been needed for subnational use, the sensible and cost-effective approach has been to use more robust measures like the share of employees  to apportion UK totals to regions ‘top down’. This approach then informs derived metrics like regional Gross Value Added, household income and investment. Errors are likely to emerge if the ratio between the indirect measure and the ‘missing’ variable of interest is different to that evidenced for the UK. Wales has relatively low levels of industrial and occupational variety (and of capital ownership), compared not just to the UK, but to most other UK regions so this is a real concern.

The Value of the Input-Output Project

For the reasons above (and indeed others) the Input Output Tables for Wales are classified as Statistics in Development: they are effectively exploratory and experimental. They are by no means the economic ‘truth’ for Wales, and they do not supplant existing ONS, HMRC or Welsh Government data that describes particular parts of the economy. Effectively the need to ‘balance’ supply and demand – and adjust data sources to do so – means that users should not assume the accuracy (or wider consistency) of any individual result. Nonetheless, the SUTs and IOTs represent the best and most recent coherent and holistic picture of the Welsh economic system.

The Tables can be used to explore the potential use of economic accounting structures to inform better policy, both directly and as they are used for socio-economic, and potentially energy and environmental modelling purposes. The Tables are a resource for third parties in Wales to do likewise, albeit always remembering the data weaknesses, the inherent constraints of Input-Output analysis, or the appropriateness of any uses to which they are put. Play, but play safely.

Perhaps more importantly, the IOT project has provided a useful reality check and gap analysis: bringing together in one place a lot of what was already known about the reliability and scope of regional economic data, and the ways in which it might no longer match what is needed for an era of increasing sub-UK economic autonomy and differentiation. Consequently, the next step is not just a mechanistic update of the IOTs to more recent data. Rather, the IOT update will occur alongside a thoughtful consideration of what data are needed to effectively manage the Welsh economy and inform wider national objectives. This will involve considering what we can do in Welsh Government, as well as how we can work with our partners and colleagues across the Civil Service to fill those gaps.