Grŵp Defnyddwyr Ystadegau Economaidd Cymru, 13 Mawrth 2025: cofnodion

• Vanessa Allis (VA), Welsh Government
• Jack Price (JP), Wales Centre for Public Policy
• Ian Halliday (IH), Scottish Government
• Peter Horvat (PH), Organisation for Economic Co-operation and Development (OECD)
• Rhys ap Gwilym (RaG), Bangor University
• Jonathan Harvey (JH), Northern Ireland Statistics and Research Agency (NISRA)
• Chris Ganley (CGa), NISRA
• Lloyd Fisher (LF), Vale of Glamorgan Council  
• Peter Davies (PD), Senedd
• Tom Nicholls (TN), Welsh Government (Chair)
• David Phillips (DP), Institute for Fiscal Studies
• Kristine Cooke (KC), NISRA
• Lee Clarke (LC), Cardiff County Council
• Kate Mulready (KM), Welsh Government
• James Carey (JC), Welsh Government
• Benjamin Harries (BH), Welsh Government
• Abigail Cole  (AC), Carmarthenshire County Council
• Megan Williams (MW), Welsh Government
• Steve King (SK), City and County of Swansea
• Stevan Croasdale (SC), Scottish Government
• Cian Siôn (CS), Welsh Government
• Sam Wilson (SW), Transport for Wales
• Heledd Rhys (HR), ONS Local
• Melanie Jones (MJ), Cardiff Business School, Cardiff University
• Edward Hunt (EH), NHS Wales
• Amann Juergen (AJ), OECD
• Luned Jones (LJ), Welsh Government
• Benjamin Harries (BH), Welsh Government

Speakers

• Jonathan Bonville-Ginn (JBG), Welsh Government
• Calvin Jones (CJ), Cardiff University
• Stevan Croasdale (SC), Scottish Government
• Martin Irvine (MI), NISRA
• James Black (JBl), Fraser of Allander Institute, University of Strathclyde
• Cerys Ponting (CP), Welsh Government
• Chris Goldsworthy (CG), Office for National Statistics (ONS)
• Jane Naylor (JN), Chief Statistician and Deputy Director Statistics and Data Division, Department for Business and Trade

Minute taker

• Mair Smith (MS), Welsh Government 

Apologies

• Alan Felstead, Cardiff University
• Chloe Gibbs, ONS
• Debbie Hudd, Welsh Government
• Martin Jennings, Senedd
• Neil Paull, Welsh Government
• Sian Price, Development Bank Wales
• Sam Sullivan, Data Cymru
• Becky Tinsley, ONS

TN opened the meeting, welcomed attendees and outlined the agenda. 

There were no actions from the meeting in October to discuss.

Simultaneous translation was not provided as the minimum quota of ten percent was not met.

JBG and CJ gave a presentation on producing Input Output Tables (IOTs) for Wales explaining how they came about, challenges with developing the tables and plans for further development.

Main messages

• JBG and CJ undertook a three-year project to produce Welsh IOTs for the reference year 2019 – chosen because it was the latest year with full data availability at the time of commencing (that wasn’t a pandemic year). 55 industries
• The IOTs are classified as Official Statistics in Development which reflects some of the issues in data source availability and because it’s the first time undertaking these methods.
• Issues when developing the tables:
  • The national economic data landscape limits our understanding of the Welsh economy.
  • Apportionment used when surveys are unable to estimate directly – however, this can be problematic where there are occupational/industrial differences between Wales and UK
• SUTs and IOTs were published 11 March: this includes a methodology document and blog from CJ, tables and an interactive tool is coming soon
• Future: WG are seeking budgetary approval to continue this work and continue to support users to make the most of the tables. The aim is to produce a new set of tables (for reference year 2022) and continue to review and improve methods based on feedback received through peer review process.

Discussion

• DP expressed an interest in the online tool as well as noting concern for the potential misuse of IOTs (to make bold claims about different sectors). DP asked what caveats, guidance and disclaimers are going to be included so misuse isn’t proliferated by the tool?
• CJ responded that extensive caveats are included within the tool acknowledging that it only covers the demand side (a link is provided to Miller and Blair textbook on advantages and disadvantages). CJ noted that the data tables will not be included within the tool and that users will be able to use their own tables.
• JBG expanded that when people open the tool, warnings will appear, and users will need to check a box to say they’ve read and understood them before accessing the tool.
• TN concluded that the project has created a platform to allow a journey towards the potential for something a more advanced. He acknowledged that misuse is a key concern and something which will be considered as part of ongoing continual review as the project evolves. 

Welcome any questions or feedback to: InputOutputTables@gov.wales or TablauMewnbwnAllbwn@llyw.cymru. 

SC gave a presentation on the work of the IOTs team in Scottish Government - their production timeline, published outputs, regular analysis, recent ad hoc analysis and IO vs computable general equilibrium models.

Main messages

• The team consist of three statisticians and produce annual updates and tables (the production timeline takes a year).
• Published 1998 to 2021 SUTs and IOTs in December.
• It’s not possible to publish the full matrix due to issues with disclosure.
• Regular annual analysis
  • High Level Carbon Assessment of the Scottish Budget (SG)
  • Tourism impacts (SG and VisitScotland)
  • Scottish Government procurement spending impacts (SG)
  • Housebuilding impacts (SG)
• Notable ad hoc analysis
  • Importance of Natural Capital to the Scottish Economy (SG – RESAS)
  • International bilateral trade modelling dataset (SG – OCEA – Policy Modelling)
  • Rail and Transport impacts (Transport Scotland)
  • Scotwind impacts (SG – Offshore Wind)
  • Commonwealth Games Legacy, 2014 and 2026 (SG – Communities)
• Static IO model is better suited where there is spare supply chain capacity or to estimate the wider impacts of existing activity.
• Supply and price change effects are better handled by Computational General Equilibrium (CGE) models.
• The Industry-by-Industry table forms the basis of the Social Accounting Matrix (SAM) at the heart of CGE
• Supply, Use and Input-Output Tables (Scottish Government)

Discussion

CJ asked if the CGE model is the same or separate to the Strathclyde model.

SC confirmed it’s the same – we help build the same SAM.

JB confirmed it’s the same SAM.

MI gave a presentation on the background of SUT and IOTs in Northern Ireland, along with example outputs of data.

Main messages

• Previously a team of three but have benefitted from a couple of new members recently. A lot of the data is produced in house within the branch. This makes it easier to drill down into detail than for other Devolved Authorities.
• Work started in 2014 with a feasibility study on the potential for produce NI specific accounts. The first set of high-level experimental SUTs (for 2012) were published in 2015. The PowerPoint documents all subsequent publications and reference years.
• For 2025 the plan is to publish SUTs and IOTs for 2021 reference year in spring and 2022 reference year in the Autumn
• Start with SUTs (product balances, which industries produce them, imports and consumption). This supports development of SAM and IOTs to cover income distribution, household spending and government transactions. And then this feeds CGE models.
• Uses of SUTs, IOTs and multipliers:
  • CGE models of whole economy
  • hypothetical extraction models (extraction of a sector of the economy to aid better understanding e.g. pandemic impact, closure of a business/sector)
  • partial equilibrium models for studying policy impacts and price changes in a particular industry
  • circular economy models e.g. to demonstrate which industries are contributing most to waste.
  • satellite accounts (extending national accounts around specific themes) e.g. tourism study commissioned to show contribution to the economy
  • impact assessments e.g. evaluating policies or potential developments and the impact on GDP or jobs.
• MI presented additional examples of specific uses of the data with external links provided in PowerPoint.
• The landscape of econometric modelling in NI is potentially changing as a result of NI fiscal developments and devolution following publication of a 2022 report More fiscal devolution for Northern Ireland (Fiscal Commission Northern Ireland).
• There is a need for better data to support this process e.g. quarterly national accounts (QNAs) could be more timely and would help monitoring of fiscal policies. Having annual SUT framework is a prerequisite for developing QNAs.
• MI provided a list of additional reading and links to other information and online tables (see PowerPoint).

Discussion

CJ expressed concern that CGE models may become used as default to answer policy questions with them being seen as the panacea to addressing the supply side blindness of IOTs when in reality, lots of accounts don’t balance. CJ suggested that with a set of robust UK IOTs then it would be good to collaborate to better understand what policy questions can be answered by the different models.

JBG added he’d appreciated getting to see the processes that Scot and NI go through to produce tables as well as the range of further development beyond WG approach to date. Most exciting is seeing how it’s been used for different policy interventions. E.g. offshore wind. Seeing range of applications can provide some inspiration to people on the call – how we can use the tool to answer some of the questions and insight into problems that we haven’t been able to do before. 

CGa reflected on “the journey” of the last 10/11 years and extended his thanks to the Scottish Government team for their support. Efforts have been made to maintain consistency so that what is produced is comparable. It’s been a journey educating users, policy side, economists, statisticians on what’s available and how and when to use certain tools and models. There is a need to demonstrate the value of what’s produced and build up reliance on the outputs, examples of its use in order to secure funding for developments. 

CGa reflected that there is a need for a team of at least three people, along with other data providers to input and support in order to deliver such a data intensive process involving sector and subsector analysis. CGa encouraged bringing policy along the journey of developments.

TN acknowledged all models have their shortcomings and that there are times when one model is more appropriate than another. Overall having a wide suite of models is valuable. 

JB gave a verbal presentation reflecting on IOT work undertaken in Wales and the presentations so far. He extended thanks for the data, as a future user, and noted various ways in which the Welsh IOTs will help better understand the Welsh economy. 

Main messages

• No modelling is perfect, that’s the nature of modelling. Brilliant thing about IO is methodology is it’s transparent in its process and methodologically simple. It’s good at offering an illustrative sense of scale and impact of industries.
• This also represents a drawback – anyone can use it, users aren’t always trained, contrast that with more complex comparisons with CGE models and econometrics which can be much less transparent and have their own sets of assumptions.
• One of the benefits of IO modelling and explaining step by step is that it’s possible to explore questions like what if we’re constrained by the supply side? How would that change the results? What if certain industries are involved or aren’t involved? How would that change the results?
  What if the firms in the industries are different to the industry average? How would that change the results?
  This can help to develop intuition towards how a policy change might impact the economy.
• JB reflected that the WG methodology doc is excellent and detailed.  Being able to see how it’s broken down by step-by-step calculations, allows users to break the IOT down into further subsectors and understand key data assumptions.
• IOTs and SUTs are important for understanding the data landscape and identifying issues, how data fits together, key missing data e.g. trade data which is an issue in regional SUTs.
• Interesting to see some of the classic challenges e.g. with key industries such as finance and agriculture, public administration etc. but also some of the Welsh specific challenges – sample size and lack of information in certain surveys that differs from Northern Ireland and Scotland.
• JB offered some suggestions of potential uses of IOT data.
  • Potential impacts of new sources of Final Demand and properties of the economy. For example, highest multipliers were in services industries e.g. software, professional services - more than manufacturing
  • Impact of existing activities. E.g. large closure of a major plant where a company is closing down. It’s possible to look at potential supply chain issues and explore risks and national impacts. It provides a framework for looking at these things and considering where might these be correct or not.
  • Combining with data about other industries. E.g. renewable energy – lots of interest in how this relates to jobs. Could experiment how IOTs can be combined with the low carbon and renewable energy survey – what are the bigger impacts of an industry.
  • Extending multipliers – can start to look at things like jobs by sex, age group – how well do they hold in the modelling. Or explore wider impacts such as carbon foot printing or impacts relating to the Well Being of Future Generations Act.
• Conclusion: IOTs can have many applications with capabilities extending through to econometric techniques e.g. IO represents a huge part of what becomes a SAM which helps build a CGE model.

TN asked that members of the group contact WG to share any examples of how they use the IOTs in future and any thoughts they have. This is an interest and WG are keen to work with users. Please email: InputOutputTables@gov.wales. 

CP gave a presentation on the Trade Survey for Wales (TSW) and its interaction with IOTs.

Main messages

• WG have undertaken TSW annually since 2019, collecting trade flows to and from Wales using data for references years 2019 to 2022. TSW was born out of EU exit referendum and frustration at lack of granular economic data for Wales. Hoping to fill evidence gaps and means of triangulating with other data – this helps to show the extent to which apportionment may or may not be effective.
• TSW is the only statistical survey of businesses undertaken by Welsh Government. Work closely with other Devolved Administrations to make sure questions are aligned even though our approaches differ.
• Setout to understand on the ground activity. Intra UK data is valuable. Seeking to produce data by business size and sector.
• TSW is a voluntary survey – 15% response rate on average. Small sizes can affect our detailed breakdowns – break down to EU/Non-EU and sectors are grouped – would like to improve on these over time.
• Gap in 2023 data due to budget constraints. Next TSW is pending financial approval and, if successful, would collect 2024 reference data.
• Have contracted out previously to IFF Research who had been delivering ASGS so had good insight into delivering financial business survey. Delivered as push to web survey with a view to reducing burden on businesses.
• Fieldwork takes place September to January. Data cleansing, weighting, imputation all takes place Jan – March and typically aim to publish mid-July.
• The results of TSW are published as official statistics in development.
• The limitations of the survey are highlighted in a Quality and Methodology report. E.g. there is a long time lag in data production. Being voluntary, the lower response rate is an issue. Incomplete coverage – the survey doesn’t cover all trade activity. As a result, it’s recommended that the results aren’t used to create trade balance.
• TSW is the only survey that collects intra-UK trade values. Prior to TSW, we didn’t know its magnitude compared to international trade.
• TSW is used by IOTs and ONS and Office for the Internal Market to gain a more detailed understanding of internal market.
• Future plans for the TSW:
  • Work closely with IOT colleagues who would like more detailed data.
  • Outsource data collection again.
  • Reduce burden on respondents and questions asked. Economic Statistics Centre of Excellence (ESCoE) recommended shortening. Would be good to do every 3 to 5 years potentially.
  • Considering a model where we work more closely with larger businesses.
  • Open to considering alternative methods of communication.
  • Looking at sampling approach and confidence intervals and how we can improve.
  • Consulting with ONS Methodology Advice Service.
  • Also involved with ONS around a proposed UK wide trade survey to collect data on a consistent basis – long way off in terms of sourcing funding and running that survey.
  • Always trying to get more granular HMRC trade in goods data and BACs from ONS to triangulate our TSW data. 

JN gave a verbal update on a project working across organisations in GSS. 

Main messages

• GSS is the community of all civil servants who collect, collate and communicate statistics. The vision statement is Strength in Numbers.
• There are so many different organisations across GSS – it’s important to consider where we can collaborate to have as much impact as possible and meet the public good.
• Part of the vision is around prioritisation of work – there are lots of cross cutting themes across GSS e.g. labour market, education, health, trade. There are a number of different departments, Devolved Administrations and bodies involved in those different topics. There is a need to ensure coordination.
• The intention is to develop and publish GSS wide work plans for particular topics, and trade is one of those. JN and CGi are working to develop a joined-up work plan covering ONS, HMRC, WG and DBT and other departments and devolved administrations. This links well to the Statistical Assembly held on 22 January, the first event of its kind bringing together users and producers of stats in the UK, helping us as producers to understand user needs.
• The work plan will list publications across GSS but also where we’re developing outputs, thinking about new trade statistics to fill gaps, improve timeliness, granularity, coherence and quality. Joining up – can we be smarter, cut costs, data share?
• Hoping to publish the work plan later this year.
• Other themes on the ONS website.

CG gave a presentation on inter-regional trade, what’s being measured, current limitations, data, potential for data in Wales and future plans for a UK wide trade survey.

Links to publications this year

• Interregional trade in goods and services: UK ITL1, 2019 and 2020, Dataset (13th January) (ONS)
• Interregional trade in goods and services: UK ITL1, 2019 and 2020, Bulletin (10th February) (ONS)
• Interregional Trade in goods and services UK: QMI (10th February) (ONS). Looking to publish 2021 data as soon as possible.

Main messages

• UK interregional trade began as part of a project led by the Fraser Allander Institute which served as a blueprint for developing the methodology. Changes to the methodology included:
  o    Using ABS data instead of Scottish reported imports data
  o    Using more granular data sources such as BACs financial transaction data and haulage data
• Key data inputs are the Devolved Administration surveys. TSW is particularly important as the only survey that reports trade with the rest of the UK disaggregated which was essential in constructing interregional trade for England.
• Needed one country that effectively reported each of these countries to produce this residual matrix and to make sense of and cross check with administrative data sources.
• A main ambition has been to ensure each of the devolved administration survey data sets is as coherent as possible, ensuring they’re measuring the same concepts and on the same basis. This has meant adjusting the industry classifications of some businesses, tweaking the weighting methodology, and imputing non-response and utilising the ABS responses.
• Comparison with devolved administration results should be approached with caution due to methodological differences. Use of the TSW data is different to how WG use the data. Inter-regional trade in goods and services is reported at the ITL one level for a number of industry groups – that coverage is not the same as that reported by the Welsh Government, and so the two estimates vary.
• A full list of industry exclusions can be found in the methods article referenced above. The industry exclusion that has the most significant impact is the financial services industry. Limited most by the industries covered by devolved administrations which means most closely aligning with that reported by NI.
• Benefit of inter-regional trade data is that it enables comparison between Devolved Nations which better allows analysis of the impact of policies or economic impacts. E.g. impact of devolved policies.

Comparing estimates

• For 2019, goods exports figures are fairly similar, but services are lower for ONS than WG. This is due to methodological differences (including imputation) and different industry coverage – both values can be correct.

Future plans 

• continuous development to improve quality of data
• larger scale development e.g. improving industry coverage (including financial services)
• UK Regional Trade Survey – working on a bid as part of next spending review to use a sample equivalent to that collected by Devolved Administrations and England.

Discussion

CJ questioned whether there is a disconnect in talking about modelling ITL1 regions when the political space is often referred to in mayoral authorities e.g. integrated Birmingham, Manchester, Teesside. Are we going to be presenting results by areas that no one cares about anymore? CJ has concerns that there are statements made about cities which shouldn’t be.

CG explained that the current geographical mappers restrict this, but it is something that could be explored. CG also noted that there is a need to take as big a region as possible to avoid suppression when reporting.

JBG reflected that CG's insights offer a real example of what we can do with more regional and geographically granular data. Important that this group do a lot of good work and keep going with this. Invest in the data infrastructure going forward.

TN thanked speakers and those who have supported the development of Welsh Government's IOTs. 

TN invited participants to get in touch if they have suggestions for themes or presentations for the next meeting. 

The next meeting will take place on 14 October 2025.