By Graeme Roy
ESCoE, in collaboration with colleagues from the Fraser of Allander Institute and the Office for National Statistics (ONS), held an online workshop in June on regional statistics. The workshop was arranged around a number of thematic sessions: ‘Improving Regional Indicators’; ‘Regional Impact of COVID-19: The View From Academia’; and ‘Measurement Innovations and Considerations for Policy Making’. The day concluded with a panel session ‘The Future of Regional Statistics’.
This blog brings together the main concepts discussed by our panel, which included: Ed Humpherson (Director General for Regulation, UK Statistics Authority); Jonathan Athow (Deputy National Statistician, Economic Statistics, ONS); and Jane Naylor (Head of Statistics, Department for International Trade). I present the panel’s reflections as a set of, sometimes closely related, challenges and / or opportunities for the economics statistics community, both now during the pandemic, and as we look further forward to other policy demands. Contrary to the tendency for compressing increasing demand for regional statistics into the broad ‘levelling-up’ agenda, our first panellist, Ed Humpherson, began his comments by outlining the importance of breaking down this increasing demand into its different constituent elements. His point being that if there are increasing demands, it is important for us to comprehensively understand these trends so as to be in a better position for producers of statistics to meet that demand. It seems sensible to adopt a similar approach here in this blog.
Coherence and strategy
Jonathan Athow made the case for a coherent and strategic approach in the production of regional statistics a key focus of his comments. One of the challenges the statistics community faces in this area is that the growth in demand is not just for sub-national statistics in one particular domain, but across cross-cutting domains. Statistics users want multi-dimensional, joined-up data, so that when analysing the economic output of a particular part of the UK, for example, they also know what industries are related to that output and how the local labour market interacts with that output etc.
Engagement was also identified as an issue here. When it comes to engagement at the subnational level, whether that is for data production, analysis, or for consultation with user groups, there is the question of how best to engage coherently across numerous and diverse devolved government, regional government, local government, private sector or third sector stakeholders?
Communicating the challenges in producing robust sub-national statistics – and how to interpret them – needs to be an ongoing investment.
All panellists agreed that there is success to be built upon. Compared to even just five years ago, ONS is now producing more regional and local statistics. But in looking to the next stage of investment, the demands in terms of scope and timeframe can often represent a call for ‘everything, now’ – so as a community we need to think carefully about feasibility, and where we want, or need to be, in the different statistical domains. Where we might want to get to with regional prices for example, might be different from where we want to get to with regional labour market statistics or with regional crime statistics etc.
There are inevitable resource ‘trade-offs’. With a constantly changing economy there is also competing demand for developing more detailed UK-wide statistics. We need to think carefully about where we put our resources to best develop and balance national and subnational statistics.
Welfare, household experiences and alternative data sources
One key area of growing demand is in measuring household welfare going beyond GDP. The original imbalances or levelling-up debate is typically one which takes the national GDP picture, and then disaggregates it. There is though now a lot of analysis which shows that this is an imperfect way of describing household welfare experiences. Ed described how the UK Statistics Authority has undertaken considerable work on subjects such as income, earnings and poverty, referencing for example a recently published review of poverty statistics which encouraged the use of administrative data to provide a better regional picture of household experiences. It was Ed’s view that a key area of future growth would be to explore more around what could be done on the ‘beyond GDP’ argument at a more disaggregate regional level.
Alongside developing ‘new measures’ of economic prosperity, a key strand for future work is around using new data sources to obtain a more accurate and timely picture of economic activity. All panellists highlighted the many opportunities in some of the new and alternative data sources available for statisticians to work with. Jonathan highlighted how, in one strand of their work, ONS have made innovative use during COVID-19 of the real-time PAYE data from HMRC. These data provide details on all payroll employees and ONS are cutting that data in new ways. They have plans to push that down to NUTS level 3 so that it provides still more detailed information. Work by Stuart McIntyre and the ESCoE team have been developing regional nowcasts of economic activity. Jane explained that at DIT, in order to provide new, more timely estimates of trade and indicators of trade, they have been looking at alternative data sources like the ONS faster indicators, the Business Impact of COVID-19 Survey (BICS) data, data on freight movements and data on shipping movements. The utilisation of such new / alternative data sources are going to be amongst the most important ways that we measure and understand what is happening at the subnational level.
Representation, identity and dissemination
National economic decision-making institutions are not necessarily good at representing the sheer variety and complexity of local interests. Notions of representation and identity are shifting. Notions of how to represent the identity of an individual, a household, or an area in statistics are shifting as well as how to represent those identities in political institutions. All the panellists highlighted how ‘economic regions’ from a statistical perspective can be problematic concepts, sometimes unhelpful agglomerations that do not really reflect the identity or political realities of the subregions that make them up. For statistics users in the local context, with people wanting access to very localised, granular statistics, a region no longer becomes the meaningful way of looking at an area.
At the personal level there are ongoing discussions about how to represent identity and statistics around ethnicity, sex and gender. Producers of statistics need to, and are trying to, adjust to the shifts occurring as a result of those reflections and understandings.
A real challenge for producers of statistics is how all this localised, personalised information are then disseminated. How can we tell the story about particular areas of the country? Interactive maps such as those produced by Ahmad Barclay were identified as one solution to this complex problem by Jonathan. People want local data, they want to be able to see that data on an interactive map where they can select different geographic areas and then be able to access all the relevant information around that.
Regional imbalances, levelling up and aligning with other policy agendas
There is an ongoing, long-standing interest in regional imbalances in the UK economy. This is the observation that the UK is particularly imbalanced, a consensus supported by available statistics, it is more imbalanced than it was in the past and much more imbalanced than other equivalent economies. The levelling-up agenda has been a key driver of developments in regional statistics. COVID-19 has also been a recent driver, having accelerated demand for timely subnational data, particularly as we are still trying to understand the pandemic and its impacts, as well as monitoring and supporting the recovery around the UK.
Policy can sometimes change quite rapidly. Developments in data and statistics however can take and require more time. Jane highlighted that this lag can sometimes be a challenge for producers of statistics. She referred to an example from the Department for International Trade, who face the challenge of how to accurately measure progress in supporting and promoting exporting across all regions of the UK in the light of the Government’s ambitions to promote international trade more broadly across the UK. At present, we do not have one consistent measure of regional trade. We have HMRC regional trade statistics, and for the trading of goods we have ONS estimates on trading services, but we cannot actually put those two data together. They are not of a consistent basis, or with the same kind of timeliness. Creating one measure of exports at a regional level is not possible at present. The existing statistics do not always align with exactly what we want to measure for policy making.
Of course, one of the issues that DIT (together with partners at ONS and ESCoE) have been grappling with is not just understanding international trade at the regional level but understanding intra-regional trade. This involves not just gaining an understanding of how much a particular region exports, say to France, but also how much does that region exports to say the South East, or how much Scotland trades with England.
This blog has attempted to condense some of the key topics of discussion from the panel session. For the purposes of brevity it has not been possible to detail all of the excellent points that were made. However, you can access all of the slides and video recordings, including of the panel session, here. Thank you to all of the presenters and panellists who took part in the workshop.