Data sources and quality assurance

1.21 CRPSF public sector expenditure data are sourced from HMT’s CRA statistics. The CRA presents statistical estimates of identifiable expenditure (spending that can be identified as having been for the benefit of a particular area of the UK) across UK nations and English regions. Identifiable expenditure in 2020-21 constituted 92% of public sector expenditure on public services.

1.22 CRPSF revenue estimates are based on the “who pays” principle, which identifies the location where the revenue is raised, that is, the location of the individual or enterprise that bears the tax burden. Taxes are not usually collected on a regional basis, which makes it difficult to identify where country or regional tax receipts should be allocated, so modelling assumptions are required to estimate public sector revenue at the country or region level.

1.23 The CRPSF published methodology information outlines the background to each of its revenue data sources, and provides a comparison with data sources used in other public finance statistics including GERS and the DoTR and Northern Ireland Department of Finance net fiscal balance reports. The detail offered in its methodology documentation supports users’ understanding of compilation process of CRPSF statistics.

1.24 The data used to compile CRPSF statistics are the same as those used to produce UK PSF statistics. This means that the procedures in place to quality assure PSF data sources apply equally to CRPSF statistics. As part of its re-assessment of the National Statistics designation of UKPSF statistics in 2015, OSR focussed on the quality assurance procedures in place for administrative data used in the production of the PSF. The 2015 review noted concerns about the quality assurance arrangements in place for administrative data.

1.25 In response ONS reviewed its quality assurance arrangements for administrative data, introducing a new approach to data assurance in 2017 which focusses on assessing the level of risk and public interest in administrative data, as set out in the UK Statistics Authority’s Quality Assurance of Administrative Data (QAAD) Toolkit. The identified level of risk is then used to determine the extent of quality assurance dedicated to each administrative dataset.

1.26 The quality assurance of each stage of the production process of CRPSF statistics is thorough with numerous data quality checks taking place at all stages of production. For example, checks are made that the correct data sources have been downloaded, that regional data have been correctly collected, and that the English regional totals sum to the total for England and that the nations’ totals sum to the UK total.

1.27 For quality assurance purposes, preliminary estimates of CRPSF statistics are shared with members of the working group. Each of the group’s members offer different perspectives on estimates of apportioned regional and country data, with HMT quality assuring expenditure estimates and HMRC assuring revenue estimates. Representatives from the Scottish, Welsh and Northern Ireland governments bring local intelligence to bear when reviewing and assuring preliminary estimates of apportioned expenditure and revenue country data.

1.28 The activity of sharing data and estimates with statistical producers and users of CRPSF statistics for quality assurance is an example of best practice. ONS’s work with its partners responds directly to the requirements of OSR’s 2019 review of the public value of statistics on public finances in a devolved UK.