Recommendations

 Recommendation 1:  ONS should commission external expertise to review its approach to the seasonal adjustment of GDP, with a particular focus on the detection of, and adjustment for, emerging seasonality. The outcomes of the review should be made publicly available increase confidence amongst users. In finalising this review, ONS have told us they have already commissioned this work.

Recommendation 2: ONS should further increase transparency in its approach to seasonal adjustment and the associated uncertainties and limitations. Further transparency and openness would increase the confidence users have in the results and could be promoted in various ways. We suggest ONS could:

• Build on the publication of the analysis of seasonality in GDP by publishing a regular review of seasonal adjustment across its main output areas of economic statistics, including not just GDP, but also prices and labour markets.
• In line with its stated intention, publish updated results for residual seasonality checks each year for main aggregated estimates as part of the annual Blue Book and Pink Book updates.
• Include, in future published reviews of seasonality, the results of investigations of the sources in the high-level component series of any potentially emerging seasonal patterns in headline GDP, including those that fall short of statistical significance.
• Provide greater clarity about diagnostic “triggers” (for example, major economic shocks such as the COVID-19 pandemic) that should prompt a review of adjustment methods and/or of the length of the periods used to assess the statistical significance of seasonal patterns.
• Publish regular assessments of whether there is emerging residual seasonality in the unrevised data contained in first releases that subsequently gets “revised away”. Such assessments would sit naturally within the annually published review of GDP revisions, but it is for ONS to determine the most appropriate route for publication.
• Provide more direct and/or signposted access to unadjusted series so that users can apply their own approaches to seasonal adjustment. This might both build confidence and, through openness to challenge, contribute to quality assurance. (We note that while this report was being finalised ONS committed to publishing non-seasonally adjusted measures of GDP.)
• Publish further assessment of the impact of different adjustment methods and diagnostic tests, compared to those currently adopted.

Recommendation 3: We recommend that ONS should continue with its plans to improve the resourcing, planning, and implementation of its approach to seasonal adjustment.

• ONS should ensure sufficient resources are allocated to undertake a broader range of investigations of residual seasonality in higher-level aggregates (for example, in quarterly income and expenditure components separately).
• ONS should ensure residual seasonality checks are embedded into the production process via a reproducible statistical process that can be automatically applied, while continuing to draw on complier’s domain-specific expertise.
• Resource requirements for seasonal adjustment should inform human resource planning more generally, including consideration of the incentives and rewards needed to recruit, retain and develop relevant specialist expertise and capability across ONS.
• ONS should make greater use of external expertise, for example by establishing an external expert group to review seasonal adjustment on an ongoing basis and by increasing engagement with other national statistical institutes.