Annexes

Annex A: Revisions analysis

A.1. In this revisions analysis overview, Mean Revisions (MR), Mean Absolute Revisions (MAR) and Relative Mean Absolute Revisions (RMAR) are discussed. Revisions for the Global Financial Crisis (GFC) and pre-COVID periods, were measured after 12, 24 and 36 months (T+12, T+24 and T+36 respectively), using the quarterly first estimates as a comparative point. For the COVID period (2020Q1-2021Q4), only the T+12 revision is available. The process is explained in the table below:

Please note that:

• The Pre-COVID period covers 2010 (Q1: January to March) to 2019 (Q4: October to December)
• The COVID period covers 2020 (Q1: January to March) to 2021 (Q4: October to December)
• The Global Financial Crisis period covers 2008 (Q1: January to March) to 2009 (Q4: October to December)

A.2. Mean Revisions (MR) show whether revisions are biased in a particular direction (positive or negative). As explained by the OECD’s Working Party on National Accounts “A positive (negative) sign indicates that, on average, earlier releases have been underestimated (overestimated).”

A.3. The average quarterly real GDP growth during the pre-COVID period, was hardly revised both one and two years after the first estimate (T+12 months and T+24 months), while at three years (T+36 months) it was revised by 0.06 percentage points (pp). In contrast, during the COVID period, on average, the first estimate of quarterly GDP growth was revised up by 0.58pp at T+12 months, compared with the the Global Financial Crisis (GFC) of 2008-9 which was revised down by 0.15pp. Figure A1 compares MR of quarterly real GDP growth between the COVID, pre-COVID and GFC periods.

A.4. Mean Absolute Revisions (MAR) measures the average revision in absolute terms, avoiding the offsetting effect from the negative and positive values. Consequenlty, it is better placed than MR to guage the size of revisions.

Figure A1: Mean revisions of quaterly real GDP growth at T+12 months, T+24 months and T+36 months

Source: ONS data from GDP in Chained Volume Measures Revision Triangles Q2 2023 Edition

A.5. Figure A2 shows MAR to early estimates of quarterly real GDP growth. The MAR in the COVID period was 0.78pp, compared with the pre-COVID and GFC periods of 0.12pp and 0.20pp respectively.

Figure A2: Mean Absolute Revisions (MAR) of quarterly real GDP growth at T+12 months, T+24 months and T+36 months

Source: ONS data from GDP in Chained Volume Measures Revision Triangles Q2 2023 Edition

A.6. The third presentation of GDP growth revisions we considered was the Relative Mean Absolute Revisions (RMAR). This contextualises the revision in the context of the growth rates, measuring the MAR in relation to the size of the growth rates. During economic turning points, such as the COVID period, growth rates are larger in magnitude and more volatile, which in turn could potentially generate larger changes and revisions. What might seem a large revision when contextualised by the size of the growth rate, might be relatively small or vice versa. Additionally, the value of RMAR can be increased by carrying out a comparison with other economic turning points. Using a Relative Mean Absolute Revisions (RMAR) during periods of structural breaks or economic turning points, can improve users’ understanding of the latest period’s revisions relative to previous periods.

A.7. Although the MAR of quarterly GDP growth was 0.78pp during the COVID period, this was also relative to much higher average GDP growth rates (6.4pp). In relative terms, the revision during the COVID period was relatively smaller, 0.12pp, compared with both the pre-COVID and the GFC periods, which were 0.26pp and 0.23pp respectively. Figure A3 displays the RMAR to early estimates of quarterly real GDP growth.

Figure A3: Relative Mean Absolute Revisions (RMAR) of quarterly real GDP growth at T+12 months, T+24 months and T+36 months

Source: ONS data from GDP in Chained Volume Measures Revision Triangles Q2 2023 Edition

Annex B: ONS revisions timetable for quarterly estimates of GDP up to September 2023

Source: ONS, National Accounts Revisions Policy: updated November 2022.

Please note ONS aims to update this revision policy in November 2023.