What should we expect from the 2021 GDP revisions?

Image: Burak K on Pexels.

South Africa’s national accounts were last rebased and benchmarked in 2014 (for years up to 2013). The next rebasing (from the current 2010 base year to 2015) should have been done in 2019, was rescheduled to 2020, but was deferred due to the unusual circumstances of the year. Publication of the 2018 supply and use tables was also deferred, and the March 2020 and March 2021 GDP estimates did not include the usual adjustments for prior years based on independent annual estimates. Statistics SA indicated in March that rebased and benchmarked GDP estimates would be published by the end of May 2021 and that the June GDP estimates (for the first quarter of 2021) and the June SARB Quarterly Bulletin would be compiled on the revised basis. But there has been a further delay – Stats SA now says the 2015-based GDP estimates will be published in August.

The rebasing of the national accounts has, in the past, been accompanied by substantial upward revisions to both the level of GDP and prior growth rates. In 2014, the upward adjustment to 2010 real GDP was 2.8 per cent. In 2004 and 2009, real GDP for 2000 and 2005 respectively were revised up by 4 per cent and 2.1 per cent.

In 1999, a more comprehensive revision was undertaken as part of the transition to the 1993 System of National Accounts, leading to an upward adjustment in the weighting of services from 55½ to 61½ per cent in 1995. The growth rate in GDP at market prices between 1993 and 1998 was revised up from 2.2 per cent a year (in 1990 prices) to 2.7 per cent a year (in 1995 prices). The 1999 methodological revisions and upward adjustments led to a 14 per cent increase in the measured level of nominal GDP in 1995.

What should we expect of the 2021 revisions?

The 2014 adjustments were largely a result of methodological changes. Partial implementation of the 2008 System of National Accounts accounted for 45 per cent of the adjustment:  accounting changes in research and development expenditure, weapons systems, and livestock, and revisions to the indirect measurement of financial intermediation services. Enhanced measurement of economic activity accounted for 1.6 percentage points of the 2010 increase of 2.8 per cent.

There might be further methodological changes this year. But the remaining 2008 SNA recommendations would mainly affect the classification of income or payments, not the overall level of activity.

Substantial upward adjustments are more likely to arise from enhanced measurement and better use of available data.

A broad indication of the likelihood of upward revisions to GDP growth over the past decade is provided by comparison of employment and national income trends. Between 2010 and 2019, non-agricultural non-household formal employment increased by 16.7 per cent, while gross value added increased by 14.9 per cent. It is unlikely that employment growth has exceeded the output trend for such an extended period – there is usually some contribution to aggregate production growth from productivity gains.

More specific evidence comes from the Quarterly Employment Statistics, which record employment numbers and “gross remuneration” for a representative sample of enterprises and organisations across the non-agricultural formal economy. The QES measure of gross remuneration increased in nominal terms by 130 per cent between 2010 and 2019, whereas “compensation of employees” (CoE) in the national production accounts increased by 96 per cent. In 2010, QES gross remuneration was 3 per cent higher than CoE; in 2019 it was 20 per cent higher. These are not strictly comparable aggregates – part of gross remuneration is recorded in the national accounts in “gross operating surplus / mixed income” of the household and unincorporated enterprise sector. But the divergence between these measures is too large to be explained by coverage differences. It is compelling evidence that overall earnings growth has been greater than the national accounts indicate.

The sectoral trends suggest where we might expect to see substantial adjustments in the new GDP series.

The GDP CoE aggregates have lagged behind QES remuneration trends in all sectors other than electricity, gas and water. The gap is widest in non-government services (largely health services, but also education, recreational activities and personal care services), followed by construction, wholesale and retail trade, hotels and restaurants, and the transport and communication sector. These industries account for a third of total output. In these sectors, the QES data imply that 2019 GDP could be understated by 20 per cent or more.

These comparisons are in nominal terms – the adjustments might be shared between real and inflation-related revisions. But it is also plausible that the coverage of the QES has widened over the past decade, so that the divergence in trends is not only a consequence of lags in national accounts estimates.

It seems reasonable to expect that there will be substantial upward adjustments to our GDP and growth estimates in the forthcoming national accounts revisions. Real GDP growth between 2010 and 2019 as currently measured was 1.5 per cent a year; the revised estimate could well be over 2 per cent a year, raising 2019 real GDP by 5 per cent or more. There would be nominal improvements in several important indicators – government debt to GDP; household debt to disposable income; perhaps the savings and investment ratios; and productivity indices. But GDP revisions will not change our unemployment rate, or measures of poverty and inequality, or the broader policy imperative of faster and more inclusive growth. They signal, rather, that the dynamics of growth and adjustment are more buoyant than we think, and that we need to understand them better.

This  article was also published by Business Day.