Statistics agencies, here and abroad, are under fire. The UK’s main statistics agency is in trouble because it cannot produce reliable figures for unemployment. The head of the Bank of England has described it as a “substantial problem” that affects the central bank’s ability to set interest rates, as well as commercial banks’ decisions over bad debt provisions and risk appetite.
The UK Office for National Statistics (ONS) now says it will take until 2027 to fix its figures. The problems stem back to the Covid-19 pandemic, when the response rate to its survey plummeted and has not recovered. Experts who have examined the data suggest the agency is undercounting employment by about 1-million people.
The debacle illustrates that even a well-funded statistics agency such as the ONS can get things wrong. It also shows the serious economic impact when statistics can’t be relied on — everything from interest rate decisions to how banks decide to allocate capital and take risk is compromised.
That is a helpful lesson to keep in mind as Stats SA grapples with its future. Its head, Risenga Maluleke, told parliament in September that it needs an extra R295m to stabilise the organisation and maintain its operational integrity. The agency has 660 vacancies out of 3,301 posts. The 2,672 employees are working on outdated computers and the empty posts can’t be filled because of a hiring freeze from the National Treasury.
Amid all this, it is natural to ponder whether Stats SA’s statistics can be relied on. Everything from local government budget allocations to national economic policy is driven by the agency’s work. It faced extensive criticism over the 2022 census, which had a significant undercount, revealed through a postenumeration survey.
The agency has strongly defended the census outcomes and they have been assessed as “fit-for-purpose” by the independent SA Statistics Council. Maluleke told parliament that the agency has an extensive internal review process to ensure the quality of data and that the public can rely on them.
But one wonders. Eyebrows are raised about the third-quarter GDP figures released last week, for instance. They flew against almost all predictions, recording an overall fall in GDP of 0.3%, while the market was expecting growth of 0.4%. The main driver of the surprise was the performance of the agricultural sector, which showed a 28.8% decline, which took 0.7 of a percentage point off the overall GDP level. While most economists were expecting some level of decline given the midsummer drought and impact on crop production, the scale was unprecedented.
Contributing to the raised eyebrows is some contradictory data. For one thing, agricultural exports during the quarter were up 5%. The other is that business confidence in the agricultural sector was heading upward in the quarter, hitting 48 against the previous quarter’s 38, as recorded in the agribusiness confidence index undertaken by the Agricultural Business Chamber (Agbiz). It went up to 58 in the fourth quarter. This pattern just doesn’t seem consistent with a sector that had more than a quarter of its production falling off in the third quarter.
Of course, these contradictory statistics don’t mean the Stats SA figures are impossible, but Agbiz chief economist Wandile Sihlobo has said we may well see a revision in the data for the third quarter in future releases. Combine the contradictory data with Stats SA’s clear operational challenges and it is only natural to expect public confidence in the figures to be easily shaken.
Revisions do happen — they are a normal part of Stats SA’s work. The agency gets late submissions of data and respondents can report corrections to figures. The GDP estimates are based on a wide variety of source data inputs. Some of these, such as Stats SA’s quarterly financial statistics, are only finalised after the release of GDP figures, so a revised version will show differences.
GDP is measured through production, expenditure and income, and each approach has different sources. Data from the SA Reserve Bank, SA Revenue Service and various government departments are included, with several surveys Stats SA undertakes. It is, all in all, a complex endeavour, which means things can go wrong.
For the financial sector, accurate statistics are important. If we can’t rely on the data we must treat it with a pinch of salt, which means adding a layer of additional caution and spending more on alternative sources of data. Banks must decide how much to provision for loans they are making, which they do through a model that estimates likely default rates. Those estimates include GDP figures, unemployment and much else.
If banks can’t rely on the statistics they must build a margin of caution into their models, which means provisioning more than they otherwise might. The result is less lending into the economy, and therefore less economic growth. Many other businesses make investment decisions on the basis of economic data — the less reliable it is, the more cautious they must be, and again, the less economic growth.
A good statistics agency costs money, but it is a wise investment given the economic impact it has.
• Stuart Theobald is chair of research-led consultancy Krutham.
This article first appeared in Business Day.
