The latest quarterly labour force survey from Stats SA is a shocker. It finds that unemployment rose to a record high above 35% in the fourth quarter despite an increase in jobs, because so many more folk were looking for work. It shows, too, that employment declined by more than 3% in 2021 even though the economy recovered from the worst of Covid to grow by almost 5%. Total job losses since the start of the pandemic stand at almost 2-million.
The picture of a jobless recovery may be plausible. Employers might have realised during the lockdown that they could produce the same with fewer workers, so retrenched as the economy opened up. Some companies just couldn’t hold on any longer as the lockdowns continued.
But even if it’s plausible, is it true? The response rate was so low — just 44% — that it’s hard to read anything off the figures reliably. What makes it even more concerning is that in the urban, economic heartland even fewer people responded to the survey — just 23% in Gauteng and 39% in the Western Cape.
To its credit, Stats SA did publish the figures with a health warning. But some experts wonder whether it should have published them at all — at least as official figures. The Statistics Council, an independent body that oversees Stats SA’s data, is understood to have called an emergency meeting following the release of the survey.
You have to feel sympathy for Stats SA. It’s had a hard time collecting data during Covid, as have many other official statistics agencies, and it has had its capacity severely eroded by years of budget cuts. SA’s troubled sociopolitical dynamics are also increasingly affecting household surveys.
The first big problem is that Stats SA switched from face-to-face interviews to telephonic ones at the start of the lockdown in March 2020 and has continued that ever since. It samples just under 30,000 households for the labour force survey, and extrapolates from that to the population as a whole, as statisticians do. Neither the fieldworkers nor the respondent households wanted to be exposed to Covid, hence the telephonic surveys, which have continued amid uncertainty about lockdown levels.
However, over time the response rates have got ever worse as contact details for the households in the sample have become outdated, and because people just don’t want to do telephone surveys. They put down the phone on fieldworkers they might have been persuaded to talk to in person. It hasn’t helped that private sector marketers are increasingly using telephone “surveys” to sell products.
The biggest problem is simply the climate of fear and distrust in the country at large. Who is going to disclose their income to a fieldworker, even an official one? Amid high levels of distrust of the government, people don’t necessarily believe that their data will be safe with an official agency. That’s always been a problem with the gated communities of suburbia, especially when it comes to questions on income. Add to that xenophobia, Operation Dudula and the climate of fear in the townships, and people are now even more reluctant to respond.
Nor is it only a problem for official surveys — independent researchers report some of the same response rate problems. Before the pandemic the response rate for the quarterly labour force survey was typically over 80% (and even early in the pandemic it was 57%). Stats SA did send out fieldworkers when the pandemic permitted to refresh phone numbers and details, but clearly that didn’t help much.
It is now at last going to switch back to in-person interviews for the next labour force survey, for the first quarter 2022. If employment jumps in the first quarter don’t be surprised — but again, it could be more about response rates than reality. The labour force survey is the only household survey that Stats SA does on a quarterly basis, but there are other, annual ones such as the general household surveys that have been similarly affected.
All of this must surely raise concerns about SA’s first census in more than a decade, which is currently in the field — and so far has reached only 70% of SA’s estimated 17-million households, with Stats SA this week launching a second “mop up” to try to get data from the rest. Where a survey is based on a sample, a census is supposed to be just that — in our case, it is supposed to count every household in SA as of the February 2 reference date. The census is essential to give us a picture of ourselves as a nation, but also as the basis of all the other sampling and statistics. The 2011 census had all sorts of undercount issues, so this one is really important.
If you can’t measure it you can’t manage it, as the saying goes. Good policy-making depends on good data, and though SA still has much better official data than many emerging markets, the question marks over the quality of some key economic statistics is a problem for policy. Just imagine if the SA Reserve Bank was the US Federal Reserve, which has an employment mandate along with its price stability mandate and relies heavily on monthly labour force statistics to guide it on the state of the economy and the decisions it must make. Labour force data based on less than half the intended sample surely wouldn’t cut it.
Finance minister Enoch Godongwana seems to be taking the statistics agency’s woes seriously and has allocated some extra funds. Other policymakers also need to be alert and to read the data critically, as should we all. On the upside, it seems at least possible that unemployment may not be quite as bad as we think it is. The survey is certainly under-measuring the urban provinces, where most of the jobs and side-hustles are. And it’s over-measuring the more rural, dormitory provinces where there are few jobs.
Even with the methods statisticians use to try to correct for this, it’s surely likely that there are a lot more people at work in SA than the labour force survey lets on.
• Joffe is editor at large.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.