Prominent academic economists are scratching their heads after the shock announcement by Stats SA that 355 000 jobs disappeared between the end of last year and the end of March this year.
There is a real possibility that it is a statistical glitch, but it is hard to know for sure either way.
The introduction of a new “master sample” last year, from which the households surveyed in the Quarterly Labour Force Survey are drawn, caused strange jumps in the numbers last year.
It was based on the Census 2011 and made the sample give more weight to Gauteng and some mining areas, among other things.
Martin Wittenberg, associate professor of economics at the University of Cape Town (UCT) and head of its DataFirst project, said he was “on the fence”.
“If I was a betting person, which I’m generally not, I would bet on a significant proportion of the lost jobs being due to the new master sample,” he told City Press.
More evidence was needed before judgement could be passed on the new numbers, he said.
According to Wittenberg, it might be that the new sample resulted in more “volatile” estimates.
“But I have no evidence to suggest that this has happened,” he emphasised.
Ingrid Woolard, another prominent UCT professor and labour economist, said that the numbers were “surprising”, but that she had not had a chance to investigate them.
There is often a drop in the number of employed people in the first quarter of the year related to the end of seasonal jobs in retail and hospitality. However, the drop was completely out of proportion this year.
Neva Makgetla from the think-tank Trade and Industrial Policy Strategies said the job loss figure “seems unreliable”, but probably reflected the true state of affairs after last year’s statistics overestimated employment.
Some of the big shifts in estimated jobs were just unrealistic, she argued.
The statistics show, for instance, that more than 10% of manufacturing jobs in the Western Cape disappeared, as did 20% of construction jobs in Mpumalanga, 11% of retail jobs in Gauteng and 17% of “business service” jobs in the Eastern Cape.
“Obviously, it seems unlikely this kind of bloodbath could have occurred without significantly more effects on communities and workers. So what is really going on? The short answer is we can’t tell, because of statistical problems,” Makgetla wrote in a policy brief.
“The Quarterly Labour Force Survey has been a high-quality survey for most of the past few years, and a credit to Stats SA. It would be desirable if Stats SA published its own analysis of what went wrong with the latest report,” said Makgetla.
Statistician-General and head of Stats SA Pali Lehohla dismissed suggestions that changes to the master sample distorted the numbers.