You are currently viewing our desktop site, do you want to download our app instead?
Moneyweb Android App Moneyweb iOS App Moneyweb Mobile Web App

NEW SENS search and JSE share prices

More about the app

Pretoria, we have a counting problem

It appears that our school maths outcomes have taken hold in our official statistics agency.
The formal sector in South Africa has not actually grown in over four years, as confirmed by other data , such as business confidence and car sales. Picture: Rogan Ward/Reuters

Yip, it looks like our school maths outcomes have taken hold in our official statistics agency.

By my count, it would seem that the formal compensation figure in South Africa should be around 19.2% higher than the figure the statistical agency uses in its gross domestic product estimates for the non-agricultural formal sector.

This is good news for the country but does not reflect well on Stats SA’s counting abilities. Let me explain further.

On Tuesday, we learnt the wonderful news that employment in our formal sector grew by 16.4% between September 2014 and September 2018. You will be forgiven for wondering how this is possible given that the economy is very weak and unlikely to grow employment at the reported rate and levels.

‘Adjustments’, not growth

It comes down to the adjustments that Stats SA has, since the early 2000s, been making to the number of people captured on non-farm payrolls. SA has adjusted the number of people it estimates to be in the formal sector (excluding agriculture) by 95% since September 2002.

Let me be clear: this is not job growth, this is simply better counting of jobs that previously existed.

Just to show how big the adjustments have been, I have compiled a table that shows when non-farm formal sector jobs were revised and by how much.

The table shows that the number of jobs in the formal sector has grown from just under five million to just over 10 million in the past 16 years.

However, as can be seen, the total number of people added by revisions to the dataset was 4.4 million which means that in effect the SA economy created fewer than one million jobs by this measure rather than 5.4 million if one was to just look at the start and end numbers.

Changes to the number of non-farm formal sector employees

Data for the month of

Old number

Revised number


Adjustment release date

Sep 2002

 4 678 609

6 510 000

1 831 391

Mar 2003

Mar 2005

 6 558 000

6 945 000

387 000

Sep 2005

Mar 2006

 7 285 000

8 059 000

774 000

Apr 2007

Sep 2014

 8 669 000

8 947 000

278 000

Mar 2015

Dec 2015

 8 992 000

9 288 000

296 000

Jun 2016

Sep 2016

 9 282 000

9 672 000

390 000

Mar 2017

Sep 2017

 9 589 000

9 716 000

127 000

Mar 2018

Sep 2018

 9 733 000

10 064 000

31 000

Mar 2019

Since September 2014, all job ‘growth’ has come from revisions to quarterly employment statistics (QES) data. In total, the revisions of the number of people employed add up to 1 422 000.

Just the revision added to the original data means that in effect SA has lost 27 000 jobs since then, and not added 1 395 000 as reported!

In other words, the revisions show that there has been no formal sector job growth. This is not all bad. It means the formal sector has been bigger all along and so too the economy.

Most of the reasons for the revisions come from clues from other government datasets that are no longer available in the public domain, such as actual Unemployment Insurance Fund (UIF) collections.

Picture this: Employee X, retrenched from Company Y, heads down to the UIF offices to collect. However, Company Y is not registered for UIF, so the Department of Labour sends inspectors to Company Y – and Bob’s your uncle – Company Y, which employs 25 people, is registered for UIF. Do you see where the mismatch in numbers comes from?

What happened to UIF data?

Since researchers have become more interested in UIF data, it has disappeared from the public eye as the department of labour and the UIF no longer publish it in their annual reports or other publications. 

So the formal sector in South Africa has not actually grown in over four years, as other data (such as business confidence and car sales) indicates. Think about it: if we really had added 1.4 million employees to the formal sector, don’t you think car sales would have shot out the lights over the last five years?

So the news stories you read about formal sector jobs growing are fairytales – but there is a silver lining here.

If more people were employed in 2002 than we thought, then adjustments need to be made to the size of South Africa’s GDP (compensation is included in this calculation).

And if our GDP was, or is, bigger than we thought, then surely the ratings agencies would have had a different view on the economy? More on this in Part 2.


Sort by:
  • Oldest first
  • Newest first
  • Top voted

You must be signed in to comment.


Entirely fascinating! WHY is UIF data being left out of annual reports by the labour dept and UIF?
Are they trying to save on printing costs or might there be something they’re trying to hide from researchers? Shouldn’t they be forced by law to INCLUDE the data in annual reports in the same way companies have to report and make public their earnings, if they’re listed?
Stats SA is still to the best of my knowledge “uncaptured” politically, so data can be trusted to be credible and professional.
But then the quality of Stats SA data probably depends on the accuracy of figures it gets from places such as the labour dept and UIF offices?

Me thinks they stealing.

No such thing as counting problem in 2019… “one for me, two for you…”

Da department of lies, damn lies and statistics !!!

Seems Stats SA is is another captured entity. A pity, as nothing now can believed that originates from them.

No surprises. We’ve known that for years.

Don’t believe a word they say. Especially around crime and HIV.

Maybe they are just sugar coating the numbers to make it more palatable?

The new Statistician General who heads up Stats SA seemed a career civil servant with no political agenda, so possibly not.

I agree with you. There is no political agenda here. That is 100% true. The fact is that StatsSA has missed large parts of the economy and ironically one of the datasets that were considered weak is now with the help of other administrative data helping show SA the real size of the formal VAT paying workforce. We are larger as an economy and that should be good news. A political appointment would have paraded that around a long time ago. But there may be something else keeping the real data from getting out there. Not at all sure what.

Maybe not but it shows that it is run by fools if Mike can pick it up so easily.

Or maybe the problem lies with US READERS; maybe Zuma was right after all when he came to power (not only figuratively) and vowed he would create all those new jobs!

isn’t there elections around the corner?

Well we all know that too many South African’s have a math problems. This is what you get when you dumb down the criteria for passing grades. WHAT THE HELL did you expect? This education problem would just get submerged “somewhere?”

There are three kinds of accountants – those that can count and those that can’t …

Meanwhile, back at the UIF, the UFiling system is down:
‘Website is temporarily unavailable
Due to a required maintenance and upgrade of the uFiling System,
the Website is temporarily unavailable.
We apologise for any inconvenience caused.
The uFiling Team’
The system has been down since early March. I recall hearing a UIF spokesman saying on radio the system would be back up 17 March and in the interim employers could still visit their local labour centre to process their monthly returns and payments.

Not sure why more of a fuss is not being made about this in the press. Could it be that many employers are not UIF compliant? Or are, but don’t use the system and prefer to wait in queues (unlikely)?

Either way, the department should be more transparent about what is happening or not happening with their UFiling system.

OK, the system is back up after a month. And it’s a brand new interface. But it appears there have been data migration issues. Names of retired/resigned (domestic) employees now appear, but no historical or current financial data on current or past (domestic) employees. One is required to re-declare all employee data. I hope they haven’t lost historical contributions, both the money and the records!

On the upside, I had only ever used the system as an employer of domestic employees, but my information as employee in my own capacity is now included on this site, so they’ve tied my registration as employer to my ex-status as employee. Somebody thought to join the dots.

I think that STATSA still suffers from their age old problem they don’t have sufficient qualified people who have studied statistics as a subject. Many years ago I engaged STATSA and took along a member of the then NPI (National Productivity Institute) and we tried to address variable in CPI, formal and informal job creation and the methodology which they used to collect data, back test data then and evaluated their results and we eventually walked away from them as their number where unreliable and their selection criteria suspect

The NASA space organization’s internet connection can transfer 91 gigabits per second….I would have run away from STATSA if I where you…..on second thought you should not have engaged with them in the first place.

One would think that when publishing an article on stats and numbers, extra effort would be made to ensure that your numbers are printed correctly. The difference between the old and revised numbers for 2018 is 331,000, not 31,000.

“Pretoria, we have a counting problem”

In South Africa, we have Pretoria and we also have Cape Town. These 2 are separate entities, as they are projected to be apparently.

no matter how figures are adjusted or manipulated, the unemployment rate in SA are a staggering 27%. The highest rate of U.S. unemployment was 24.9 percent in 1933. That was during the Great Depression.

So the record of major gov agencies proving their incompetence/ability to steal continues unabated.

Shameful from Stats SA and UIF but not one bit surprising.

Can’t wait for part 2!

Mike, if you think this is bad, you should see the RTMC’s road traffic stats. Since 2000, they only include data on fatal crashes, not all crashes (in other words, they are reporting on 1% of crashes). Major statistical rates are no longer included, and critical items like crashes by street element have been missing for 8 years. The stats are so bad they would be useless for policymaking if they were even used for that purpose, which they haven’t been for 2 decades.

Stats SA issues standards with which statistics released by organs of state have to comply. The RTMC stats certainly doesn’t. Now it seems Stats SA isn’t complying with its own standards. Scandalous.

End of comments.





Follow us:

Search Articles: Advanced Search
Click a Company: