PALI LEHOHLA | The Vulture Culture of numbers: shield official statistics from the arithmetic of self-interest

Twenty-nine years after its change of name in 1998 from the Central Statistical Service (CSS) to its new identity, Statistics South Africa (Stats SA) stands beyond any shadow of doubt as one of the best-run institutions of the state. It is a paragon of trust and a steadfast upholder of the United Nations Fundamental Principles of Official Statistics (UNFPOS). Yet as we stand on this precipice of credibility, the winds of disinformation are gathering, driven not by the state, but by private interests masquerading as public insight.

In a webinar held on December 11 this year, the Friends of Official Statistics (FOS) — a global group of veteran retirees from statistical operations — convened to consider the existential threats to the institution of UNFPOS. To foreground our discussion, Hermann Habermann, a veteran of the US Federal Statistics System (FSS) and former director of the United Nations Statistics Division, led with a paper titled “The Trauma of the Federal Statistical System”.

His preamble was chilling: “Since January 2025, the United States Federal Statistics System has been undergoing severe and significant trauma. Other countries are experiencing similar experiences ... the activity in the US provides a test case to examine what responses, if any, the FSS can employ in the face of turmoil.”

It was as though Habermann was offering the comfort of company to South Africa. While Stats SA has, to its benefit, largely escaped raw vitriol from political principals, it has received a spectrum of vengeance exclusively from white men in the private and academic sectors.

Albeit by all counts the best institution of the state in the service of statecraft, Stats SA has not been immune from these sporadic attacks. The record shows this antagonism has been notably racialised, emanating from white male professionals. It is not immediately clear why the attacks have taken this specific racial character, but we must differentiate the attackers, lest we diffuse excellence into a bottomless pit of mediocrity.

First, there are the intellectual career critics, most notably professors Rob Dorrington and Tom Moultrie of the University of Cape Town. These men have been the sharpest critics of the demographics Stats SA produces. After every census, these two professors — who often had an inside lane in evaluations — would inevitably produce a minority report.

In the censuses of 2011 and 2022, they repeated the skepticism that the veteran Prof Dorrington displayed in 1996 and 2001. A notable exception occurred in 2011, when they pressured me to postpone the release of the Census results to satisfy their intellectual curiosity. I refused. They had abandoned ship for personal work-related reasons, making it impossible for them to camp at Stats SA for the prolonged period required to evaluate the census properly. I decided they had gone Awol. The Statistics Council and I were not short of independent expert evaluators mobilised locally and internationally. While their critiques were rooted in science, they were often crafted in the laboratory of imagination, far removed from the lived reality of the count.

The second group of experts are those who commit fatal errors in interpreting data within a relational database. Here, we find the late Mike Schussler and Loane Sharpe.

Schussler, a great friend of mine who read Stats SA data from back to front, would often unleash interesting critiques but suffered from schoolboy limitations. One of the most ridiculous missives Schussler threw at me involved arguing against sacrosanct rules of counting. He counted the frequency of visits into South Africa as a basis for estimating the population of Lesotho. Alas, the headline screamed: “More than the entire Lesotho population cross borders into South Africa.”

Making such a rudimentary error displays ignorance of the rules of counting in a relational database structure: the difference between one-to-one, one-to-many, and many-to-many relationships. Schussler counted the event (the border crossing) rather than the entity (the person). After buying me a lousy meal at the airport in Bloemfontein, Schussler would say to me, “Pali, you know without Stats SA I am out of business, mate.” We laughed loudly. But the error remained.

Sharpe of Adcorp committed a similar sin with his so-called “Adcorp Index of Employment”. Like Schussler, he failed to use his recruitment brokerage data correctly to contest national numbers. Sharpe’s index was counting the number of jobs (contracts) his labour brokerage offered and translating that directly to individuals. Suddenly the number of employed people appeared vastly inflated. I had to lay down the law, and Sharpe’s index withered away like ether.

In both instances, the “shotgun, own-data-on-the-shelf” approach could not stand the test against the robust methodologies of the mighty Stats SA. But these men never give up. They reincarnate their interests in different forms. Nature allows no vacuum.

Recently a new breed of men has entered the arena of contesting national numbers. This time, the assault is led by multibillion-rand executives: Gerrie Fourie of Capitec, and the chorus of numerical dissonance joined by Alan Knott-Craig jnr and Magnus Rademeyer of Fibertime. With eyes set on a JSE listing in 2027, they seek leverage by rubbishing Stats SA population numbers.

Unlike the earlier crop of academic critics, these latter-day critics display a brand of arithmetic driven purely by self-interest and the aggregation of their subjects. They look at the black population and see only a share of low-quintile income.

This is the “Economics of Aggregation”. Each Please Call Me message is a rand in the kitty of Knott-Craig; each ATM transaction is a rand for Fourie. Be it from the 8-million recipients of the R350 grant or the 17-million social assistance recipients, they see transactions, not people. A Please Call Me generates R28m for Knott-Craig; an ATM entry generates millions for Fourie. This is the economics of the marginalised — the vultures feed on them.

These two businessmen, I am sure, would have an interesting party full of champagne as they compare their mendacious numerology inspired by dollar signs. But it is a serious abuse of statistics

The Vulture Culture Scenario is not the brazen extortion of a Vusimuzi “Cat” Matlala as learnt from the Madlanga commission; it is the silent bip-bip on an ATM and a mobile phone. These men are mesmerised by aggregation. To this, Fourie imputes an unemployment rate that should not exceed 10% (ignoring that a transaction does not equal a job), and Knott-Craig projects a population of 95-million — a staggering one-third more than the current official population.

Let’s interrogate Knott-Craig’s 95-million figure using basic demographic consistency, something clearly absent from their “AI” models.

If the population were indeed 95-million, the demographic structure of the country would have to fundamentally shift. Currently Stats SA records about 14-million children in school. If the population were 95-million, keeping the current demographic pyramid, the population attending school would be about 22-million.

Knott-Craig must show us where the missing 8-million schoolgoing children (ages six to 18) are hiding. Our school registers and the statistician-general’s census sit neatly on top of each other, confirming 97% of South African children in this age group are in school. You cannot hide 8-million children. They are not in the registers; they are not in the classrooms. They exist only in the fibre-optic delusions of business leverage.

Furthermore, Knott-Craig must produce the birth records. A population of 95-million would imply a number of annual births closer to 1.8-million, rather than the 1.2-million recorded by the department of home affairs and Stats SA. Where are the 600,000 extra babies born every year? Are they born without certificates, without clinics and without footprints?

Perhaps Knott-Craig should also produce the death certificates. A population that size would imply at least 300,000 more deaths annually than recorded. And the labour force? According to Knott-Craig’s “naught numbers”, the labour force should be 36-million, not 24-million.

These two businessmen, I am sure, would have an interesting party full of champagne as they compare their mendacious numerology inspired by dollar signs. But it is a serious abuse of statistics. When you overlay it with artificial intelligence, as they have done, it exposes not the power of AI but the debilitating weight of “natural stupidity” that these very successful business people possess.

Here is the crux of the matter. Before the 1976 Soweto student uprising, blacks were significantly invisible to whites — a feature that continues to this day, despite black expenditure being the economic majority. Pineteh Angu elaborates on this phenomenon in the paper “Being Black and Non-Citizen in South Africa: Intersecting Race, White Privilege and Afrophobic Violence”.

Despite black expenditure surpassing that of whites by 2022 — suggesting the preponderance of the 80% black population (accounting for 62% of expenditure) against the 7% white population (accounting for 25% of expenditure) — this economic significance remains invisible in terms of ownership. Why? Because black expenditure is simply aggregated by white people in terms of rands and cents into white business and white riches.

These white men can only count blacks in their token form of revenue streams, rather than recognising the profound development deficits they suffer. Knott-Craig counts them through fibre laid — 284,000 points counted, from which he absurdly extrapolates 95-million people. Fourie counts them in ATM transactions.

This is not how you use secondary data. This is the arithmetic of erasure.

The statistician-general has just provided the poverty numbers of the nation this past Friday. They are harrowing. But it is not clear what message these numbers will convey to Fourie and Knott-Craig, who see only potential IPO value in the masses.

One thing is clear, as Steve Biko said more than five decades ago: “Black man, you are on your own.” The government of national unity (GNU) will not rescue you. Instead it has seemingly emboldened white arrogance in the midst of black misery. The vibe analysis of the GNU — where dry grass is painted green so the livestock will graze — aligns perfectly with this falsification of numbers.

When Habermann shared his seminal paper on the effect of politics on the American Federal Statistical System, he sounded a warning. The conduct of these white South African men towards Stats SA is not an isolated aberration. It is a pandemic that threatens to engulf the world — a place where the blind obliterate the light, and where the truth of our condition is traded for the fiction of a balance sheet.