Statistician-general Risenga Maluleke says Stats SA faced huge hurdles delivering the disputed 2022 census, which has come under scrutiny for having “the largest undercount in history”.
He cited budget cuts as a contributor to the poor quality of the census, saying R6bn was required to undertake a comprehensive survey but Stats SA ended up receiving only R3bn from the National Treasury.
After months of defending the quality of its 2022 census data against criticism from researchers, Stats SA admitted this week that an undercount of 31% had occurred in the data collected on mortality, fertility, labour and income, prompting it to withhold this information from its report. Flawed census data could materially affect financial allocations to municipalities and provinces which need to be based on reliable data.
Maluleke insisted Stats SA was transparent about the undercount and that the data had been withheld rather than withdrawn.
“We haven’t withdrawn data. How can we withdraw something that we never made public? We never withdrew any data — we didn’t release it.”
He said work on the 2022 census began in 2018, but Stats SA was rocked by financial constraints and other operational disruptions including the Covid-19 national lockdown, the 2021 unrest in KwaZulu-Natal, and flooding in that province and other regions. Some field workers, he said, had to travel via tractor or donkey cart to complete their surveys.
Stats SA has not engaged at all with the substantive concerns raised. Instead, they have simply insisted ... that the data are not cause for concern and are fit for purpose.
— UCT demographer Tom Moultrie
“We asked [National Treasury] for R6bn. We were given R3bn. When we were implementing in the field we realised that was not enough. We spent up to (about) R3.8bn. We needed double the amount but [lack of funding] was not the main reason for the undercount.”
In the 2011 census, the undercount was about 14.6% and in 2001 was 17.6%, according to data.
The announcement of the undercount prompted concerns about the quality of the 2022 data, which the government relies on for expenditure planning and resource allocation. Maluleke maintained that the Statistics Council, which assesses Stats SA’s data, said what was released in the 2022 census report was fit for purpose.
But the council, to which Stats SA accounts, did raise concerns about some of the data.
“Last year the Statistics Council ... brought their own local and international experts ... and advised us not to release the fertility data’,” he said of the unreleased data that contained the undercount.
Census 2022 was conducted via self-enumeration and telephonic surveys; but some communities, including gated communities, were not comfortable with field workers visiting them late. Households that could not be reached as no-one was home also could not be surveyed.
UCT demographers Rob Dorrington and Tom Moultrie were the first to raise the alarm about the census, calling it not fit for purpose as there was a huge 31% undercount of the population compared with an undercount of 14% in 2011, which was already high by global standards.
The statistician-general, who at first dismissed Dorrington and Moultrie's concerns, acknowledged on Friday that “we have just recorded the highest undercount in the history of our census-taking as a country”.
He said the two academics were allowed to make inputs on the data before the Statistics Council did so, but after meeting with Stats SA they indicated that no further engagement was required.
“I respect them so much. They are demographers in their own right. They were asked by the Statistics Council to be part of the process and they stayed out. For any census you have in South Africa, there will be issues. But if you are free to be part of the process and you don’t participate ... the process cannot wait for two people.”
Moultrie told Business Times that when the initial results of the 2022 census were released in October 2023, he and Dorrington used standard demographic methods to evaluate the data against the results from previous censuses and their detailed knowledge of births and deaths in South Africa between 2011 and 2022.
“While we were able to publish our study only in early July 2024, we alerted Stats SA to our principal concerns in December 2023. In the period thereafter and working with the data that were in the public domain, we subsequently identified other areas of concern that were published in July.”
He and Dorrington were specifically concerned about the implied in-migration into South Africa among people aged 50 and older, as this was an age group not usually prone to migration. Another concern was “glaring implausibilities in the implied growth of the population at a municipal level”.
He said, “To date, Stats SA has not engaged at all with the substantive concerns raised. Instead, they have simply insisted — without providing evidence to the contrary — that the data from the census are not cause for concern, and are fit for purpose.”
For any census you have in South Africa, there will be issues. But if you are free to be part of the process and you don’t participate ... the process cannot wait for two people
— Statistician-general Risenga Maluleke
He said he and Dorrington did not feel at all vindicated by Stats SA’s decision to not release certain aspects of the data, but he said the announcement adds weight to their concerns that the data collected were not fit for purpose.
“A census is often described as the most complex logistical exercise undertaken by governments during peacetime. Had we been asked, we would have strongly advised against trying to run a census in the middle of a pandemic,” Moultrie added.
He said Stats SA must try to fully understand what went wrong with the exercise and aspects of the planning that should be fully investigated to ensure that lessons are learnt for the next big count.
A Stats SA delegation told parliament’s portfolio committee on planning, monitoring and evaluation on Friday that the institution conducted the census to the best of its ability but was hobbled by a 20% vacancy rate and a strained budget.
DA MP Katherine Christie asked Stats SA to release the data on mortality, fertility, labour and income, even as a 10% sample, so that other demographers and statisticians could compare it with their own data to test the accuracy of the census report.
“These stats are vital for knowing [things like] the distribution of income between national, provincial and local government and how many vaccines need to be distributed [and where].”
Solly Molayi of the social statistics chief directorate at Stats SA said that after compiling the census, they had conducted a data quality assessment to get extra assurance that it was of good value, and that is when the undercount was discovered.
He said other data, namely the income and expenditure survey, the immediate population estimates, and the super labour force surveys, would be released through the year and would contain much of the data that the 2022 census lacked.
“Remember, a census is not perfect. We are not proud of the 31%. We are the first ones to acknowledge that there were challenges, but we are also saying we have tools we can use to make adjustments, as we have done before.”
Trade & Industrial Policy Strategies senior economist Neva Makgetla said the undercount was “very high” by international standards and appeared to be due to the shift to digital counting methods for much of the population.
She said while the census was still the largest survey available, the numbers directly affect the distribution of the budget according to the constitution and unreliable data could result in underfunding in regions where the count was understated.
“It would be worth allocating more resources from the national budget to undertake
in-depth surveys in areas where population growth trends changed sharply according to the 2022 census... The cost would be high but the cost of not ensuring legitimate population figures, especially for budgeting and economic policy initiatives, would be even higher.”
Institute for Security Studies head of futures and innovation Jakkie Cilliers said the organisation updated its forecasts on South Africa’s development prospects without the 2022 census data, relying instead on internal forecasts from their international futures platform.
“We are not demographers, so rely on others for their analysis of the census data, and the consensus there seems to point to the need for a comprehensive review by experts or to set the results aside,” said Cilliers.
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.