After the Bell: The 2022 census needs to be redone

For most people, the census is vaguely interesting for the oddities it reveals about a country. The comedian Conan O’Brien quipped that the US Census Bureau had reported that American homes were 650 square feet larger today than they were in 1980. Unfortunately, so are most Americans.

Generally, the census comes and goes in a torrent of unwanted advertising and people with clipboards. However, SA’s 2022 census was different. It was — wait for it — controversial. I am not making this up.

Part of the problem was that the census, carried out every decade, coincided with the Covid-19 pandemic, making the process much more difficult. Statistics South Africa (Stats SA) realised this and wanted to delay it. However, the Treasury decided the department would not be allowed to roll over its budget.

Anyway, the process continued, with an attempt to gather at least some of the information online. Incomplete results were released last year, but this was quickly followed by some very trenchant criticism of the findings.

Once the census is taken, which is specific to a certain day, there is something called the post-enumeration survey, which measures the accuracy of the census. The main problem it picks up is whether people or households were missed or counted twice.

When this survey came out, it turned out that only 69% of South Africans were counted — a pretty huge miss, because it suggests that nearly 20 million people (for the sake of accuracy, 19,228,525 people) were not counted.

There is normally an undercount in SA (as there is all over the world) because it’s always difficult to pinpoint exactly where every South African is at a given moment. However, this undercount was huge: in 1996, 2001 and 2011 the undercounts were 10.7%, 17% and 14.6%, respectively. After the 2001 undercount, the then Statistician-General Pali Lehohla said Stats SA had “lost the plot” and promised to do better. He did — sort of.

But oddly, in this case, the current Statistician-General, Risenga Maluleke, staunchly defended the census. He came out with guns blazing, saying critics represented a minority view in a scientific world that widely endorsed the project. Stats SA has an advisory board made up of industry experts, local and international, and they had authorised the publication of the data.

The 2022 census would not be redone. “We stand by the census numbers, we are solid, we have put them through a battery of tests,” said Maluleke.

The most prominent objections were raised in a technical report prepared for the South African Medical Research Council. In this report, Tom Moultrie, a professor of demography; and Rob Dorrington, professor emeritus, both at the University of Cape Town, have described Census 2022 as just an estimate, not a credible enumeration.

Their biggest gripe with the census result was that the total population of SA, reported at 62.3 million, represents an overestimate by a million or more! They also suggested there might be a big underestimate of migrants; there were some incoherent and implausible results at a subnational level; and they questioned the adjustments Stats SA had made due to the huge undercount.

I’ll just give one example of what they found and questioned, but there are many others at the local, national and provincial levels.

Notionally, the city of Johannesburg grew by only 8% over the period between the current and the previous census. However, the adjacent metros of Ekurhuleni and Tshwane grew by 28% and 38%, respectively. According to the census, Cape Town is now very close to being SA’s largest city, with 4.77 million people, compared to Joburg’s 4.8 million. No doubt the Mother City has grown and there has been a lot of semigration, but Joburg had 4.4 million people in the previous census and Cape Town 3.7 million — that’s a very big gap to close.

The problems with this dispute are just beginning to come out. Recently Stats SA published a Census in Brief document and did an odd thing: it noted that census data on income, labour, demography, fertility, mortality and some aspects of housing were being excluded because of “reporting-related biases”.

Labour and income are covered by other surveys, but the exclusion of fertility and mortality is pretty disastrous. These numbers are crucial for macro-planning on everything from energy, health systems, education, grants — you name it.

One of the authors of the report mentioned above, Moultrie, tweeted this week that the exclusions were a “travesty”, particularly given the Statistician-General’s defence of the census.

“The size and the scope of the census usually permits all kinds of users (economists, demographers) to explore income, employment and demographic indicators to a fine level of granularity. It provides one of the key pegs for paramterising population projections which feed into a host of policy and programme initiatives... None of this, seemingly, will be possible. It’s past time that we got to the bottom of what happened in the 2022 census,” he wrote.

The problems don’t stop there. Poor old Joburg is now just about to take up arms in a battle to save its budget. My former colleague Carol Paton points out that financial transfers from the national fiscus to municipalities are determined largely by population size, indicators such as income and poverty data, and demographic trends. But Census 2022 implies that Johannesburg is 20% smaller than previously thought and on which budgetary estimates for the past 10 years have been calculated.

Part of the problem is that the national government doesn’t seem to take the issue very seriously, and you can see that by its declining budget granted to the census process. Researcher Neva Makgetla pointed out in a column in Business Day last year that in both 2001 and 2011, funding for the census equalled about 0.25% of the national budget; in 2022 the figure dropped to 0.18%.

“In real terms, the census cost R50 per person in 2022, compared with R75 in 2011. The total budget for Stats SA fell from R200 per citizen in 2011 to R80 in 2022,” she calculated.

I hate to say this, but I think they need to redo the whole thing. If you can’t measure it, you can’t manage it. DM