Why everyone is wrong about the white farm murder rate in South Africa
A deep dive into numerators and denominators
Addendum: This piece originally made an incorrect claim that the Myburgh’s head of household rate could not be compared with the individual murder rate. I acknowledge that you can make the comparison on statistical level, however it does not make sense within the context of the farm murder problem in South Africa.
Afrikaans refugees clad in shorts and flannel apparel teem into Dulles Airport. Trump lambasts South African President Ramaphosa for allowing a white genocide to take place on his watch. Someone mysteriously edits Grok’s internal code to force the AI to bring up farm murders and white genocide at every opportunity. Fact checking groups, and legacy media rush to put out articles “debunking” claims of farm murders. The South African farm murder discourse has gone global.
I have always believed that the issue was overblown. Yes, these murders are particularly gruesome, but to suggest that white farmers were disproportionately targeted seemed absurd to me. I scoffed when the Democratic Alliance, South Africa’s second largest political party, stated without evidence that farms were four times more dangerous than other parts of the country. I was then quite surprised when credible South African journalists, Nathan Geffen and Tim Cohen, wrote that white farmers were murdered at rates higher than the national average.
Their assertions stem from a piece by James Myburgh of Politics Web. He calculated that in 2016, white farmers were murdered at a rate of 108 per 100,000. This is more than triple the national average of 34 per 100,000 in the same period. However, fact checking organisation, Africa Check, quickly disputed his figure. They argued that due to a lack of data, the question was actually “impossible” to answer.
The ensuing debate between Africa Check and Myburgh became the basis for many claims about farm murders going forward. The Washington Post, the BBC, and Al Jazeera, routinely cite Africa Check’s view on the issue. And, as previously mentioned, James Myburgh’s work is referenced by prominent South African journalists, as well as Afrikaans special interests groups like Afriforum.
So who is correct? In this post we are going to take a journey into the world of numerators and denominators, and see why both sides get it wrong. I will argue that it is possible to estimate a reasonable figure for white farm murders in South Africa. However, it is nowhere near the 108 murders per 100,000 that James Myburgh claims. Instead in some years it hovers slightly above the national murder rate and in others slightly below, with an overall downward trend.
But first, we need to define what a farm murder is.
What is a farm murder?
The closest thing to an official definition of a farm murder can be found in the South African Police Service (SAPS) Rural Safety Strategy. In this document they define what constitutes an attack on rural people and what is meant by a farm. You can read the document for yourself, but the important takeaways are:
The location is not limited to commercial farms, but also smallholdings. This is important, as 40% of farm murders take place on smallholdings.
It is not just farm owners who are counted in these statistics, but anyone who is present on the farm or smallholding.
Murders that result from social arguments, domestic violence, or liquor abuse are excluded from this definition.
This definition is not perfect, but it is used by stakeholders on both sides of the political aisle. Afriforum, a special interest group representing Afrikaners, the Transvaal Agricultural Union, and SAPS make use of this definition when reporting farm murder statistics.
So we have an agreed upon definition of farm murders and multiple groups report murders using this definition. So why is the issue so complicated? It all has to do with numerators and denominators.
A journey into numerators and denominators
To calculate the murder rate of white farmers in a given year, we need a numerator and a denominator:
The total number of white people killed on farms in a given year (numerator).
The total number of whites on farms in a given year. (denominator).
Whereby, we are using the SAPS definition of farms to include smallholdings.
Unfortunately, we can’t just plug some numbers into the numerator and denominator and call it a day. SAPS does not provide a breakdown of farm murder victims by race, meaning we don’t know what the numerator should be. Similarly, there is no exact census category for “whites living on farms according to the SAPS definition of farms”, meaning we don’t know the denominator.
This is one of the reasons Africa Check says that the farm murder rate is unknowable. However, James Myburgh of Politics Web argues that just because we can’t figure it out exactly, doesn’t mean we can’t work out a rough estimate. He writes:
While these “fact-checkers” have raised legitimate points around the methodology for calculating the murder rate for white farmers, it is unclear why instead of trying to come up with a better estimate they went with the assertion that this was, alone among all the many great mysteries of the universe, fundamentally unknowable.
Snark aside, Myburgh is correct. It is possible to estimate the murder rate of white farmers in South Africa.
James Myburgh vs Africa Check
James Myburgh tries to calculate the murder rate in the 2016/17 financial year. He uses data from the Transvaal Agricultural Union (TAU), which kept records of farm murder victim names and locations in 2016/17. From these names he infers the race of the victims. He states that there were 74 farm murders in the 2016/17 period. Of these 74 murders, 66 were white farmers and/or their families, and 51 were just the white farmer alone.
I was unable to check this data with the TAU, however these numbers are broadly in line with other estimates. SAPS recorded a similar number of victims for the 2016/17 financial year. Furthermore, in a 2011 study, the Institute for Security Studies found that 89% of farm murder victims were white, which is similar to Myburgh’s ratio of white victims. Finally, someone compiled a public list of 60 farm murder victims in 2016/17. The racial distribution of this list, seems to also be congruent with the TAU data that Myburgh cites.
So we have a numerator. What about a denominator?
For this he relies on the 2016 Community Survey Statistical Release. The Survey lists the number of white households that partake in agriculture (143,361). Importantly, it also disaggregates these households by where the farming takes place. This is crucial, as some households are “back yard farmers”, meaning they could just be growing tomatoes on their balcony. The survey shows that 37.9% of white agricultural-households have farming activity that takes place on a farm or smallholding. This is the area most congruent with the SAPS definition of a farm.
Myburgh also notes that the there was only an 87% response rate to the question of where the agricultural activity takes place. He argues that the response rate is likely to be close to 100% for white farmers, giving him a total of 47,2721 white farmer households. He then notes that the average household size for whites is around 2.642, meaning that there are about 125,000 white farmers and their families in these households.
So we have a numerator of 66 whites killed on farms and a denominator of 125,000 whites on farms. Which gives a murder rate of 52.8 per 100,000. This is higher than the South African murder rate of that year, which was 34 per 100,000.
Myburgh then takes his argument further. Instead of just examining the number of whites killed on farms, he wants to know the murder rate of “white farmers alone”. He thus considers all the cases where white farmers were killed without their family members (51 times), and divides this by the total number of white farmer households. This amounts to 108 per 100,000, which is nearly triple the murder rate of normal South Africans!
Africa Check then attempted to “debunk” Myburgh’s findings, arguing that his denominator was flawed3. They allege, based on their own correspondence with Stats SA (South Africa’s statistical agency), that:
The survey was never intended to be an agricultural survey, therefore it may not be representative of agricultural households.
Because of low response rates in the place of agricultural activity (farmland vs back yard), the proportions listed in Table 7.28(b) should not be taken as indicative.
There are problems with Myburgh’s figures, but I’m not convinced by Africa Check’s argument, and neither is Myburgh, he notes:
The Community Survey is quite representative (1 in 12 South African households are sampled). To justify this, Myburgh compares the 2016 community survey numbers with the 2011 census numbers and reveals that they are broadly in line with each other.
Given that the survey is representative, we can apply the proportions in Table 7.28(b) to the number of white agricultural households. Furthermore, if you distribute the non respondents equally among all the groups, you arrive at 54, 354 white farmer households, which would place his white farmer murder rate at 93 per 100,000. Still much higher than the national average.
With hindsight, I do believe that you should use the 54,354 figure and not 47,272 for the number of white farmer households. The 2022 census reports 50,534 white farmer households. Given the longterm decline in the number of white farmer households in South Africa, it is highly unlikely that the figure dipped in 2016 and then recovered in 2022.
So is that case closed? Do we accept Myburgh’s “conservative” estimate of 94 per 100,000? Not quite.
Heads of households
I agree with Myburgh that Africa Check’s “debunking” of his estimates was flawed. However, there is still a problem with Myburgh’s calculation. There are two reasons.
First, I am not sure that it is epistemically correct to think about farm murders in terms of heads of households. The reason Myburgh wants to, I assume, is because he is focused on murders of the white farmer profession. However, most households will have more than one farmer in the household (often multiple members of a given family will participate in farming activities). So to assume that the number of households is equal to the number of farmers is going to underestimate the number of white farmers.
Second, I believe we should consider the family members of white farmers. This is because most people conceptualise farm murders as attacks against farmers and their families. For example, in 2016/17 a farm murder took place where four members of the Smuts family were brutally tortured and killed. It would be erroneous to say that there was only one farm murder victim in this killing. This is why SAPS looks at all people killed on farms, because there is a contention that the white farming community is being threatened.
So, what happens if we use Myburgh’s data in the “correct” way and look at white farmers and their families. We have 66 white victims, 54,345 households (instead of 47,272), and a household size of 2.64, which gives us 46 per 100,000. This is still above the national murder rate of 34 per 100,000 in 2016/17, but below Myburgh’s estimate of 53. So Myburgh was directionally correct, but his magnitude was off.
What about current data?
So that was 2016, what about now? Well, we can make some good inferences using the 2022 census and SAPS data. In 2022, there were 49 farm murders. If we assume the racial breakdown of victims has remained constant, this would amount to around 44 white victims. In terms of white households involved in agriculture on farmland, the number dropped to 50,534 in 2022. Meaning the white farmer murder rate for 2022 was around 33 per 100,000, which is lower than the national murder rate in 2022 of 40 per 100,000.
Even if we assume that all the farm murder victims in 2022 were white (which is undoubtedly false), we would arrive at a murder rate of 36 per 100,000. This would still be lower than the national murder rate. And most importantly, regardless of which way you look at it, the farm murder rate has come down since 2016!
In conclusion
After Trump and Ramaphosa’s oval office meeting, members on both sides of the debate rushed to use dodgy statistics to score political points. CNN compared the number of farm murder victims with the total number of murder victims in the country, completely ignoring denominators4. Joel Pollack, editor of Breitbart news called the meeting, “the best thing to happen to South Africa since the end of Apartheid”.
Historically, this has been a debate filled with vitriol and emotion. Part of this is certainly fuelled by the fact that farm murders are usually accompanied by exceptional levels of violence. This discourse has also had actors who have continually strived to find the truth, despite the fact that data reporting in South Africa makes this difficult. On this note, I want to emphasise that the purpose of this piece is not to “dunk” on Africa Check and James Myburgh, who both consistently put out high quality work on crime in South Africa.
Instead, the purpose of this piece was to simply investigate the farm murder rate for white South Africans and compare it to the national average. In 2016, I found this rate to be above the national murder rate. In 2022 it was below. A more nuanced picture than both sides care to admit, but nothing akin to a white genocide.
Finally, if you have any questions, are curious about a source, or found that I have made some terrible mistake, please let me know in the comments.
Next week, I intend to publish a piece on the economic effects of immigrants coming to South Africa. If that interests you, please consider subscribing.
Some extras for those who are interested:
Areas of focus
I only focused on Myburgh’s and Africa Checks arguments, however many others have attempted to calculate farm murder rates. The reason I only focus on Africa Check and James Myburgh are as follows:
Many other studies do not report all their calculations, so they are difficult to properly engage with. For example, Chris de Kock argues the denominator should be between 120,000 and 240,000, but I could not find a source for his denominator.
The Democratic Alliance simply stated without citing anything that being a farmer was the most dangerous profession in South Africa.
Many studies don’t use the SAPS definition of a farm and only limit their scope to commercial farms, such as the back of the envelope calculation done by The Inclusive Society.
Most attempts at calculating farm murder rates simply take the CNN approach and compare the total number of farm murders, or white farm murders with national murder totals. These are not even worth engaging with.
Recent SAPS Data
Recent data released by SAPS for the first quarter of 2025 showed that there were six farm murders and five of those murdered were black. If we extrapolate this to rest of the year, the white farm murder rate comes down massively! However, I’m skeptical of using these figures to make any projections. This is the first time that SAPS has ever released a breakdown of victims of crime by race and this quarter could be an outlier. I believe we should wait and see what the yearly figures say.
Across his two articles, Myburgh actually comes to two different figures: 47,272 and 47,218. Based on his logic, the former is correct. I assume the latter was a typo.
A hangover from the Apartheid, is low levels of racial mixing in South Africa. It can be assumed that in a household with a head, the other occupants will be white 95% (I guess) of the time. Especially on farms.
Actually, their main issue with Myburgh’s work was that he was not transparent with his sources. However, we are not going to get into that debate today, we are just here to look at the statistics!
This is really bad. Imagine a society with 100,000 people. 10 are black and the rest are white. Imagine there are 1000 murders in a given year, and all the black people are killed. Based on CNN’s logic, only 1% of the murder victims are black, this is despite the fact that all the black people were killed!
great article, wow, enlightening
Independent, unbiased journalism is dead. This article is SUCH a breath of fresh air. I'm equally confused on land expropriation. Is it as benign as some down play it to be or is it a gateway to anarchy for an ill intentioned future leader?