South Africa

South Africa

South Africa’s mysterious murder rate

South Africa’s mysterious murder rate

South Africa's murder rate is high by global standards. But the increase of the last three years is particularly worrying. There isn’t yet a good explanation for this recent uptick. It may be that we are seeing a renewed decline in state legitimacy. It may also reflect the significance of the change in the types of murders being committed. By ANINE KRIEGLER and MARK SHAW.

The latest national statistics released by the South African Police Service (SAPS) put South Africa’s murder rate at 33 per 100,000 people. This is high by global standards. But there are three important things about our murder rate that you may not know: It saw a massive surge to 1994; it declined rapidly until 2012; it has started creeping up again since then. Different factors are probably at play for each.

The global average rate of intentional homicide is roughly six per 100,000. Rates in Western Europe are less than half that, but this has not always been the case. In the Middle Ages, homicide rates in England and Western Europe were 10, 50, even 100 times higher than they are today. The city of Oxford, now famous for its university and the porcine proclivities of its elite, was known in the 14th century for its university and for being an utter bloodbath, with an estimated 110 homicides per 100,000 people.

Highly variable data quality means international comparisons can be deceptive, but the best available estimates suggest the highest murder rates today are found in Southern Africa and Central America. Honduras leads the race by a fair margin at about 90 per 100,000, and Venezuela (54), Belize (45), and El Salvador (41) are also considered to have extremely high murder rates. Although not quite at Central American rates, South Africa’s rate of above 30 is considered high. Of the roughly 437,000 deaths due to intentional homicide in the world in 2012, about 3.7% of them happened in South Africa. This although we make up only about 0.7% of the world’s population. There is already a large body of research on why we are such a violent society, providing reasons including a history of violence and brutalisation, patchy and illegitimate policing, the impact of apartheid on families and the education system, enduring inequality, racism, the availability of firearms, and so on. There is a lot more to South Africa’s murder rate, however, than its high level. For one thing, it has been much higher.

For all the popular narrative of societal decline, the reality is that this country has almost certainly never been as violent as it was in 1993-1994. Determining the extent of fatal violence prior to that is a difficult task. The SAPS only came into being in 1995, as an amalgamation of the old South African Police (SAP) and the 10 other policing agencies of the former Bantustans. Segregation demanded that each was nominally policed by its appropriate ethnic group, but white supremacy demanded that each remain subordinate and serve the security interests of white South Africa. As such, the Bantustan police forces were ill-trusted, ill-trained, ill-resourced, ill-led, and overall ill-placed to do much crime prevention, detection, or recording. Some did not keep databases of reported crime statistics at all, while others’ databases were critically compromised. Even in nominally white South Africa, it is likely that a number of homicides of black people were not recorded, especially in the poorly-policed township areas. So the homicides of almost half the South African population didn’t make it into any official stats.

Yet it is possible to make some educated guesses based on the data we have. Here we have a graph showing the number of murders recorded and reported by the SAP and later SAPS from 1975 to the latest release of statistics:


The most obvious feature here is the sudden spike in 1994. This is easily explained, as it was then that the statistical net was abruptly expanded to include the additional 16-million-odd people living in the Bantustans. Indeed, the spike is almost certainly a considerable underestimation, as policing and proper record-keeping didn’t instantly reach all of those places which were previously ignored. Had the true murder rate stayed constant from 1994, it would not have been surprising to see the official rate rise over the following decade or so, as the new SAPS consolidated its reach and improved its crime information systems. All reported crime in South Africa was registered on a central system as of 1995, but the early years were thought to have considerable under-counts, partly as a result of negative attitudes from the stations, which didn’t see the operational use of this data in combatting crime. A series of improvements were made to the crime registration and analysis systems, including during and just after the stats moratorium in 2000-2001. As a result, crime statistics prior to about 2003 should be treated with caution, and thought of as only a broad indication of what might have been the reality.

So the huge spike to 1994 is almost certainly entirely a data quirk. The standard fix for the problem of changing population sizes for crime statistics is to present the data not as raw figures but as rates per 100,000. Again, this isn’t easy to do in the SA historical context. Estimates of population size and distribution were of poor quality and politically fraught. But based on SAP annual reports and estimates of the number of people living in ‘white’ South Africa from the Central Statistical Service (now Statistics South Africa) and from the Institute for Race Relations, the following is a decent guess for the murder rate in those parts of the country served by the SAP/SAPS. It does not include the Bantustans, except insofar as the SAP jurisdiction was a reasonable sample of the rest of the country. Data as of 2003 is of the best quality, that from between 1994 and 2003 is only fairly reliable, and that from before 1994 is a pretty rough estimate.


Here the spike in 1994 is eliminated, and more meaningful features of the data are better revealed. First, it appears that at least that part of South Africa within the SAP’s jurisdiction had murder rates above 20 – high by global standards – since at least the 1970s. Our high levels of violence are by no means a post-apartheid development. This gives credence to those long historical explanations.

But although murder rates have been high for at least the last 40 years, they have not been constant. They surged through the 1980s and 1990s to a peak in 1993 of about 78 per 100,000. This made South Africa one of the most fatally violent places (for which there are reasonably reliable data) in the world. Although figures from the Bantustans are not directly available, we know that their inclusion in 1994 saw the recording of an additional 9,365 murders among an additional population of about 16-million, with the result that the total Bantustan murder rate that year can be estimated at about 58 per 100,000 – and remember, this is probably a large underestimate. It isn’t clear what proportion of the homicides around this peak period were related to political conflict, but some estimates have it in the region of a fifth to perhaps a quarter. Comparative research suggests periods of political upheaval also raise the rates of crimes not directly related to politics, partly as a result of a decline in state legitimacy. In other words, the huge hump around 1994 would be considerably smaller if political killings were excluded, but would by no means be eliminated.

The second important feature of the murder rate is that (despite improved reach and recording practices) it declined steadily between 1994 and 2012, down again to levels roughly in line with those seen in the 1970s. The average person in the country was less than half as likely to be murdered last year as they were 20 years ago. In 1994, the SAPS recorded 26,832 murders in the country, or about 74 a day. Last year, they recorded 17,805, or about 49 a day, and this in a population that has grown since then by about 40%. The murder rate has halved.

As Oxford and Honduras exemplify, rates can vary a great deal over space and time, but a decline of the magnitude and speed seen in South Africa over the last 20 years is not the norm. It’s almost in line with the plummeting crime rates in New York City in the 1990s, which sparked thousands of books, academic articles, and opinion pieces, proposing explanations varying from zero-tolerance policing, through abortion policies, to the developmental impact on children of lead pollution from car exhausts. Unlike in New York City, rates of other crimes in South Africa have not fallen in concert with murder rates (although many have declined a good deal in at least the last 10 years). But murder is enough of a special case that its trends deserve particular attention. It is (understandably) deeply feared, and a key element in people’s feelings about crime and their safety. Beyond that, murder is often considered an indicator of the nature and magnitude of violence more broadly at play. Unlike many other crimes, it is readily measurable, clearly and quite consistently defined, well-reported, and overall a loose but reasonable proxy for insecurity. Murder figures in South Africa correlate extremely well with total robbery figures since 2003/2004 and very well with figures for a basket of crimes including aggravated and common robbery, assault with grievous bodily harm, and burglary – both residential and non-residential. The large increases in house and business robberies run clearly counter to this trend, but they remain relatively rare, such that the decline in the more common crime types maintains the correlation.

There does appear to be a level on which people are aware of the change in murder rates, lest you fear that this is all a manipulation of the stats by the SAPS. Victims-of-crime surveys are a means of bypassing some of the myriad possible problems with police crime data by going into people’s homes and asking them directly about the crimes they have experienced, whether they reported them to the police and why, how fear of crime affects their lives, and so on. Successive national surveys between 2003 and 2014 indicate that murder has dropped steadily from fourth to seventh in the rankings of what respondents considered to be the most common crime, and dropped from first to fourth for the most feared.

The murder rate remains high by global standards, but the decline of the last 20 years is significant. Socio-political turmoil around the country’s transition placed extraordinary upward pressure on the murder rate, and this pressure appears to have abated. The decline has likely also been encouraged by the consolidation of state power and legitimacy, and possible improvement in the social and economic conditions that drive violence.

In this context, the third feature of the data – the increase of the last three years – is particularly worrying. Unlike the generally high levels, the surge to 1994, and the recovery since, there isn’t yet a good explanation for this recent uptick. It may be that we are seeing a renewed decline in state legitimacy. It may also reflect the significance of the change in the types of murders happening. Police docket analysis in the late 1990s indicated that the vast majority of murder dockets were ‘social murders’ – mostly starting as an argument between people who knew each other and of which at least one was intoxicated, developing into a physical fight ending up in murder. A docket analysis in 2007/2008, however, found that only 65.4% were social murders, and evidence presented to the Khayelitsha commission still more recently suggested that the proportion of social murders in that jurisdiction may be closer to 50%. Instead, more and more murders are committed by strangers, and relate to other crimes – especially aggravated robbery. It may be that although social murders continue to decline, the increase in aggravated robbery rates is driving a new growth in instrumental murder. One thing to note is that aggravated robbery rates are relatively manipulable by good policing, using crime prevention, crime detection, and crime intelligence tools.

In short, our murder rate of 33 remains high. It seems that we have recovered from the surge fairly typical of transitional societies. The recent increase is something new and worrying. If we could recapture the downward trajectory, we could reduce the murder rate over the next decade to 20 per 100,000. This would still be high, but it would be a significant achievement and would represent a large reduction in overall violence in the country. DM

Photo: Members of the South African Police raid Alexandra Mens Hostel during a midnight exercise aimed at searching for weapons used during xenophobic violence, in Johannesburg, South Africa, 23 April 2015. EPA/KEVIN SUTHERLAND.


Please peer review 3 community comments before your comment can be posted


This article is free to read.

Sign up for free or sign in to continue reading.

Unlike our competitors, we don’t force you to pay to read the news but we do need your email address to make your experience better.

Nearly there! Create a password to finish signing up with us:

Please enter your password or get a sign in link if you’ve forgotten

Open Sesame! Thanks for signing up.

We would like our readers to start paying for Daily Maverick...

…but we are not going to force you to. Over 10 million users come to us each month for the news. We have not put it behind a paywall because the truth should not be a luxury.

Instead we ask our readers who can afford to contribute, even a small amount each month, to do so.

If you appreciate it and want to see us keep going then please consider contributing whatever you can.

Support Daily Maverick→
Payment options

Daily Maverick Elections Toolbox

Download the Daily Maverick Elections Toolbox.

+ Your election day questions answered
+ What's different this election
+ Test yourself! Take the quiz